Amino Acid Racemization (AAR) Applications

1. Introduction

Amino acid racemization (AAR) has been applied extensively as a method of relative and quantitative dating by evaluating the degree of postmortem conversion of the chiral forms of amino acids from the biological (L-enantiomers) to the nonbiological (D-enantiomers). In modern samples, the L-amino acid configuration is almost exclusively present i.e. D/L=0. Amino acids with a single chiral carbon center—including aspartic acid, valine, and leucine—undergo racemization until dynamic equilibrium is reached at D/L=1.

For the past 60 years, the development and diverse applications of amino acid racemization has garnered considerable interest and a large body of literature on the subject has been amassed. AAR dating has been suggested as a cost-effective and rapid preliminary dating technique to identify qualitative relative age information in the analysis of a large number of samples, with the possibility of independent calibration by a separate geochronological technique. As a geochronological method, AAR dating has been widely employed as a standard chronostratigraphic tool in Quaternary research. AAR dating has been applied to a diverse array of fields ranging from geology and planetary science, paleontology, archeology, history, forensic science, anthropology, and astrobiology. Current areas of amino acid interest include AAR dating, preservation of ancient proteins and amino acids, diagenesis of amino acids both through geologic time and on short time scales, stable isotope chemistry of amino acids, the stability of amino acids at extreme conditions, and extraterrestrial amino acids. Following the seminal compilation on the topic Biogeochemistry of Amino Acids [1], and advances in the field were again reviewed in Perspectives in Amino Acid and Protein Geochemistry [2]. A number of review articles on the topic of AAR dating were produced in the years between, including those authored by Wehmiller [3-5], Masters [6], Miller and Brigham-Grette [7], Mitterer [8], Murray-Wallace [9], Rutter and Blackwell [10], Kaufman and Miller [11], Johnson and Miller [12], and Wehmiller and Miller [13], among others. In 2013 a topical issue of the journal Quaternary Geochronology was published on AAR, edited by Penkman and Kaufman [14].

2. Early Development of AAR and Bone-dating Controversy

The preservation of amino acids most stable during high temperature laboratory heating experiments in biomineral fossils over extended periods of geological time was first proposed by Philip Abelson from the Carnegie Institute in Washington DC [15]. The fossil samples in this study included fossilized fish from the Devonian Period, more than 300 million years old. Hare and Abelson [16] in 1968 first reported the AAR reaction as a method of age assessment by separating L-isoleucine from its epimer, D-alloisoleucine, and demonstrating a correlation between increased concentrations of D-alloisoleucine with increasing fossil age. The geochronological application of AAR gained greater popularity in the late 1960s and 1970s with influential works from Hare and Mitterer [17, 18], Bada with others [19-21], Bada and Prostch [22], Kvenvolden [23], Davies and Treloar [24], and Williams and Smith [25], among others. Starting in the early 1970s, AAR gained new popularity when studies (e.g. Wehmiller and Hare [26], Bada et al. [27]) correlated burial depth with the D/L ratio in marine sediments. The calibrated AAR geochronology method as a dating tool was brought to light due to the efforts of Bada and coauthors [19] in 1972, particularly when applied to Quaternary bones, who proposed that AAR included several advantages over radiocarbon dating, such as significantly smaller sample size requirements and a greater age range [22].

However, confusion and controversy struck in 1974. On the basis of aspartic acid racemization dating, a study on California Paleo-Indian skeletons suggested humans were present in North America at least 50,000 years ago. Perhaps most sensational was the estimate of ca 70,000 years to a human skeleton from Sunnyvale, California [28, 29]. These results were sensational, as the previous estimate was 12,000 years. A human skull called the Del Mar skull used in the study which was recovered in 1929 from a sea cliff near San Dieguito River in Southern California, USA, even appeared on an episode entitled “Earth Visitors” of the television program In Search Of. The episode quoted the Bada et al. [29] study’s age estimate that the skull’s age was 48,000 years old, suggesting that ancient civilizations had existed long before the arrival of Native Americans. However, Bischoff and Rosenbauer [30] of the United States Geological Survey applied radiometric uranium series dating (thorium-230 and protactinium-231 decay systems) of human bone samples from the Del Mar and Sunnyvale sites to constrain the skeletal remains to 11,000 and 8,300 years, respectively. The results of Bischoff and Rosenbauer revealed a crucial discrepancy; the Del Mar and Sunnyvale humans were relative newcomers compared to the estimates of 48,000 and 70,000 years based on AAR.

