alpha-crystallin polymers and polymerization

International Journal of Biological Macromolecules

22 (1998) 253–262

a-Crystallin polymers and polymerization: the view from down under

R.C. Augusteyn *

The National Vision Research Institute of Australia, 386 Cardigan St, Carlton, Victoria 3053, Australia

Abstract

Several models have been proposed for the quaternary structure of a-crystallin. Some suggest the subunits are arranged inconcentric shells. Others propose that the subunits are in a micelle-like arrangement. However, none is able to satisfactorilyaccount for all observations on the protein and the quaternary structure of a-crystallin remains to be established. In this review,factors contributing to the assembly and polymerization are examined in order to evaluate the different models. Consideration ofthe variations in particle size and molecular weight under different conditions leads to the conclusion that a-crystallin cannot bea micelle or a layered structure. Instead, it is suggested that the protein may be assembled from a ‘monomeric’ unit comprisingeight subunits arranged in two tetramers with cyclic symmetry. The octameric unit is proposed to be disc-like particle with adiameter of 9.5 nm and a height of 3 nm. The larger particles, chains and sheet-like structures commonly observed are assembledfrom the octamers. Structural predictions indicate that the polypeptide may be folded into three independent domains which havedifferent roles in the structural organization and functions of the protein. It is suggested that the tetramers are stabilized throughinteractions involving the second domain (residues 64–104) while assembly into the octamers and higher polymers requireshydrophobic interactions involving the N-terminal domain. Deletion of parts of this domain by site directed mutagenesis revealedthat residues 46–63 play a critical role in the assembly. Current research aims to identify the specific amino acids involved. © 1998Elsevier Science B.V. All rights reserved.

Keywords: a-Crystallin; Quaternary structure; Critical role

1. Introduction

As the major protein of the ocular lens in mostvertebrate species, a-crystallin is certain to makeimportant contributions to the optical, chemicaland physical properties of the tissue. Knowledgeof the protein’s structure would be of considerablevalue in understanding these contributions andhow the lens fulfills its functions.

It has not yet been possible to crystallize theprotein for X-ray analysis and it is too large forcomplete structural determination with NMRspectroscopy. Therefore, various solution studieshave been used to probe the structure. Because of

the proteins’ polydispersity and heterogeneity, in-terpretation of data has not always been straightforward. This has led to quite disparate viewpointsand vigorous debate on the structure and proper-ties of a-crystallin. We are pleased to have beenable to contribute to this debate over many years.

In this overview, I will present some of ourviews on the structure of this important and fasci-nating protein as well as the results of some recentexperiments. For details of work in other labora-tories, the reader is referred to additional articlesin this volume, a recent review from our labora-tory [1] and the excellent book edited by Bloemen-dal [2]. Undoubtedly, our views will differmarkedly from those expressed elsewhere in thisvolume. Perhaps this reflects differences in per-spective between ‘down under’ in the SouthernHemisphere and ‘up there’ in the North.

* Tel.: +61 3 93497480; fax: +61 3 93497473; e-mail:[email protected]

0141-8130/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved.

PII S0141-8130(98)00023-3

R.C. Augusteyn / International Journal of Biological Macromolecules 22 (1998) 253–262254

2. Particle size

a-Crystallins can be isolated in various forms,ranging from particles to sheet-like structures. Allare large enough to be observed with the electronmicroscope. Examples are shown in Fig. 1.

Heterogeneous populations of apparently spher-ical particles with diameters of up to 20 nm arecommonly observed [1–3]. The size may be relatedto age since the protein isolated from young tis-sues is smaller at 10–15 nm (Fig. 1a). The particlesexhibit substantial surface features but the originsof these features have not been identified. In someimages, it appears that the subunits may be ar-ranged in coils. More uniform particles, of similarappearance but with smaller diameters (9.3–9.5nm), are obtained with proteins assembled frompurified aA or aB subunits [4], with am-crystallin[3] and with the closely related small heat shockproteins [5]. The aB particles are particularly uni-form and should be amenable to image enhance-ment techniques. Because of shrinkage duringsample preparation, particles may be up to 10%larger in solution.

The particles can assemble into long chain-likeaggregates. Fig. 1b shows such a chain whichcontains at least six and possibly ten, of the 10–15nm diameter particles. Similar chain-like assem-blies are observed with the 9.5 nm species but theyappear to be shorter, containing only 3–4 particles[4]. Relatively few chains are usually observedbecause they are fragile and fragment duringpreparation of the sample grids for electron mi-croscopy. However, if they are first stabilized bycovalent crosslinking with glutaraldehyde, underconditions where there is no increase in the molec-ular mass of the proteins, large numbers can beobserved. We have demonstrated using a combina-tion of electron microscopy and light scatteringmeasurements, that the polydispersity of isolateda-crystallin populations is related to variations inthe length of these chains [4,6].

A sheet-like form of the protein (Fig. 1c) can beisolated from lens extracts by avoiding damagedue to physical trauma such as that associatedwith centrifugation and homogenization [7]. Thesheet appears to consist of small particles of 2–2.5nm diameter, presumably subunits, packed side byside. It has been reported that similar structures(pancake-like) can be created by exposure of theprotein to elevated hydrostatic pressure [8]. Renat-uration of denatured subunits at elevated tempera-tures generates fragmented sheets [9].

