Essay Writing Service

Investigation of Neurochemicals’ Role in Glioblastoma and Neural Stem CellsDoes a History of Prior Concussions and Subconcussive Impacts Affect Neurocognitive Performance Involving Memory in Youth Athletes?

Investigation of Neurochemicals’ Role in Glioblastoma and Neural Stem Cells: Characterization of Dopamine Receptor D4 Inhibition in Glioblastoma Stem Cells Abstract Neurochemicals that mediate synaptic communications between mature neurons are now appreciated to have a critical role in neurogenesis and cancer.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of


Hypothesis: Youth athletes with an increased frequency of previous concussions and subconcussive impacts will display poorer neurocognitive performance in tasks related to memory (compared to those without a history of concussion and fewer subconcussive impacts).

Methods: A literature search was conducted using PubMed and Google Scholar. Using certain exclusion criteria, several cross-sectional, retrospective cohort and case-control studies deemed to have an evidence level of 2 or 3 were chosen and evaluated. Neurocognition measured by computerized testing was compared across youth athletes with and without a history of concussion.

Results:   The majority of the studies reviewed suggested that there is no effect of prior concussion on memory-related neurocognitive performance. However, two studies found subtle differences in the Verbal Memory composite score of the ImPACT battery in athletes with previous concussions. One study found that athletes with repetitive subconcussive impacts displayed poorer ImPACT composite scores for Visual Motor Speed and Reaction Time, both of which comprise tasks involving working memory. Other studies demonstrated athletes with previous concussions reporting more severe post-concussion symptoms.

Conclusions: These findings, although inconclusive, are consistent with previous studies that show subtle but not definitive evidence that multiple concussions can play a role in neurocognitive deficits in youth athletes. Potential future studies should utilize multiple modalities of neurocognitive testing to account for different cognitive domains. This would include more multiply concussed athletes who are often underrepresented, and define the severity and temporality of prior concussions. Further longitudinal research is required to determine the extent to which multiple brain traumas affect post-concussion symptoms.

Word Count: 253

Keywords: traumatic brain injury, brain concussion, multiple concussions, athletic injuries, adolescent, child, neurocognition, cognitive function, attention, memory

Ultramini Abstract: This paper determined the effect of prior subconcussive and concussive impacts on neurocognitive performance in youth athletes by reviewing current literature. It was found that most assessments via ImPACT and ANAM showed no difference in performance, while some demonstrated an association between increased frequency of impacts and poorer cognitive performance involving memory.


Traumatic brain injury (TBI) is defined by the Centers for Disease Control and Prevention (CDC) as “a disruption in the normal function of the brain that can be caused by a bump, blow, or jolt to the head, or penetrating head injury” (CDC, 2017). Sports-related TBI is increasingly becoming an important public health concern and has been described as an epidemic by the CDC. The exact incidence is unknown and quite often underreported, but it has recently been estimated that 1.6 million to 3.8 million sports-related TBIs occur in the United States annually (Daneshvar, Nowinski, McKee, & Cantu, 2011; Yue et al., 2016; Langlois, Rutland-Brown, & Wald, 2006), leading to over 500,000 emergency department (ED) visits and more than 60,000 hospitalizations (Yue et al., 2016). While the majority of the focus in the media has been on collegiate or professional athletes, the pediatric population contributes the most participants to contact and collision sports. Over 44 million youth participate in sports annually, who now begin at earlier ages and play multiple sports throughout the year. This increases the potential risk for TBIs, concussions, a type of mild TBI, and subconcussive injury, a term used to describe neural dysfunction after sport exposure in the absence of symptoms (Giza et al., 2013). About 60-80% of sports-related TBI ED visits are by pediatric patients, primarily adolescents (Kannan, Ramaiah, & Vavilala, 2014), and it is estimated that high school sports alone are responsible for at least 250,000 concussions per year (Kimbler, Murphy, & Dhandapani, 2011).

Most youth athletic activities involve some risk for concussion, the highest risk sports being those that engage in interpersonal contact and collisions such as rugby, hockey and American football (Pfister, T., Pfister, K., Hagel, Ghali, & Ronksley, 2016). There is substantial concern surrounding the short and long-term effects of concussions, particularly in children who suffer repetitive and subconcussive injuries. Some experimental models and clinical studies have suggested that multiple injuries may have cumulative and enduring impacts (Mannix et al., 2013; Meehan, Zhang, Mannix, & Whalen, 2012; Guskiewicz et al., 2003; Prins & Hovda, 2003). However, many of these studies pertain to the effects in collegiate and professional athletes, who likely sustain a higher number and severity of injuries than high school and younger athletes. It is inappropriate and insufficient to apply findings from adults to children due to the difference in injury mechanisms, and also the differences in age that determine biomechanical properties, intracranial water content, intracranial blood volume, and overall myelination within the CNS (Giza, Mink, & Madikians, 2007; Bauer, Fritz, & Harald, 2004; Gefen, A., Gefen, N., Zhu, Raghupathi, & Margulies, 2003). Therefore, there is a need for more studies focused on pediatric sports-related TBI.

The neurocognitive effects after suffering a sports-related concussion have been well studied in adults, but again there is a lack of studies documenting these effects in children and adolescents. An additional problem is the lack of standardized measures of specific cognitive domains. Approximately 40% of US high schools with athletic trainers use a computerized test battery to assess neurocognitive function after concussions in pediatric athletes (Meehan, d’Hemecourt, Pollins, Taylor, & Comstock, 2012). Of those using computerized testing, the majority used the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT Inc., Pittsburgh, PA) battery, although the Automated Neuropsychological Assessment Metrics (ANAM developed by the US Department of Defense), CogSport (CogState Ltd, Melbourne, Australia) and HeadMinder (ImPACT Applications, Inc., Axon Sports, LLC) tests are also used (Echemendia et al., 2013).

The ImPACT battery is a web-based computer-administered neurocognitive test developed to assess sport-related concussion in youth, collegiate and professional athletes. It includes six tests/modules that yield five composite/domain scores for visual memory, verbal memory, visual motor processing speed, reaction time, and impulse control, as well as a total symptom score based on 22 common post-concussion symptoms (ex. headache, dizziness) rated from 0 (none) to 6 (severe) on the Post-Concussion Scale (PCS) (ImPACT Applications, Inc., 2017). Details regarding the specific modules and composite scores are provided in the Appendix in Tables 1 and 2, and the Post-Concussion Scale is shown in Table 3.

The ANAM test is a collection of computerized test batteries designed for serial testing of cognitive functions including speed and accuracy of attention, memory, and thinking abilities. It includes 31 test modules testing Simple Reaction Time, Code Substitution, Code-Substitution Delayed, Continuous Performance Test, Mathematical Processing, Matching to Sample, Spatial Processing, Sternberg Memory Procedure, and Procedural Reaction Time (Cernich, Reeves, Sun, & Bleiberg, 2007).

There is also a paucity of research on the effects of sustaining repetitive subconcussive injuries or previous concussions on neurocognitive function in children and adolescents, resulting in conflicting results from the literature that does exist. Multiple concussions have been associated with risk of future concussions, and prolonged symptoms and recovery time in adolescents (Eisenberg, Andrea, Meehan, & Mannix, 2013; Terwilliger, Pratson, Vaughan, & Gioia, 2016; Miller et al., 2016). Re-injury to the brain before full recovery has occurred has been hypothesized to increase impairment and prolong neurometabolic recovery, as shown in animal models (Prins, Hles, Reger, Giza, & Hovda, 2010) and in a small sample of three adult male athletes (Vagnozzi et al., 2008). High school athletes with a history of two or more concussions have shown higher ratings of cognitive, physical and sleep symptoms than athletes with a history of one or no previous concussions (Schatz, Moser, Covassin, & Karpf 2011). Moser, Schatz, & Jordan (2005) found that high school athletes with recent concussions performed significantly worse on measures of attention and concentration than youth athletes with no concussion, especially those with a history of two or more concussions. However, it is unclear if a consistent quantifiable negative impact on neurocognition is seen with a history of concussion, particularly in the pediatric population.

The purpose of this review was to determine whether a history of sports-related TBI has an effect on neurocognitive performance in children and adolescents. The hypothesis was that youth athletes with previous concussions or subconcussive injuries would display poorer neurocognitive performance on tasks relating to memory compared to those without previous injury.


All of the referenced literature included in this review was identified using two databases, PubMed and Google Scholar. Medline medical subject headings (MeSH) and search terms included combinations of “traumatic brain injury/brain concussion”, “multiple concussions” “child/pediatric/adolescent”, “neurocognition”, “cognitive function”, “attention” and “memory”. Studies were included if they were original research on the neurocognitive effects of sports-related TBI or concussion in children published in English within the last five years. Studies were restricted to only sports-related TBIs rather than including more general studies of TBIs, as the latter involve various mechanisms and injuries. Due to the focus on the pediatric population in sports, studies only included research on children ages 5 to 22 years and were excluded if they included mixed age cohorts and did not report child data separately from adult data. Studies that did not involve testing neurocognitive performance relating to memory were excluded. Publications that compared computerized testing of neurocognitive performance of previously concussed with not previously concussed children were included. Review articles, case reports, and meta-analyses were excluded, as well as studies containing less than 15 subjects.


Six of the eight studies chosen were cohort studies, three of which were cross-sectional, three of which were retrospective in design. The other two studies were matched case-control studies. Seven of the studies used ImPACT to assess neurocognitive performance, six of which compared preseason baseline scores of previously concussed and non-concussed groups, while the last examined the subconcussive effects of high and low contact sports. The remaining study used different compared pre- and post-season neurocognitive function with ANAM.

In assessing the validity and reliability of ImPACT in a sample of symptomatic concussed high school and collegiate athletes, a study by Schatz and Sandel (2013) reported the online version of ImPACT showed 91% sensitivity and 69% specificity. Within that same sample, the measure possessed 95% sensitivity and 97% specificity for athletes suspected of denying their symptoms (Schatz & Sandel, 2013).

Mannix et al. (2014) retrospectively evaluated the association of prior concussion on baseline computerized neurocognitive testing with ImPACT. The study sample consisted of 6005 male and female high school athletes from Maine, with a mean age of 16. Subjects who reported a concussion within 26 weeks of testing, history of epilepsy, or history of brain surgery were excluded. The athletes completed baseline preseason testing with ImPACT and were divided into groups based on the number of previously sustained concussions. The sports represented included football, soccer, basketball, ice hockey, lacrosse, track and field, wrestling, baseball, field hockey, cheerleading, volleyball, swimming and softball. 85.3% of the athletes had never sustained a concussion, 10% reported a history of 1 concussion, 2.9% sustained 2 concussions, 1.1% reported 3 prior concussions, and 0.6% sustained 4 or more concussions. Simple linear regression showed that increasing frequency of previous concussions was associated with decreased baseline composite scores in Verbal memory (p = 0.039), increased scores on Impulse control (p = 0.002), and increased total scores on the PCS (p < 0.001).  However, with multivariate modeling accounting for predictors such as age, gender, history of attention-deficit/hyperactivity disorder (ADHD) or learning difficulties, history of headache or migraine treatment, and history of psychiatric condition, only the relationship between Impulse Control and number of prior concussions remained. Therefore, the number of previous concussions is not a strong predictor of ImPACT scores.
The paper entitled “Effects of two concussions on the neuropsychological functioning and symptom reporting of high school athletes” by Tsushima, Geling, Arnold, & Oshiro (2016) used a retrospective archival search to select 483 high school athletes aged 14-18 who were administered ImPACT during preseason baseline testing. Excluded from the study were participants whose primary language was not English, those with a history of learning disability or special education, those who were tested less than seven days after a concussion, and those with three or more prior concussions (Tsushima et al., 2014). Athletes were divided into three groups based on number of self-reported previous concussions: no concussion (84.7%), 1 concussion (12%) and 2 concussions (3.3%). The sports represented included football, soccer, wrestling, basketball, baseball, judo, cheerleading, volleyball, track, softball, field hockey, boxing and paddling. There were no significant differences in ImPACT composite scores (p > 0.25) or the Total Symptom score (p = 0.195) between the three groups, even when adjusted for age. It was found that athletes with a history of two concussions were on average older compared to those without a history of only one previous concussion.

Brooks et al. (2017) conducted a cross-sectional cohort study to determine whether or not measureable differences in cognitive functioning or symptom reporting exist in high school football players with a history of multiple concussions. A total of 5232 male adolescent football players age 14 to 18 from Maine were assessed using the ImPACT battery and symptom ratings were obtained from the PCS. Athletes were excluded if they had a concussion within 6 months before baseline testing, history of meningitis, epilepsy or brain surgery, and testing completed in a language other than English. Based on the number of self-reported injury history, the athletes were stratified into 0 (80%), 1 (14%), 2 (4.1%), 3 (1.3%), or ≥4 (0.6%) prior concussions. There were no significant differences in ImPACT performance across the 5 groups when examining the independent contribution of concussion history (p = 0.396).  However, there was an association between greater symptom scores and a higher number of prior concussions across all of the symptom score subdomains: cognitive-sensory, sleep-arousal, vestibular-somatic, and affective (p < 0.001).

A report by Brooks et al. (2013) investigated the potential cumulative effects of prior concussions on neurocognitive functioning. In this study 643 male and female Bantam and Midget ice hockey players, ages 13 to 17.9 years old, were recruited from the most elite divisions of hockey leagues in Edmonton, Alberta, Canada. Exclusion criteria consisted of having English as a second language, having had a concussion within the past six months, attention or learning problems, and an injury or chronic illness that prevented return to play at the beginning of the season. All athletes completed ImPACT baseline testing as part of a larger study on outcome from concussion (Brooks et al., 2013). Previous concussions were self-reported by the athlete and parent/guardian using a pre-season questionnaire (PSQ). Groups were made for those with no (62%), one (30.8%), and two or more (7.2%) previous concussions. The number of previous concussions as rated on the PSQ was significantly correlated with higher levels of PCS symptom ratings (p = 0.002), but not with cognitive abilities measured by ImPACT (p > 0.05) (Brooks et al., 2013).

Barker et al. (2017) performed a matched case-control analysis to determine the effects of multiple concussions on neurocognitive scores and symptoms using ImPACT. From the Nova Southeastern Sports Medicine Clinic database, exclusion criteria and case matching was applied to a pool of 26,240 male and female athletes, leaving 204 participants aged 10-19. Each concussion group contained 68 participants to equally represent those with 0, 1, and 2 prior concussions. Exclusion criteria included a self-reported history of treatment for substance abuse, psychiatric disorder, special education enrollment, repeated years of schooling, diagnosis of ADHD, learning disability, autism, speech therapy; different first language than test administered; and history of brain surgery (Barker et al., 2017). A variety of sports were represented, including football, lacrosse, soccer, wrestling, gymnastics, swimming, cheerleading, basketball, baseball, and tennis. Comparisons of the groups’ ImPACT composite scores showed no difference for Verbal Memory, Visual Memory, Visual Motor, Reaction Time, or Impulse Control. There was, however, a significant difference in the PCS total symptom score between those who had no previous concussions and those with two previous concussions, with the total symptom score of the two-previous concussion group being more than twice the score of the zero-concussion group (p < 0.05). Those with one prior concussion did not significantly differ from the other groups in symptom reporting.

The report by Iverson, Echemendia, LaMarre, Brooks, & Gaetz (2012) examined whether a history of three or more concussions is associated with poorer performance on neurocognitive testing or with greater reporting of subjective symptoms during athletes’ baseline preseason evaluation. From an archival database containing 786 male athletes who had undergone baseline ImPACT testing, 26 athletes aged 17 to 22 with a self-reported history of three or more concussions were identified. These athletes were matched to controls who had not sustained a prior concussion on age, education, self-reported learning problems, special education, and repeated grades. Most athletes were also matched based on the sport they played, the position played if possible, and the school or organization attended (Iverson et al. 2012). The sports included football, soccer, ice hockey, lacrosse, wrestling, and water polo. The compared ImPACT scores showed no difference between the groups except for the Verbal Memory composite, in which athletes with three or more prior concussions performed more poorly than those without concussion history (p = 0.028). The PCS symptom score differences were not significant (p = 0.13).

Another study by Tsushima, Geling, Arnold, & Oshiro (2016) explored the neuropsychological effects of repetitive, subconcussive impacts in youth sports that do not result in a diagnosable concussion. In this study 282 non-concussed male high school athletes (grades 8 to 12) from public schools were divided into two groups based on the known rates of concussions among various sports: the high contact sport group, consisting of football players and the low-contact sport group, including wrestling, soccer, baseball, judo and baseball. The assumption was made that more subconcussive head traumas occur in sports in which more concussions are reported (Tsushima et al., 2016). This increased frequency of injury can be likened to having a history of more previous concussions, while the lower frequency of injury can be likened to having fewer or no previous concussions. Baseline testing using ImPACT composite scores and symptom reporting using PCS was completed prior to the start of the athletes’ respective seasons and compared to testing after the season was finished. Analyses using t tests comparing the high-contact and low-contact groups showed lower scores in Visual Motor Speed (p < 0.001) and Reaction Time (p < 0.001) in the football players, but no differences in Verbal Memory, Visual Memory, and no significant differences in PCS total symptom scores (p = 0.069).

The study entitled “Examining neurocognitive function in previously concussed interscholastic female soccer players” by Forbes, Glutting, & Kaminski (2014) conducted a cross-sectional assessment of neurocognitive performance as related to repetitive subconcussive blows from purposeful heading. A total of 210 interscholastic female soccer players were recruited from four metropolitan U.S. high schools and equally split into two groups: never-been concussed (control) and previously concussed (experimental) prior to the beginning of the season. The Automated Neuropsychological Assessment Metrics (ANAM) version 1.0 was used to measure pre- and post-season neurocognitive function and a Concussion Symptom Checklist was completed to assess number and severity of 17 symptoms rated from 0 to 6. The ANAM test consists of five subset components: simple reaction time, continuous performance test, math processing, matching to sample, and Sternberg memory. These components are further explained in Table 4 of the Appendix. The number of headers was also recorded for each player during each competitive game. The average number of headers was approximately 24 for both the control and experimental groups. The average number of previous concussions in the experimental group was 1.3. There were no significant differences found in neurocognitive performance in any component of the ANAM between the two groups or in reported symptoms (all p values > 0.05). Therefore, the authors concluded that purposeful heading in soccer is more than likely not detrimental in terms of neurocognitive performance and long-term persistent concussion-related symptoms, even when players had a previous history of concussion (Forbes et al., 2014).



Overall, the results of this review do not strongly support the hypothesis that sustaining prior concussions negatively impacts objective neurocognitive function in adolescents that undergo computerized testing. However, it is premature to definitively conclude that there is no association, as two studies suggested a negative impact on Verbal Memory composite scores (Mannix et al., 2014; Iverson et al., 2012), and one study demonstrated lower scores for Visual Motor Speed and Reaction Time, both of which contain tasks involving working memory (Tsushima, Geling, Arnold & Oshiro, 2016).

The varying absence and presence of differences amongst neurocognitive function seen in these studies is not necessarily unprecedented, as past studies have also reported mixed results. Earlier studies containing high school athletes proposed that a history of two or more prior concussions display greater impairment on neuropsychological and memory tests than athletes with a history of only one concussion (Collins et al., 1999; Collins et al., 2002). Other studies containing high school and collegiate athletes have suggested measureable cognitive deficits from multiple prior concussions (Gaetz, Goodman, & Weinberg, 2000; Iverson, Gaetz, & Collins, 2004; Moser, Schatz, & Jordan, 2005; Schatz, Moser, Covassin, & Karpf, 2011), yet still others suggest no effects (Broglio, Ferrara, Piland, & Anderson, 2006; Bruce & Echemendia, 2009; Collie, McCrory & Makdissi, 2006; Iverson, Brooks, Lovell & Collins, 2006; Thornton, Cox, Whitfield, & Fouladi, 2008).

There are several speculated explanations for the apparent lack of association between concussion history and obvious testable neurocognitive deficits despite this association being proven in adults. Due to the focus on a younger population, it is possible that these results could represent recovery and neuroplasticity, a shorter duration of overall exposure to concussions and repetitive head trauma, less intense exposure (e.g. fewer hits, lesser force of impact, slower ball velocities) than in professional and collegiate athletes, or a combination of those factors (Brooks et al., 2016).

The study by Tsushima, Geling, Arnold, & Oshiro (2016) suggested that participating in high-contact sports such as football may affect neurocognitive functioning in youth athletes due to repetitive subconcussive head trauma. This may be due to incomplete neurological development, as animal and clinical research has shown that immature adolescent brains are more predisposed to head injury (Field, Collins, Lovell & Maroon, 2003). Additionally, it is possible that athletes who experience subconcussive impacts were actually concussed. The concept of a concussion spectrum calls into question the diagnostic approach to concussions, which mainly involves symptom analysis instead of concrete brain imaging (Forbes, Glutting & Kaminski, 2014).

The finding that different neurocognitive deficits were observed in different cognitive domains across studies makes test standardization both necessary yet problematic. Any type of testing must be standardized in order to compare results between individuals, but perhaps the computerized tests currently being used do not encompass all aspects of the neurocognitive domains tested. It is difficult to pinpoint specific domains as they often involve overlapping subdomains, some of which are not always tested. There are many ways to test “memory”, and it appears as though that may be key to ascertaining the effect multiple concussions have on specific aspects of memory. Every reviewed study used computerized testing of neurocognition, either ImPACT or ANAM for the purpose of standardizing performance assessment. ImPACT is relatively easy to administer, and research in the past decade has shown that it is a reliable, valid and practical approach to the neurocognitive assessment of mild TBI in high school and collegiate athletes (Schatz & Ferris, 2013; Maerlender et al, 2013).

ANAM is not as widely used as ImPACT, but has proven suitable for concussion surveillance and demonstrated validity, reliability as well as sensitivity to mild TBI (Cernich, Reeves, Sun, & Bleiberg, 2007).

There are many advantages of computerized testing, such as easy and rapid administration to large groups (and in different languages if needed), inclusion of adaptive test procedures, provision of more accurate and precise measurement of time-based responses (e.g. reaction time, response latency); consistent administration across various settings; immediate results relating to clinical diagnosis and prognosis, and finally easy collection, storage, access and sharing of data (De Marco & Broshek, 2016). Even with these advantages, the findings of this review support the need for future studies to use different testing modalities, as the neuropsychological test measures in this research only account for a handful of neurocognitive functions related to memory (visual memory, working memory) contained in the ImPACT battery and ANAM battery.

For example, Moore et al. (2015) investigated the effects of concussion history on children’s neurocognitive processing, with an emphasis on attention and cognitive control, during flanker performance. Cognitive control is associated with a wide range of processes and cannot be restricted to a particular cognitive domain, but the core functions that constitute cognitive control are working memory, inhibition and cognitive flexibility (Diamond, 2013). The presence of impairments in cognitive control may be attributed to deficits in attention, memory, language comprehension and emotional processing (Mackie, Van Dam & Fan, 2013).

It is worth noting that the ImPACT and ANAM test batteries are screening tools designed to assess cognitive symptoms to help diagnose concussions in an acute setting. These tests might not possess the sensitivity and specificity to allow for detection of chronic changes in cognition (Lovell, Collins, Iverson, Johnston & Bradley, 2004).

An important finding in four of the seven studies that used ImPACT testing is that adolescents with a history of prior concussions or repetitive subconcussive trauma had more self-reported symptoms (such as dizziness, fatigue, trouble falling asleep, difficulty concentrating, difficulty remembering) as assessed by the Post-Concussion Scale symptom portion of ImPACT, even when more than six months post-injury. Whether a true causal relationship exists between previous concussions and more severe symptoms needs to be further explored in adolescents and should be considered when monitoring recovery and determining the risk of sustaining a subsequent concussion.Unexpectedly, the number of symptoms reported did not correlate with testable neurocognitive abilities, suggesting that the current screening tools used are not able to accurately measure the subjective effects experienced by those with multiple concussions, and that perhaps multiple testing modalities should be included in future studies.

There were certain limitations and weaknesses in the reviewed studies that may have led to the inconclusive results of this review. First, although most studies had a large number of participants in total, all studies had small sample sizes of previously concussed athletes. As the number of previous concussions increased, the corresponding number of athletes decreased. For example, in the study by Iverson et al. (2012), their database of 786 subjects only contained 26 individuals with a history of three or more injuries. This limitation is and has been difficult to overcome due to the lack of a large cohort of multiply concussed athletes. Future studies would benefit from representing more previously concussed athletes to match the high number of controls to increase the statistical power of the studies.

Second, none of the studies defined the injury severity characteristics of prior concussions, age at which the prior concussions occurred, or the time interval between multiple concussions.Consequently, the severity of concussions was unknown. Though most studies only included athletes if they had not sustained a concussion within the past six months, the average time since last concussion and time between concussions was unaccounted for. This is important to know due to the concept of a vulnerable window, during which repeat concussion results in worse outcomes and prolonged recovery in children (Eisenberg, Andrea, Meehan & Mannix, 2013). There is preliminary evidence for a potential association between a second blow to the head sustained within 24 hours of an initial concussion and more significant and persistent post-concussion symptoms in adolescent student-athletes (Terwilliger, Pratson, Vaughan, Gioia, 2016).

Third, the primary outcome of all of the studies reviewed was a rapid computerized screen of neurocognitive abilities (either ImPACT or ANAM) and does not represent a more complex or comprehensive evaluation of neurocognitive functioning, such as paper and pencil testing from a battery of neuropsychological tests. Computerized testing is useful for standardized scores yet is in itself limiting, as it does not approach neurocognitive deficit detection from all angles. There is a general lack of studies assessing post-concussion neurocognition in children in other ways, and it is possible that a different test battery would lead to results that differ from those of this review. Furthermore, studies did not include other methods of investigation such as functional magnetic resonance imaging (fMRI) that may reveal persistent neurological effects from prior multiple concussions through the detection of hemodynamic changes associated with neuronal activity. Researchers have documented fMRI changes in athletes with persistent post-concussion symptoms but no difference in neurocognitive task performance compared to controls. (Chen et al., 2014). This suggests that regional brain activation shown by fMRI during a task may be more sensitive to cumulative concussive injuries than test performance. Thus, the conclusions from this review are limited to an absence of neurocognitive effects based solely on the computerized batteries that were administered.
Fourth, each individual tested could have a myriad of psychological, biological, and social factors that could either intensify or reduce the severity or onset of serious head injury consequences. Some of these were accounted for, but the same inclusion and exclusion criteria did not apply to all of the studies.

Fifth, the studies were not able to account for variables that may have affected neuropsychological functioning such as test effort and school aptitude. Some athletes tend to “tank” or “sandbag” their baseline performances in a way that makes post-concussion return-to-play easier to pass (Erdal, 2012). No studies reviewed contained effort tests to assess test motivation in the participants.

Lastly, the studies’ common use of cross-sectional design renders them less powerful and convincing than a longitudinal study. In addition, the retrospective reporting of concussions can be influenced by normal human biases, especially if concussions occurred many years prior.

Going forward, studies would benefit from recruiting larger samples of concussed athletes and using multiple investigative techniques of neurocognitive testing to assess a broader set of measures and domains. For example, the concurrent use of multiple batteries of computerized tests with imaging to view brain activity during task performance could add another dimension to the studies performed thus far. Future studies should be longitudinal in nature, wherein athletes are followed over several years to monitor recovery, repeat testing, document additional injuries, and identify those who stop playing sports due to injuries. Certain studies in this review indirectly attributed the group differences in reporting more symptoms to a history of concussions, but a longitudinal study might establish whether there is a direct relationship, or if there are factors that mediate that relationship. Prospective studies are also required to determine whether the lack of adverse cognitive effects found here is carried throughout an athlete’s lifetime, or if the cumulative effects of concussions show up as even longer-term symptom sequelae. Of particular importance is the potential association between playing high school football and the increased risk of neurodegenerative disorders such as chronic traumatic encephalopathy and Alzheimer’s disease later in life. Another area for further research is exploring the plausible relationship between increased subjective symptom scores and long-term development of depression, anxiety, or other mental health problems.