Other high-profile reports containing controversial age estimates appeared while the technique was still being developed, including Bowen and Sykes[31] and Bada et al. [29, 32]. The combination of spectacular results with deep inconsistences led to skepticism and distrust within the community. Many considered AAR “some kind of joke,” in the critical view of paleoanthropologist Milford Wolpoff [33]. In 1974 Hare wrote a critical review in the MASCA newsletter (Museum Applied Science Center for Archaeology Newsletter) of the University of Pennsylvania of the inconsistencies to-date and stated that bones – especially bones obtained from warm environments—were unreliable substrates for AAR testing. Hare even lamented that this controversial body of work had damned the whole process, even when applied correctly, due to “that nonsense with bones” [33].

Prior to the controversial 1973 [22] and 1974 [29] papers, Bada and coauthors published aspartic AAR data on samples obtained almost exclusively from Old World archaeological or paleontological contexts. In 20 cases a direct comparison could be made between the calibrated AAR bone dating data and the age determined by radiocarbon dating (Bada [29] and references therein). The average age difference between radiocarbon and the AAR ages was in reasonable agreement at 1,500 years, with a range of 100 to 5,000 years. Furthermore, the racemization half-life of aspartic acid in bone at 20°C was reported to be ca 15,000 years in 1975 by Bada & Schroeder [34]  and thus offered one of the fastest rates of racemization among stable amino acids. Bada and coworkers utilized aspartic acid, seeing this as a potential solution to the difficulty in establishing the temperature dependency of isoleucine. The aspartic acid D/L ratio and the determined age of the calibration sample apparently allowed a first order rate constant (k_asp) to be calculated. Using this k_asp value and the aspartic acid D/L ratios, the AAR age could then be calculated for other bones with the same temperature history by the following equation

tyrs=ln⁡1+D/L1-D/L-ln⁡1+D/L1-D/Lt=02kasp

Considering the close agreement of the two dating techniques and the agreement of the temperatures calculated from the rate constant with average air temperatures of the archeological sites at the time, Bada argued that there were only two critical variables. Time and environmental temperature critically impacted amino acid racemization in bone, provided that the bone samples were non-contaminated and had not experienced anomalous heating [29]. The calibration value for the Del Mar skull was based on the skeleton from Laguna Beach, California, which had been previously radiocarbon dated to 17,150

±1,470 years before present using conventional beta-decay counting. The extent of racemization indicated by the aspartic acid D/L ratio in the Laguna Beach skeleton and the calibrated age, were used to date both the Del Mar among other skeletons, including the Los Angeles (Baldwin Hills) skeleton to an estimated Pleistocene age. The AAR age of 26,000 years for the Los Angeles Man skeleton was in reasonably good agreement with the radiocarbon age of >23,600 years [35] determined by the conventional radiocarbon method [36].  Thus, the calibration had been considered reasonable to determine AAR dates of other California Paleoindian skeletons found at Del Mar and Sunnyvale. Despite the apparent success of AAR dating bones from Old World contexts, the application to New World skeletal samples generated intense controversy regarding the antiquity of humans in the New World and the validity of AAR as a dating technique.

These controversial findings prompted new efforts to date the samples with other techniques, including R.E. Taylor [35], as well as others [30, 37, 38]. A decade after the original publication using AAR to date the skeletal remains found near La Jolla, Del Mar, and Sunnyvale, Bada et al. [32] revisited the former study using accelerator mass spectrometry (AMS) to determine and reassess the radiocarbon ages of the amino acids.  A portion of the original amino acid extracts containing 5-10 mg amino acids from the controversial 1984 study had been preserved through freezing or as dried residue [32], as well as other additional amino acids isolated from previously radiocarbon dated skeletons, were subjected to AMS analysis. While the original AAR studies suggested that people were present in North America during the Upper Pleistocene, the AMS data indicated that some of the Californian skeletons were instead Holocene. With the newfound viability of AMS analysis, a number of other studies re-evaluated the accuracy of AAR bone dating [37, 39-41]. In every study, the age determined by radiocarbon AMS was significantly lower, as much as by an order of magnitude, than the original AAR estimated age. The age from the Laguna skeleton was found to be significantly lower, merely 5,000 years ago as opposed to 50,000, and the Los Angeles skeleton ca 3,500 years. As a result of these AMS re-evaluations it has been established that no directly dated human skeleton from North America has an age of 11,000 years or greater.

