11. Amino acid racemization in biominerals: the imp act of protein degrada tion and l oss Matt hew J. Co ll ins and Michae l S. Ri l ey Amino acid racemization I (AAR) presents a con- undrum. Ever si nce H are (1971), it has bee n appre- ciated (hut there is a dear inte rrelat ionship between AAR and protein di:lgencsis! (c.g .. Wehmiller, 1980; Miu crcr, 1993). Experimelltal studies of diag- enesis ha\'c revealed nlllrked diffcrcllt't's in the nileS of hydrolysis of peptide bonds, racemization :Ind decomposition of co nSti tutnt ami no lIcids <lnd the loss of organic fractions from different biomincntls. Yet observed ralCs of racemiza ti on in fossils (with Ihe exception of asp<lTtic acid) displllY surprisi ngly little variation between amino acid, genera or bio- mineral type. Thi s minimal vOl rilllio n enables AAR to be tre:ued as iI black box, in which the under- lying processes can be ignored. The black-box approach, however successful, leavcs many qucstions lillanswered. exposes resu lt s to chaUenge fr om those who query illlerprelati ons based upon AAR, and is perhups partly responsible for the limited impact IllHdc by the uppro;leh in the wider arehiteological illtd geologic;11 cOllllllu nity. Recent reviews of prOtein diagenesis (B;lda. 1991; Millerer, 1993; WchmiUer. 1993: Bada. 1998: Collins et al" 1998;1: Bada et aI., 1999) permit a renewed consideration of thc relationship betwccn diagenesis and racemization. It is therefore to th e process of racemization that lVe \Urn fir st and con- s ider how Ihe increasingly complex pallern of kinetics from free <tm ino acids via proleins to the organic (ruction of biominerals brings into play, first, a considenttion of protein structure and. sub- sequently. biomineral ultrastrueturc lind orga ni ;w- tion. RACEMIZATION Ra ce mization of Free Amino Acids Th ere is broad agreement over the rdative rates and rate constants for thc racemiz<ttion of frcc amino acids differences between studies are in large p.lrt due to variations in Ih e pH co nditi ons used (Bada and ShOll, 1980). Ami no acids with ;tli- phatic side chains display :1 4-6-fold variation in rate lit pH 7.6, betwccn Ihc most rilpid (Phe) and the slowcst (Vltl) (e.g .. Bada li nd ShOll, 1980; Smith and Sol. 1980). At the samc pH. Wo nacotl (1979) / rc po rt s a va riation in rale of over 1 50-fold between Ser and Glu . At pH 9, Liardon and Lederman n (1986) observed a 90-1'01<1 variation in rate between Ser and Val. The varialion in relative rates of racemization lind their pH dependence is broadly in agreement with the mechanism proposed by Neuberger (1948), who suggested tlwt und er basic conditions rltcemization procC\,"<!s vi<t abstraction of H lind the formation of a planar carbanion. This is illu- strated by the correlation between relative r:ltes of racemi7.ation and thc electron-withdrawing capacity of the si(le·clmin (cstimltted from the Taft inductive constan t): the more elect ron -wilhdmwing the side- chain. the greater will be Ihe 5Htbilizatio n of the cilfbanion and thus the more reltdily is this formed (B<tda and Shou, 1980; Smit h .md de Sol. 1980; Li<lrdoll ,tnd Ledermann. 1986). Unfo rtun<ltciy, r..tcemi7.alion o f free amino acids has liule bearing upon observed raccmization in foss il biominerals. Protein Decomposition and Ra cemiza ti o n Rate Unl ike their free counterparts, relative rates of race- mization between constituent amino acids in bio- minerals display surpri si ngly little var iation (e.g., Llljoie el al.. 1980; Goodfriend and Meyer, 1991 ; but sec Csapo el al .. 1990). Wor k by Stcinberg and Bacia (198 1. 1983), Minerer and Kriausa. kul (1984), Moir and Cr;twford (1988), Smith and Baum (1987), Gaines (tnd Bada ( 1988) and Scpetovet al. (1991 ) have highlighted the impor- tance of diketopiperazinc (D KP), formed during dl'Composition, in the kinelics of racemization of dipeptides and prot ei ns. Overall rates of raccmiza- tion are more than ten times faster in dipcptides than for fr ee amino acids. where<ts in OK P itself, Tiltes nrc estil1l<tted to be over two orders of mHgni- tude greater. Although studies at basie p H do revea l some evi- dence of racemi zou ion of internal residues (e.g., Fridkin et aI., 1970), there is little evidence for sig- nificant racerni7 .a tion within polypeptides at neutral pH. Two lines o f ev idence support this contention: first ly, limited racemiz.atio n in high-molecular. weight fractions (e.g., Kimber 'lIld Griffin 1987; Kimber and Hare, 1992: Kaufman and Miller, 1995) and melanoidins (Rafalska et aI., 1991): sec- ond, the unusu .. 1 behavior of Asx (Asp plus Asn). Lajo ie ct al. (1980) observed It "er}' signifi c;t1l( posi ti ve intercept in Ihe slope of DIL Leu vs. olt. A sx . becllUSC there is significant racemization of houlld Au resi dues (e.g., Gei ge r and Clarke. 