In conclusion, there are still inconsistencies regarding the effects of multiple prior concussions on computerized neurocognitive performance in youth athletes, though the evidence shows that prior concussive and subconcussive history being a very weak predictor of lower neurocognitive scores for tasks relating to memory. Although these results are somewhat reassuring given the current concern over possible long-term cognitive effects, they still highlight the need for more studies using different modalities of neurocognitive testing in pediatric athletes and designing studies to be longitudinal in nature. The implications of these findings are important not only for the management of youth athletes with a history of multiple concussions, but also for the diagnosis of concussions, monitoring recovery, return-to-play guidelines, and implementing new policies regarding rules of youth sports to minimize the amount of traumatic brain injuries.






































Barker, T., Russo, S. A., Barker, G., Rice, M. A., Jeffery M. G., Broderick, G., &                             Craddock, T. J. (2017). A case matched study examining the reliability of using                            ImPACT to assess the effects of multiple concussions. BMC psychology 5(1), 14.


Bauer, R., & Fritz, H. (2004). Pathophysiology of traumatic injury in the developing                            brain: an introduction and short update. Experimental and ToxicologicPathology56(1), 65-73.

Broglio, S. P., Ferrara, M. S., Piland, S. G., & Anderson, R. B. (2006). Concussion                            history is not a predictor of computerised neurocognitive performance. British Journal of Sports Medicine40(9), 802-805.

Brooks, B. L., Mannix, R., Maxwell, B., Zafonte, R., Berkner, P. D., & Iverson, G. L.                            (2016). Multiple past concussions in high school football players: are there                                          differences in cognitive functioning and symptom reporting?. The American journal of sports medicine44(12), 3243-3251.

Brooks, B. L., McKay, C. D., Mrazik, M., Barlow, K. M., Meeuwisse, W. H., & Emery,                            C. A. (2013). Subjective, but not objective, lingering effects of multiple past                            concussions in adolescents. Journal of neurotrauma30(17), 1469-1475.

Bruce, J. M., & Echemendia, R. J. (2009). History of multiple self-reported                             concussions is not associated with reduced cognitive                                                                      abilities. Neurosurgery64(1), 100-106.

Centers for Disease Control and Prevention. (2017).  Traumatic Brain Injury &                             Concussion. Retrieved from

Cernich, A., Reeves, D., Sun, W., & Bleiberg, J. (2007). Automated                neuropsychological assessment metrics sports medicine battery. Archivesof Clinical Neuropsychology22, 101-114.

Collie, A., McCrory, P., & Makdissi, M. (2006). Does history of concussion affect                            current cognitive status?. British Journal of Sports Medicine40(6), 550-551.

Collins, M. W., Grindel, S. H., Lovell, M. R., Dede, D. E., Moser, D. J., Phalin, B. R.,                            … & Sears, S. F. (1999). Relationship between concussion and              neuropsychological performance in college football players. Jama282(10), 964-              970.

Collins, M. W., Lovell, M. R., Iverson, G. L., Cantu, R. C., Maroon, J. C., & Field,                            M. (2002). Cumulative effects of concussion in high school                                                                      athletes. Neurosurgery51(5), 1175-1181.

Daneshvar, D. H., Nowinski, C. J., McKee, A. C., & Cantu, R. C. (2011). The                             epidemiology of sport-related concussion. Clinics in sports medicine30(1), 1-              17.

De Marco, A. P., & Broshek, D. K. (2016). Computerized cognitive testing in the               management of youth sports-related concussion. Journal of childneurology31(1), 68-75.

Echemendia, R. J., Iverson, G. L., McCrea, M., Macciocchi, S. N., Gioia, G. A.,               Putukian, M., & Comper, P. (2013). Advances in neuropsychological assessment               of sport-related concussion. Br J Sports Med47(5), 294-298.

Eisenberg, M. A., Andrea, J., Meehan, W., & Mannix, R. (2013). Time interval               between concussions and symptom duration. Pediatrics, peds-2013.

Erdal, K. (2012). Neuropsychological testing for sports-related concussion: how athletes can sandbag their baseline testing without detection. Archives of clinicalneuropsychology27(5), 473-479.

Field, M., Collins, M. W., Lovell, M. R., & Maroon, J. (2003). Does age play a role                            in recovery from sports-related concussion? A comparison of high school and              collegiate athletes. The Journal of pediatrics142(5), 546-553.

Forbes, C. R., Glutting, J. J., & Kaminski, T. W. (2016). Examining neurocognitive              function in previously concussed interscholastic female soccer players. AppliedNeuropsychology: Child5(1), 14-24.

Gaetz, M., Goodman, D., & Weinberg, H. (2000). Electrophysiological evidence for              the cumulative effects of concussion. Brain Injury14(12), 1077-1088.

Gefen, A., Gefen, N., Zhu, Q., Raghupathi, R., & Margulies, S. S. (2003). Age dependent changes in material properties of the brain and braincase of the              rat. Journal of neurotrauma20(11), 1163-1177.

Giza, C. C., Kutcher, J. S., Ashwal, S., Barth, J., Getchius, T. S., Gioia, G. A., … &              McKeag, D. B. (2013). Summary of evidence-based guideline update: Evaluation              and management of concussion in sports Report of the Guideline Development              Subcommittee of the American Academy of Neurology. Neurology80(24), 2250              2257.

Giza, C. C., Mink, R. B., & Madikians, A. (2007). Pediatric traumatic brain injury:              not just little adults. Current opinion in critical care13(2), 143-152.

Guskiewicz, K. M., McCrea, M., Marshall, S. W., Cantu, R. C., Randolph, C., Barr,              W., … & Kelly, J. P. (2003). Cumulative effects associated with recurrent              concussion in collegiate football players: the NCAA Concussion              Study. Jama290(19), 2549-2555.

ImPACT Applications, Inc. (2017). ImPACT version 2.0: Clinical user’s manual.                             Retrieved from

Iverson, G. L., Brooks, B. L., Lovell, M. R., & Collins, M. W. (2006). No cumulative              effects for one or two previous concussions. British Journal of SportsMedicine40(1), 72-75.

Iverson, G. L., Echemendia, R. J., LaMarre, A. K., Brooks, B. L., & Gaetz, M. B. (2012).              Possible lingering effects of multiple past concussions. Rehabilitationresearch and practice2012.

Iverson, G. L., Gaetz, M., Lovell, M. R., & Collins, M. W. (2004). Cumulative effects              of concussion in amateur athletes. Brain injury18(5), 433-443.

Kannan, N., Ramaiah, R., & Vavilala, M. S. (2014). Pediatric neurotrauma. International journal of critical illness and injury science4(2), 131.

Kimbler, D. E., Murphy, M., & Dhandapani, K. M. (2011). Concussion and the Adolescent Athlete. The Journal of Neuroscience Nursing: Journal of theAmerican Association of Neuroscience Nurses43(6), 10.

Langlois, J. A., Rutland-Brown, W., & Wald, M. M. (2006). The epidemiology and              impact of traumatic brain injury: a brief overview. The Journal of head traumarehabilitation21(5), 375-378.

Lovell, M. R., Collins, M. W., Iverson, G. L., Johnston, K. M., & Bradley, J. P. (2004).                            Grade 1 or “ding” concussions in high school athletes. The American journal ofsports medicine32(1), 47-54.

Mackie, M. A., Van Dam, N. T., & Fan, J. (2013). Cognitive control and attentional              functions. Brain and Cognition82(3), 301–312.

Mannix, R., Iverson, G. L., Maxwell, B., Atkins, J. E., Zafonte, R., & Berkner, P. D.              (2014). Multiple prior concussions are associated with symptoms in high school              athletes. Annals of clinical and translational neurology1(6), 433-438.

Mannix, R., Meehan, W. P., Mandeville, J., Grant, P. E., Gray, T., Berglass, J., … &              Peters, N. V. (2013). Clinical correlates in an experimental model of repetitive              mild brain injury. Annals of neurology74(1), 65-75.

Meehan III, W. P., Zhang, J., Mannix, R., & Whalen, M. J. (2012). Increasing recovery time between injuries improves cognitive outcome after repetitive mild              concussive brain injuries in mice. Neurosurgery71(4), 885-892.

Meehan, W. P., d’Hemecourt, P., Collins, C. L., Taylor, A. M., & Comstock, R. D.              (2012). Computerized neurocognitive testing for the management of sport-related              concussions. Pediatrics129(1), 38-44.

Miller, J. H., Gill, C., Kuhn, E. N., Rocque, B. G., Menendez, J. Y., O’Neill, J. A., …              & Ferguson, D. (2016). Predictors of delayed recovery following pediatric sports              related concussion: a case-control study. Journal of Neurosurgery:Pediatrics17(4), 491-496.

Moore, R. D., Pindus, D. M., Drolette, E. S., Scudder, M. R., Raine, L. B., & Hillman, C. H. (2015). The persistent influence of pediatric concussion on              attention and cognitive control during flanker performance. Biologicalpsychology109, 93-102.

Moser, R. S., Schatz, P., & Jordan, B. D. (2005). Prolonged effects of concussion in              high school athletes. Neurosurgery57(2), 300-306.

Pfister, T., Pfister, K., Hagel, B., Ghali, W. A., & Ronksley, P. E. (2016). The incidence of concussion in youth sports: a systematic review and meta-              analysis. Br J Sports Med50(5), 292-297.

Prins, M. L., & Hovda, D. A. (2003). Developing experimental models to address traumatic brain injury in children. Journal of neurotrauma20(2), 123-137.

Prins, M., Hales, A., Reger, M., Giza, C., & Hovda, D.(2010).Repeat traumatic brain              injury in the juvenile rat is associated with increased axonal injury and cognitive              impairments. Dev. Neurosci. 32, 510.

Schatz, P., & Sandel, N. (2013). Sensitivity and specificity of the online version of              ImPACT in high school and collegiate athletes. The American Journal of SportsMedicine41(2), 321-326.

Schatz, P., Moser, R. S., Covassin, T., & Karpf, R. (2011). Early indicators of enduring symptoms in high school athletes with multiple previous              concussions. Neurosurgery68(6), 1562-1567.

Terwilliger, V. K., Pratson, L., Vaughan, C. G., & Gioia, G. A. (2016). Additional post-concussion impact exposure may affect recovery in adolescent       athletes. Journal of neurotrauma33(8), 761-765.

Thornton, A. E., Cox, D. N., Whitfield, K., & Fouladi, R. T. (2008). Cumulative concussion exposure in rugby players: neurocognitive and symptomatic              outcomes. Journal of Clinical and Experimental Neuropsychology30(4), 398-              409.

Tsushima, W. T., Geling, O., Arnold, M., & Oshiro, R. (2016). Are there subconcussive              neuropsychological effects in youth sports? An exploratory study of high-and              low-contact sports. Applied Neuropsychology: Child5(2), 149-155.

Tsushima, W. T., Geling, O., Arnold, M., & Oshiro, R. (2016). Effects of two concussions on the neuropsychological functioning and symptom reporting of              high school athletes. Applied Neuropsychology: Child5(1), 9-13.

Vagnozzi, R., Signoretti, S., Tavazzi, B., Floris, R., Ludovici, A., Marziali, S., … &              Lazzarino, G. (2008). Temporal window of metabolic brain vulnerability to              concussion: a pilot 1H-magnetic resonance spectroscopic study in concussed              athletes—part III. Neurosurgery62(6), 1286-1296.

Yue, J. K., Winkler, E. A., Burke, J. F., Chan, A. K., Dhall, S. S., Berger, M. S., … &              Tarapore, P. E. (2016). Pediatric sports-related traumatic brain injury in United              States trauma centers. Neurosurgical focus40(4), E3


Table 1. ImPACT Neuropsychological Test Modules (ImPACT Applications Inc., 2017).

Test Module & Description Ability Areas Tested
  1. Word Discrimination

-Twelve target words presented twice
-Recall is tested for a 24-word list with 12 target words and 12 non-target words from the same semantic category as the target word

Immediate and delayed memory for words
  1. Design Memory

-Twelve target designs presented twice
-Recall is tested for 24 designs, 12 target and 12 non-target designs

Immediate and delayed memory for designs
  1. X’s and O’s

-First is a distractor task: subject is asked to click the left mouse button if a blue square is presented and the right mouse button if a red circle is presented

– Then subject is presented with a random assortment of illuminated X’s and O’s and tested for memory of their location after the distractor task re-appears

Attention, concentration, working memory, reaction time
  1. Symbol Match

-Recall of 9 common symbols matched to a number

Visual processing speed, learning and memory
  1. Color Match

-First, the subject clicks a red, blue or green button as they are presented on the screen to avoid effects of color blindness

-Then a word is displayed on the screen in the same colored ink as the word or in a different colored ink

-Subject is instructed to click in the box as quickly as possible only if the word is presented in the matching ink

Focused attention, response inhibition, reaction time
  1. Three Letters

-Subject clicks on numbered buttons in backward order starting with 25, then is asked to remember three consonant letters displayed on the screen

-Numbered grid re-appears and subject clicks the numbered buttons in backward order again

-After 18 seconds, the numbered grid disappears and subject recalls the three letters using the keyboard

Attention, concentration, working memory, visual-motor speed

Table 2. ImPACT Composite Scores Composition (ImPACT Applications Inc., 2017)

Composite Score Comprised of:
Verbal Memory (higher score = better performance)
  • Total % correct score from Module 1
  • Total correct hidden symbols from Module 4
  • % total letters correct from Module 6
Visual Memory (higher score = better performance)
  • Total % correct score from Module 2
  • Total correct memory score from Module 3
Visual Motor Speed (higher score = better performance)
  • Total number correct /4 during interference of Module 3
  • Average counted correctly x3 from countdown phase of Module 6
Reaction Time (lower score = better performance)
  • Average correct RT of interference stage of Module 3
  • Average Correct RT /3 of Module 4
  • Average Correct RT of Module 5
Impulse Control (lower score = better performance) 

*Used as a measure of test validity

  • Total errors on the interference phase of Module 3
  • Total commissions from Module 5

Table 3. Post-Concussion Symptom Scale (ImPACT Applications Inc., 2017).

Symptom Minor Moderate Severe
Headache 1 2 3 4 5 6
Nausea 1 2 3 4 5 6
Vomiting 1 2 3 4 5 6
Balance Problems 1 2 3 4 5 6
Dizziness 1 2 3 4 5 6
Fatigue 1 2 3 4 5 6
Trouble Falling Asleep 1 2 3 4 5 6
Sleeping More Than Usual 1 2 3 4 5 6
Sleeping Less Than Usual 1 2 3 4 5 6
Drowsiness 1 2 3 4 5 6
Sensitivity to Light 1 2 3 4 5 6
Sensitivity to Noise 1 2 3 4 5 6
Irritability 1 2 3 4 5 6
Sadness 1 2 3 4 5 6
Nervousness 1 2 3 4 5 6
Feeling More Emotional 1 2 3 4 5 6
Numbness or Tingling 1 2 3 4 5 6
Feeling Slowed Down 1 2 3 4 5 6
Feeling Mentally Foggy 1 2 3 4 5 6
Difficulty Concentrating 1 2 3 4 5 6
Difficulty Remembering 1 2 3 4 5 6
Visual Problems 1 2 3 4 5 6

Instead of zero, subjects check a box indicating they are not experiencing the symptom.

Table 4. Subset components of ANAM used by Forbes, Glutting & Kaminski (2016).

Component Description
Simple reaction time (SRT) -Measures response time (ms) to a stimulus presented at various time intervals
Continuous performance test (CPT) -Measures attention and concentration 

-Participant must continuously monitor letters and identify whether the current letter displayed is the same or different from the immediately preceding letter

Math processing (MTH) -Measures mental processing speed and mental efficiency 

-Participants solve a three-step simple addition or subtraction equation and identify whether the solution is greater than or less than 5

Matching to sample (MSP) -Measures visual memory 

-Participants recall a checkerboard matrix after 5s and match it to the original matrix design

Sternberg memory (STN) -Measures working memory 

-Participants memorize a string of 6 letters and subsequently recall whether or not a presented letter belongs to that six-letter string

Table 5. Summary of the reviewed studies reporting findings in previously concussed groups compared to non-concussed groups.

Study Authors Testing Modality Cognitive Domains Affected No. of Post-Concussion Symptoms Significance
(p value)
“Multiple prior concussions are associated with symptoms in high school athletes” Mannix et al. ImPACT -Verbal Memory (decreased) 

-Impulse Control (increased)

Increased -Verbal Memory
(p = 0.039) 

-Impulse Control
(p = 0.002)
-Symptoms (p < 0.001)

“Effects of two concussions on the neuropsychological functioning and symptom reporting in high school athletes” Tsushima, Geling, Arnold & Oshiro ImPACT None Same -ImPACT Scores
(p > 0.25)
-Symptoms (p = 0.195)
“Multiple past concussions in high school football players: are there differences in cognitive functioning and symptom reporting?” Brooks et al. ImPACT None Increased -ImPACT Scores
(p = 0.396)
-Symptoms (p < 0.001)
“Subjective but not objective, lingering effects of multiple past concussions in adolescents” Brooks et al. ImPACT None Increased -ImPACT Scores
(p > 0.05)
-Symptoms (p = 0.002)
“A case matched study examining the reliability of using ImPACT to assess the effects of multiple concussions” Barker et al. ImPACT None Increased -ImPACT Scores
(all p > 0.16) 

-Symptoms (p = 0.01)

“Possible lingering effects of multiple past concussions” Iverson et al. ImPACT Verbal Memory Same -Verbal Memory
(p = 0.028)
-Symptoms (p = 0.13)
“Are there subconcussive neuropsychological effects in youth sports? An exploratory study of high-and low-contact sports” Tsushima, Geling, Arnold & Oshiro ImPACT -Visual Motor Speed (increased)
-Reaction Time (increased)
Same -Visual Motor Speed, Reaction Time
(p < 0.001)
-Symptoms (p = 0.69)
“Examining neurocognitive function in previously concussed interscholastic female soccer players

depend on neurochemical signals for proliferation and survival. Given the close similarities between GBM stem cells and neural stem cells (NSC), and their existence in the rich neurochemical milieu of brain, neurochemical signaling may profoundly impact tumor growth. To test this hypothesis, I interrogated a library of 680 neuroactive compounds in patient-derived GBM neural stem cells (GNS) against normal human NSC and human fibroblasts cells to identify selective compounds. I found that compounds modulating dopaminergic, serotonergic and cholinergic signaling pathways selectively affected GNS and NSC growth compared to fibroblast cells. I identified ten compounds that showed more than 8-fold selectivity and in particular, dopamine receptor D4 (DRD4) antagonists (L-741,742 and PNU 96415E) selectively inhibited GBM growth in vitro and in vivo, and exhibited synergism with the current chemotherapeutic drug, temozolomide (TMZ). Primary GBM tumors and GNS cells expressed functional DRD4 receptor and antagonism of DRD4 led to disruption in its downstream effectors PDGFRβ/ERK1/2, and mTOR signaling.  Furthermore, DRD4 antagonism disrupts endolysosmal function compromising the autophagy-lysosomal degradation pathway leading to autophagic flux inhibition with accumulation of autophagic vacuoles, autophagy specific substrate p62, cholesterol and ubiquitinated proteins, followed by G0/G1 cell cycle arrest and apoptosis. The identification of DRD4 antagonists as GNS selective compounds revealed autophagy-lysosomal system as a vulnerable and important system for GBM cell survival and suggests modulation of dopamine signaling may hold therapeutic potential. Interestingly, DRD4 antagonists L-741,742 and PNU 96415E were also identified as hits in the high content differentiation screen. I further characterized DRD4 antagonist induced differentiated cells, which showed VGLUT1 positive cells with increased expression of NERUOG2CTIP2 and ETV1 identifying them as glutamatergic cortical deep layer V neurons. Finally, DRD4, when inhibited in normal NSC, initiate differentiation, while in GNS cells, they are vulnerable and undergo cell death.

List of Abbreviations

5-HT- 5 hydroxytrptamine /serotonin 6-OHDA-6 hydroxydopamine ADHD- attention deficit hyperactivity disorder AML-acute myeloid leukemia AMPA-α-amino-3-hydroxy-5-mehtyl-4isoxazolepropionic acid βArr2- β arrestin 2 BDNF- brain derived neurotrophic factor bFGF-basic fibroblast growth factor bHLH-basic helix-loop-helix BIO- 6-bromoindirubin-3’-oxime BLBP-brain lipid binding protein BMP4- bone morphogenetic protein 4 BSA- bovine serum albumin CAMKII- Ca2 /calmodulin dependent protein kinase II cAMP- cyclic adenosine monophosphate     CD- cluster of differentiation CDKN1A-cyclin dependent kinase inhibitor 1A cDNA- complementary DNA ChAT-choline acetyltransferase CI-combination index CIMP-CpG island methylator phenotype CNS-central nervous system COMT-catechol-o-methyl transferase CpG- cytosine and guanine separated by a phosphate CSC- cancer stem cell CTIP2- coup-tf-interacting protein 2/BCL11B CXCR4-chemokine (C-X-C motif) receptor 4 DAPI-4’6-diaminido-2-phenylindole DARPP-32- dopamine and cAMP regulated phosphoprotein-32/PPP1R1B DAT-dopamine transporter DG-dendate gyrus DKK1-dickkopf 1 DMEM- dulbecco’s modified essential medium DMSO- dimethyl sulfoxide CDKN1B-cyclin dependent kinase inhibitor 1B DNA-deoxyribonucleic acid dNTP-deoxyrnucleotide DQ- dequenched DR-dopamine receptor DRD4-dopamine receptor D4 DRIP-dopamine receptor interacting protein E-epinephrine EGF-epidermal growth factor eGFP- enhanced green fluorescent protein EGFR-epidermal growth factor receptor ERK-extracellular signal regulated kinase ETV1-ets variant 1 FACS-fluorescence activated cell sorting FBS-fetal bovine serum FDR-false discovery rate FGF-fibroblast growth factor FOXG1- forkhead box G1 G0-Growth phase 0 G1-Growth phase 1 GAPDH- glyceraldehyde 3- phosphate dehydrogenase GBM-glioblastoma GFAP-glial fibrillary acidic protein GFP- green fluorescent protein Gi- G protein, inhibitory GIRKS-G protein coupled inward rectifying potassium channel GLAST-glutamate-aspartate transporter GNS-glioblastoma derived neural stem cells GPCR- G protein couple receptor Grb2-growth factor receptor bound protein 2 Gs-G protein, stimulatory GSEA-gene set enrichment analysis GSK3- glycogen synthase kinase 3 IC- inhibitory concentration IDH-isocitrate dehydrogenase IGFR- insulin-like growth factor receptor IP3- inositol triphosphate IPC-intermediate progenitor cells KLF2-kruppel like factor 2 GABA- gamma-aminobutryic acid KLHL12- kelch-like family member 12 L1CAM-L1 cell adhesion molecule LAMP1- lysosomal associated membrane protein 1 LC3- microtubule associated protein 1A/B light chain 3 LDA-limiting dilution assay L-DOPA-L-3,4-dihydroxyphenylalanine LGE-lateral ganglionic eminence MAOB- monoamine oxidase B MAP2- microtubule associated protein 2 MAPK-mitogen activated protein kinase Mash1- mammalian achaete scute homolog1/ASCL1 MDM2- mouse double minute 2 homolog MET-mesenchymal epithelial transition factor MGE-medial ganglionic eminence MGMT- O-methyl-guanine-methyltransferase mRNA-messenger RNA mTOR-mechanistic target of rapamycin NCAM-neural cell adhesion molecule Nck-NCK adaptor protein NE-norepinephrine NEC-neuroepithelium cells NEUROG2- neurogenin 2 NF1- neurofibromatosis type I Ngn-neurogenin NMDA-N-methyl-D-aspartate NS-neural stem cells NSC-neural stem cells NSG-NOD scid gamma (NOD.Cg-Prkdcscid Il2rg tm1Wjl/SzJ) OB- olfactory bulb p53-tumor protein 53 PARP- poly ADP ribose polymerase PCR-polymerase chain reaction PDGFRA- platelet derived growth factor receptor A PDGFRβ-platelet derived growth factor receptor-β PFA-paraformaldehyde nAChR- nicotinic acetylcholine receptor PI3K-phosphoinositide-3-kinase PKA-protein kinase A PLC-phospholipase C PLO-poly-L-ornithine PPI-protein phosphatase 1 PTEN- phosphatase and tensin homolog PVDF-polyvinylidene fluoride membrane Ras- rat sarcoma viral oncogene RB- retinoblastoma RFP- red fluorescent protein RGC-radial glial cells RMS-rostral migratory stream RNA- ribonucleic acid rpS6- ribosomal protein S6 RPS 18- ribosomal protein S18 RTK- receptor tyrosine kinase SATB2-special AT rich sequence binding protein 2 SCID- severe combined immunodeficiency PI-propidium iodide SEM-standard error of the mean SGZ-subgranular zone shRNA-short hairpin RNA SOX2- Sry-box2 SVZ-sub ventricular zone Tbr- t-box brain protein TCGA-The cancer genome atlas TEM-transmission electron microscopy TH-tyrosine hydroxylase TMZ- temozolomide TPH- tryptophan hydroxylase U2-OS-human osteosarcoma cells VGLUT1- vesicular glutamate transporter 1 VNTR-variable number of tandem repeats VTA-ventral tegmental area VZ-ventricular zone WNT-wingless
SDS-sodium dodecyl sulfate
Chapter 1 Introduction

1.1 Glioblastoma

Glioblastoma (GBM) is the most common malignant brain tumor, which accounts for 15.6% of all primary brain tumors and 45.2% of primary malignant brain tumors (Ostrom et al., 2013). The incidence of GBM increases with age from 75-84 years and affects more men than women. GBM can occur in any part of brain including brain stem and cerebellum, but occur more commonly in frontal and temporal lobes. Based on the clinical history, GBM are often classified as primary or secondary. Primary GBM arise de novo with no previous brain lesion and grow very rapidly, and they represent 90-95% of GBM and occur more commonly in older patients. Secondary GBM accounts for only 5-10% of GBM and occurs in younger patients and with an average interval of 4-5 years for disease progression from lower grade astrocytoma to GBM(Ohgaki and Kleihues, 2005).  Primary and secondary GBMs are usually indistinguishable on histological grounds but they show very different genetic alteration and genomics. The histopathological characteristic of GBM includes highly cellular, poorly differentiated pleomorphic astrocyte-like cells with high mitotic activity and nuclear atypia, microvascular proliferation and necrosis with pseudopalisading cells(Furnari et al., 2007). GBMs show inter- and intratumoral heterogeneity, which are properties making them exceedingly difficult to treat.

1.1.1 Glioblastoma therapy

GBM patients are diagnosed based on symptoms, neurological signs and neuroimaging. Since GBM grows rapidly, the most common symptoms are caused by increased pressure in the brain that includes headache, nausea, vomiting, but depending on location of the tumor they can have different symptoms. The current standard care of treatment includes maximum surgical resection, radiotherapy and concomitant or adjuvant chemotherapy with temozolomide (TMZ). Despite this multi modal approach, the median survival time remains 14.6 months with two-year survival rate of 26.5% and five-year survival rate less than 5%(Stupp et al., 2005). Patients whose tumors present epigenetic silencing of the DNA repair enzyme O-methyl-guaninemethyltransferase (MGMT) through promoter methylation show better outcome with TMZ treatment with overall median survival length of 21 months from 14 months of patient with unmethylated tumors(Hegi et al., 2005). Unfortunately, tumors progress in all patients regardless of the tumor characteristics.  Recurrence may be treated with repeat surgery and concomitant treatment with antiangiogenic drugs such as bevacizumab(Friedman et al., 2009). Other chemotherapeutic drugs such as carmustine and lomustine provide marginal benefits. The dismal prognosis of the current therapies is attributed to the highly infiltrative nature of the tumor and the highly heterogeneous nature of the tumor cells which have different molecular profiles. Thus GBM remains an incurable disease with few therapeutic advances and demands more effective, targeted therapy.