Despite the waxing and waning of the method’s credibility, investigations on the mechanisms of racemization, origins of homochirality, and pioneering applications of the technique has continued to the present-day.

3. Chromatography section / separations

Since the amino acids and the D-enantiomer in particular usually occur at trace levels in archaeological or paleontological samples, methods for enantiomeric separation and detection need to be sensitive, and as always when numbers of samples exist, the methods should be fast, inexpensive, and rugged. For example, a recent article concerning the identification of artificially aged (forged) silk using amino acid enantiomers as biomarkers used a capillary electrophoresis- mass spectrometry (CE-MS) method using 30 mM (+)-(18-crown-6)- 2,3,11,12-tetracarboxylic acid (18-C-6-TCA) in water as a background electrolyte [42]. Figure 1 shows electropherograms and samples of the silk. This method was fast, reliable, relatively inexpensive and only used micrograms of sample. Moini and Rollman have reported a portable (battery-operated) CE-MS to be applied to future on-site analyses of amino acid racemization [43].

In general, amino acid enantioseparation methods can include capillary electrophoresis (CE)  [44, 45], gas chromatography (GC) [46-52], micellar electrokinetic chromatography (MEKC) [53, 54], high-performance liquid chromatography (HPLC) [55, 56], and ultra-performance liquid chromatography (UPLC) [57, 58] as well as others (references provided here are for example purposes only, as many others exist.) However, HPLC separation methods appear to prevail in the literature for AAR archaeological or paleontological dating purposes.

Both direct and indirect chiral separation methods can be used in the determination of amino acid enantiomers. Indirect LC methods require appropriate chiral derivatization agents that react to form two diastereomers which can then be separated on achiral stationary phases. Direct methods require a suitable chiral selector, either as a chiral stationary phase or as a chiral mobile phase additive. Derivatization of the amino acids is also utilized to improve the selectivity of the separation and/or the sensitivity of detection [55]. Commonly used in situ derivatizations used in liquid chromatography with mass spectrometry detection were reviewed in 2016 by Baghdady and Schug [59].  Finally, the advantages and disadvantages of direct and indirect chiral separation methods including the possibility of racemization are clearly summarized in the work of Ilisz et al. [55].

AAR use in dating has a number of difficulties due to the nature of the samples. As succinctly stated by Torres et al., “Many of the problems that we have encountered have arisen from not considering what is referred to as the site effect, which includes taphonomy, time-averaging, anthropogenic influence and geological evolution” [60]. Contaminants from bacteria, seepage from the external environment, etc., would “lower and distort the expected extent and pattern of racemization for the various amino acids” [61] as illustrated in  Figures 2 and 3. Intrinsic factors that affect AAR dating include different racemization rates of the amino acids due to side chain substituents, ease of hydrolysis of the peptides, or the ability to undergo secondary reactions [62]. Differing rates of racemization are known to occur between species of mollusks, perhaps due to the varying bond strengths of the amino acids in different proteins within the shell, or differences in protein stability within the crystalline matrices [7, 63]. Extrinsic factors include temperature most importantly, as well as solvent effects, catalytic effects, and pressure [10].  Studies try to limit the effects of extrinsic factors by careful choice of samples that are thought to approximate closed systems versus open ones, although of course time and temperature history is unchanged. One example is the isolation of intracrystalline amino acids from mollusk shells, where intracrystalline material is protected from external influences, such as pH and free water, and behaves as a closed system during diagenesis [64, 65]. Eggshells are also considered as an approximate closed system, provided that the eggshells were not treated with fire [66-69]. Bone is porous and thus susceptible to degradation and contamination by recent amino acids and has been found be an open system with original amino acid loss and exogenous components [60, 61]. Dentine has also been used as a source for AAR, with a study of modern black bear and fossil bear dentine reporting different Asp racemization between temperature-induced and time-induced racemization [70]. Samples obtained from caves, which have relatively uniform thermal and humidity profiles, would be advantageous in terms of having similar taphonomic effects [71-74].  Torres et al. demonstrated that by adding an additional dialysis step to the sample preparation, AAR showed a satisfactory correlation between dentine Asp and other dating methods for material from modern humans and for Pleistocene cave bears, horses and Neanderthals [60]. Chromatograms showing the effect of the dialysis step are shown in Figure 4 and the dating correlation is shown in Figure 5.