1987: Colli ns et HI" 1999. (or revicw). T he absence of a positi\'e offset in any other ltmino ll ci d (except occa- sionally Pro l ) when plolled against o IL Leu sug- gests that in all ot he r ca.scs, bound residues nlcemize equally slowly if at all. The vHfiation ill relative racemization wtes of both fr ee alllillo acids and dipeptides is not apparent in biominerals. implying that thc rate-limiting step in thc observed ractmizoilion rates must be sought elsewhere. Modelling Racenlization It is common to report estimHted Tnte constants for racemization in biominerals by treating the net racemization as a single reaction with its own rate constant; we will prefix estimates or this type km to dislinguish Ihem from underlying nile constants. The relationship between and the various underlying rates (i.e .. a scries of rate con- stants for hydrolysis, nlcemization. and decomposi- tion) is potentially very co mpl ex. The morc tha t is understood regarding the factors influencing mce- miull i on , namely the potentia! for non-equilibrium ratios (Moir and Crawford, 1988: Fujii et :II.. 1994). Ih e impnct of conformation (Fujii et al" 1994: van Duin and Collins. 1998) and the wide mnge of rate con stants observed for free amino and dipcp- lid es (ltbove). the more surprising is the relatively small varilliion in Ihe rat es lind simplicity in the pltllerns of racemizmion (alt hough size fracti ons do display mor e convoluted kinetics: Kimber and Griffin, 1987). The approximation of early of racemization in biomineral proteins to first-order re\'Crsible kinetics means t ha t k't;, is widely reported the rate of racemizlltion. In recent yenrs, mathe- malicaltTansformations (Mitlerer and Kriausakttl, 1989; Murray-Wallace and Kimber, 199 3) or poly- nomial functions (Gillard et aI., 1991: Wehmillcr. 1993) ha ve been used to represclllthe net 'rate' over a wide range of L> IL values but such approaches ignore the underlying reactions. In order to utilizc k7J sa fcly. it is necessary to come 10 terms with the processes that infiuence it: these processes can be broadly tcrmed 'protein diagenesis'. It is this approach developed by Kri:tus.1ku lund Miuerer (1980:1. b). Mittercr and Kr ia usakul (1984) :tIlel Wehmill er (19110, 1982. 1990. 1993). whi ch has provided the most widely accepted concep lua lmodel to ex pl ai n the nonlinear kinetics (olher models exist. e.g .. Julg. 1984: Saint- Martin. 1991). We have already reviewed evide ll ce dcmonstrltling that the ratc of racemization varies dcpending upon the statc of protein degmdation, namely (I) slow racemiZation of peptide-bound residucs (2) fa st racemization of ei ther (a) terminlt! or (b) DKP residues. and (3) moderate rates of racemization of free amino Th .. models of Mill ercr and K ri:tUsakul and of Wehmi11er r.llionalize the overall mcemization rate in terms of protein degradation. at its simplest ill three stages (figure 11·1). as ou tlin cd below. I. Initially. racemi7..;tlion is slow bcc<tuse Ihe over- whelming majority of amino acids <Ire in the form of peptide-bound residucs in l ong polypep- ti de chains. 2. Over time. liS hydrolysis of pe ptidc bonds leads to the solubilization of insoluble proteins and the generation of oligomers. dimers. and DKP. the nlte of racemiZlllion is accelemted. 4 3. Finally. the the controlling faclOr. rcfiected in a decli ni ng rale of V;;t. This conceptual model. hereafter refcrred 10 as the "three· box modd" , is 10 expill in the used 10 dcrive it. The example of non · linear raccmization lIsed by Kriausakul and Mittcrer (19110b) of M('1'f( '11(I I' i(l heated at 15TC dis- plays a final rate of epimcrization less than half the rate of free isoleucine (lie) from the same study (figure 11-2 ). In the simple three-box model. the slowcst possible kti. is as is observed for hc,lIed pcptides and proteins ill riJro and Mil1ercr. 1980b). However. an adequale expla. nation of epimerizlltion of lie in M('l'cClwria and. by implication. in other bi ominerals. ha s 10 al'Counl for ;In in the laller stages of which is slower that .. INADEQUATE MODE LS OF RACEMIZATION Two processes l11ily aceount f or the slower-than- anticipated rHle ofracemi;o:ation: (1) slol\' hyd ro lysis of a residual peptide-bound fraction or (2) prefer. ential loss of free (highly racemized) amino acid s. Each of these two possibilities will nolV be co nsid- ercd. HydrolySis In a previous paper ( Ri ley alld Collins. 1994) we modeled a polypeptide comprising Po amino acid residues in ter ms o f ra ndom chain scission. leading to an initial increase in N-tcrrui ni illld an overall decreltsc in the mean lenglh of residual polypep- tides, uhinwtely leading to thc liberution of rrce amino Hcids (equation 11 -1). The weight fraction. y,. of x- mers allime I. is gi\' cl1 by for,\'< Po forx = Po ( 11-1 ) whcre 0' is the probability of bond breakitge given by 0' = I - exp( -kt). and k is the rate of hydrolysis .,.,-, .. ,"".,