1.1.2 Causes of glioblastoma

Like in many cancers, the cause for GBM is not known. But there are many studies that point to genetic mutations associated with different cells. The genetic mutation may be inherited by the environment or both. Only 5% of primary brain tumors are associated with inherited genes alone. GBM may also occur in the course of genetic diseases such as neurofibromatosis type I(Broekman et al., 2009) and tuberous sclerosis(Padmalatha et al., 1980). Exposure to certain chemicals such as petrochemicals, pesticides and formaldehyde poses a higher risk of developing brain tumor. Ionizing radiation and electromagnetic fields also increases the risk of developing brain tumor(Spinelli et al., 2010). There may be many environment and genetic factors that can cause GBM but in most cases, the cause is not known.

1.1.3 Genetics and molecular biology of glioblastoma

In the past two decades with the advancement in technologies to evaluate genetic and epigenetic changes, there has been tremendous influx of data describing the genomic alterations in many cancers including GBM. GBM being one of the most molecularly complex tumors, it was the first solid tumor type to undergo comprehensive genomic, epigenetic, transcriptional analysis by The Cancer Genome Atlas (TCGA), a US government funded project involving multiple institutions. In this effort, TCGA initially analyzed 206 and then 543 primary GBM samples to define and validate the core biological pathways deregulated in GBM and classify GBM into four molecular subgroups (Brennan et al., 2013; Cancer Genome Atlas Research, 2008; Verhaak et al., 2010). The main common alteration in gene coding sequence appears to target three main signaling pathways in GBM. RTK/RAS/PIU3K signaling pathway

Receptor tyrosine kinases (RTK) are primary mediators of signal transduction and are often deregulated in many cancers including GBM. The main alteration of this pathway seen in GBM is the mutation or amplification of epidermal growth factor receptor (EGFR) present in 45% of all primary GBM, and amplification of platelet derived growth factor A (PDGFRA) present in 13% of GBM and MET present in 4% of GBM(Cancer Genome Atlas Research, 2008). Both EGF and PDGF play an important role in normal and tumor gliogenesis. Activation of EGFR and PDGFR signaling can enhance GBM growth through stimulation of the RAS/RAF/MEK/ERK pathways. An increased activity of RAS is observed in all GBMs however mutation in this gene is very rare. The up regulation of RAS could be from the activation of the upstream regulator such as EGFR and PDGFR and also from the loss of function of neurofibromatosis type I (NF1), present in 18% of GBM, which is a negative regulator for RAS pathway(Nissan et al., 2014). PI3K/AKT/PTEN/mTOR pathway is also activated in GBM through activation in RTK signaling (EGFR) and lesion in PIK3R1/PIK3CA and mutation or loss of PTEN which is present in 36% of GBM, a negative regulator of PI3K. p53 pathway

p53 is a tumor suppressor and a transcription factor that coordinates cells response to stress by regulating genes involved in apoptosis, DNA repair and metabolism(May and May, 1999). This pathway is highly disrupted in GBM through mutation/deletion in TP53 (27.9%) and deletion of CDKN2A (ARF) (55%), which encodes for two distinct proteins (p16INK4a and p14ARF) that act as negative regulators of the cell cycle. p53 pathway is also affected indirectly through amplification of murine double minute 2(MDM2;11%) and MDM4(4%) which is an E-3 ubiquitin ligase complex that can repress p53 function through its exertion of degradative control.  TP53 alterations are mutually exclusive with amplification of MDM family genes and CDKN2A. RB pathway

This pathway plays a key role in the regulation of cell cycle and proliferation. The RB gene encodes for the retinoblastoma (RB) phosphoprotein. In quiescent cells, RB is in a hypophosphorylated (active) state and bound to E2F, preventing transcription of genes required for progression through the G1/S cell cycle phase. In proliferating cells, RB is phosphorylated (inactive) by CDK-cyclin complexes enabling E2F release and subsequent promotion of G1/S transition. The RB pathway is altered in 78% of primary GBMs either directly by mutation, deletion, or promoter methylation at the RB locus, or indirectly through alteration in RB positive and negative regulators. The most frequent event for this pathway was deletion in the cyclindependent kinase inhibitor 2A(CDKN2A)/CDKN2B locus on chromosome 9p21 (55% and 53% respectively) followed by amplification of the cyclin dependent kinase 4(CDK4) locus (14%). P16INK4A inhibits the association of CDK4/6 with cyclin D, which promotes G1/S transition. This complex phosphorylates RB, facilitating release of bound E2F, a G1/S transcription factor. A loss of p16INK4a allows CDK4/6 and cyclin D association and subsequently promotes the G1/S transition. IDH1 mutation

In parallel to TCGA studies, Parson et al reinforced identification in the above genes and pathways in GBM.  They also discovered mutations in a metabolism-related gene called IDH1IDH1 gene encodes for isocitrate dehydrogenase I, an enzyme that catalyzes the oxidative decarboxylation of isocitrate to α-ketoglutarate and reduces NAD and NADP to NADH and NADPH respectively. IDH1 gene is mutated in 12% of GBMs. Mutation in IDH1 almost always affects the R132 codon and are more frequently detected among low grade glioma and secondary GBMs (70-75%) and very rare in primary GBM (5%), and IDH1 mutant GBMs are associated with increased overall survival (Parsons et al., 2008). Mutant IDH1 is reported to alter metabolism by favoring reduction of α-ketoglutarate to 2-hydroxyglutarate (2HG)(Dang et al., 2010), which in turn inhibits DNA and histone demethylase, hypermethylating large number of loci, ultimately altering gene expression leading to tumor growth.

1.1.4 Molecular classification of glioblastoma

GBMs are historically classified by clinical presentation as either primary or secondary based on the evidence of preexisting lower grade glioma. Recent genomic analysis also validated these two clinical groups as distinct molecular groups and has also identified extensive patient to patient inter-tumoral heterogeneity, further redefining histopathological classification of the disease. There have been several gene expression studies classifying GBM into several molecular subtypes including the initial description from Phillips and colleagues (Phillips et al., 2006), but the most recent and consensus classification comes from the integrated genomic and copy number analysis on a large cohort of 200 adult GBM from TCGA identified four different molecular subtypes: proneural, neural, classical and mesenchymal correlating to abnormalities in PDGFRA, IDH1, EGFR and NF1 respectively (Verhaak et al., 2010). Proneural subtypes are mostly characterized by abnormalities in PDGFRA and IDH1, associated with younger age and secondary GBMs. This subgroup is also enriched for genes involved in neural development such as DCX, DLL3, ASCL1, TCF4, SOX genes and oligodendrocyte development such as PDGFRANKX2-2 and OLIG2. The neural subtype lacked a distinctive genetic profile and display gene expression similar to those found in normal brain tissue, with expression of neuronal markers such as NEFL, GABRA1, SYT1 and SLC12A5, many have dropped this group from the subtypes. The classical subtype was defined by the presence of most common genomic aberration seen in GBM, with 93% of samples harboring chromosome 7 amplification and chromosome 10 deletion accompanied with EGFR amplification and homozygous deletion of CDKN2A-p16 Ink4a/ARF locus. This subgroup also lacks additional abnormalities in TP53, NF1, PDGFRA or IDH1 and expressed neural precursor and stem cell marker gene from Notch and Sonic hedgehog signaling pathways. The mesenchymal subgroup is characterized by abnormalities in NF1, mutation/deletion or low levels of NF1 mRNA expression, with high expression of mesenchymal (CHI3L1 and MET) and astrocytic (CD44 and MERTK) markers. Furthermore, this classification system is further refined based on characterization of DNA methylation patterns, as proneural GBM into glioma-CpG island methylator phenotype (GCIMP) positive and G-CIMP negative GBM subsets, which corresponds strongly with IDH1 mutations(Brennan et al., 2013; Noushmehr et al., 2010). Only the G-CIMP subset of proneural subgroup was reported to have better prognosis, with other subgroups being highly similar.

1.2 Developmental neurobiology, neurogenesis, neural stem cells

Understanding early development of the mammalian brain and neurogenesis, the process by which neurons are generated from the neural stem cell or progenitor cells, has great importance for regenerative medicine, and also provides insight into understanding the origins and molecular mechanisms governing GBM. In early development of the vertebrate embryo, the neural fate is induced in ectoderm by the underlying notochord to give rise to the neural plate. The neural plate subsequently folds to become the neural tube and undergoes patterning for future distinctive central nervous system (CNS) region. Thus neural tube is specified to generate the prosencephalon, mesencephalon and rhombencephalon, which give rise to the future forebrain, midbrain and hindbrain respectively. The prosencephalon is comprised of two parts, the telencephalon, which gives rise to the cerebrum and the diencephalon, which generates the thalamus, hypothalamus and the posterior portion of the pituitary gland.

1.2.1 Neurogenesis in the developing cortex

The most studied neurogenic niche in the developing CNS is the cerebral cortex. The mammalian cerebral cortex has six layers and each layer contains neurons that share a characteristic morphology, connection and gene expression pattern(Hevner et al., 2003). The neurons of the cerebral cortex are broadly divided into two categories; projection neurons that transmit signals to other cortical or subcortical regions with the excitatory neurotransmitter glutamate, and interneurons that regulate local circuitry with the inhibitory neurotransmitter gamma ammunobutyric acid (GABA). Although these two neuron populations intermix within the mature cortex, they are generated from different regions of the telencephalon. Early in development, just like the rest of the CNS, the cerebral cortex starts from a simple neuroepithelial sheet at the anterior neural tube called telencephalon. The excitatory projection neurons arise from the neuroepithelium cells (NEC) of the dorsal telencephalon (pallium), while the inhibitory interneurons are generated in the ventral telencephalon (subpallium), which then migrate tangentially into the dorsal telencephalon consisting of immature cortex(Anderson et al., 1997; Gorski et al., 2002; Molyneaux et al., 2007; Rakic, 1978). Importantly, this precise coordination of neurogenesis and cell migration is controlled by both intrinsic and extrinsic factors. For example, the proneural bHLH transcription factors neurogenin2 (Ngn2) are expressed in the dorsal telencephalon and are required for proper specification of projection glutamatergic neurons, whereas Mash1/Ascl1, which is predominantly expressed in the ventral telencephalon, is required for proper specification of GABAergic interneurons(Britz et al., 2006; Guillemot and Joyner, 1993; Hand et al., 2005; Parras et al., 2002). The cytoarchitecture of the neocortex can be defined by its glutamatergic cell components. In the developing mouse cortex, the dividing NECs comprise the entire neuroepithelium at embryonic days 8-9. NECs are bipolar, with an apical and a basal process contacting the ventricular and pial surfaces respectively. They also express the SRY-related HMG-box transcription factor Sox1(Zhao et al., 2004). NECs divide symmetrically to self renew and to generate an adequate pool of founder progenitors. The initial proliferative phase affects both lateral and radial extension and has significant impact on the final surface area and thickness of the cortex(Florio and Huttner, 2014). At the onset of neurogenesis, embryonic day 10 to 12, these NECs undergo a transition to become radial glial cells (RGCs). The transition of NECs to RGCs are instructed through several extrinsic factors such as Notch1 and FGF signaling and intrinsic factors such as Sox1(Gaiano et al., 2000; Sahara and O’Leary, 2009; Suter et al., 2009). These RGCs maintain NECs features with adherens junctions and apical-basal polarity with long basal process that span the entire thickness of cortex. These RGCs express glial markers such as GLAST (Glutamate-aspartate transporter) and BLBP (Brain lipid-binding protein). RGCs are the principal progenitor cells of the cerebral cortex and it also serve as scaffolds for the orientated migration of later born neurons through their elongated processes(Noctor et al., 2001; Rakic, 1972). They can self renew by symmetric division but primarily undergo asymmetric neurogenic division, which produce a new RGC and either a neuron (direct neurogenesis), or an intermediary type of progenitor cell called intermediate progenitor cells (IPC) that then give rise to neurons (indirect neurogenesis)(Hansen et al., 2010; Kowalczyk et al., 2009; Noctor et al., 2004). Ultimately at the last stages of neuron production, RGCs undergo terminal symmetric division giving birth to two neurons. The prevalence of indirect neurogenic divisions increase markedly as neurogenesis progress. IPC are transiently amplifying progenitors characterized by the expression of the transcription factor Tbr2(Sessa et al., 2008). IPCs delaminate from the ventricular zone (VZ) to settle in the subventricular zone (SVZ), where they divide symmetrically to self renew before undergoing a terminal division that give rise to two neurons. The final neuronal output is sequentially impacted by the initial pool size of founder progenitors NECs, by progressive switch from symmetric autoreplicative to neurogenic division, and finally the duration of the neurogenic phases, which presents significant variation between species. In the mouse, SVZ progenitors comprising of IPCs undergo at most two rounds of division (Noctor et al., 2004), whereas in human and other primates, they undergo significantly more rounds to division(Fietz et al., 2010). This is the major difference observed between lissencephalic species such as rodents and gyrencephalic species such as humans. Additionally, the primate SVZ is subdivided into two layers of filamentous structure: the inner SVZ (ISVZ), which is juxtaposed to the VZ, and the more basal or outer SVZ (OSVZ) (Hansen et al., 2010). Secondly, the ISVZ contains only Tbr2 positive IPC as in the mouse SVZ, the OSVZ contain additional type of progenitor cells that are similar to ventricular RGCs and are called basal RGCs or oRGC. These cells are distinguished by being unipolar with only an apical or a basal process(Betizeau et al., 2013). oRGCs are originally generated from ventricular RGCs by asymmetric division most frequently characterized by horizontal or sometime an oblique division plane. They can in turn give rise to new oRGCs and IPCs as well. As neurogenesis begins, RGCs undergo successive rounds of asymmetric, self renewing divisions, giving rise to all the diverse subtypes of cortical excitatory neurons in a defined temporal sequence. The first cohort of terminally differentiated neuron are pioneer layer I neuron consisting of Cajal-Retzius (CR) cells that forms a transient structure called preplate (PPL).  Cajal-Retzius cells express secretory glycoprotein called Reelin (Reln).  Layer I Cajal-Retzius neurons serve as essential scaffold for the construction of the neocortical cytoarchitecture regulating radial migration in subsequent-born projection neurons through diffusive cues. Approximately 75% of these scaffolding cells are eliminated during early postnatal period, a proportion of Cajal-Retzius cells survive in the postnatal neocortex(Soda et al., 2003). Following the disposition of the preplate cells consisting of Cajal-Retzius cells, RGCs begins to produce projection layer neurons through sequential round of cell cycles. New neurons are successively generated and migrate radially using RGCs fibres from their place of origin at the ventricular surface past the pre-existing neurons to occupy the more superficial layers thus resulting in an inside-out lamination of the neocortex. Thus the neuronal birthdate is highly correlated with final laminar fate, in which neurons that occupy the same radial position are typically generated within the same temporal window and share common projections targets. In this way deep layer (DL) neurons are generated early, which include layers V and V1 that consist mainly of corticofugal projection neurons that project to subcortical targets. Specifically, layer V neurons project to the brainstem and spinal cord, and express Fezf2 and Ctip2(Chen et al., 2008; Molnar and Cheung, 2006) and subset of these neurons express Er81/Etv1(Yoneshima et al., 2006). In turn layer VI neurons establish corticothalamic projection neurons and express Tbr1, Zfpm2 and Sox5(Hevner et al., 2001; Kwan et al., 2008)The upper layer (UL) neurons are generated later, which includes layer II/III projection neurons and layer IV thalamorecipient neurons processing higher order information through intracortical connections. Layer II/III express the transcription factor Cux1/2, Brn1/2, Satb2 and project their neurons to the ipsilateral and contralateral cortex, thereby establishing bilateral cortical connections(Britanova et al., 2008; Nieto et al., 2004). Layer IV neurons are recipient cells for thalamocortical inputs and act as gateway for processing information from periphery sensory organs. In contrast to other layer neurons, Layer IV neurons have very limited specific markers that include Rorb, Unc5d and Kcnh5 (Eag2)(Schaeren-Wiemers et al., 1997; Zhong et al., 2004). After neurogenesis, RGCs switch to gliogenesis, producing astrocytes and oligodendrocytes (Noctor et al., 2001). Some of these RGC precursors also stay behind and eventually contribute to the pool of adult neural stem cells (Bonfanti and Peretto, 2007).

1.2.2 Neurogenesis in the adult brain

Neurogenesis persists in adult mammals in specific brain regions called neurogenic niches. In adult mouse brain, the main neurogenic regions are the subependymal zone of the lateral ventricles, also known as ventricular-subventricular zone (V-SVZ) and the subgranular zone (SGZ) of the dendate gyrus in the hippocampus(Altman and Das, 1965; Doetsch et al., 1999; Ming and Song, 2011). In the human adult brain, these neurogenic regions have been also shown to be active, where V-SVZ regions are thought to contribute  new neurons to the striatum and the SGZ region contributes new neurons to the dendate gyrus (Eriksson et al., 1998; Ernst et al., 2014). This addition of new neurons to the complex circuitry of the adult brain reveals its crucial function in memory, behavior and regeneration, which is the focus of intensive research. The neural stem cells (NSC) in V-SVZ are identified as astroglial cells (B1 cells), similar to RGCs in embryonic development(Doetsch et al., 1999; Mirzadeh et al., 2008). In both embryonic and adult brain, NSCs are specialized form of glia that resides in neurogenic niches(Kriegstein and Alvarez-Buylla, 2009). In contrast, the embryonic NSCs are inherently transient and continually changing their developmental potential and location over time, where as the adult VSVZ NSCs are more stable and present in a well defined niche that includes ependymal cells, mature vasculature, axonal terminals. This unique niche provides multiple regulatory controls for the production of neurons within fully assembled adult brain. Similar to RGCs, B1 cells retain epithelial feature that has apical processes that contact the ventricle and end feet on blood vessels(Kriegstein and Alvarez-Buylla, 2009). This elongated structure allows B1 cells to bridge all compartments of the V-SVZ.  Consistent with their astrocytic morphology and ultrastructure, B1 cells express glial markers such as glial fibrillary acidic protein (GFAP), glutamate aspartate transporter (GLAST) and brain lipid binding protein (BLBP). Recently, it has been shown that B1 cells can exist in either quiescent or activated state(Codega et al., 2014; Mich et al., 2014), where quiescent B1 cells do not express Nestin, an intermediate filament protein that has long been considered as marker of NSC while the activated B1 cells express Nestin.  These activated B1 cells give rise to transit amplifying precursors (C cells), which then divide symmetrically approximately three times before becoming migratory neuroblasts (A cells)(Doetsch and Alvarez-Buylla, 1996; Ponti et al., 2013). These type A cells then divide once or more while on route to the olfactory bulb and migrate within a network of interconnecting path, forming the rostral migratory stream (RMS), where they differentiate into different subtypes of interneurons. Adult hippocampal neurogenesis is not extensively characterized as in the V-SVZ neurogenesis. The generation of DG is unique from developmental point of view. While V-SVZ is seen as a continuation of the embryonic ventricular zone (VZ) of the telencephalon, the formation of DG involves generation of a dedicated progenitor cells source away from the VZ and in close proximity to the pial surface. This additional proliferative zone remains active during postnatal stages and eventually becomes the SGZ, which is site of adult hippocampal neurogenesis (Altman and Bayer, 1990; Pleasure et al., 2000). In SGZ of dendate gyrus, the adult NSCs are also radial glial like cells called type-I cells that populate the border between the hilus and the inner granule cell layer, which express markers of NSC and divide rarely(Seri et al., 2001).  When activated type-1 cells give rise to intermediate progenitor cells, which exhibit limited number of cell division before generating neuroblast(Berg et al., 2015). These neuroblasts migrate tangentially along SGZ and differentiate exclusively into granule neurons, which are then integrated into the hippocampal circuits(Sun et al., 2015). Like RGC in the embryo, NSCs in both V-SVZ and SGZ express GFAP, Nestin and Sox2 and they directly contact the blood vessels. However, both adult NSC populations have restricted potential each generating a unique neuronal subtype and one type of glia as shown recently by genetic fate mapping and clonal lineage tracing method. In adult hippocampus in vivo, NSCs in the SGZ give rise to neurons and astrocytes but not oligodendroctyes(Bonaguidi et al., 2011), while NSC in V-SVZ has shown to generate neurons and oligodendrocytes (Calzolari et al., 2015; Ortega et al., 2013). Therefore, in adult brain whether endogenous NSC with an intrinsic tri-potent potential exist or not, or whether they are intrinsically tripotent but suppressed by niche remains a fundamental unanswered question.

1.3 Cancer stem cells, brain tumor stem cells, clinical relevance

Cancer is caused by a series of aberrant genetic alteration and epigenetic modifications that over time results in the loss of cell cycle control and DNA damage checkpoints leading to an increased cell proliferation potential. In the process of this transformation from a normal to a cancerous cell state, the cell acquires specific characteristics including indefinite replicative ability, sustained proliferative signals, evasion of cell death and the immune system, development of neoangiogenesis, and ultimately an ability to invade and metastasize(Hanahan and Weinberg, 2011). Cells within the individual tumor undergo natural selection and diverge in a process of cancer evolution. It has been shown that cancer tissues are heterogeneous similar to organs, with multiple cell types that interact with each other and with extracellular matrix. They are heterogeneous not just in cellular morphology and tumor histology, but heterogeneous in cell surface markers, genetic abnormalities, growth rates and response to therapy. Much evidence points to the existence of multiple tumor cell subpopulations within single cancers. The cancer stem cell model of cancer development proposes that only a subpopulation of cancer cells have the potential to generate new tumors containing heterogeneous populations of cancer cells, where as the stochastic model proposes that all cancer cells have ability to give rise to new tumors.

1.3.1 Cancer stem cell model

The idea that cancers retain features of embryological development has been proposed more than 150 years ago, and the modern idea of cancers having caricatures to normal tissue organization has been shown with the seminal studies of teratocarcinoma by Pierce (Pierce et al., 1960), and in mammary carcinoma by Rudland (Bennett et al., 1978). All these studies suggested that tumor cells that were more differentiated were generated by “tumor stem cells” similar to normal tissue stem cells. A study by Pierce and Wallace in 1971 indicated the presence of a cellular hierarchy in the tumor, where they found that undifferentiated malignant cells give rise to benign well differentiated cells(Pierce and Wallace, 1971). Based on these studies, it was believed that cancer stem cells (CSC), with deregulated self-renewal and differentiation were responsible for tumor initiation and progression. Early studies in leukemia, based on normal studies in the hematopoietic system, showed clear evidence that majority of leukemia blast were postmitotic and it needed to be replenished from a small population of highly proliferative cells(Clarkson et al., 1967). Later with the development of fluorescence-activated cell sorting techniques (FACS), combined with refinement in xenograft techniques in immune deficient mice, and quantitative methods to measure tumor propagating potential, set the stage for the first purification of tumor initiating cells or cancer stem cells (CSC). In 1994 Lapidot and Dick used CD34 and CD38 to prospectively isolate CSC in acute myeloid leukemia (AML). They demonstrated that only CD34 CD38 subpopulation of AML cells gave rise to leukemias when injected in severe combined immunodeficiency (SCID) mice(Lapidot et al., 1994). This was the first prospective isolation of a CSC. They later confirmed that CD34 CD38 grafts recapitulate the heterogeneity of the patient samples from which they were derived, further demonstrating that AML is organized as a hierarchy with CD34 CD38 CSC at the apex(Bonnet and Dick, 1997). This initial studies in AML set the foundation for the identification and prospective isolation of CSC in many cancers including breast(Al-Hajj et al., 2003), brain(Singh et al., 2004), pancreatic(Hermann et al., 2007), prostate(Collins et al., 2005), head and neck(Prince et al., 2007) and colorectal(O’Brien et al., 2007).  All these studies demonstrated that tumors are hierarchically organized- that not all cells have the potential to initiate and propagate tumor growth, similar to developmental hierarchies seen in normal tissue where stem cells reside at the apex and are responsible for generating progeny that contribute to cancer growth. CSCs are defined functionally as a potently tumorigenic, self-renewing population in vivo with the ability to differentiate and generate mature cell types reflective of the original tumor phenotype. Of note, the definition and characterization of a CSC has no implications regarding its tumor cell-of-origin. For many cancers, CSC represent distinct populations that can be prospectively isolated from the rest of the tumor cells and can be shown to have clonal long term repopulation and self renewal capacity(Clarke et al., 2006). There is also some evidence suggesting that certain cancer cells exhibit plasticity by reversibly transitioning between a stem and non-stem cell state although this remains debated. Regardless, the stemness state is the main contributor to tumor growth and survival after therapy. It has been hypothesized that conventional cancer therapy reduces the tumor bulk but fail to prevent tumor recurrence and complete remissions due to incomplete eradication of CSC population. Therefore, CSC represents a potential target for better therapeutic intervention (Figure 1.1). However, not all cancers may follow a stem cell hierarchy(Quintana et al., 2008).  Figure 1.1. Cancer stem cell model and treatment approaches

1.3.2 Brain tumor stem cells

The first prospective isolation of human neural stem cells (NSC) using the CD133 cell surface marker (Uchida et al., 2000), prompted a search for brain tumor cells that shared similar characteristics to normal NSC.  Using a similar premise as in leukemia CSCs, which shared many properties with their normal hematopoietic stem cells, the assay condition used for identification and characterization of normal NSC were used to address if self-renewing multipotent cells could be isolated from primary brain tumors. Like normal NSCs, subpopulations of cells isolated from primary brain tumors also form neurosphere (clonal) colonies that are passagable when cultured at low densities in serum free medium containing epidermal growth factor (EGF) and basal fibroblast growth factor (bFGF)(Ignatova et al., 2002). Notably, these clonally derived neurospheres demonstrated core properties of self-renewal and ability to differentiate into neurons, astrocyte and oligodendrocytes and can regenerate tumor when injected subcutaneously or intracranially into immunodeficient mice (Galli et al., 2004; Hemmati et al., 2003; Ignatova et al., 2002; Singh et al., 2004). Dirks and colleagues, in 2003, demonstrated that these stem cell characteristic were prospectively found to exist exclusively in the subpopulation of tumor cell expressing the CD133 stem cell marker(Singh et al., 2003). These primitive cells expressed stem cell markers like nestin, and has multipotent differentiation potential similar to normal NSC.  When primary patient tumor cells were injected in vivo, as few as 100 CD133 cells had formed tumors while as many as 1×105 CD133 tumor cells could not form tumors. Importantly, CD133 fraction recapitulates the histopathological features and cellular heterogeneity of the patient original tumors. These results led to identification of CSC in gliomas being initially defined by expression of cell surface marker CD133. However, subsequent studies showed that subpopulation of CD133 cells are also able to form tumors, and not all tumors express CD133, demonstrating the patient to patient heterogeneity of gliomas (Beier et al., 2007; Wang et al., 2008). There are other reports indicating additional markers enriched for CSC subpopulation including CD15(Son et al., 2009), integrin α6(Lathia et al., 2010), CD36, A2B5(Ogden et al., 2008), L1CAM(Bao et al., 2008) and CXCR4(Ehtesham et al., 2009).  Although all these markers further enhanced our understanding of CSC function and regulation, no single marker can definitively identify or define CSC. Due to this lack of definitive markers, functional validation is essential to ensure that the enriched cells truly exhibit the functional characteristic of stem cells. Various method both in vitro and in vivo are employed to assess the stem cell characteristic of enriched cells such as self renewal and ability to reproduce the complexity and heterogeneity of the original tumor. The gold standard for CSC determination is the ability of a limiting dilution of cells to recapitulate the complexity of the original patient tumor when transplanted orthotopically.  Finally, brain tumor stem cells can be enriched in serum free medium growth condition similar to normal NSC, as adherent monolayer culture on a poly-L-ornithine (PLO) /laminin matrix, and are potently tumourigenic with as few as 100 cells capable of initiating tumors that recapitulate heterogeneity of patient tumor(Pollard et al., 2009). We called these cells as GBM derived neural stem cells (GNS) in this thesis.