Most of the AAR HPLC determinations in the literature use the method detailed in the seminal paper by Kaufman and Manley [75] in 1998, with a recent check of Google Scholar citations showing that this publication has been cited 297 times.  This paper detailed an automatic HPLC method utilizing precolumn derivatization of the AAs to form fluorescent diastereomers which were separated by reversed phase HPLC and fluorescence detection, yielding a detection limit in the sub-picomole range. Sample size was in the range of milligrams.  This reported method involved a hydrolysis procedure of the sample solution which minimized the amount of induced racemization and gave good recovery of amino acids [75].  The precolumn derivatization was achieved with o-phthaldialdehyde and N-isobutyryl-L-cysteine (a chiral thiol) to yield fluorescent diastereomeric derivatives of the D and L amino acids. This derivatization was automated and performed online prior to each injection.  Separation was by a reverse-phase column packed with an end-capped C18 stationary phase and detection was by fluorescence. Figure 6 shows an example chromatogram using the original technique described by Kaufman and Manley[75].

While this chromatographic method has been modified and adapted with advances in technology such as ultra-high pressure LC or modification for greater resolving ability of the stationary phase, the method remains the workhorse for the determination of AAR in fossils, probably due to its low cost and ruggedness. Instrumentation and stationary phases improvements have in turn led to better resolution of enantiomers as demonstrated by Penkman et al. [76] in Figure 7. However, other separation methods have been developed that require less sample and no derivatization.  For example, capillary electrophoresis with mass spectrometry (CE-MS) has been used for baseline separation and detection of D/L mixtures of amino acids from human bones, using a bare fused-silica capillary and (+)-tetracarboxylic acid 18-crown-6 as a chiral CE background electrolyte [77]. Ion electropherograms using this method are shown in Figure 8. Since this CE-MS method is almost 1000 times more sensitive than radiocarbon dating, only a small amount of bone sample is required compared to the quantity that the HPLC technique consumes [77].

One result of the longevity of the HPLC method is the large database of AAR results obtained for the geochronological samples and the various interlaboratory comparisons of the AAR results.  For example, the University of Delaware AAR Database (UDAARDB) contains AAR and other geochronological data obtained from all analytical techniques from coastal Quaternary sites for over 25 years so that the data will be available for future use [78]. Interlaboratory comparisons of AAR have been reported for over nearly 30 years, with one by Wehmiller in 2013 [79]. In this study, previously prepared hydrosylate samples of Pleistocene mollusk and eggshell were distributed to laboratories that used ion-exchange liquid chromatography, HPLC, or GC. Agreement within 10% was found for all amino acid D/L values for the five amino acids most commonly used in geochronological applications (Asx, Glx, Leu, Val, and A/I). This report also provided inter-method comparisons with quantitative regressions that could be used when results from one method are compared with those from another [79]. Another interlaboratory comparison included the different sample preparation techniques to evaluate measurement bias and reported that D/L values biases of up to 30% were not uncommon [80]. The authors suggest that the bias as well as precision be included in AAR uncertainty estimates and called for future reference materials to be developed.

Prior to the mid-1990’s IEC was widely used for the determination of D/L enantiomers [81]. A recent paper reports converting the IEC values for allo-isoleucine:isoleucine to corresponding HPLC D/L values for comparison, finding correlations for Asx, Glx, and others [81]. Regression equations were provided for conversion, enabling the large literature base of AAR results from IEC to be compared and integrated with AAR results from RP chromatography [81].

It is interesting to the authors that relatively few papers publish examples of chromatograms, or resolution values or other data showing the separation of the enantiomers. While the emphasis is on the use of AAR for dating, and the mathematical calibration is important, it seems that including resolution values would give an immediate indication of the efficiency of the separation and the ease of calculating the D/L ratios.