1.3.3 Clinical relevance of cancer stem cell model

In cancers that follow a CSC model, there are considerable clinical implications of functional differences between tumorigenic and non-tumorigenic populations. There exists a close relationship between CSCs, tumorigenesis and drug resistance in glioblastoma (Bao et al., 2006; Chen et al., 2012), breast cancer(Diehn et al., 2009; Li et al., 2008) and many other cancers, therefore targeting these CSC is hypothesized to enable more effective cancer therapy. The potential importance of a CSC model is realized in cancers where tumors expressing higher levels of a stem cell signature is highly predictive of patient outcome as shown in breast(Liu et al., 2007), colon(Merlos-Suarez et al., 2011), glioblastoma (Murat et al., 2008) and leukemia(Eppert et al., 2011), medulloblastoma(Vanner et al., 2014).Similarly, brain cancer patients whose tumor cells self-renew and form tumourspheres in vitro have worse outcomes(Laks et al., 2009; Pallini et al., 2008). Cells with long term propagating potential will be positively selected and may therefore increase in frequency with cancer progression and greater stemness features may reflect more advanced and aggressive disease(Kreso and Dick, 2014). Another emerging feature of CSC is the resistance to conventional therapies in multiple cancers including GBM, where CSC fractions are more resistant to therapy compared to non-CSC(Chen et al., 2012; Ishikawa et al., 2007; Kreso et al., 2013). As a result, stem cells are enriched after therapy and are likely the cause for relapse. These emerging evidence of stemness to prognosis and therapy failure implicate that therapeutic targeting of determinants of stemness might be an effective means to eradicate CSC and prevent relapse. However, targeting only self renewing CSC may not to be sufficient if non-stem cells have considerable proliferative potential or can revert to stem cell state. For example, in mouse glioma, genetic ablation of quiescent nestin temozolomide resistant cells prolong survival however when combined with ablation of cycling cells showed greater benefit (Chen et al., 2012). In this thesis, we attempt to identify compound that selectively effect human GBM derived neural stem cells (GNS) compared to normal human NSCs or fibroblasts.

1.4 Neurotransmitters/Neurochemicals

Neurotransmitters or neurochemicals are endogenous chemical messengers that are synthesized by a neurons and released into a synapse upon stimulation. They can transmit signals to a target cells; neurons, muscle cells or another effector cells, through binding to their respective receptors. Traditionally, neurotransmitters can be divided into three types based on the chemical structures as amino acid, monoamine and peptides. Amino acid neurotransmitters include glutamate, gamma aminobutryric acid (GABA) and glycine. Monoamine neurotransmitters include dopamine, norepinephrine, epinephrine, serotonin, histamine and melatonin. Peptides include substance P, neurotpeptide Y, opioids etc. Further, there are purines that includes adenosine triphosphate (ATP), adenosine, ATP and others including acetylcholine. In this thesis we refer to all neurotransmitters as neurochemicals, as the term neurochemical appears more inclusive to all fast synaptic neurotransmitters as well as slow acting neuromodulators, which generally mediates neurotransmission through second messenger via metabotropic receptors (eg. dopamine) and has long lasting effects and diffuse into large areas of the brain.

1.4.1 Neurochemicals in normal neurogenesis

The thought that neurochemicals are primarily associated with synaptogenesis in mature neurons has been challenged and a crosstalk between neurochemicals and neurogenesis is beginning to emerge, suggesting neurochemicals are involved in the formation of neurons, not just in the function of neurons. There are several reports identifying new roles of neurochemicals in cell fate determination in a wide range of species both within and outside the CNS. Neurochemicals and their receptors are present and functionally important in organisms without a nervous system. For example, GABA and glutamate have been shown to regulate cell behavior in sponges (Ellwanger et al., 2007), and the antagonistic relationship between GABA and glutamate signaling in spore induction in Dictyostelium was established prior to the evolution of synaptic communication in CNS(Fountain, 2010). Furthermore in mammals, GABA signaling in early embryos was seen long before the onset of neurogenesis(Andang et al., 2008). After the onset of developmental neurogenesis, neurochemicals form the chemical environment of neural cells impacts neurogenesis including proliferation, migration and differentiation in the developing telencephalon, ventral midbrain and retina(Heng et al., 2007; Kim et al., 2006; Martins and Pearson, 2008). Based on all these observations we can hypothesize that ancient function of neurochemicals, prior to the emergence of their role in neurotransmission, is critical to brain development in regulation of neurogenesis and later plasticity. Neurochemicals in developmental neurogenesis

Cortical development involves precise coordination of neurogenesis and cell migration, where dorsally derived projection neurons migrate radially and ventrally derived interneurons migrate tangentially to populate the cortical plate as described earlier. Importantly, this tightly orchestrated process of neurogenesis and cell migration is crucial to brain development and later functions, where defects in cortical development may lead to mental retardation and changes in neurotransmitters levels have been observed in inherited complex neurological disorders such as autism(Lam et al., 2006), schizophrenia(Rehn and Rees, 2005) and epilepsy(Ben-Ari, 2006).  This tightly coupled and coordinated neurogenesis and cell migration is influenced by both intrinsic and extrinsic factors. Neurochemicals comprise the chemical environment of developing cortex and expression of functional neurochemical receptors have been reported in migrating neurons regulating the migration of both projection neurons and interneurons(Benitez-Diaz et al., 2003; Heng et al., 2007; Lujan et al., 2005; Nguyen et al., 2001). Specifically, amino acid neurotransmitters comprise the chemical environment of cortical progenitors and implicated in the regulation of their proliferation and cell cycle exit(Haydar et al., 2000). Several studies showed role of glutamate signaling in regulating cell migration by ventricular zone (VZ) cells. Dissociated cells from cortical VZ cells showed expression of functional N-methyl-D-aspartate (NMDA) receptors, and pharmacological manipulation of NMDA receptors has demonstrated its control over radial migration of VZ cells(Behar et al., 1999). Furthermore, the NMDA receptor role in controlling radial migration from VZ cells was demonstrated using embryonic slices. Similarly, GABA signaling also demonstrated a role in regulating VZ cell migration. GABA is present in gradient fashion from low to high from VZ to the cortical plate (CP) and its receptor GABAA receptor has been detected in VZ as well as CP cells(Behar et al., 2000). Pharmacological manipulation using GABA A-C antagonist on embryonic brain slices impedes radial migration of VZ cells into the SVZ and showed GABAB receptor is required for regulating migration from SVZ to CP and GABAA receptor is required for arrest of cell migration at the CP(Behar et al., 2000). These studies showed cells of cortical VZ differentially integrate GABA and glutamate signaling through specific receptor subtype for the control of cell proliferation and radial migration (Figure 1.2A). Additionally, glycine receptor signaling may also influence migrating neurons in developing cortex(Flint et al., 1998). Neurochemicals also regulate the tangential migration of interneurons in the developing telencephalon. There is increasing evidence of role of GABA signaling in the regulation of tangential migration by embryonic interneurons (Cuzon et al., 2006; Lopez-Bendito et al., 2003). GABA is present in the main migratory routes of the developing cortex and the embryonic interneurons express GABA receptors. Studies using tissue transplantation and brain slice cocultures demonstrated the importance and requirement of GABAA receptor signaling in migrating interneurons to cross the corticostriatal junction to populate the cortex(Cuzon et al., 2006). Furthermore, GABA signaling through its GABAB receptor also influences the behavior of tangentially migrating interneurons as well as their choice of migratory route into the cortex. Besides GABA, other neurotransmitters such as dopamine have demonstrated to modulate tangential migration. Dopamine in the basal forebrain through its D1 receptors promote migration of GABAergic interneuron into the cortex, while D2 receptors have the opposite effect, impairing their migration from medial ganglion eminence (MGE) and caudal ganglion eminence (CGE)(Crandall et al., 2007; Ohtani et al., 2003). Dopamine, also through its D1and D2 receptors, have opposing effects on proliferation of neuronal progenitors from the lateral ganglion eminence. Altogether, these data suggest role of dopamine through D1 and D2 receptor in regulation of crucial cellular steps required for development of telencephalon.  Figure 1.2. Neurochemicals in developmental neurogenesis. A. A schematic diagram showing GABA and glutamate’s role in radial migration of newborn excitatory neurons from radial glial cells (RGC) at ventricular zone to cortical plate (CP). B. Dopamine and GABA regulate tangential migration of interneurons generated in medial ganglion eminence (MGE) to cortical plate. Neurochemicals in adult neurogenesis

Adult neurogenesis, confined to the SVZ of lateral ventricle and SGZ of dendate gyrus, is the life-long continuous production and functional integration of new neurons into existing neuronal network of CNS. In these locations, the brain can modify responses to external stimuli, as well as to learn and remember. Neurogenesis is under precise spatial and temporal control that can be modulated by both internal and external stimuli. Neurochemicals such as dopamine, serotonin and acetylcholine that are secreted by small group of neurons can affect neuronal activity through large brain areas. They have been shown to have long-range effects through neuronal projections into the SVZ. Neurochemicals such as GABA and glutamate, primarily confined to the synapse and responsible for fast synaptic transmission, have demonstrated effects on SVZ proliferation and neurogenesis (Berg et al., 2013). However, extra niche source of these neurochemicals remain to be explored. Furthermore, there is also accumulating evidence that diseases and pathologic and physiologic states such as Alzheimer’s disease, seizures, sleep and pregnancy influences SVZ cell proliferation. Dopaminergic neurons originating from the substantia nigra have been shown to extend its projections into SVZ in rodents and primates(Freundlieb et al., 2006; Hoglinger et al., 2004). Dopamine receptors are also expressed in the precursor cells in the SVZ cells predominantly type C cells and type A neuroblast (Diaz et al., 1997; Hoglinger et al., 2004). As dopaminergic neurons project its axon into SVZ and the precursor cells in SVZ express dopamine receptors, it is conceivable that dopamine released from these afferents controls neurogenesis. Dopamine can impact neurogenesis at several developmental stages and in regions including the adult SVZ. Ablation of dopaminergic neurons in rodents by injection of selective neurotoxin such as 6hydroxydopamine (6-OHDA) or 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine(MPTP), resulted in decreased SVZ proliferation and OB neurogenesis, while administration of the dopamine agonists or precursor levodopa restores SVZ proliferation to near normal(Baker et al., 2004; Hoglinger et al., 2004; Winner et al., 2009; Yang et al., 2008). However, there are conflicting reports on actual effects on dopamine on SVZ cell proliferation and neurogenesis (Berg et al., 2011; Kippin et al., 2005; Wakade et al., 2002). In vivo pharmacological manipulation using agonists or antagonists of selective dopamine receptors led to significant increases or decreases in SVZ neurogenesis in rodents as measured using injections of the S-phase marker bromodeoxyuridine (BrdU). In a study in which the dopamine receptor antagonist haloperidol was administered for 14 days an increase in proliferation and in number of label retaining cells (stem-like cells) in a dopamine D2 receptor dependent manner was shown(Kippin et al., 2005). Another study using D3 receptor ligand reported no effect. There are discrepancies noted between studies and it may be due to the age and species of the animals, selectivity of the agonists or antagonists, activation of distinct subtype of dopamine receptors, at different developmental stages, and dopamine signaling may have different effect on stem cells and transient amplifying cells type C cells. There are reports suggesting that the dopamine-induced activation of SVZ neurogenesis is in part mediated through EGF dependent (Lao et al., 2013; O’Keeffe et al., 2009) or ciliary neurotrophic factor (CTNF)(Emsley and Hagg, 2003; Yang et al., 2008), both of which are known to promote proliferation in the SVZ. Considering the role of substantia nigra in reward, addiction and movement, we can speculate that SVZ cell proliferation and neurogenesis may be regulated by addictive behaviors and new movement based learning paradigms. Furthermore, postmortem studies in human have identified dopaminergic fibers in contact with epidermal growth factor receptor (EGFR)-positive cells in the SVZ, which were presumably type C cells(Borta and Hoglinger, 2007; Hoglinger et al., 2004). Parkinson’s disease (PD), which is characterized by the loss of dopaminergic neurons in the substantia nigra, showed decreased neurogenesis in an animal model generated by ablating dopaminergic neurons in substantia nigra. In addition, examination of postmortem brains of PD patients have also shown decreased number of proliferative cells in the SVZ(Hoglinger et al., 2004). Serotonin (5-HT), another monoamine, is synthesized by neurons in the raphe nuclei, which regulates many aspects of behaviors including mood, sleep, appetite, reproductive activity and cognition. There are seven families of 5-HT receptors (5-HT1-7) and all except 5-HT3 are G protein coupled receptors. 5-HTis a ligand gated cation channel. There are several reports indicating serotonin and its receptors regulating cell proliferation and neurogenesis in SVZ(Banasr et al., 2004; Brezun and Daszuta, 1999). A recent study has demonstrated that serotonergic axons originating from raphe nuclei form a dense plexus covering most of the wall of lateral ventricles contacting both ependymal cells and type B1 cells(Tong et al., 2014). B1 cells also express 5HT 2C and 5A receptors, and subraependymal release of 5HT increases SVZ cell proliferation, which is mediated primarily by 5HT2C receptor. Ablation of raphe nuclei results in decreased SVZ proliferation indicating 5HT axons directly interact with NSC to regulate neurogenesis(Tong et al., 2014). Also, pregnant mice have higher levels of serotonergic innervation in the SVZ that has been suggested to be responsible for increased SVZ proliferation. Interestingly, serotonergic projections originating from the raphe nucleus are also found in the DG. Depletion of these neurons leads to decreased proliferation in the DG and is rescued by grafting of fetal raphe neurons suggesting serotonin has stimulating effect on neurogenesis in DG(Brezun and Daszuta, 1999). GABA is the main inhibitory neurotransmitter released primarily by interneuron in the adult brain. It acts through activation of ionotropic ligand gated GABAor GABAC receptors and Gprotein coupled GABAreceptors. The SVZ is located along the striatum, which is predominantly composed of GABAergic neurons. In addition, nitric oxide containing GABAergic striatal neurons project into the SVZ. These neurons may provide an activity dependent GABAergic control of SVZ neurogenesis. Furthermore, vesicular GABA transporter expression (VGAT) has been reported in the SVZ in agreement with the presence of GABAergic inputs form the striatum. GABA has significant impact on several phases of SVZ neurogenesis such as proliferation of astrocyte like stem cells and type A neuroblasts, neuroblast migration. Migrating neuroblasts spontaneously release GABA in a non-vesicular fashion, which tonically activate GABAA receptors on progenitor cells(Liu et al., 2005; Nguyen et al., 2003). This GABA dependent depolarization of neural precursors inhibits cell proliferation and neuronal differentiation thus functioning in part as negative feedback signal derived form neuroblast, down-regulating their own production. This inhibition of stem cell proliferation by GABA A receptor is mediated through a mechanism involving phosphorylation of the histone variant H2AX(Fernando et al., 2011). The same mechanism was also observed in the GABA mediated control of the embryonic stem cell proliferation. Interestingly, type B1 and C cells secrete the diazepam-binding inhibitor protein (DBI), which competitively inhibits GABA from binding to GABA receptors, increasing the proliferation and neurogenesis of SVZ(Alfonso et al., 2012). In the adult dendate gyrus, tonic GABA release from parvalbumin-positive interneuron maintains quiescent state of adult NSC through γ2 subunit containing GABAA receptor(Song et al., 2012). Glutamate is an excitatory neurotransmitter, GABA’s counterpart, and is also known to affect SVZ neurogenesis. Glutamate signals through ionotropic AMPA, kainite and NMDA receptors as well as metabotropic glutamate receptors mGlur1-8 receptors. In the SVZ, glutamate receptors such as mGluRs and kainate have been observed on Type A neuroblast cells but not on Type B and C cells(Di Giorgi-Gerevini et al., 2005; Platel et al., 2010). The source of glutamate appears to be astrocytes in SVZ, which is found to express vesicular glutamate transporter 1 (Vglut1)(Platel et al., 2010). Glutamate in SGZ is better characterized. Glutamatergic input in DG comes from dendate granule cells, neurons in layer II of the entorhinal cortex and contralateral hilar mossy cells(Kumamoto et al., 2012; Witter, 2007). NMDA and AMPA receptors were not detected on RGL cells but in vivo administration of agonist of NMDA receptor reduces proliferation while an antagonist increases proliferation of progenitor cells in DG(Cameron et al., 1995; Kitayama et al., 2003). Ablation of entorhinal cortex leads to increased proliferation suggesting the glutamate source for the DG. Acetylcholine was the first neurotransmitter discovered and is produced by cholinergic neurons present in small number but its afferents are spread throughout the brain. Cholinergic signaling is important in the modulation of various brain states including learning, memory consolidation, attention and sleep. Its receptor includes ionotropic nicotinic and metabotropic muscarinic ACh receptors. Cholinergic input involving a population of choline acetyltransferase(ChAT) neurons are reported to be present in SVZ niche and these neurons are morphologically distinct from other striatal neurons(Paez-Gonzalez et al., 2014). Optogenetic manipulation of subependymal ChAT neurons have demonstrated that they are sufficient to increases neurogenic proliferation, partly through synergizing with fibroblast growth factor (FGF) receptor activation. Studies have shown that lesion in cholinergic inputs decreases the number of newly born neurons in the OB. Similarly, activation of cholinergic signaling with donepezil enhanced the survival of newly born OB neurons(Kaneko et al., 2006). In contrast, knockout of nicotinic beta-2 animals show increased survival of newly born neurons in the OB. Thus, acetylcholine has a complex effect on neuronal survival in the OB. In dendate gyrus, cholinergic input from the medial septum and fibers expressing choline acetyltransferase has been observed in close proximity to progenitors(Kaneko et al., 2006). Muscarinic acetylcholine receptors have been seen on RGLs (type B cells) and ablation of cholinergic neurons in the adult brain lead to reduced proliferation in the SGZ(Mohapel et al., 2005). Taken together, all these studies indicated that the SVZ and SGZ are recipients of brain neural circuit and neural activity can influence adult neurogenesis (Figure1.3). However, the understanding on activation of different neuronal pathways and the regulation of neuronal production remains to be elucidated. Here in this thesis, we attempt to understand neurochemical’s influence on directing human NSC fate by using a small molecule approach in vitro.  Figure 1.3. Neurochemicals regulate adult neurogenesis in SVZ. A schematic diagram showing the ventricular-subventricular zone organization of astrogial neural stem cells (B1), which give rise to activated B1 (B1a) that generates transit amplifying cells (C) and then migrating neuroblasts (A). A rich network of different neurons projects their axons into SVZ neurogenic niche such serotonergic axon (5HT) contacting B1 cells and ependymal cells(E), choline acetyltransferase(ChAT) neurons and dopaminergic terminals(DA) extending into SVZ region contacting type C and A cells.

1.4.2 Neurochemicals in cancer

All organs and body parts are innervated by peripheral nerves, which connects it to CNS and orchestrate tissue homeostasis by releasing neurochemicals. A tumor is not a completely independent entity; it depends strongly on its microenvironment. The communication between tumor cells and the microenvironment drives the process of tumor progression. Tumor growth and progression depends highly on the formation of blood vessels (angiogenesis) and lymphatic vessels (lymphanangiogenesis) in tumor microenvironment(Alitalo et al., 2005; Folkman et al., 1971). Likewise, the role of infiltrating nerves in tumor microenvironment is beginning to emerge. Aside from the role of nerve fibers attracting cancer cell to invade and migrate, it has been proposed that cancer cells stimulate their own innervation termed as neoneurogenesis (Palm and Entschladen, 2007). Tumor cells can exploit the neurochemicals released by nerve fibers to stimulate tumor growth and dissemination, and alternatively tumors cells can stimulate neurite growth by releasing neurotrophic growth factor and axon guidance molecules(Ayala et al., 2006). The neuronal influence on tumor growth was initially described several decades ago and the presence of nerve endings within the tumors has been shown in bladder(Seifert et al., 2002), prostrate(Magnon et al., 2013; Ventura et al., 2002), breast(Tsang and Chan, 1992) and pancreatic(Kayahara et al., 2007). But the landmark studies that clearly demonstrated the involvement of nervous system in tumor progression comes from the work of Magnon and colleagues, who demonstrated that autonomic nerve sprouting in prostrate tumors is essential for prostrate cancer progression(Magnon et al., 2013). Both sympathetic and parasympathetic nerves were required throughout the prostrate cancer development in mouse. The early phase of tumor development was preventable by sympathectomy or genetic deletion of β-adrenergic receptors and tumor were also infiltrated by parasympathetic cholinergic fibers that promotes cancer dissemination. Furthermore, in human prostrate cancer, the density of nerves was directly correlated to the Gleason score and tumor progression. Another study on gastric cancer in a mouse model also showed that denervation of the stomach by surgical or pharmacological neurotoxic agents dramatically reduced tumor incidence and progression, through inhibition of WNT signaling and subsequent suppression of stem cell expansion mediated through cholinergic signaling(Zhao et al., 2014). There are also numerous studies that showed role of neurochemicals and their receptors in stimulating cancer cell growth through activation of corresponding signaling pathways. Cancer incidence and progression seems to be strongly dependent on psychosocial factors. Stress can induce the release of neurochemicals that will further influence tumor development and progression(Antoni et al., 2006; Thaker et al., 2006). The most studied neurochemicals for their role in cancer growth and progression is catecholamine, stress mediators. The most common catecholamine, epinephrine (E) and norepinephrine (NE), secreted mainly from adrenal medulla and sympathetic nerves respectively, have been linked to stress induced tumor growth and progression. NE and E exert their functions through α and β adrenoreceptors. These neurochemicals can modulate cell proliferation, survival and migration through the activation of β-adrenergic receptors shown in ovarian(Thaker et al., 2006), breast(Badino et al., 1996), and colon(Masur et al., 2001) cancer models. Dopamine, another catecholamine neurochemical precursor in synthesis of NE and E, has an opposite effect to that of epinephrine and norepinephrine. Dopamine is reported to have inhibitory effect on cancer growth such as breast and colon cancer(Chakroborty et al., 2008; Chakroborty et al., 2004). However, dopamine by itself doesn’t effect the proliferation and survival of cancer cells rather its inhibition is mediated indirectly through inhibiting VEGF- induced angiogenesis and prolactin secretion, a growth factor for cancer cells (Basu et al., 2001; Clevenger et al., 1995). These effects of dopamine are mainly mediated by dopamine D2 receptor, which are expressed in tumor endothelial cells. However, the role of dopaminergic system in cancer growth is not clear and has no consensus. Acetylcholine is the neurochemical of parasympathetic nerve and its activity is mediated by nicotinic and muscarinic acetylcholine receptors. The role of cholinergic signaling in regulation to cancer was first demonstration of the potential role of autonomic nervous system in cancer. Cigarette smoking is involved in causing many cancers and nicotine is the one of the main component that is related to addiction. Nicotine induces its biological effects through binding to nAChRs. Nicotine and its byproduct, tobacco specific nitrosamine 4(methylnitrosamino)-1-(3pyridyl)-1-butanone(NNK) were shown to influence cell proliferation through nAChRs in lung(Schuller, 1989), colon(Wong et al., 2007) and breast cancers(Jimenez and Montiel, 2005). Similarly, amino acid neurotransmitters like glutamate and GABA have been shown to influence many cancers. The discovery of the neural impact in prostrate and gastric cancer sheds a new light on the role of autonomic neurotransmitters on tumor cell growth, as nerve fibers release these neurochemicals directly into the vicinity of cancer cells to stimulate their survival, proliferation and ability to spread. However, how generalize the neural involvement in cancer progression is not well understood. Brain tumors originate in a neurochemical rich milieu, but very little is known about its impact on brain tumor growth. In this thesis, we investigate neurochemicals’ role in brain tumors using a small molecule approach.

1.5 Dopamine and its receptors

Dopamine is a monoamine neurotransmitter, which control a variety of functions in human CNS including cognition, emotion, motor activity, feeding. Dopamine also plays an important physiological role in regulating retinal processes, olfaction, hormonal regulation, renal functions and cardiovascular functions (Carlsson, 2001; Missale et al., 1998; Vallone et al., 2000). Dysregulation of dopaminergic system is linked to several pathological conditions such as Parkinson’s disease, schizophrenia, attention-deficit hyperactivity disorder (ADHD), bipolar disorder and depression(Niznik and Van Tol, 1992; Seeman, 2006). In the human CNS, dopamine is predominantly synthesized in the neurons of four major dopaminergic pathways namely nigrostriatal, mesolimbic, mesocortical and tuberoinfundibular pathways(Anden et al., 1964; Dahlstroem and Fuxe, 1964). Dopamine is synthesized from the amino acid tyrosine in a two step enzymatic process, where the first limiting reaction is the conversion of tyrosine into L-3,4-dihydroxyphenylalanine(L-DOPA) catalyzed by tyrosine hydroxylase (TH). The second step involves production of dopamine by decarboxylation of DOPA by aromatic L-amino acid decarboxylase(AADC)(Vallone et al., 2000). Dopamine also serves as a potential intermediate substrate in the biosynthesis of catecholamine such as epinephrine and norepinephrine in other neurons and adrenal medulla. Unlike excitatory or inhibitory neurochemicals, dopamine is a neuromodulator that alters the response of a target neuron to other neurochemicals and that can alter synaptic plasticity. Dopamine does not mediate fast synaptic transmission but modulates it by triggering slow acting effects through signaling cascades.

1.5.1 Structural characteristics of dopamine receptors

Dopamine exerts its physiological function in the CNS and periphery by five distinct but closely related dopamine receptors, which belong to a family of G protein coupled receptors (GPCR) namely D1, D2, D3, D4 and D5 (Beaulieu and Gainetdinov, 2011). Biochemical studies on dopamine receptors based on the stimulation of cyclic adenosine monophosphate (cAMP) production and ligand binding assays led to the classification of dopamine receptors into two subgroups. In 1978, Spano and group demonstrated that dopamine receptors could exits in two distinct subgroups and that only one subgroup was positively coupled to adenylate cyclase(AC)(Spano et al., 1978), and this subsequently led to classification of D1 and D2 subtypes based on ability to modulate cAMP production and pharmacological properties(Kebabian and Calne, 1979). Based on their structural, pharmacological and biochemical properties, dopamine receptors are now classified as D1-like receptors (D1 and D5) and D2-like receptors (D2, D3 and D4)(Andersen et al., 1990; Niznik and Van Tol, 1992). The subfamilies of D1-like and D2-like receptors share high level of homology of their transmembrane domain and have distinct pharmacological properties. The D1-like receptors activate adenylate cyclase (AC) through Gαs/olf family of G protein to increase cAMP levels and are found exclusively in postsynaptically on dopamine-receptive cells. The D2-like class of receptors inhibits AC through Gαi/o family of G proteins and inhibits cAMP production and are expressed in both postsynaptically on dopamine target cells and presynaptically on dopaminergic neurons(Rondou et al., 2010; Sokoloff et al., 2006). The two subtypes are also different at the level of genetic structure, primarily in the presence of introns in their coding sequences. The genes encoding D1-like receptors lacks introns and share 80% homology in their transmembrane domain, while genes encoding D2-like receptors are interrupted with introns; six introns in D2, five in D3 and four in D4(Gingrich and Caron, 1993). D2 receptor shares a 75% homology with the D3 receptor and 53% homology with the D4 transmembrane domain(Missale et al., 1998). The NH2-terminal domain of all dopamine receptors has similar number of amino acid residues and carries a variable number of consensus N-glycosylation sites; D1 and D5 possess two sites, D2 has four and D3 has three and D4 has only one potential N-glycosylation sites. The COOHterminal tail for the D1-like receptors are about seven times longer than the D2-like receptors. The D1-like receptors has short third intracellular loop, typical for receptors interacting with Gstimulatory (Gs) proteins to stimulate cycle AMP production, whereas the D2-like receptors has a long third intracellular loop typical for receptors interacting with G-inhibitory (Gi) protein, to inhibit cyclic AMP production. The third intracellular loop is the responsible for the G protein coupling and signal transduction(Missale et al., 1998; Pivonello et al., 2007). The genetic organization of D2-like receptors with presence of introns provides the basis of receptor splice variants. For example, the D2 receptors exist in two main variants, the long isoform D2L and the short isoform D2S, generated by an alternative splicing of an 87bp exon between intron 4 and 5.(Giros et al., 1989; Monsma et al., 1989) The two D2 isoforms differ in the presence or absence of 29 amino acid residues in the third intracellular loop and has similar pharmacological but different functional characteristics. The D3 receptor was reported to have splicing variant of non-functional proteins. The D4 receptor shows the existence of polymorphic variations within the coding sequence, having 48bp repeats of two (D4.2), four(D4.4), seven(D4.7) or eleven(D4.11) fold repeat sequence(Van Tol et al., 1992). Therefore, the D4 receptor variants differ in their length of the third cytoplasmic loop and have two, four, seven or eleven repeat of the same insert of 16 amino acids residue in their protein structure. The D5 receptor has two related pseudogenes, which share a 95% homology with gene and encode for truncated nonfunctional forms of the receptor.

1.5.2 Functional characteristics of dopamine receptors

Dopamine receptors mediate the effect of dopamine and dopaminergic compounds through a number of different signaling mechanisms. In general, dopamine receptor function is known to signal through G-protein that contain nucleotide binding Gα subunit and a heterodimeric Gβγ subunits, which mediates signaling through modulating the levels of cAMP and protein kinase A (PKA) activity. The D1-like receptors are coupled to Gαs/olf proteins and stimulate the activity of AC and production of cAMP. In contrast, the D2-like receptors are coupled to Gαi/o proteins and inhibits the production of cAMP(Kebabian and Greengard, 1971). Growing evidence suggests dopamine receptors can signal through other pathways independent of AC modulation depending on the brain area, physiological and pathological conditions. cAMP mediated dopamine receptor signaling

cAMP mediated signaling is the canonical pathway activated by dopamine receptors. cAMP is an important second and ubiquitous messenger for many signaling pathways and can influence various effectors such as PKA and other exchange proteins activated by cAMP (Epac1 and EPac2). Among all the PKA substrates affected by dopamine receptors such as CREB, ionotropic glutamate receptors (AMPA and NMDA), ion channels, the most extensively studied molecule is the 32-kDa dopamine and cAMP regulated phosphoprotein (DARPP-32/PPP1R1B). DARPP-32 is a multifunctional phosphoprotein that acts as an intermediate component of multiple neurochemical signaling including dopamine(Svenningsson et al., 2004). Phosphorylation of DARPP32 at Thr34 by PKA induces the inhibition of the protein phosphatase 1(PP1)(Hemmings et al., 1984). In contrast, phosphorylation of DARPP-32 on Thr75 by cyclin dependent kinase 5(CDK5) induces the inhibition of PKA causing a feedback loop in the activation of the PKA(Bibb et al., 1999). The direct regulation of DARPP-32 phosphorylation at Thr34 by dopamine receptor has been shown in vivo studies where in D1 expressing neuron, enhanced dopamine receptor stimulation results in increased phosphorylation of DARPP-32, whereas stimulation of D2 receptor reduces the phosphorylation of DARPP-32 at Thr34, as a consequence of reduction in the PKA activation(Bateup et al., 2008). Furthermore, the activation of PKA or inhibition of PP1 mediated by PKA, may cause direct changes in the phosphorylation of two subunits of ionotropic glutamate receptors GluA1 (AMPA) and GluN2B (NMDA), which are involved in glutamatergic signaling(Greengard et al., 1999) (Figure 1.4). The recent interesting development of cAMP mediated dopamine signaling is its regulation of mRNA translation mechanism, where either D1 receptor activation or blocking D2 receptor by haloperidol increased the phosphorylation of the ribosomal protein S6 (rpS6) on Ser235/236 and Ser240/244(Santini et al., 2009; Valjent et al., 2011). The rpS6 is a component of the small 40S ribosomal subunit implicated in mRNA decoding. RpS6 is phosphorylated at multiple site comprised between Ser235 and Ser247 by p70rpS6 kinase (S6K), which is a major downstream effector of the mammalian target of rapamycin complex1(mTORC1). Phosphorylation of rpS6 at the dual site Ser 235/236 occurs also independently of mTORC1, via the p90 ribosomal S6 kinase(RSK), which are activated by the ERK pathway. In both studies, increased phosphorylation of rpS6 was mediated through activation of PKA and subsequent inhibition of PP1 by DARPP-32, and activation via D1 receptor was dependent on mTOR pathway while haloperidol study was independent of mTOR and ERK pathway. The mechanism of dopamine receptor mediated regulation of rpS6 on mRNA translation is still in an early stage. cAMP independent dopamine receptor signaling

Dopamine receptors can also exert their biological effect through alternative signaling pathways such as transactivation of receptor tyrosine kinases (RTKs), direct interaction regulating calcium channels, Na -K -ATPase, Gβγ mediated activation of PLC, and regulation of L- and N-type calcium channels, as well as G protein coupled inwardly rectifying potassium channels (GIRKS), Gαq coupled regulation of PLC and IP3 (Beaulieu and Gainetdinov, 2011). Interestingly, all G protein mediated signaling converge on, among many other targets, phosphorylation of two subunits of ionotropic glutamate receptors GluA1 and GluN2B, which are involved in glutamatergic transmission(Jenkins et al., 2014)(Figure 1.3). D2-like receptor can signal through alternative mechanisms involving the multifunctional adaptor protein β-arrestin 2 (βArr2). Arrestins are family of four molecular adaptor proteins that were originally characterized for their role in mediating GPCR desensitization and internalization(Lohse et al., 1990). In addition, βArr1 and βArr2 are known to act as molecular scaffolds for kinases and phosphatases. Dopamine D2 receptor is shown to signal through βArr2 mediated mechanism in the regulation of serine/threonine kinase Akt and glycogen synthase kinase (GSK3) supported by direct in vivo biochemical observations in pharmacological and genetic models of enhanced dopaminergic neurotransmission(Beaulieu et al., 2005). D2-like receptors are also known to activate cell proliferation related pathways such as mitogen activated protein kinase (MAPK)/ERK1/2 pathway shown in various cell lines(Beom et al., 2004), and ERK pathways can also be activated through the D2R-β-arrestin complex once the receptor is internalized(Lefkowitz and Shenoy, 2005). Dopamine receptors are also known to transactivate many RTKs. RTK are a major family of cell surface receptor involved in many functions in both neuronal and non-neuronal cell types. RTK activation generally lead to activation of several downstream signaling pathways such as PI3K/Akt, Ras/MAPK and PLC mediated signaling. In addition to direct activation by their ligands, RTKs can also be transactivated by GPCRs. The dopamine receptors are reported to transactivate RTK such as the platelet derived growth factor β receptor (PDGFRβ), insulin-like growth factor 1 receptor (IGFR1), epidermal growth factor receptor (EGFR) and BDNF receptor, neurotrophic tyrosine kinase, receptor type 2 (TrKB)(Chi et al., 2010; Iwakura et al., 2008; Swift et al., 2011). However, the mechanisms by which dopamine receptors transactivate RTKs are not fully understood. Release of EGF appears to be play a role in transactivation of EGFR by D2 receptor (Yoon and Baik, 2013). RTKs are coupled to signaling mechanism that can elicit different cellular responses, which are beyond direct effect of G protein or arrestin mediated cellular responses. For example, although stimulation of D2 receptor leads to βArr2 dependent inactivation of Akt, the opposite has been seen where Akt is activated by D2 receptor stimulation attributed by transactivation of IGFR and concomitant activation of PI3K-mediated signaling by IGFR. A                                                                                B  Figure 1.4. D1-like and D2-like receptor mediated signaling.

1.5.3 Dopamine D4 receptor

In 1991, Van Tol and group first identified and cloned the dopamine D4 receptor (DRD4) gene(Van Tol et al., 1991).  Subsequently, DRD4 gene was reported to contain considerable variation in human gene and documented to be among the most highly polymorphic genes in human genomes(Chang et al., 1996; Lichter et al., 1993; Van Tol et al., 1992). Because of its highly polymorphic nature and its association with many neurological disorder and behavior, there was great interest in this gene. DRD4 is located on the chromosome11 (11p15.5), contains four exons and transcription site at 400-500bp upstream, whereas the promoter sequence is located further upstream of the transcription site(Kamakura et al., 1997; Van Tol et al., 1991). The DRD4 gene contains a large number of polymorphism in its coding sequence and most extensive in exon3 region that codes for the third intracellular loop (IC3) of the receptor(Van Tol et al., 1992). This polymorphism consists of a variable number of 48bp tandem repeats (VNTR) existing as 2-11 repeats denoted as D4.2 to D4.11, the length of each variant is (no of repeats X 16 amino acids). The DRD4 allele frequencies of the different variants are very heterologous where the most frequent variant is D4.4 (64%) followed by D4.7 alleles (21%) and the D4.2 (8%)(Chang et al., 1996). However, there is a difference in the allele frequency among different population. For example, in Asian population D4.7 allele occur occasionally (<1%), whereas D4.2 allele is more frequent (18%). Interestingly, D4.7 alleles are reported to be five to ten times younger than the common D4.4 alleles but have increased in frequency by positive selection(Ding et al., 2002). Dopamine D4 receptor signaling

D4 is a D2-like receptor primarily known to activate through Gαi/o protein inhibiting AC and thereby reduces intracellular concentration of the second messenger cAMP. The VNTR polymorphism in the IC3 of D4 receptor appears to be important region for AC coupling and G proteins(Kazmi et al., 2000; Oldenhof et al., 1998). For example, D4.7 receptor variant has two to three fold lower potency for dopamine mediated coupling to AC than the D4.2 and D4.4 receptors(Asghari et al., 1995). While D4.10 receptor is two to threefold more potent in AC coupling than the D4.2 receptors(Jovanovic et al., 1999).  Activation of D4 receptor inhibits AC and reduces cAMP level, which is an important and ubiquitous second messenger that further inhibits downstream effectors such as PKA and DARPP-32.  DARPP-32 is an intermediate in many signaling pathways activated by various neurochemicals, integrating their action and provides link to other effectors and transcription factors(Le Novere et al., 2008). Furthermore, D4 receptor activation has been shown to activate NFkB, an important transcription factor that plays a role in inflammation(Zhen et al., 2001), to induce kruppel-like factor-2 (KLF2), a critical regulator of quiescence in T-lymphocytes(Sarkar et al., 2006) and c-Fos expression, which upon dimerization with c-jun forms the transcription factor AP-1 that activates transcription of gene involved in proliferation, differentiation, defense against invasion and cell damage(Bitner et al., 2006) (Figure 1.5). D4 receptor also modulates many signaling pathway in addition to AC, including phopholipases, ion channels, MAP kinases and NA /H exchange.  Many of these pathways are regulated by Gβγ protein subunits that are released by receptor activation of Gαi/o proteins. D4 receptor activity influences the intracellular calcium levels through different mechanism depending on the cell types(Rondou et al., 2010), and increase in calcium levels leads to the activation of CaMKII, which in turn regulate other targets such as AMPA receptors. D4 receptor also influences different types of potassium channels. D2 and D4 receptor interact with G protein coupled inwardly rectifying potassium channel;kir3 (GIRK), an important regulator of cellular excitability, the opening which reduces the firing rate of neurons(Lavine et al., 2002). Dopamine was shown to stimulate D4.2, D4.4 and D4.7 receptors and modulate GIRK currents through Gi/o proteins in Xenopus oocytes(Wedemeyer et al., 2007). D4 receptors can also induce arachidonic acid release (Piomelli et al., 1991)and affect the activity of Na /H exchangers, which regulate intracellular pH, extracellular acidification and cell volume(Felder et al., 1990). D2-like receptors are also reported to activate cell proliferation related pathways such as mitogen activated protein kinase (MAPK) signaling, more specifically the extracellular signal regulated kinase ERK1 and ERK2(Narkar et al., 2001). Furthermore, this ERK activation upon dopamine D4 receptor stimulation is dependent on transactivation of PDGFβ receptor(Oak et al., 2001) and D4 receptor can also transactivate intracellular PDGFβ receptors(Gill et al., 2010). However, there was no difference in the magnitude or duration of ERK activation amongst different D4 variants (D4.2, D4.4 and D4.7). In vivo studies have demonstrated that this transactivation of PDGFβ receptor by D4 receptor leads to depression of excitatory glutamatergic transmission mediated through NMDA receptors(Kotecha et al., 2002; Oak et al., 2001). Although the precise mechanism of D4 receptor transactivation of PDGFβ receptor is not clear, these finding suggests an important role of RTKs in the regulation and communication of dopamine and glutamate signaling in CNS, defects of which have been implicated in neurological disorders such as ADHD and schizophrenia. Dopamine D4 receptor also modulate GABA signaling in pyramidal neurons of prefrontal cortex by decreasing the functional GABAA receptor level at the plasma membrane by decreasing its transport through actin depolymerization(Graziane et al., 2009; Wang et al., 2002). Dopamine receptor interacting protein (DRIP) have been characterized in D2-like receptors, where GPCR like dopamine receptors are concentrated in a specialized distinct microdomain such as lipid rafts or caveolae to enhance the speed and specificity of GPCR signaling by recruiting different signaling components to form dynamic complexes that contributes to fine tuning of the downstream signaling complexes. Dopamine D4 receptor contains a proline rich sequences in the polymorphic regions of IC3, which can interact with SH3 domain containing proteins(Oldenhof et al., 1998). A strong interaction has been detected with Grb2 and Nck, two adaptor proteins capable of recruiting multi-protein complexes to the receptor and influencing cell proliferation, cell movement, axon guidance etc. Mutant receptor with removal of this putative binding site can still bind dopamine and G protein but fails to couple with AC. Furthermore, the SH3 domains of the D4 receptor are involved in control of receptor internalization. D4 receptor also interacts with another DRIP such as GIRK, which modulates ion passage in response to D4 stimulation. Interestingly, a more specific DRIP to D4 receptor but not to other dopamine receptor subtypes is the BTB-Kelch protein KLHL12, that binds specifically to the polymorphic region of the D4 receptor and function as a substrate specific adaptor in a Cul3-based E3 ubiquitin ligase complex for subsequent ubiquitination of the D4 receptor(Rondou et al., 2008) (Figure 1.5). Dopamine D4 receptor is also reported to form oligomers that plays a role in receptor biogenesis and different polymorphic variant has different affinities(Van Craenenbroeck et al., 2011).  Figure 1.5. D4 receptor mediated signaling. Regulation of D4 receptor

Receptor functions are generally regulated by variety of systems including agonist induced receptor desensitization and internalization, regulation by gene expression and biosynthesis of the receptor. Similar to GPCRs, the functionality of dopamine receptor is regulated by agonist induced desensitization and internalization of receptor. Studies of D2-like receptors have shown that continuous stimulation with agonists leads to phosphorylation of the D2 receptor, leading to uncoupling of G proteins and subsequent arrestin recruitment and internalization that is dynamin dependent endocytosis(Macey et al., 2004). However studies on desensitization of D4 receptor are rare and studies on several cell lines indicated that D4 receptor is not strongly regulated through desensitization or internalization mechanism compared to other GPCR such as γ2 adrenergic receptors(Cho et al., 2006; Spooren et al., 2010). The mechanisms regulating DRD4 gene expression are far from being understood. Some studies suggested up regulation of DRD4 in striatum of schizophrenia patients, although other studies showed decreased expression(Helmeste and Tang, 2000; Oak et al., 2000). Furthermore, folding efficiency is demonstrated to be rate limiting in biogenesis of the receptor(Van Craenenbroeck et al., 2011) and antipsychotic drug can function as pharmacological chaperones in upregulating receptor expression by stabilizing it in the ER(Van Craenenbroeck et al., 2006). Expression profile of D4 receptor

Dopamine receptors have broad expression patterns in the brain and in the periphery. All dopamine receptors are present in all targets of dopamine in CNS. D4 receptor is found in various brain regions but is less abundant than D2 receptor and distribution of D4 expression overlaps with D2 receptor expression suggesting a redundancy of these two receptor subtypes. D4 receptor is most abundant in retina(Cohen et al., 1992), cerebral cortex, amygdala, hippocampus, hypothalamus but sparsely in the striatum as assessed by Northern blot and RTPCR (Matsumoto et al., 1995; Valerio et al., 1994), in situ hybridization (Lidow et al., 1998; Meador-Woodruff et al., 1994; Meador-Woodruff et al., 1997) and ligand binding assay(Defagot et al., 2000; Primus et al., 1997) and immunohistochemistry(Mrzljak et al., 1996; Rivera et al., 2002). Interestingly, these studies found D4 receptor in both pyramidal and non-pyramidal cells of the cerebral cortex particular layer V, and in the hippocampus. Outside of CNS, D4 receptor is found in the cardiac atrium, lymphocytes and kidney. Physiological role of D4 receptor in the brain

Dopamine is not a simple excitatory or inhibitory neurotransmitter but a neuromodulator, a mediator of slow synaptic transmission that alters the response of the target neurons to other neurotransmitters thus altering synaptic plasticity. Dopamine through binding to D4 receptor can control the activity of glutamate receptors (NMDA and AMPA), which mediate corticostriatal neurotransmission. Dopamine D4 receptor has been reported to have the unique ability to perform phospholipid methylation that can affects ion channels, important for modulation of neuronal firing activity(Kuznetsova and Deth, 2008), where impairment in methylation contributes to disorders of attentions. The D4 receptor knockout mouse has been generated to study the in vivo role of D4 receptor(Rubinstein et al., 1997). The D4 mutant mouse showed a reduction in locomotion and behavioral response to novelty(Dulawa et al., 1999; Glickstein and Schmauss, 2001). This finding is of great interest because few studies have suggested association between D4 VNTR polymorphism (D4.7) and human personality trait of novelty seeking(Benjamin et al., 1996; Ebstein et al., 1996). D4 mutant mice also showed enhanced activity of cortical pyramidal neurons, and experiment with pharmacological agents indicated that D4 receptor has inhibitory role in frontal cortex glutamatergic activity(Rubinstein et al., 2001). D4 mutant mouse also showed more sensitivity to the stimulation of locomotor activity induced by ethanol, cocaine and methamphetamine although the mutant mice have hypoactive locomotor phenotype(Rubinstein et al., 1997). In general, the specific physiological roles played by D4, D3 and D5 receptors in the brain are largely unknown. Dopamine D4 receptor related diseases

DRD4 has been widely implicated in neurological and psychiatric disorders such as ADHD, schizophrenia, Tourette’s syndrome and novelty seeking. A great deal of interest was shown towards DRD4 following reports when DRD4 level was found to be six fold higher in post mortem brains of schizophrenia patients compared to control (Seeman et al., 1993) and a high degree of affinity of antipsychotic drug clozapine was observed in DRD4 (>10fold) compared to other D2 like receptor(Van Tol et al., 1991). Dopamine D4 receptor polymorphism is strongly associated with ADHD, where D4.7 variant showed an increased risk of developing ADHD where as D4.4 is protective(Faraone et al., 1999; LaHoste et al., 1996; Li et al., 2006). D4.7 variant is also associated with other neurological complex behaviors such as Tourette’s syndrome(Comings et al., 1999) and personality trait of novelty seeking(Ebstein et al., 1996). A single nucleotide polymorphism in the promoter region of D4 gene C-521T is also reported to be associated with novelty seeking(Ronai et al., 2001). However, there are other studies that failed to confirm the association with novelty seeking(Jonsson et al., 1997). The D4 receptor has been also linked to schizophrenia, a complex neuropsychiatric disorder(Catalano et al., 1993; Lung et al., 2002; Talkowski et al., 2008). There are other accumulating evidences that suggest association of dysbindin, neuregulin1, dystrobrevin-binding protein1(DNTBP1) as a risk factor for schizophrenia. The observation that D2 receptors are blocked by classical neuroleptic compounds led to the theory that dopamine plays an essential role in the pathogenesis of schizophrenia. However, many studies following possible association between D4 and schizophrenia have not led to a strong links and failed to demonstrate the correlation between D4 polymorphism and clozapine response(Rao et al., 1994; Tanaka et al., 1995). Regardless, a lot of evidence supports the hypothesis that reduction in positive systems following treatment with antipsychotic is mediated through blocking D2 receptors. Although exclusive blockade at the D4 receptor may not be sufficient for antipsychotic action, it might result in an improved symptomatic profile in combination the D2 receptor blockade.

1.6 Hypothesis and specific aims

Neurochemicals are now being appreciated to have an important role in neurogenesis, in the formation of neurons other than its function in synaptic transmission. Given the close similarities between GBM stem cells and neural stem cells, and the emergence and growth of GBMs in the rich neurochemical milieu of the brain, these neurochemical signaling pathways may profoundly impact tumor growth. I hypothesized that interrogation of neurochemical space and its influence in patient derived GBM neural stem cell (GNS) using small molecules would reveal many novel regulatory mechanisms, which could then be exploited for therapeutic applications. Specific Aim 1: To interrogate neurochemicals’ role in GBM and identify GNS selective compounds. A library of 680 neuroactive compounds was screened against a panel of patient derived GNS cells, human neural stem cells (NS) and fibroblasts to identify GNS selective compounds. Selective compounds were further validated in vitro and in vivoSpecific Aim 2: To characterize the mechanism of action for the GNS selective compounds and determine the mechanisms that confer the selectivity. The mechanism of action for the GNS selective compound was determined through an unbiased approach using genome wide expression array and a phospho-kinase array. Selective compounds were validated for its target by genetic knockdown experiments. Specific Aim 3: To investigate if modulation of neurochemical signaling can direct NS cells fate into specific neuronal lineages. A high content screen was developed to capture neurochemicals influence on NS cells differentiation into neurons and astrocytes, and proliferation. Human NS cells were screened with a library containing 680 neuroactive compounds to identify compounds with differentiation potential and further validate and characterize the fate of differentiated cells. Chapter 2 Interrogation of neurochemicals’ role in glioblastoma stem cell growth (I conducted all experiments and data analysis described in this chapter except the in vivo experiment, where I got assistance from Lilian, Kevin, Michelle and Milly with animal handling, and Veronique Voisin and ChangJiang Xu with TCGA data analysis)

2.1 Introduction

Glioblastoma (GBM) is the most common malignant primary brain tumor in adults and has proven resistant to all therapeutic strategies attempted to date. The median survival time for GBM patients is 15 months even with standard care of treatment including surgery, radiation and chemotherapy (Furnari et al., 2007; Stupp et al., 2005). The alkylating agent temozolomide (TMZ) is the only chemotherapeutic of any benefit, and it is effective only transiently in a subset of patients (Brennan et al., 2013; Hegi et al., 2005). Therefore, there is an urgent need for identification of novel targets and better therapeutic approaches for treatment of GBM. A prerequisite to identifying new therapeutics is a better understanding of the diversity of mechanisms that govern GBM growth. GBM growth is initiated and maintained by small subpopulations of tumorigenic cells termed GBM stem cells, which have a phenotype similar to normal neural stem cells (NSCs)(Beier et al., 2007; Galli et al., 2004; Singh et al., 2003; Singh et al., 2004). GBM stem cells contribute to tumor progression and resistance to therapy (Bao et al., 2006; Chen et al., 2012), such that longterm disease control is likely to require elimination of this driver population, in addition to the more differentiated tumor bulk. These GBM stem cells are best prospectively identified from fresh tumors and interrogated directly in vivo, but tumorigenic cells that show similar properties to directly isolated cells can be grown in a defined media (herein called GBM derived neural stem cells, GNS cells) which then can be used for screening(Pollard et al., 2009). A deeper understanding of the regulatory mechanisms that govern the proliferation and survival of GBM stem cells will be essential to developing rational new therapies.  In a previous screen of an unbiased bioactive small molecule library on mouse NSCs, neurochemical signaling pathways significantly impacted proliferation and survival of normal NSC populations (Diamandis et al., 2007). As tumorigenic GBM cells display many characteristics of NSCs, this observation raised the intriguing possibility that known neuromodulators might also affect tumorigenic GNS cells. Neurochemicals are endogenous chemical messengers that mediate the synaptic function of differentiated neural cells in the mature CNS. Recent studies suggest an important role of neurochemicals, for example gamma-aminobutyric acid (GABA) and glutamate, in regulating NSC fate both in early development (Andang et al., 2008; Martins and Pearson, 2008; Schlett, 2006) and in adult neurogenesis (Berg et al., 2013; Ge et al., 2006; Hoglinger et al., 2004; Song et al., 2012). GABA regulates adult mouse hippocampal NSCs by maintaining their quiescence through the GABAA receptor (Song et al., 2012), yet can also promote embryonic NSC proliferation (Andang et al., 2008), suggesting context specific functions. These effects may reflect influences of local or more distant neuronal activity on the NSC niche. Consistent with this idea, dopamine afferents project to neurogenic zones and depletion of dopamine decreases the proliferation of progenitor cells in the adult subventricular zone (SVZ) through D2-like receptors (Hoglinger et al., 2004). Dopamine is also present in early neuronal development in the lateral ganglionic eminence (LGE) and modulates LGE progenitor cell proliferation (Ohtani et al., 2003). Serotonin signaling similarly contributes to the SVZ NSC niche (Tong et al., 2014). Neurochemicals and their receptors have also been implicated in the growth and progression of many non-CNS cancers (Dizeyi et al., 2004; Schuller, 2008; Weddle et al., 2001). The mechanisms whereby neurochemicals affect cancer growth is not understood, but given that GBM arises in the rich neurochemical milieu of the mature CNS, it is highly plausible that neurochemical pathways may play a role in promoting GBM growth and tumour progression. Consistent with this proposition, a very recent study on optogenetic manipulation of cortical neuronal activity in a mouse GBM xenograft model showed a definite influence of neuronal activity on GBM growth (Venkatesh et al., 2015). As well, a very recent study suggests antidepressants may affect survival of lower grade models of GBM(Shchors et al., 2015). The neurochemicals that make up the rich extracellular milieu of the brain that GBM arises from have not been exploited for therapeutic applications. Modulation of neurochemical signaling offers great promise for GBM therapy because large number of agonists and antagonists have been developed against various neurochemical signaling pathway, which are selected to access CNS and overcome blood brain barrier, and many are clinically approved for neurological disorders that may be readily deployed for GBM therapy. In this chapter my aim was to interrogate the role of neurochemicals in regulating the growth of GBM stem cells. I hypothesized that a systematic survey of known neuroactive compounds against GBM stem cells could reveal novel targets and regulatory mechanisms, outside of traditional chemotherapies. Importantly, the extensive previous development of neurochemical modulators for psychiatric disorders and other CNS-specific diseases would allow the immediate redeployment of approved drugs for oncologic applications. Towards to this end, neurochemical modulators from all different classes were screened against primary in vitro cultures of human GBM-derived NSCs, referred to as GNS cells (Lee et al., 2006; Pollard et al., 2009) and human fetal NSC (NS) and human fibroblast (BJ) to identify selective compounds. Identification of GNS selective compounds will reveal the vulnerability of GBM stem cells that can be exploited for therapeutic purposes.

2.2 Materials and Methods

Cell culture and primary tumor cells

GNS and NS lines were grown as adherent monolayer culture on Primaria tissue culture dish coated with poly-L-ornithine (PLO) and laminin, in a serum free medium supplemented with epidermal growth factor (EGF) and basal fibroblast growth factor (FGF) as described previously (Pollard et al., 2009). Neurocult media (Stem Cells Technologies) was supplemented with 10ng/µl EGF, 10ng/µl FGF, N2 and B27. Cells were passaged by enzymatic digestion using Accutase. Primary patient samples were obtained with approval from research ethics committees and appropriate patient consent. Tumor cells were freshly dissociated from patient sample in artificial cerebrospinal fluid followed by treatment with enzyme cocktail (1.33mg/ml trypsin, 0.67mg/ml hyaluronidase and 0.1 mg/ml kynurenic acid) at 37oC(Singh et al., 2003; Singh et al., 2004). BJ fibroblast, Daoy and C8-D1A and U-2 OS (ATCC) were maintained in DMEM with 10% FBS.

Compound Library

The neurotransmitter library was purchased from BIOMOL international (now integrated into Enzo life sciences). The library contains 680 compounds covering thirteen classes of neurotransmitters; adrenergics, cholinergics, dopaminergics, serotonergics, ionotropic glutamatergics, metabotrophic glutamatergics, GABAergics, purinergics and adenosine, histaminergics and melatonin ligands, opioids and sigma ligands. The compounds were supplied in DMSO at 10mM concentration in 96-well medium deep plates and stored at -80oC. All compounds for retesting were purchased from Tocris Bioscience.

Chemical screen

Cells were seeded in laminin and poly-L-ornithine (PLO) coated 384-well black clear bottom plates at a density of 2000 cells per well. Compounds were added at a concentration of 5 µM using Biomek FXP automation workstation and incubated with cells for five days at 37oC.  Cell viability was assessed by measuring Alamar Blue incorporation according to the manufacturer’s protocol (Invitrogen). Percent growth inhibition was calculated relative to the control DMSO wells.

Secondary screen/dose response curve

The potency and selectivity of hits from primary screen was tested in 8-point two- fold dilution series ranging from 50 µM-0.39 µM concentrations with more lines of GNS, NS and fibroblast. Experimental conditions were same as in primary screen. IC50 was calculated based on an approximate observed value. Fold selectivity was calculated as IC50 of BJ/IC50 of any GNS cells with lowest IC50.

Combination/ synergy screen

2000 GNS cells were seeded in 96 well plate and treated with combination of 6-point 2-fold dose series of either L-741,742 (6.25 µM-0.39 µM) or PNU 96415E (25 µM-1.56 µM) with 10-point 2-fold dose series of temozolomide (100 µM-0.39 µM) in 60-point combination doses.  Cells were incubated with combination of two drugs for five days and checked for cell viability by Alamar blue assay.  Combination index (CI) plot and CI value was calculated for 5-point dose series in each combination using program COMPUSYN. Data points taken for COMPUSYN analysis are temozolomide (100, 50, 25, 12.5 and 6.25 µM) in combination with either L-741,742 (6.25, 3.12, 1.56, 0.78 and 0.39 µM) or PNU 96415E (25, 12.5, 6.25,3.12 and 1.56 µM).

Patient derived xenograft

All mouse procedures were approved by the Hospital for Sick Children’s Animal Care Committee. For subcutaneous xenograft, 2×105 GNS cells in 200 µl of PBS and matrigel (1:1) were injected subcutaneously into flanks of non-obese diabetic/severe combined immunodeficient (NOD/SCID) female mice. 8 mice (2 tumors per mouse except for one mouse in control group that has one tumor) were maintained for each group; control (15 tumors), L741,742(16 tumors) and PNU 96415E (16 tumors). L-741,742 and PNU 96415E were dissolved in 40% 2 hydroxy β-cyclodextrin (Sigma).  Mice were treated three days after tumor implantation. Both L-741,742 and PNU 96415E were treated at 20mg/kg i.p for 5days on two days off until end point. Control group was injected with 40% 2 hydroxy β-cyclodextrin. Tumor growth was monitored with microcalipers until tumor volume reached 17mm in any one tumor from any group and all mice were sacrificed at the same end point. Dissected tumor volume was measured and mass was determined by weighing. For intracranial xenografts, 5×103 GNS cells were injected in the forebrain of NSG mice. Mice were treated one week after tumor implantation with L-741,742 (25mg/kg) five days on two days off for two weeks. TMZ (25 or 12.5 mg/kg) was treated for the first week through gavage. Mice were sacrificed when presented with dome head and neurological symptoms. For the combination experiment, data from two experiments were pooled and performed statistical analysis using Prism.

In vitro Limiting dilution assay (LDA)

Limiting dilution assay was performed as described previously(Tropepe et al., 1999). Primary tumors were dissociated into single cell suspension and seeded in 96 well plate with 10 point-2 fold serial dilution starting from 2000cells -4 cells/ well, 6 wells for each dilution per plate. Each well was score for neurosphere formation after 14 days of incubation. Percent of wells not containing spheres for each cell density was calculated and plotted against the cells per well and regression lines were plotted and x-intercept value at 0.37 was calculated at 95% confidence interval using Sigma Plot, which gives the number of cells required to form at least one neurosphere.

Short hairpin construct and transfection

5 µg of short hairpin targeting DRD4 (RHS4533-EG1815; TRCN0000014453; Thermo Scientific) or control shRNA construct targeting eGFP (RHS4459;Thermo Scientific), and another set of short hairpin against DRD4 from Origene (TL313054A&B) or control non effective scramble shRNA (TR30021) were transfected in 1X10GNS cells using the Amaxa Nucleofector kit (VPG-1004) and Nucleofector II electroporator (Amaxa Biosystem) according to manufacturer’s protocol.  After 24h transfection, cells were briefly selected with puromycin for 48h and then seeded for proliferation assay without selection. I maintained two wells from each transfection after electroporation; one well was lysed to check for knockdown by western blot and the other well seeded for a proliferation assay.

Western blots

Cells were lysed in a denaturing lysis buffer with protease and phosphatase inhibitor as described previously. Protein fragments were separated on sodium-dodecyl-sulfate (SDS) polyacrylamide gel by electrophoresis and transferred onto polyvinylidene fluoride (PVDF) membrane. PVDF membranes were blocked with 5%milk or bovine serum albumin (BSA) for one hour followed by incubation with primary antibody overnight at 4oC and secondary antibody for one hour at room temperature. Secondary antibodies were conjugated to horseradish peroxidase and detected with chemiluminescence. Western blots were performed using following antibodies; anti-DRD4 antibody at 1:750 (Millipore# MABN125), anti-β-actin at 1:10,000 (Sigma), anti-active caspase3 at 1:200 (Abcam#ab2302).


Primary patient tumor samples or mouse GBM xenograft tumors tissues were fixed overnight in 4%PFA, paraffin embedded and serial sectioned. Sections were deparaffinized and rehydrated through an alcohol gradient to water for antigen retrieval in 10mM citrate buffer pH 6.0 in a microwave pressure cooker. Endogenous peroxide activity was blocked with 3%(v/v) peroxide in methanol for 15 min at room temperature and nonspecific binding was blocked with 2% (v/v) normal goat or horse serum (vectorlabs) in PBS with 2% (w/v) BSA for 30 min. Primary antibodies; anti-DRD4 at 1:750 (Millipore# MABN125), anti-LC3B at 1:1500 (Cell Signaling #3868), anti-p62 at 1:1000(BD Bioscience) and anti-mono and polyubiquitinylated protein conjugates (FK2) at 1:1000 (Enzo Life Sciences) was incubated overnight @ 4°C. Appropriate secondary antibody; biotinylated, avidin-linked peroxidase), DAB (Vectastain Universal Elite ABC kit, Vectorlabs) and Alexa-568 were used to detect binding of the primary antibody. Normal rabbit serum was used for control sections.

cAMP assay

GNS cells (G362) was seeded at a density of 5000 cells/well in 96-well plate and incubated overnight. Cells were either untreated (DMSO) or pretreated with 30µM of A412997 for 30 minutes and added 30µM of forskolin, and incubated for 20 minutes. Cells were lysed and followed as per protocol Cyclic AMP XPTM Assay kit (Cell Signaling #4339).

2.3 Results

2.3.1 Identification of GNSUselective compounds

To identify compounds that selectively inhibit the growth of GNS, I established proliferation assays for three different cell types: GNS cells, human NS and the BJ human fibroblast cell line. GNS cells are patient-derived tumor cells which display many characteristics of normal NS cells including expression of the markers Nestin and SOX2, and the ability to self-renew and to partially differentiate (Pollard et al., 2009). Thus NS cells serve as a well-matched control for their neoplastic GNS counterparts.  To eliminate compounds with non-specific cytotoxic effects, I defined NS-selective hits as those that target both NS and GNS cells, but not fibroblasts. I then filtered for compounds that showed more activity towards GNS cells compared to NS cells, and termed these as GNS-selective compounds.

  1. library of 680 neuroactive compounds was screened against three GNS lines, two NS lines and the BJ fibroblast line at a concentration of 5µM for five days. Primary hits were defined as compounds that caused greater than 20% growth inhibition compared to the DMSO control. The total hit rate in all cell populations ranged from 2.6-6.5% (Figure 2.1A). Of all the neurochemical classes, compounds known to modulate dopaminergic (27%), cholinergic (17%), adrenergic (18%) and serotonergic (9%) pathways were enriched in the total hits, suggesting that these pathways may play a specific role in regulating NS cell growth (Figure 2.1B).  These pathways were also the main enriched hits when normalized to each neurochemical class, defined as the number of hits per number of compounds in each class (Figure 2.1C). Additionally, sigma receptors showed high activity in my screen and interestingly sigma receptor genes showed the highest expression compared to other neurochemical receptors (Figure 2.1C and Figure 2.2).

  1. C


Figure 2.1. Identification of GNS-selective compounds

A. An outline of the primary and secondary screens. Primary screen data is presented in a heatmap showing growth inhibition of each compound (5µM) across the six cell lines screened. Secondary screens were done in dose series to determine the fold selectivity (IC50 of BJ/IC50 of any NS or GNS cells with the lowest IC50). B. Percentage of different neurochemical classes enriched in the total hits. C. Percent activity (hits) of each neurochemical class. Number of hits/total number of compounds present in the library of each class.  Figure 2.2. Gene expression pattern of different neurochemical receptors across various GNS and NS cell lines. A heat map of neurochemical receptor genes for 13 different classes across panel of GNS and NS cells. Gene expressions of sigma receptors are highlighted with an arrow. From the primary screen hits, 29 compounds were chosen that showed a selective effect on GNS and NS cells compared to fibroblasts and were retested in a dose response series (0.39-50µM) in the same cell populations as in the primary screen (Table 2.1). From this secondary screen, ten compounds were selected that showed more than 8-fold selectivity towards GNS and NS cells compared to fibroblasts. These ten NS-selective compounds were PNU-96415E, L-741,742, Ifenprodil tartrate, LY-165,163, MDL-72222, Tropanyl 3,5-dimethylbenzoate, N,N-Diethyl-2(4-(phenylmethyl)phenoxy)ethanamine, (±)-Tropanyl-2-(4-chlorophenoxy)butanoate, MG-624 and Ivermectin (Table S2.1, Figure 2.3). The ten NS-selective compounds were enriched for dopaminergic, serotonergic and cholinergic classes (Figure 2.3) suggesting these pathways as potential targets for GBM. To further validate selectivity, each compound was retested in three further non-NS control cell lines, namely Daoy cells (human medulloblastoma), U-2 OS (human osteosarcoma) and C8-D1A (mouse astrocyte cells). The ten compounds were 8-128 fold more active against NS or GNS cells compared to BJ fibroblasts (Table S2.1). Notably, three of the compounds PNU 96415E, L-741,742 and ifenprodil tartrate, showed 8-fold selectivity against GNS cells compared to NS cells (Table S2.1) and were termed GNS-selective compounds. Two of these compounds, PNU 96145E and L-741,742, represent DRD4 antagonists and were chosen for further investigation. Table 2.1: List of 29 hits retested from the primary screen to identify NS selective compound compared to fibroblast.

DRUG NAME Fold Selectivitya Class Activity
5-Iodotubercidin 6 fold Adenosine Adenosine kinase inhibitor
Ifenprodil tartrate 16 fold Glutamatergic/ Adrenergics NMDA receptor antagonist/Alpha1 adrenergic antagonist
RS 17053 HCl 4 fold Adrenergics Alpha 1A adrenergic
(±)-Isoproterenol HCl Not active Adrenergics Beta adrenergic agonist
(-)-Cyanopindolol hemifumarate Not active Adrenergics Beta adrenergic antagonist
Ivermectin 16 fold Cholinergics Allosteric modulator of alpha 7 nicotinic receptor
MG-624 64 fold Cholinergics Alpha 7 nicotinic antagonist
(-)-N-Phenylcarbamoyleseroline Not active Cholinergics Choline acetyltransferase inhibitor
(±)-Tropanyl-2-(4chlorophenoxy)butanoate 32 fold Cholinergics Stimulates acetylcholine release
R( )-SKF-81297 0 Dopaminergics D1 dopamine agonist
R(-) Propylnorapomorphine HCl 2 fold Dopaminergics D2 dopamine agonist
L-741,742 HCl >8 fold Dopaminergics Dopamine D4 antagonist
Fluphenazine 2HCl 4 fold Dopaminergics Dopamine antagonist
Thioridazine HCl 6 fold Dopaminergics Dopamine antagonist
3-a-[(4-Chlorophenyl) phenylmethoxy] tropane HCl 2 fold Dopaminergics Dopamine uptake inhibitor
GBR-12909 4 fold Dopaminergics Dopamine uptake inhibitor
GBR 13069 2HCl 4 fold Dopaminergics Dopamine uptake inhibitor
GBR 12935 2HCl 4 fold Dopaminergics Dopamine uptake inhibitor
PNU 96415E >16 fold Dopaminergics Dopamine D4 receptor antagonist
Astemizole 0 Histaminergics H1 Histamine antagonist
N,N-Diethyl-2-[4- (phenylmethyl)phenoxy]ethana mine >64 fold Histaminergics Histamine antagonist
BNTX maleate 2 fold Opoids Delta 1 opioid antagonist
LY-165,163 >16 fold Serotonergics Serotonin 5-HT 1A agonist
SB 216641 HCl 2 fold Serotonergics Serotonin 5-HT 1B antagonist
MDL-72222 8 fold Serotonergics Serotonin 5-HT 3 antagonist
Tropanyl 3,5-dimethylbenzoate >32 fold Serotonergics Serotonin 5-HT 3 antagonist
RS 39604 HCl 2 fold Serotonergics Serotonin 5-HT 4 antagonist
1-Methyl-4-[2-(2-naphthyl)ethenyl]-pyridinium iodide 6 fold Sigma receptor Sigma receptor ligand
cis-(±)-N-Methyl-N-[2-(3,4dichlorophenyl)ethyl]-2-(1pyrrolidinyl)cyclohexamine 2HBrb 16 fold Sigma receptor Sigma receptor ligand
  1. Fold selectivity= IC50 of BJ/IC50 of NS or GNS with lowest number.

“>” Indicates IC50 not seen even at highest concentration tested (50 µM) in BJ cells thus may be more selective than the number indicated.

  1. Not available for further retest

 Figure 2.3. The ten NS selective compounds and their IC50 (µM) across different cell lines. The ten NS selective compounds are grouped under their neurochemical classes.  

2.3.2 DRD4 antagonists selectively inhibit GNS growth and reduce clonogenic potential of primary GBM tumor cells.

I next re-tested PNU 96145E and L-741,742 along with other commercially available DRD4 antagonists (L-745,870 and PD 168568) for their effects on a larger panel of six GNS and four NS lines. All of the DRD4 antagonists showed selectivity toward GNS cells with differing potency (IC50), in the order of L-741,742 (1.5-6.2µM)> L-745,870(3.1-6.2µM) > PNU 96415E(6.25µM) > PD168568 (25-50µM). L-741,742 and PNU 96415E are specific DRD4 antagonists and showed the greatest selectivity towards GNS cells (Figure 2.4). PNU 96415E displayed robust selectivity towards GNS cells compared to NS cells and non-NS control cells, the latter of which were not sensitive even at the highest concentration tested (50µM) (Figure 2.4A). L-741,742 was the most potent GNS inhibitor but showed variable effects on different GNS cells. L-745,870 exhibited selectivity towards GNS but was less potent (Figure 2.4C), while PD 168568 showed modest selective effect at higher concentration (data not shown). A                                                                          B  C                                                                          D 

Figure 2.4. DRD4 antagonists are GNS selective

  1.              Percent cell viability of 4 Non-NS control cell lines, 3 NS and 6 GNS lines upon treatment with PNU 96415E in dilution series from 0.39µM-50µM.  Controls n= 3 mean ± SEM, NS lines n=5-15 mean± SEM, GNS lines n=3-12 mean± SEM.
  2.               Percent cell viability of BJ fibroblast, 3 NS and 5 GNS lines upon treatment with L-741,742 in dilution series from 0.39µM-50µM. BJ n=3 mean ± SEM, NS lines n=3-11, mean ± SEM, GNS lines n=3-7, mean ± SEM.
  3.               Percent cell viability of 3 NS and 3 GNS lines upon treatment with L-745,870 in dilution series from 0.39 µM-50 µM. NS lines n=3-11 mean ± SEM, GNS lines n=3-11 mean ± SEM.
  4.              Phase contrast image of differential response of GNS (G411) cells and NS (hf5205) cells and fibroblast (BJ) upon L-741,742 (10 µM) treatment after 5 days. Scale bar (100µm)

To confirm that the effect of DRD4 antagonism was not merely specific to GNS cell lines, L741,742 and PNU 96415E were tested in freshly isolated GBM patient tumor cells using a primary in vitro limiting dilution assay (Figure 2.5). Tumor samples were dissociated into a single cell suspension and directly seeded prior to treatment with L-741,742 (10µM), PNU 96415E (25µM) or DMSO control for 14 days before scoring wells for presence/absence of neurosphere colonies. A massive reduction in frequency of colony forming cells after treatment with L-741,742 (40-83 fold reduction) and PNU 96415E (19-29 fold reduction) (Figure 2.5) was observed. These data strongly suggest that both L-741,742 and PNU 96415E inhibit the clonogenic potential of fresh primary tumor cells, and may therefore effectively target the stem cell population in each patient tumor. A                                                                          B               C 

Figure 2.5. DRD4 antagonists reduced the clonogenic potential of primary patient tumor cells

A-C. Linear regression plot of in vitro LDA for freshly dissociated GBM patient tumor (GBM686, GBM648, GBM677) treated with L-741,742 (10µM), PNU 96415E (25 µM) and DMSO. Representative phase contrast image of neurospheres at day 14 in well seeded with 2000 cells. Scale bar (100µm).

2.3.3 DRD4 antagonists are synergistic with TMZ

Next, I evaluated the effect of DRD4 antagonists in conjunction with the commonly used chemotherapeutic agent TMZ to assess the clinical potential of this drug pair combination. Synergy was tested in both G362 and G481 cells using the combination of TMZ with either L741,742 or PNU 96415E.  L-741,742 and PNU 9641E exhibited striking synergism with TMZ in vitro GNS cells (Figure 2.6 A-B).  The degree of synergism was quantified using the combination index (CI) method (Chou, 2010), for which a CI value of 1 indicates additivity, a value of < 1 indicates synergism and a value > 1 indicates antagonism.  The lowest CI value for L-741,742 in combination with TMZ in G481 and G362 was 0.28 and 0.29 respectively, and for PNU 96415E in combination with TMZ was 0.32 and 0.56 respectively (Figure 2.6 C-D). Based on these in vitro data, both DRD4 antagonists might enhance the therapeutic efficacy of TMZ in patients. A                                                                  B  C                                                                         D 

Figure 2.6. DRD4 antagonists are synergistic with temozolomide

A&B. Growth inhibition plot for G362 and G482 cells treated with TMZ in combination with L741742 or PNU 96415E respectively. C&D. Combination index plot for TMZ with L-741742 (L7) or PNU 96415E (PNU) in G362 and G482 cells respectively. Combination index (CI) plotted against fractions affected (Fa) analyzed using software COMPUSYN.

2.3.4 DRD4 antagonists inhibit GBM xenograft growth in’vivo

To validate the effects of L-741,742 and PNU 96415E in vivo, I first validated their effect in a subcutaneous tumor model, where GNS cells were injected into the flanks of immunocompromised NOD scid gamma (NSG) mice and treated with PNU 96415E (20mg/kg), L741,742 (20mg/kg), or vehicle until tumors reached our institutional volumetric cutoff of 17mm in any one mouse (Figure 2.7A). The effect of PNU 96415E and L-741,742 treatments were tested by three different measures.  Measurement of tumor volume over time course revealed much slower growth in the treated groups compared to the vehicle control group (Figure 2.7 B). The average tumor weight at the end point was reduced by 44.3% with PNU-96415E treatment and 40.9% with L-741,742 treatment (Figure 2.7C). A  B                                                                                     C  D                                                                               E  F                                                                      G  Figure 2.7. DRD4 antagonists inhibit GBM xenograft growth in vivo.  A. A schematic of in vivo subcutaneous xenograft model. B. Growth curve of subcutaneous implanted tumor (G362) over period of time, Control n= 15, mean ±SEM, PNU 96415E n=16, mean ±SEM, L-741,742 n=16, mean ±SEM. ** p value <0.005, *p value <0.05 unpaired onetailed t-test. C. A dot plot showing tumor mass of each tumor from the three treatment groups at end point. Control n= 15, mean ±SEM, PNU 96415E n=16, mean ±SEM, L-741,742 n=16, mean ±SEM. Significance analyzed by t-test unpaired one-tailed. D. Linear regression plot of in vitro LDA for an in vivo treated tumors. Average of each group was taken for the plot, neurospheres scored for 18 wells (6 wells from each tumor). E. Immunohistochemistry staining and quantification for anti-p62 and anti-FK2 (ubiquitin conjugate) and LC3 in in vivo treated tumor. Scale bar-p62 and FK2 (11µm) and LC3(50 µm). N=3 sections, mean ±SEM. F. DRD4 antagonists inhibit xenograft tumor growth in vivo.  Kaplan-Meier survival curve of intracranial xenograft mice (G362) treated with L-741,742 (25mg/kg). N=6 per group. Significance was performed using log rank (Mantel-Cox) test.  G. Kaplan-Meier survival curve of two pooled intracranial xenograft experiment (G362) treated with L-741,742 in combination with TMZ and TMZ alone. Significance was performed using log rank (Mantel-Cox) test. N=18 per group and censorship/endpoint at day 182. Mice censored were sacrificed from other non-brain tumor symptoms. Control and treated tumors were then dissociated and subjected to primary in vitro limiting dilution assays to determine if L-741,742 and PNU-96415E affected the clonogenic capacity of the in vivo treated tumor cells. A substantial reduction in frequency of colony forming cells was observed in both treatment groups compared to controls (Figure 2.7D), suggesting a reduction in the stem cell fraction of treated tumors. The colony forming cell frequency was reduced in treated groups by 4-7 fold, from 1 in every 11 cells in the vehicle treated tumors to 1 in every 43 cells in PNU 96415E treated tumors or 1 in every 76 cells in L-741,742 treated tumors (Figure 2.7D). I further validated the effect of PNU 96415E in vivo using an independent GNS cell line (G411) and observed a similar reduction in tumor growth rate and end-point size (Figure S2.1). Additionally, increased level of both p62, ubiquitinated protein substrate(FK2) and LC3was observed in the treated tumors (Figure 2.7E) confirming the in vivo mechanism. To better assay the clinical potential of DRD4 antagonists, the effect of L-741,742 was validated in an intracranial xenograft model. Mice were treated one week after tumor implantation when the tumor size is substantial (Figure S2.1-C). The DRD4 antagonist alone had a modest but significant benefit in survival over control (Figure 2.7F). However, when combined with TMZ, data from two in vivo experiments demonstrated that L-741,742 significantly improve survival compared to TMZ alone (Figure 2.7G). Together, these data from two different human patientderived xenograft models provide a strong basis for further exploration of DRD4 antagonism and /or inhibition of autophagic flux as a new avenue for GBM therapy.

2.3.5 Primary GBM tumor and GNS cells express functional DRD4 receptor, with higher expression linked to a worse prognosis.

To determine if DRD4 antagonists exert their effects directly through the DRD4 receptor, I first confirmed that DRD4 was expressed in both GNS and NS cells by western blot (Figure 2.8A). I further confirmed that GBM patient tissue samples also expressed DRD4 as shown by both immunohistochemistry and western blots (Figure 2.8B-C, S2.2). DRD4 is therefore expressed on GNS cells but appears to be expressed also on tumor bulk and non-tumorigenic progeny. To assess DRD4 function in GNS cells, I measured a known downstream readout of the receptor activity. DRD4 is a dopamine D2-like receptor that inhibits adenylate cyclase and decreases cAMP levels (Rondou et al., 2010). GNS cells were treated with forskolin to activate adenylate cyclase, and then assessed whether activation of the DRD4 receptor by the DRD4-specific agonist A412997 could block the forskolin-induced cAMP response. Forskolin treatment in GNS cells increased cAMP concentration by 2.1 fold and pretreatment with DRD4 agonist A412997 blocked this response by 31.3%, confirming that DRD4 functions as expected in GNS cells (Figure 2.8D). Primary tumor and tumor-derived GNS cells thus express DRD4 and can respond to DRD4-dependent signals. A                                                                                     B  C                                                                                       D 

Figure 2.8. Primary GBM and GNS cells express functional DRD4 receptor

A&B. Western blot analysis for anti-DRD4 and anti-β-actin across different NS and GNS lines, and primary GBM patient tumor samples respectively. CImmunohistochemistry staining for DRD4 in primary patient tumor samples (GBM742). Scale bar (100µm and 50µm) respectively. D. Fold change of cAMP levels in G362 cells treated with forskolin (30 µM) alone and pretreatment with DRD4 specific agonist A412997 (30 µM) followed by forskolin treatment. N=3, mean ±SEM. In order to probe the clinical relevance of DRD4 expression I probed The Cancer Genome Atlas (TCGA) data on GBM gene expression and found that patients expressing highest DRD4 level have shorter survival than those patients expressing lower DRD4 level (Figure 2.9A-C). A similar pattern was seen with TH (tyrosine hydroxylase) expression (Figure 2.9D), the ratelimiting enzyme for dopamine synthesis. Furthermore, DRD4 was not methylated in GBM samples (Figure 2.9E), suggesting active expression. A                                                       B                                                   C  D                                                                             E  Figure 2.9. The TCGA data on GBM sample show patient with highest DRD4 expression

has worst prognosis, and DRD4 gene is not methylated

A. Kaplan-Meier overall survival (OS) curve of patients with highest DRD4 expression (RNAseq) compared to remaining low DRD4 population using one cut off. B. Kaplan-Meier overall survival (OS) curve of patient with the highest and lowest DRD4 expression (RNAseq) using two cut offs. C. Histogram representing the DRD4 expression distribution across all GBM patient and its cutoff used for OS curve. D. Kaplan-Meier overall survival curve of patient showing highest tyrosine hydroxylase (TH) compared to the remaining low expression population using one cut off. E. Density function of the beta-value of the DRD4 related CpG site across all patients. The 2 gray curves are positive control used to localize the positive methyaltion peak: density function of MGMT related CpG site and density function of all sites for patient TCGA-02-2007

2.3.6 Loss of DRD4 function suppresses GNS growth

To validate DRD4 as a therapeutic target in GBM and determine if loss of its function phenocopies the effect of PNU 96415E and L-741,742, I performed shRNA-mediated knockdown experiments and measured the effect on cell proliferation. I tested five lentiviral shRNA constructs out of which only one shRNA-DRD4 constructs (shRNA-DRD4-4) caused consistent knockdown at 72 hours post transfection (Figure 2.10A). I confirmed reduced DRD4 expression after transduction of the shRNA-DRD4 construct in two separate GNS lines (Figure 2.10B-C). This knockdown was accompanied by a significant reduction in proliferation compared to control shRNA-eGFP transfected cells (Figure 2.10D-E). Moreover, DRD4 knockdown cells are less sensitive to DRD4 antagonists (L-741,742) than control shRNA-eGFP transfected cells validating selectivity of the compounds to DRD4 (Figure 2.10F). Altogether, these results confirm the inferred role for DRD4 function in GNS cell growth.  A  B                                                                   C            D                                                                  E  F  Figure 2.10. Loss of DRD4 function phenocopies DRD4 antagonist’s effects. A. Western blot analysis for anti-DRD4 and β-actin in GNS (G362) cells transiently transfected with various short hairpins against DRD4 post 72h. B&C. Western blot analysis for anti-DRD4 and anti-β-actin in G362 and G482 cells, transiently transfected with shRNA against DRD4 and eGFP at 72h post transfection. D&E. Cell viability assay for G362 and G481 cells transiently transfected with shRNA-DRD4 and shRNA-eGFP measured over 5days. N=3, mean ±SEM. * p<0.005, ** p<0.0005, unpaired one-tailed t-test. F. Percent growth inhibition of G362 cells transiently transfected with shRNA-DRD4 and shRNA-eGFP treated with L741,742 for 3 days. N=3, mean ±STDEV.

2.4 Discussion

Neurochemicals and modification of their receptors influence neural stem cell behavior both during development and adulthood, and they may mediate neuronal activity-related modification of the production of neurons and glia.  The fact that the brain contains cell populations capable of extensive proliferative capacity also suggests a link between dysregulation of neurogenesis and the emergence of CNS tumors.  Brain tumors contain subpopulations of cells that resemble stem cells functionally, but which show dysregulated self renewal ability and differentiation capacity, and resistance to current therapies.  Understanding the stemness behavior of brain tumors is predicted to lead to new strategies to treat these treatment refractory diseases. The parallels between normal neural stem cells and their neoplastic counterparts suggest that modification of neurochemicals may also influence the proliferation and survival of brain tumor stem cells. In this thesis, I investigated whether neurochemical signaling activity influences the growth and survival of human GNS cells. This study represents the first systematic and unbiased interrogation of all neurochemical classes on human GNS cell growth. Of the 13 neurochemical classes, compounds modulating dopaminergic, serotonergic and cholinergic pathways predominantly affected GNS cells. Each of these neurochemical classes have been previously implicated in regulating NS cells during brain development and adult neurogenesis (Diaz et al., 1997; Hoglinger et al., 2004; Mohapel et al., 2005; Tong et al., 2014; Whitaker-Azmitia, 1991). Importantly, compound modulating sigma receptors have high activity in the screen and have relatively the highest gene expression level compared to other neurochemical receptors suggesting sigma receptors could potentially influence the GNS growth. I identified ten compounds (PNU-96415E, L-741,742, Ifenprodil tartrate, LY-165,163, MDL-72222, Tropanyl 3,5-dimethylbenzoate, N,N-Diethyl-2-(4-(phenylmethyl)phenoxy)ethanamine, (±)-Tropanyl-2(4-chlorophenoxy)butanoate, MG-624 and Ivermectin) that showed selective effective in NS and GNS cells compared to other non-NS control cells suggesting these ten compounds could have potential therapeutic value and may reveal new therapeutic targets. In particular, I have demonstrated that the specific DRD4 antagonists L-741,742 and PNU 96415E selectively inhibit the growth of GNS cells and reduce the colony forming potential of freshly dissociated GBM cells suggesting these compound may potentially target the stem cell population of the tumor. Both these compounds inhibited tumor growth in in vivo patient-derived xenograft model and are synergistic with current chemotherapeutic drug, temozolomide. Importantly, primary patient tumor and patient tumor derived GNS cells expressed functional DRD4 and knockdown of DRD4 phenocopied DRD4 antagonist, suggesting DRD4 could be potential target for GBM. Dopamine is a catecholamine synthesized by neurons predominantly in the midbrain region but have diffuse cortical projections. Dopamine signals are transmitted through five G proteincoupled receptors classified into two groups: the D1 like receptors (D1 and D5) that stimulate adenylate cyclase, and the D2 like receptors (D2, D3 and D4) that inhibit adenylate cyclase. Dopamine signaling is dysregulated in diverse neurological and psychiatric diseases including Parkinson’s, schizophrenia and drug addiction. In addition to serving a key neurotransmitter function, dopamine has been implicated in regulation of endogenous neurogenesis during brain development (Borta and Hoglinger, 2007; Diaz et al., 1997; Ohtani et al., 2003) and adult neurogenesis in the SVZ by activating D2-like receptors on transit amplifying progenitor cells (Coronas et al., 2004; Hoglinger et al., 2004; Lao et al., 2013). In post-mortem brains of Parkinson’s patients, a disease characterized by depletion of dopamine, a reduction of proliferating cells in the SVZ has been observed (Hoglinger et al., 2004). These reports suggest an important role for dopamine signaling in the regulation of normal neurogenesis. Little is known about how modification of dopamine signaling can affect brain tumor stem cell behavior. Here, this thesis demonstrates an important role for dopamine-DRD4 signaling in the growth and survival of GNS cells. The primary screen uncovered a number of DRD2 specific antagonists (thioridazine, trifluoperazine) as hits. Consistently, a recent genome wide shRNA screen in a GBM cell line (U87MG) identified a role for DRD2 in GBM growth(Li et al., 2014). However, DRD4 antagonists showed better selectivity for GNS cells in my screen compared to DRD2 antagonists (Figure S2.3).  Epigenetic suppression of DRD4 in pediatric CNS tumors has been reported (Unland et al., 2013), however DRD4 is not methylated in TCGA GBM samples and mutations in DRD5 and DRD3 genes occur in a small fraction of GBM samples (Brennan et al., 2013). The TCGA data on GBM gene expression show patients expressing highest DRD4 level have shorter survival than those patients expressing lowest DRD4 level. A similar pattern was seen with TH (tyrosine hydroxylase) expression, the rate-limiting enzyme for dopamine synthesis. Finally, dopamine receptor antagonists may have activity against other cancer stem cell types, including in leukemia (Sachlos et al., 2012) and lung cancer (Yeh et al., 2012). In conjunction with my findings, these studies suggest that dopamine receptor antagonists will prove to be useful probes for the study of GBM growth and survival, and potentially other cancer types. Finally, this interrogation on the influence of neurochemical space in GBM by small molecule approach revealed dopaminergic, serotonergic, cholinergic and sigma receptor as potential targetable pathways in GBM and in particular, the dopamine D4 receptor.

Supplemental Figures

Table S2.1. IC50(µM) of the ten NS selective compounds across four Non-NS control lines, three NS and three GNS lines and their fold selectivity.

Fold Selectivit y (NS selective )a Fold Selectivit y (GNS selective )b
Ifenprodil tartrate 50 50 50 25 12.5 12.5 3.12 3.12 3.12 0.39 128 8
LU 741,742,HCl >5 0 50 >5 0 12.5 12.5 12.5 12.5 6.25 6.25 1.56 >32 8
PNU 96415E >5 0 >5 0 >5 0 >50 25 25 12.5 6.25 3.12 1.56 >32 8
LYU165,163 >5 0 >5 0 >5 0 >50 3.12 6.25 3.12 3.12 6.25 3.12 >16
MDLU72222 50 50 >5 0 12.5 12.5 6.25 6.25 6.25 6.25 6.25 8
Tropanyl 3,5 dimethylbenz oate >5 0 >5 0 50 25 12.5 1.56 3.12 1.56 1.56 3.12 >32
N,NUDiethylU 2U[4U (phenylmethy l)phenoxy]eth anamine >5 0 50 25 25 0.78 1.56 3.12 0.78 0.78 1.56 >64
(±)UTropanylU 2U(4U chlorophenox y)butanoate 50 50 25 12.5 3.12 6.25 3.12 3.12 1.56 0.39 128
MGU624 25 25 25 6.25 1.56 0.78 0.39 0.78 0.39 3.12 64
Ivermectin 12. 5 12. 5 12. 5 12.5 0.78 3.12 1.56 1.56 1.56 1.56 16

a. NS selective= IC50 of BJ/IC50 of NS or GNS with lowest number. b. GNS selective= IC50 of any NS with highest number/IC50 of GNS with lowest number. “>” Indicates IC50 not seen even at highest concentration tested (50 µM) thus may be more selective than the number indicated. A                                                                          B           C 

Figure S2.1. DRD4 antagonists inhibits xenograft tumor growth in vivo

A. Growth curve of subcutaneous implanted tumor (G411cells) over period of time, measured from day 12 after implantation until end point when any tumor reached 17mm in size. Control n= 16, mean±SEM and PNU 96415E n=15, mean±SEM. B. Average tumor weight of control and PNU 96415E treated group at the end point. Control n= 16, mean±SEM, PNU 96415E n=15, mean±SEM. C. H&E staining of intracranial tumor (G362) at week 1 tumors (2 samples) compared to week 3 tumor at end point, when the tumor kills mice. Scale bar-100µm. A  Figure S2.2. Primary GBM patient tumor express DRD4. A.Immunohistochemistry staining of DRD4 in primary patient tumor samples. Scale bar-50µ m.  Figure S2.3. DRD2 specific antagonists do not show GNS selectivity. A. Percent cell viability of GNS (G362), NS (hf5205) and fibroblast (BJ) treated with thioridazine (0.39-50 µM) after 5 days incubation. B. Percent cell viability of fibroblast (BJ), NS (hf5205&hf5281) and GNS (G362&G411) on treatment with L-741,626 at various concentration (0.78-100 µM) after 5days of treatment. N=3, mean±SEM C. Percent cell viability of fibroblast (BJ), NS (hf5205&hf5281) and GNS (G362&G411) on treatment with haloperidol at various concentrations (0.78-100 µM) after 5days of treatment. N=3, mean±SEM Chapter 3 Characterization of mechanism of action for DRD4 antagonists in glioblastoma stem cells (I conducted all experiments and data analysis described in this chapter except gene expression data analysis done with the help of Veronique Voisin)

3.1 Introduction

Glioblastoma (GBM) exists as a functional hierarchy containing a subpopulation of cells with a stem cell phenotype that drives the tumor growth and contributes to treatment resistance. So, targeting this subpopulation of cells in addition to bulk tumor is required for a long-term disease. Currently there is no effective treatment other than the alkylating agent TMZ, which is effective transiently in a subset of patient with many side effects and even with that, the median survival remains 15 months. Therefore, a novel therapeutic approaches based on targeting stem cell population of tumor, in combination with TMZ is urgently required. This subpopulation of cells can be modeled from GBM patient tumors in a defined culture system, which share many similar characteristics to normal neural stem cell (NS) expressing SOX2 and Nestin, that we referred to as GBM derived neural stem cells (GNS). In the first chapter, I screened neurochemical library in these patient derived GNS cells and identified ten compounds that selectively killed GNS and NS cells compared to fibroblast that I termed as NS selective compounds. Interestingly, two (L-741,742 and PNU 96415E) out of ten NS selective compounds are dopamine D4 receptor (DRD4) antagonists and they showed more selectivity towards GNS compared to normal NS cells. Importantly, primary patient tumor samples and tumor derived GNS cells expressed functional DRD4 and loss of DRD4 function inhibits GNS growth. Therefore, understanding the mechanism of DRD4 antagonist activity in GNS cells will reveal the regulatory mechanism that makes GNS cells more susceptible to these compounds. Therefore, I hypothesized that characterizing the mechanism of action of DRD4 antagonists will reveal the vulnerability of GNS cells that we can exploit this for therapeutic application. DRD4 is a highly polymorphic gene consisting of variable number of 48bp tandem repeats (VNTR) in the exon 3 region that codes for intracellular loop 3(IC3)(Lichter et al., 1993; Van Tol et al., 1992). This repeat can vary from 2 repeats (D4.2) to 11 repeats (D4.11). D4.7 repeat is associated with ADHD, Tourette’s syndrome and novelty seeking. Although there are growing number of genetic association studies of DRD4 variants with many neurological disorders and behaviors, the molecular mechanism of these associations are not known. DRD4 is a D2-like receptor which not only inhibits cAMP production but also activates ERK pathway through transactivation of PDGFR-β, modulates calcium and potassium channels, and decreases functional GABAA receptor levels(Rondou et al., 2010). I therefore, took an unbiased approach to characterize the mechanism of DRD4 antagonism in GNS cells by performing genome wide expression array and phospho-kinase array to further understand this receptors mechanism.

3.2 Materials and methods

Gene expression profiling

GNS cells were treated with PNU 96415E (25 µM) for 0h (Control) 24h and 48h and cells lysed for RNA at each time points using RNeasy kit (Qiagen). RNA extracted from the samples was hybridized on Affymetrix Human Gene 1.0 ST arrays using standard protocol (TCAG, Toronto, Ontario, Canada). RMA background correction, quantile normalization and log2 transformation were applied to the CEL files using the Bioconductor affy package (R 3.0.1, affy package version 1.38.1). Batch correction was applied using ComBat function from sva (3.6.0) and gene annotations were retrieved using hugene10sttranscriptcluster.db (8.0.1). Genes were ranked based on the average log fold change (log FC) of the 2 treated GNS (G411 and G362) at 24h or 48h to vehicle (0h) samples. The data were analyzed using GSEA(Subramanian et al., 2005) with parameters set to 2000 gene-set permutations and gene-sets size between 8 and 500. The genesets included in the GSEA analyses were obtained from KEGG, MsigDB-c2, NCI, Biocarta, IOB, Netpath,  HumanCyc, Reactome and the Gene Ontology (GO) databases, updated October 14_2013 ( An enrichment map (version 1.2 of Enrichment Map software(Merico et al., 2010)) was generated for each comparison using enriched gene-sets with a False Discovery Rate < 0.02% and the overlap coefficient set to 0.5.

Western blots

Cells were lysed in a denaturing lysis buffer with protease and phosphatase inhibitor as described previously. Protein fragments were separated on sodium-dodecyl-sulfate (SDS) polyacrylamide gel by electrophoresis and transferred onto polyvinylidene difluoride (PVDF) membrane. PVDF membranes were blocked with 5%milk or bovine serum albumin (BSA) for one hour followed by incubation with primary antibody overnight at 4oC and secondary antibody for one hour at room temperature. Secondary antibodies were conjugated to horseradish peroxidase and detected with chemiluminescence. Western blots were performed using following antibodies; anti-DRD4 antibody at 1:750 (Millipore# MABN125), anti-phospho-p44/42 MAPK (Erk1/2) at 1:1000 (Cell Signaling), anti-p44/42 MAPK(Erk1/2) at 1:1000(Cell Signaling), anti-β-actin at 1:10,000 (Sigma), anti-LC3B at 1:1000 (Cell Signaling #3868), anti-p62 at 1:1000(BD Bioscience), antiLAMP1 at 1:2000(Developmental Studies Hybridoma Bank), anti-mono and polyubiquitinylated protein conjugates (FK2) at 1:1000 (Enzo Life Sciences), anti-phospho-PDGFRβ(Tyr751) at 1:1000(Cell Signaling), anti-PDGFRβ at 1:1000(Cell signaling), anti-phospho-S6 at 1:2000 (Cell signaling), anti-S6 at 1:2000 (Cell Signaling), anti-active caspase-3 at 1:200 (Abcam#ab2302).

Transmission electron microscopy

Cells were harvested, pelleted and fixed in 2% glutaraldehyde in 0.1M sodium cacodylate buffer, rinsed in buffer, post-fixed in 1% osmium tetroxide buffer, dehydrated in a graded ethanol series followed by propylene oxide, and embedded in EMBed 812 resin. Sections 100nm thick were cut on an RMC MT6000 ultramicrotome, stained with uranyl acetate and lead citrate and viewed in an FEI Tecnai 20 TEM.

Tandem mRFPUGFPULC3 reporter assay

Cells were transfected with mRFP-GFP-LC3 construct using Amaxa Nucleofector kit (VPG1004) and were seeded in 6 well plates with 25mm coverslips. After 24h post transfection, cells were treated with compounds for 24 or 48h as indicated. After treatment, coverslip with live cells were transferred onto an attofluor cell chamber with cultured medium and imaged using Quorum spinning disc confocal microscopy. Autophagy flux was assessed by counting cells mRFP GFP LC3 (yellow puncta) which represents autophagosomes, and mRFP GFPLC3 (red puncta) which represents autolysosomes.

Filipin staining

Cells were treated for 24 or 48h as indicated and visualized for cholesterol accumulation using filipin staining (Sigma-Aldrich, F9765). After fixing cells with 4% paraformaldehyde, cells were incubated with 1.5mg/ml glycine for 10min and incubated with 50mg/ml filipin for 2-3h at room temperature. Images were captured using Quorum spinning disc confocal microscopy.

DQURed BSA assay

Cells were grown in 6 well plates with 25mm coverslip. After treatment, cells were pulsed with DQ-Red BSA (Molecular Probes D-12051) for 1h and chased for 4h before imaging live cells using Quorum spinning disc confocal microscopy.

PhosphoUkinase array

A human phospho-kinase antibody array was purchased from R&D systems (Cat# ARY003). This array contains capture antibodies for 43 kinases in duplicate on nitrocellulose membrane. GNS and NS lines were treated with L-741,742(10 µM) and PNU 96415E (25 µM) along with DMSO control for 24h and cells processed according to the manufacturer’s protocol. Signal intensity was quantified using ImageJ.

Cell cycle analysis

GNS (G362 and G411) cells were incubated with L-741,742(10µM) and PNU 96415E(25µM) for 24h and 48h. Cells were accuatased and cell pellet washed and resuspended in 500µl of PBS. Cells were fixed with 4.5ml of cold 70% ethanol while vortexing gently and stored at 4oC for at least 2h. Wash cells with PBS and incubated with 0.5ml of propidium iodide (20µg/ml)and RNaseA (0.5µg/ml) and analyze through flow cytometry.

Apoptosis assay

Cells were seeded at 2500 cells/well in 96 well clear bottom black plates and treated with compounds for 24 or 48h as indicated. Apoptosis assay was performed using Apo-ONE Homogenous Caspase3/7 assay kit (Promega-G7790) as per protocol and plate measured for fluorescence after 4h incubation.

Accession number

The GenBank accession number for the PNU 96415E treated GNS microarray data described in this thesis is GSE62714

3.3 Results

3.3.1 Effect of DRD4 antagonism on global gene expression patterns

To determine the potential mechanism of action for a DRD4 antagonist on GNS proliferation and survival, I studied global gene expression profiles with and without DRD4 inhibition.  Two GNS lines (G362 and G411) were treated with PNU 96415E (25 µM) for 24h and 48h and analyzed for differential effects of PNU 96415E on gene expression. Gene set enrichment analysis (GSEA) was used to identify pathways enriched in differentially regulated genes upon PNU 96415E treatment. Genes that were down regulated at 48h were enriched in 172 gene sets that are highly connected, as categorized into 25 main biological functions including DNA replication, chromatin remodeling, DNA repair, cell cycle, and RNA splicing (Figure 3.1A). Overlap of the top down-regulated genes (fold change <-1.5) in both G362 and G411 cell lines revealed genes involved in DNA replication and cell cycle phase transitions (Figure 3.1C-D). For genes up-regulated upon PNU 96415E treatment, I observed enrichment in 45 gene sets (FDR =< 0.002) that comprised 14 main pathways including lipid/cholesterol biosynthesis, autophagic vacuoles and lysosomes (Figure 3.1B). Overlap of the top up-regulated genes (fold change>1.5) uncovered pathways involved in cholesterol biosynthesis after 24h and autophagic vacuole formation after 48h (Figure 3.1E-F). This expression analysis suggested that the genes involved in DNA replication and cell cycle progression were inhibited by DRD4 antagonism, while genes involved in lipid metabolism and autophagy were activated. A                                                                             B                   C                                                                               D 

E                                                                               F

   Figure 3.1. Effect of DRD4 inhibition with PNU 96415E on gene expression pattern. A& B Gene set enrichment map of pathways containing genes down regulated and up regulated respectively upon PNU 96415E treatment. Colored circles (nodes) represent gene-sets (pathways) that were significantly enriched in the comparison treated versus control samples (FDR <=0.002).

  1.               Enrichment map showing each gene-set and label contained in ‘DNA replication’ functional theme from the down regulated map (FDR <=0.001).  Right: Expression at 24h and 48h (logFC <= -1.5) of genes included in the DNA replication gene-set (GO:0006260, FDR =0.0005).
  2.              Enrichment map showing each gene-set and label contained in ‘Cell Cycle phase/ transition functional’ theme from the down-regulated map (FDR <=0.001). Right: Expression at 24h and 48h (logFC <= -1.5) of genes included in the Cell cycle gene-set (KEGG:HSA04110, FDR =0.0005 ).
  3.               Enrichment map showing each gene-set and label contained in ‘Lipid/ cholesterol/ biosynthesis’ theme from the up regulated map (FDR <=0.001). Left: Expression at 24h and 48h (logFC >= 1.5) of genes included in the Cholesterol biosynthesis gene-set (HUMANCYC%PWY66-5, FDR =0.0005).
  4.                Enrichment map showing each gene-set and label contained in ‘Autophagic vacuole/ lysosome’ theme from the up regulated map (FDR <=0.001). Left: Expression at 24h and 48h (logFC >= 1.5) of genes included in the autophagic vacuole gene-set (GO:0005776, FDR =


3.3.2 DRD4 antagonism causes massive accumulation of autophagic vacuoles and cholesterol

Prompted by the pronounced up-regulation of autophagy genes in response to DRD4 inhibition, autophagy status was assessed in GNS cells. Conversion of LC3-I (microtubule associated protein 1 light chain 3-I) to LC3-II serves as a hallmark for autophagosome formation. L741,742 (10 µM) and PNU 96415E (25 µM) treatment in GNS cells (G411 and G362) caused an increase in levels of LC3B-II consistent with accumulation of autophagosomes (Figure 3.2A). I also observed increased LC3B puncta in GNS cells upon treatment, with more than 50% cells showing large LC3B puncta after 48h, indicating the presence of autophagosomes (Figure 3.2B). Accumulation of autophagosomes was further corroborated by transmission electron microscopy. L-741,742 and PNU 96415E treatment in both G411 and G362 caused the formation of large autophagic vacuoles containing various cellular fragments. (Figure3.2C-D). A                                                                      B  C                                                                         D 

Figure 3.2. DRD4 antagonism causes accumulation of autophagic vacuoles

A. Western blot analysis for anti-LC3B and anti-βactin in G411 and G362 cells treated with PNU 96415E(25 µM) and L-741,742(10 µM) at indicated time points. B. Immunofluorescence staining for LC3B puncta in G362 and G411 cells treated with PNU 96415E (25 µM) and L741,742 (10 µM) at 48h. Scale bar =17µm. Quantification of LC3B puncta cells in each group (cells counted >200 cells). N=3, mean ±SEM, unpaired one-tailed t-test. C&D. TEM images showing large autophagic vacuoles in G362 and G411 cells treated with L-741,742(10 µM) and PNU 96415E (25 µM) compared to control DMSO at 48h. Arrows indicate enlarged autophagic vacuoles. Scale bar-100nm(C) 500nm(D). As genes involved in the cholesterol biosynthesis were also up regulated upon DRD4 inhibition and cholesterol accumulation is associated with autophagy impairment as reported in the context of Niemann Pick Type C disease (Vance, 2006), I next analyzed the cholesterol level in GNS cells with a filipin assay. Upon treatment with L-741,742(10 µM) and PNU 96415E(25 µM), GNS cells showed accumulation of cholesterol in large puncta, compared to the diffuse pattern observed in control cells (Figure 3.3). Together, these experiments revealed a massive accumulation of autophagic vacuoles and cholesterol accumulation in GNS cells after DRD4 receptor antagonism. 

Figure 3.3. DRD antagonism causes cholesterol accumulation

Filipin staining for free cholesterol in G362 and G411 cells treated with L-741,742(10 µM) and PNU 96415E(25 µM) compared to control DMSO at 48h. Scale bar-11µm To test whether autophagosome formation is specific to GNS cells, I assessed autophagosome formation by analyzing LC3-II levels of both fibroblasts and NS cells after treatment. While DRD4 antagonism at 48h did show a slight increase in LC3-II level in both cases, this was not to the same extent as seen in GNS cells (Figure 3.4 A-B). It is also interesting to note that GNS and fibroblast cells have very different basal levels of autophagy measured by LC3-II status under their normal growth conditions (Figure 3.4C), which likely explains differential vulnerability and dependence on autophagy for survival. A                                                                     C  B 

Figure 3.4. Autophagy levels in fibroblast, NS and GNS cells upon DRD4 antagonism

A&B Western blot analysis for anti-LC3B and anti- β-actin across different cell lines, GNS (G362&G411), NS (hf5205) and Fibroblast (BJ) upon treatment with L-741,742(10µM) and PNU 96415E(25µM) at 48h, and quantification of western blots (LC3-II/ β-actin). N=3, mean ±SEM. C. Western blot analysis for anti-LC3B and anti- β-actin across different cell lines, GNS (G362, G411&G179), NS (hf5205, hf5281&hf6539) and Fibroblast (BJ) at basal level.

3.3.3 Autophagosome accumulation is due to an inhibition in autophagic flux

An increase in LC3-II levels, and autophagosome number, can result from either the induction or inhibition of autophagic flux at a later stage in the pathway. Autophagy flux can be measured by assessing LC3-II turnover in the presence or absence of inhibitors of lysosomal degradation such as chloroquine. In chloroquine treated cells, an autophagy inducer would increase LC3-II levels, where as an autophagy blocker would not change LC3-II levels. In the presence of chloroquine, L-741,742 and PNU 96415E treatment did not increase LC3-II levels compared to control, despite the fact both drugs increased LC3-II levels when administered alone (Figure 3.5). These data suggest that the effect of DRD4 antagonism on LC3-II levels is a result of impaired flux. A                                                                                B                   

Figure 3.5. Accumulation of autohagic vacuole is due to block in autophagic flux

A&B. Western blot analysis for anti-LC3B, anti-p62 and anti-β-actin in G411 and G362 cells respectively treated with L-741,742(10 µM) or PNU 96415E (25 µM) in the presence and absence of chloroquine (30 µM) at 48h. Western quantification for LC3B-II was done using βactin as control. N=3, mean ±SEM. I then assessed autophagy turnover, first by assaying for the autophagy-specific substrate p62. As predicted for a block in autophagic flux, p62 accumulated along with the increase in LC3B-II in L-741,742 or PNU 96415E treated GNS cells (Figure 3.6 A-B). Consistent with an impairment in autophagy, I also observed an increase in undegraded ubiquitinated protein conjugates in treated cells (Figure 3.6 A-B) and further noted an increased level of LAMP 1 (lysosome associated membrane protein 1) (Figure 3.6 A-B) and LysoID reactivity (Figure S3.1), both of which indicate an increase in lysosome accumulation due to impaired autophagic flux. A                                                                                          B   

Figure 3.6. Accumulation of autophagy substrates upon DRD4 antagonism

A&B Western blot analysis for corresponding anti-LC3B with anti-p62, anti-LAMP1, anti- mono & poly ubiquitinated protein conjugates (FK2) and anti-β-actin in G411 and G362 cells respectively treated with L-741,742(10 µM) and PNU 96415E (25 µM) at indicated times points. I further validated flux inhibition by performing an image-based colocalization analysis of a tandem mRFP-GFP-LC3 construct (Kimura et al., 2007). In this experiment, GFP fluorescence is quenched by the acidic pH within a lysosome, allowing us to differentiate between autophagosomes (GFP positive and RFP positive – yellow) and autolysosomes (GFP negative and RFP positive – red). Treatment of transfected GNS cells with DRD4 antagonists showed an accumulation of yellow puncta, indicating autophagosome accumulation (Figure 3.7 A). As a negative control, I used rapamycin, an autophagy inducer, which showed a higher ratio of red puncta and as a positive control I used chloroquine, which lead to a higher ratio of yellow puncta (Figure 3.7A).  I also noted the ratio of yellow versus red puncta at 48h (Figure 3.7A) increased from that seen at 24h after treatment (Figure S3.2), suggesting an inhibition in autophagic flux over time during DRD4 antagonism. To validate the specificity to GNS cells, I further monitored flux inhibition in fibroblast and NS cells with the same assay. Interestingly, I did not see a dramatic increase in yellow puncta upon DRD4 antagonist treatment in either case (Figure 3.7B). A                                                                                                   B  Figure 3.7. Autophagic flux measurement using confocal analysis of mRFP-GFP-LC3 upon DRD4 antagonism. A. Confocal analysis of G411 cells expressing tandem mRFP-GFP-LC3 reporter treated with Rapamycin(500nM), Chloroquine(30µM), L-741,742(10µM) and PNU 96415E(25µM) at 48h, and quantification of ratio of red puncta indicating autolysosome(AL) versus yellow puncta indicating autophagosome(AP). N=3, mean± SEM** p value-0.0006 and 0.0003 in L-741,742 and PNU 96415E espectively. Scale bar-10µm. B. Confocal analysis of Fibroblast (BJ) and NS cells (hf5205) expressing tandem mRFP-GFP-LC3 reporter treated with L-741,742(10µM) and PNU 96415E(25µM) at 48h. Scale bar-7µm. To confirm that the impairment of the autophagy-lysosomal degradation pathway induced by L741,742 and PNU 96415E was mediated through DRD4, I assessed autophagic flux after shRNA knockdown of DRD4 in GNS cells. Consistent with this, increased levels of LC3-II was observed in DRD4 knockdown cells compared to sh-eGFP transduced controls (Figure 3.8).  This increase in LC3-II was accompanied by accumulation of p62, LAMP1 and ubiquitinated protein conjugates, all consistent with a block in autophagic flux (Figure 3.8, S3.3). I further validated this effect with another short hairpin from different source (Origene) and noted similar results (Figure S3.3B). Taken together, these data strongly suggest that the cytotoxicity observed after DRD4 antagonism in GNS cells is due to a block at a later stage of autophagy that results in massive accumulation of autophagic vacuoles. 

Figure 3.8. Autophagic flux inhibition in DRD4 knockdown cells

Western blot analysis for corresponding anti-DRD4, anti-LC3B, anti-p62, anti-mono and poly ubiquitinated protein conjugate (FK2) and anti-β-actin in transient transfected sh-DRD4 and sheGFP G362 and G411 cells post 72h.

3.3.4 Inhibition of autophagic flux after DRD4 antagonism is due to a disruption in lysosomal function.

To further understand the inhibition in flux, I assessed lysosomal function with a DQ BSA assay. DQ Red BSA is a derivative of BSA that is labeled with a self-quenched red fluorescent dye. Upon proteolysis by lysosomal proteases, the fluorescence is dequenched and can be detected by microscopy. Compared to control NS and fibroblast cells, GNS cells treated with DRD4 antagonists displayed a reduction in red puncta number per cell, suggesting very low dequenching of the fluorescent BSA due to compromised lysosomal function (Figure 3.9) These data demonstrate that autophagic flux impairment in GNS cells treated with DRD4 antagonists is due to an inhibition in lysosomal function. 

Figure 3.9. Disruption in endolysosomal function upon DRD4 antagonism

Confocal analysis for red puncta indicating dequenched BSA in GNS cells (G411) NS cells (hf5205) and fibroblast (BJ) treated with L-741,742(10µM) and PNU 96415E(25µM) at 48h. Quantification of red puncta per cell in each treatment. N=3, mean±SEM, unpaired one-tailed ttest. Scale bar- 7µm for GNS and 13µm for NS and fibroblast.  

3.3.5 DRD4 antagonism causes a disruption in PDGFRβUERK1/2 and mTOR signaling.

To determine how DRD4 receptor antagonism in GNS cells may mechanistically mediate the striking cellular effects observed, I studied the phosphorylation status of 43 kinases and substrates implicated in various signaling pathways in GNS cells versus NS cells using a dot blot assay. Cells were treated with L-741,742 (10µM) and PNU 96145E (25µM) for a period of 24h and protein lysates were harvested and assessed with a phosphoprotein antibody array (Figure 3.10 A). I identified 18 phosphoproteins that exhibited a decrease in phosphorylation upon treatment in GNS cells (Figure 3.10 C-D). ERK1/2 was one of the top hits in the array, with a 40% reduction compared to the untreated control. DRD4 is known to activate ERK1/2 by transactivation of platelet derived growth factor receptor β (PDGFRβ) (Gill et al., 2010; Oak et al., 2001). The selectivity of DRD4 antagonism to GNS cells was reflected in the much more modest changes in phospho-profile of NS cells after treatment (Figure3.10B). A                                                                 B  C                                                                D 

Figure 3.10. Effect of DRD4 antagonism on phospho-kinases activity

A&B. A dot blot containing 43 phosphoproteins in duplicates after exposure to lysate of G362 and hf5205 cells respectively treated with L-741,742 (10 µM) and PNU 96415E (25 µM) and DMSO for 24h. C. Signal intensity of each spot corresponding to each phosphoprotein (average of two spots) that changed upon treatment compared to DMSO. Signal intensity was quantified using ImageJ.  D. Ratio of control versus L-741,742 and PNU 96415E for each phospho-proteins that changed upon treatment in GNS (G362 cells). I validated the effect of DRD4 antagonism on ERK1/2 phosphorylation by western blot at various time intervals and observed a decrease in ERK1/2 phosphorylation over time in GNS cells but not in NS cells and fibroblasts (Figure 3.11A, S3.4), along with a concordant decrease in PDGFRβ phosphorylation (Figure 3.11B). I also confirmed that transient DRD4 knockdown decreased ERK1/2 phosphorylation in GNS cells compared to control shRNA-eGFP transfected cells (Figure 3.11C). These biochemical data suggest that the DRD4 antagonists act on target and that DRD4 regulates GNS cell growth in part through the PDGFRβ- ERK1/2 pathway. A                                                                         B  C  Figure 3.11. Downregulation of PDGFR-β/ERK1/2 signaling pathway upon DRD4


A. Western blot analysis for anti-phospho-ERK1/2, anti-ERK1/2 and anti β-actin in G362 cells treated with L-741,742 (10 µM) at indicated time points. B. Western blot analysis for antiphospho-PDGFR-β, anti-PDGFR-β and anti β-actin in G362 cells treated with L-741,742 (10 µM)  at indicated time points. C. Western blot analysis for anti-phospho-ERK1/2 and anti- ERK1/2 in G362 and G481 cells transiently transfected with sh-DRD4 and control sh-eGFP post 72h. I further validated the phospho-array data on down regulation of TOR (Figure 3.10C), which is well known to regulate autophagy both at early induction as well as later during lysosome biogenesis (Jewell et al., 2013). I observed a dramatic decrease in phospho-S6, a downstream effector of mTOR, in GNS cells after 48h of treatment suggesting downregulation of the pathway (Figure 3.12). There is no change in the Phospho-ERK1/2 and Phospho-S-6 in treated fibroblasts and NS cells (Figure 3.12) indicating that disruption in these signaling pathways confers GNS sensitivity to DRD4 antagonists.   Figure 3.12. Differential effect of DRD4 antagonism on ERK1/2 and TOR pathway in

different cell lines

Western blot analysis for anti-phospho-ERK1/2 and anti-ERK1/2, anti-phospho-S6 and anti-S6, and anti-β-actin in GNS cells (G362&G411), fibroblast (BJ) and NS (hf5205) cells treated with L-741,742(10 µM) and PNU 96415E (25 µM) for 48h.  

3.3.6 DRD4 antagonists trigger a G0/G1 phase arrest and apoptosis

As DRD4 antagonists inhibit proliferation of GNS cells accompanied by decreased expression of cell cycle genes (Figure 3.1), I sought to determine the effect of DRD4 antagonists on the cell cycle. Flow cytometric analysis of DNA content in G411 and G362 cells treated with either L741,742 or PNU 96415E revealed a G0/G1 arrest in a time dependent manner (Figure 3.13, S 3.5). 

Figure 3.13. DRD4 antagonist causes G0/G1 phase arrest

A. Cell cycle analysis of G411 and G362 cells after treatment with L-741,742 (10 µM) and PNU 96415E (25 µM) at 48h. After 48h of treatment with DRD4 antagonists, there was an induction of caspase 3/7 activity (Figure 3.14A) suggesting an increase in apoptosis. I also observed that the caspase-3 and -7 activation was seen only after 48h of treatment indicating autophagic flux inhibition precedes apoptosis. This data is further corroborated by the presence of cleaved PARP at 48h (Figure 3.14B) and an increase in active caspase 3 staining (Figure 3.14C). Combined, these data suggest DRD4 inhibition leads to cell cycle arrest followed by apoptosis. A                                                                   B  C 

Figure 3.14. DRD4 antagonists causes apoptosis

A. Fluorescence read out for caspase 3/7 activity in G411cells treated with L-741,742 (10 µM) and PNU 96415E(25 µM) for 48h, and  doxorubicin (1µM) for 24h as positive control. N=3, mean±STDEV. B. Western blot analysis for apoptosis marker anti-cleaved PARP in G362 treated with L-741,742 (10 µM) and PNU 96415E (25 µM) at indicated time points, and doxorubicin(1µM) at 24h as control. C. Immunofluorescence staining and quantification for antiactive caspase-3 in G411 and G362 cells treated with L-741,742 (10 µM) and PNU 96415E (25 µM) for 48h.Scale bar- 11µm. Furthermore, GNS cells treated with an autophagy inhibitor (chloroquine) or ERK1/2 pathway inhibitor (PD 0325901) could mediate G0/G1 arrest (Figure S3.5B), demonstrating the connection between DRD4 antagonism-autophagy inhibition and cell cycle arrest.

3.4 Discussion

There is growing evidence that suggests that dopamine has a role in regulation of endogenous neurogenesis during brain development and adult neurogenesis.  However, little is known about how modification of dopamine signaling can affect brain tumor stem cell behavior. Here this thesis demonstrates an important role for dopamine-DRD4 signaling in the growth and survival of GNS cells mediated through modulating the autophagy-lysosomal system. I have demonstrated that inhibition of dopamine D4 receptor impairs autophagy/lysosomal degradation pathway resulting in massive accumulation of autophagic vacuoles accompanied with an increased accumulation of autophagic substrate p62, ubiquitinated protein, cholesterol and lysosomal cargo, arresting GNS cells at G0/G1 followed by apoptosis. I further demonstrated that this impairment in autophagic flux is due to the disruption in the endolysosomal function. Furthermore, inhibition of DRD4 suppressed its known downstream PDGFR-β /ERK signaling pathway, where inhibition of PDGFR-β is reported to inhibit GBM self renewal and tumor growth further supporting the data(Kim et al., 2012). Autophagy is a dynamic cellular process that involves lysosomal degradation of unnecessary or damaged cellular components into cellular constituents, which then recycles back into cytosol to maintain cellular homeostasis. The autophagy-lysosomal degradation pathway system appears to be critical for the progression and/or maintenance of many cancer types(Kenific and Debnath, 2014). For example, autophagy is important for breast cancer stem cell maintenance (Choi et al., 2014; Gong et al., 2013) and pancreatic cancer stem cells(Yang et al., 2015). In a number of cancer studies, including for GBM, autophagy inhibition with chloroquine appears to augment the efficacy of anticancer therapies (Sotelo et al., 2006). My data suggests that autophagy plays an essential role in GNS cell growth and survival, and that dopamine signaling through the DRD4 receptor is required to maintain the autophagy-lysosomal degradation system. I have shown that GNS cells are vulnerable to disruption of this pathway compared to fibroblasts and NS cells. I hypothesize that potential differences in the basal activity and dependence of GNS cells on autophagy-lysosomal pathway compared to normal NS cells and fibroblasts may account for the susceptibility of GNS cells to DRD4 antagonists.  Consistently, it has recently been reported that GBM cells are sensitive to the vacuolization agent vacquinols (Kitambi et al., 2014) and a breast cancer stem cell selective compound salinomycin is also reported to confer selectivity through inhibiting autophagic flux (Yue et al., 2013). Thus, identification of DRD4 antagonists as GNS selective compounds revealed the autophagy-lysosomal system as a vulnerable and important system for GNS cell survival and maintenance, and modulation of dopamine DRD4 signaling may be of therapeutic benefit in GBM patients. Impairment of the autophagy-lysosomal pathway is also intimately associated with neurodegenerative diseases in which neurochemical signaling is dysregulated, including Parkinson’s Disease, Alzheimer’s Disease and Niemann-Pick Type C1 disease (Nixon, 2013; Sarkar et al., 2013).  Autophagy has been implicated in the maintenance of adult NSCs (Wang et al., 2013; Yazdankhah et al., 2014) and suppression of autophagy in NSCs during development in mice causes neurodegenerative diseases (Hara et al., 2006). This study suggests new insights into the role of DRD4 signaling in neurological disorders such as ADHD, bipolar disorder and schizophrenia, which are reported to be associated with DRD4 polymorphism (LaHoste et al., 1996; Rondou et al., 2010).  As I have shown in GBM, impaired DRD4 function in these neurological diseases could disrupt autophagic flux by a similar mechanism. Further investigation of the link between DRD4 signaling, autophagy impairment and cell survival in neurodegenerative diseases, cognitive brain disorders and GBM is warranted. In particular, further exploration of the role of neurochemical signaling pathways specifically dopaminergics, serotonergics and cholinergics pathways in GBM may yield novel therapeutic approaches to treat this intractable disease.

Supplemental figures

Figure S3.1. Increased number of lysosome upon DRD4 inhibition

Fluorescence staining of lysoID-Red (lysosomes marker) in live GNS cells (G411 and G362) treated with L-741,742 (10 µM)   and PNU 96415E (25 µM) at 48h treatment. Scale bar-25µm.  Figure S3.2. Autophagy flux measurement at 24h. Confocal analysis of GNS cells (G411) expressing tandem mRFP-GFP-LC3 reporter treated with Rapamycin(500nM), Chloroquine(30µM), L-741,742(10µM) and PNU 96415E(25µM) at 24h, and quantification of ratio of red puncta (RFP -GFP-LC3) indicating autolysosome(AL) versus yellow puncta (RFP -GFP -LC3) indicating autophagosome(AP). Scare bar-10µm. A                                                                       B G362                         G411                     Figure S3.3. Autophagic flux inhibition in DRD4 knockdown cells. A. Western blot analysis for anti-DRD4, anti-LC3B, anti-p62, anti-LAMP1, anti-mono and poly ubiquitinated protein conjugates (FK2) in an independent experiment G362 and G411 cells transiently transfected with shDRD4 and sh-eGFP (Open Biosystems) post 72h. B. Western blot analysis for anti-DRD4, anti-LC3B, anti-p62, anti-mono and poly ubiquitinated protein conjugates (FK2) in G411 cells transiently transfected with an another set of shRNA against DRD4 from Origene; shDRD4-1, shDRD4-2, and non-effective scramble shRNA, post 72h.

  1. B

 C  Figure S3.4. Phospho-ERK1/2 downregulation upon DRD4 inhibition. A-B. Western blot analysis for anti-phospho-ERK1/2, anti-ERK1/2 and anti β-actin in hf5205 and G362 cells respectively treated with L-741,742 (10 µM) at indicated time points. C. Western blot analysis for anti-phospho-ERK1/2, anti-ERK1/2 and anti β-actin in G411 cells treated with L-741,742(10 µM) and PNU 96415E(25 µM) at indicated time points. A  B  Figure S 3.5. G0/G1 arrest upon DRD antagonism, MEK inhibition and autophagic flux

inhibition by chloroquine

Cell cycle flow cytometry analysis of GNS cells (G362 and G411) treated with L-741,742(10 µM) and PNU 96415E (25 µM) at 24h treatment. B. Cell cycle flow cytometry analysis of GNS cells (G362 and G411) treated with Chloroquine (30 µM) and PD 0325901 (1 µM) at 24h treatment. Chapter 4 Directing neural stem cell fate with neurochemical modulators

4.1 Introduction

Neural stem cells (NSC) are multi-potent cells that have the capacity to self renew and differentiate into all major cell types of the CNS namely neurons, astrocytes and oligodendrocytes. In the adult brain, neurogenesis occurs throughout life in the two neurogenic zones (SVZ and SGZ), providing new neurons and glia to maintain brain homeostasis and to respond to disease or injury(Kempermann and Gage, 1999). The neurogenic compartments of the brain are incredibly dynamic and an increasing number of physiologic and disease associated factors regulate NSC proliferation and differentiation(Ming and Song, 2011). Therefore, understanding the biology and regulatory mechanism of NSC and neurogenesis holds a great promise for the treatment of neurodegenerative diseases and understanding many developmental neurological disorders. Neurochemicals and their receptors aside from their role in synaptic communication are now starting to be appreciated to influence NSC both during development and adult neurogenesis, and they may mediate neuronal activity-related modification of the production of neurons and glia. NSCs express variety of neurochemical receptors and responds to neurochemicals signaling (Berg et al., 2013). Neurochemicals such as dopamine, serotonin and acetylcholine have demonstrated to activate NSC in SVZ region(Freundlieb et al., 2006; Hoglinger et al., 2004; Paez-Gonzalez et al., 2014; Tong et al., 2014). While GABA, the major inhibitory neurochemical, promotes NSC quiescence by blocking cell cycle progression(Fernando et al., 2011), limits NSC division through negative feedback mechanism in SVZ(Liu et al., 2005), and in SGZ, reducing GABA signaling results in NSC activation and symmetrical division(Song et al., 2012). Small molecule screening is a complimentary approach to genetic screening in probing biological systems to reveal novel genes and pathways. High content screening is a powerful tool to interrogate the potential of small molecules in many parameters simultaneously. This approach is ideal for NSC differentiation screen where we can interrogate proliferation, as well as differentiation into neurons, astrocytes and oligodendrocytes, as well as cell morphology, all parameters in one screen (Figure 4.1). Identification of a compound that can direct NSC fate into specific neuronal lineages will have a great therapeutic impact in treating neurodegenerative disease and understanding many neurological disorders such as Parkinson’s disease, Alzheimer’s disease, and autism. In chapter 2, I identified ten compounds that showed selective effects towards NS and GNS cells compared to fibroblast that I called as NS selective compounds. Neuromodulation of serotonergic, dopaminergic and cholinergic classes have been implicated in regulating adult neurogenesis. Therefore, I hypothesized that interrogation of neurochemicals in directing NSC fate would identify a number of compounds that would reveal novel NSC regulation that can be exploited for therapeutic purposes.  Figure 4.1 Schematic for interrogation of neurochemicals in human NS cells.

4.2 Methods

High content screen

Human NS (hf5205) cells were seeded in 384-well black µ clear bottom coated with laminin and poly -L- ornithine(PLO) at a density of 1X103. The compound library was added at approximately 5µM using Biomek FXP automation workstation with pin tool. Cells were incubated in Step-I medium (N2, B27, 5ng/ml FGF) without EGF for first four days and replaced with step-II medium (1:1 Neurobasal:Neurocult, B27,1/4 N2) without FGF for next four days. At day 8th, cells were fixed with 4% paraformaldehyde (PFA) and blocked with 5% normal goat serum (NGS) and incubated with primary antibody against anti-Beta III tubulin (1:250 MAB1637),, anti-GFAP(1:1000, DAKO) and DAPI for overnight at 4oC. 25 images were taken from different areas of each well by Evotek Opera system and data were analyzed using Acapella software program using average signal intensity with set cut offs using DAPI stain as reference for cells. BMP4 and BIO (GSK3b inhibitor) were used as positive controls for differentiation.


Human NS cells differentiated on coverslips were fixed with 4% PFA and permeabilized with 0.3% Triton X100, block with 5% goat serum, and incubated with primary antibody over night at 4oC. Antibodies Beta-III tubulin (1:250 MAB1637), GFAP (1:1000, DAKO), VGLUT1(1:1000, Synaptic system) GABA (1:1000, Sigma). Appropriate fluorescent-conjugated secondary antibodies were used at 1:500 for 1h at room temperature. Coverslip mounted with fluorescent mounting medium (DAKO) containing DAPI (1:1000) and cells imaged using Leica microscope.


Total RNA from cells was isolated using RNeasy kit (Qiagen) and 1µg of total RNA was used for cDNA generation. Relative abundance of specific mRNA was then assessed using SsoFast Evagreen Supermixes (Bio Rad) run on PTC-200 thermal cycler (Bio Rad). The relative abundance of each gene expression was assessed using the delta-delta-CT method with value normalized to their respective GAPDH levels. The following primers were used for qPCR: CTIP2-F: TCC GAG CCG GTG GAG ATC GG, CTIP2-R: GCA CGG CCC TGC AAT GTT CTC, FOXG1-F: CGG CTC CCT CTA CTG GCC CA, FOXG1-R: ATG GGG TGG CTG GGG TAG GC, VGLUT1-F: GGC CAG ATC GCG GAC TTC CT, VGLUT1-R: CAA CAG CAG CGT GGC TTC CA, NEUROG2-F: ACC ACA AGC AGC TTC GCG TTA, NEUROG2-R: CGG GTC TCG TGT GTT GTG GTG

Single cell qPCR (Fluidigm)

Differentiated human NS (hf5205) at week 3 was accutased and cells were resuspended in NS medium containing 1µg per ml of propidium iodide(PI) and filtered through 40µm nylon cell strainer followed by live single cell sorting into 96 well qPCR plate on BD cell sorter. Cells were sorted into 10µl of preamplification mix containing 40nM of all primers for the 48 genes, and the following components of CellsDirect One-Step qRT-PCR kit(Invitrogen) as directed in protocol: 2X reaction mix, SuperScript III RT/platinum Taq mix. After sorting, samples were reverse transcribed and preamplified for 18 cycles. Preamplified samples were diluted(2X) with TE buffer and stored at -20oC. Sample and assay, primer preparation for Fluidigm Dynamic arrays was done according to the manufacturer’s recommendation. Samples were mixed with 2X assay loading reagent (Fluidigm Corp), 20X EvaGreen and TaqMan gene expression master mix. The Fluidigm Dynamic arrays were primed and loaded on the IFC controller and qPCR experiments were run on a Biomark system for genetic analysis. The primer sequences for Fluidigm experiment are taken from a published paper in Nature Medicine by Dolmetsch RE group (Pasca et al., 2011).

Gene expression profiling

Hf5205 cells were differentiated with 3µM of L-741,742 in two-step growth factor withdrawal protocol and RNA was extracted using RNAeasy kit(Qiagen) for control and treated samples at 3 weeks treatment. RNA extracted from the samples was hybridized on Affymetrix Human Gene 1.0 ST arrays using standard protocol (TCAG, Toronto, Ontario, Canada). RMA background correction, quantile normalization and log2 transformation were applied to the CEL files using the Bioconductor affy package (R 3.0.1, affy package version 1.38.1). Batch correction was applied using ComBat function from sva (3.6.0) and gene annotations were retrieved using hugene10sttranscriptcluster.db (8.0.1). Genes were ranked based on the average log fold change (log FC) of the L-741,742 differentiated human NS and control differentiated cells at 3 weeks.

4.3 Results

4.3.1 Identification of compound that directs NSC fate

Identification of compounds that can direct NSC fate into specific neuronal lineages could have a great therapeutic impact in regenerative medicine, potentially treating many neurologic diseases through improved generation of appropriate cell types. To achieve this goal, I performed a differentiation screen in human NS cells by high content imaging. High content screening is both quantitative and qualitative. I employed it to measure differentiation of human NS into neurons and astrocytes, and proliferation. The default differentiation protocol for NS cells is performed using two-step growth factor (EGF and FGF) withdrawal protocol, where we sequentially remove EGF in the first week and FGF in the second to third week. This two-step growth withdrawal protocol generates 5-10% early immature neurons and 10-20% astrocytes. I miniaturized the differentiation screen for human NS in an accelerated assay over 8 days in 384 well plates, first four days without EGF and second four days without FGF. Neuronal differentiation was measured by immunostaining with Beta-III tubulin, a pan neuronal marker and astrocytic differentiation by GFAP staining, and proliferation by DAPI. 25 images were taken from different areas in each well by Evotek Opera system and data were analyzed using Acapella script program using average signal intensity with set cut offs (Figure 4.2). BMP4 and BIO (GSK3b inhibitor) were used as a positive control for differentiation. 

Figure 4.2. An illustration of high content differentiation screen protocol

Based on previous screen from chapter 2, I once again interrogated the neurochemical potential for differentiation in NS cells by high content imaging as described. I tested if neurochemicalmodulating compounds can push NS cells to differentiate in 8 days where a default differentiation protocol takes 2-3 weeks. In this high content differentiation screen, I found 22 compounds that showed increase in beta III tubulin staining compared to control indicating neuronal differentiation and 17 compounds that showed an increase in GFAP staining indicating astrocytic differentiation (Figure 4.3). Since the background for this screen was high, I validated a few of the top hits and confirmed five compounds namely McN A-343, (-)-N- Phenylcarbamoyleseroline, ifenprodil tartrate, L-741,742 and PNU 96415E that promote neuronal differentiation ranging from 12-40% increase when compared to control (2-9%) (Figure 4.4). Three of these compounds ifenprodil tartrate, L-741,742 and PNU 96415E were initially identified as the ten NS selective compounds and showed differentiation potential, which further validated my previous findings.   

Figure 4.3. Summary of high content differentiation screen data

 Figure 4.4. Validation of hits that promotes neuronal differentiation in 2-step growth withdrawal condition.  

4.3.2 Compounds specifying mature neuronal lineages

I next determine if these five validated compounds have the potential to specify specific mature neuronal lineages. I performed immunocytochemistry for a series of antibodies against different classes of mature neurons based on their neurotransmitters including glutamatergic (VGLUT1), dopaminergic(TH), GABAergic(GABA) and cholinergic (AChT, ChAT). Interestingly, three of the compounds L-741,742, PNU 96415E, Ifenprodil tartrate showed positive staining for VGLUT1 indicating glutamatergic neuronal commitment (Figure 4.5 & Table 4.1). VGLUT1 is a vesicular glutamate transporter required for transport of glutamate into synaptic vesicles of glutamatergic neurons.  Two of the hit compounds L-741,742 and PNU 96415E are dopamine D4 receptor antagonists and led to an increase in VGLUT1 positive staining (20-40%) compare to control (0-5%). To confirm DRD4 antagonist potential in promoting glutamatergic neurons, I further tested L-741,742 in two additional human NS lines and showed its potential to promote VGLUT1 positive glutamatergic neurons in other NS cells as well (Figure 4.6A). Additionally, L-741,742 differentiated cells also showed positive staining for GABA indicating GABAergic neurons (Figure 4.6B). I did not see positive staining with any of the other markers tested. Thus L-741,742 promotes neuronal differentiation giving rise to mature neurons of both glutamatergic and GABAergic lineages. 

Figure 4.5. Compounds promoting glutamatergic neurons

Percent quantification of VGLUT1 positive in Hf5205 differentiated with L-741,742, PNU 96415E and Ifenprodil.         Table  4.1  Summary  of  compounds  promoting  specific  neuronal lineages.  A  B  Figure 4.6. L-741,742 promotes

With Our Resume Writing Help, You Will Land Your Dream Job
Resume Writing Service, Resume101
Trust your assignments to an essay writing service with the fastest delivery time and fully original content.
Essay Writing Service, EssayPro
Nowadays, the PaperHelp website is a place where you can easily find fast and effective solutions to virtually all academic needs
Universal Writing Solution, PaperHelp
Professional Custom
Professional Custom Essay Writing Services
In need of qualified essay help online or professional assistance with your research paper?
Browsing the web for a reliable custom writing service to give you a hand with college assignment?
Out of time and require quick and moreover effective support with your term paper or dissertation?