How any of these forms relate to the in vivostructure of the protein remains to be determined.It seems probable that all arise through differentfragmentation (or recombination) pathways fromthe same, as yet, unidentified in vivo form. There

Fig. 1. Forms of a-crystallin observed by electron mi-croscopy: a, ac-crystallin particles: the size of the particles(arrowed) may be compared with the 10 nm bar; b, a chain-like aggregate of a-crystallin: the ends of the chain areindicated with arrows; and c, fragmented sheet-like structuresin lens extracts. The bar in all figures represents 10 nm. Themicrograph in Fig. 1c was taken by Dr J.F. Koretz ofRennselaer Polytechnic Institute.

R.C. Augusteyn / International Journal of Biological Macromolecules 22 (1998) 253–262 255

Fig. 2. Estimates for the average molecular weight of bovine a-crystallin published since 1950. All reported data for the molecularweight (�) from the last 40 years are included, except those for isolated nuclear proteins because of the uncertain effects of agingon the molecular weight. Data are also presented for the minimum molecular weights (�) which were obtained with stableparticles isolated by subfractionation of native populations, or produced by partial dissociation, or created by re-association ofdenatured polypeptides.

are many observations which indicate that theaggregate structure of the protein is significantlyaffected by variations in the conditions used fordisrupting lens cells [1,2,10]. However, once re-moved from the lens, a-crystallins appear to beexceptionally stable.

It is quite obvious that a-crystallins, as isolated,are populations of polymers of some smaller‘monomeric’ unit but the size of this monomerremains uncertain. It may be the almost sphericalparticle of 9.5 nm diameter observed with reconsti-tuted proteins or a still smaller molecule. It mayeven be the subunit itself as suggested by somedata and embodied in the micelle hypothesis[23,24]. We have devoted considerable effort toidentification of this monomer and the arrange-ment of the subunits in a-crystallins, i.e. the qua-ternary structure. We have also explored thefactors contributing to the polymerization of thismolecule.

3. The molecular weight

Apart from the subunit size, the most importantinformation needed for elucidation of a protein’squaternary structure is the minimum molecularweight of the aggregate. This has proved to bevery elusive since a-crystallins are commonly iso-

lated as very heterogeneous and polydisperse pop-ulations with broad molecular weightdistributions. It appears that physical and chemi-cal modifications of the proteins during aging ofthe lens and variations in handling conditions, allcontribute to generation of the heterogeneity andpolydispersity [1,2].

Molecular weights ranging from 500 kDa to 15mDa have been reported for ‘native’ a-crystallins,i.e. proteins isolated from the lens [1,2]. In general,‘young’ a-crystallins, such as the newly synthe-sized protein and those isolated from fetal lensesor from the cortex of adult lenses, fall at thebottom end of the range. As shown in Fig. 2,published molecular weight estimates for theseyoung proteins have ranged from 0.5 to 1.5 mDa.The older proteins, such as those from the lensnucleus are much more heterogeneous and muchlarger, at up to 15 mDa [1,2].

There appears to have been a gradual decreasein the published estimates of the MW, from anaverage of over 1000 kDa in the 1950s to under600 kDa some 40 years later. This can probably beattributed to improvements in the methods usedfor isolating and studying the protein. If the trendcontinues, estimates of the average MW coulddrop to below 500 kDa by the turn of the century!We believe this would still be well above the trueminimum.


Fig. 3. Physicochemical parameters for a-crystallins fraction-ated by size on Sepharose CL-4B: a, Elution profile for thea-crystallin peak; b, Proportions of aA (�), aAm(�), aB (�)and aBm (�) polypeptides; c, Molecular mass distributiondetermined by static light scattering; d, The ratio of static todynamic scattering parameters (RG/RH); e, Particle diametersfrom electron microscopy; and f, Particles per chain fromelectron microscopy.

Surprisingly, the consensus still appears to bethat the MW of a-crystallin is a precise 800 kDa,despite the well documented polydispersity[1,2,6,10–14]. This figure is almost universallyquoted and used in the interpretation of data andin the formulation of new models. As discussedbelow, there are many indications that the proteinis a stable polymer of some smaller ‘monomeric’molecule. Although this molecule has not yet beenidentified, it is probably larger than a subunit.

All native and reconstructed a-crystallins arepopulations of chemically identical but differentlysized aggregates. They are not in equilibrium witheach other since their molecular masses show nosensitivity to large variations in protein concentra-tion [6,14,29]. Furthermore, populations of narrowsize ranges which are stable for at least 12 monthsat 20°C, can readily be isolated by gel permeationchromatography [14]. When any of these proteinsare concentrated and rechromatographed, all eluteat their original volumes, indicating that the sizesdo not change through association or dissociation.

We have examined the reasons for the protein’spolydispersity using a combination of light scatteranalysis and electron microscopy of bovine a-crys-tallins [6,14]. The size distribution of the corticalproteins is quite narrow with the majority of spe-cies in the range 600–1000 kDa [6,11–14]. Bycontrast, nuclear a-crystallins are known to con-tain substantial amounts of aggregates over 1000kDa. Therefore the nuclear proteins were fraction-ated into differently sized populations and exam-ined in detail [6]. Some of the data obtained arepresented in Fig. 3.

The high molecular weight area from the elutionprofile obtained with bovine lens nuclear proteinsis shown in Fig. 3a. The peak absorbance occursat about 900 ml but substantial amounts elutedfrom 400 ml. Electrophoretic analyses indicatedthat essentially all of the proteins eluting before1000 ml were a-crystallins and these containedvariable amounts of shortened polypeptides withapparent masses of 19 kDa (aBm) and 17 kDa(aAm). The proportions, shown in Fig. 3b, revealthat 25–40% of nuclear a-crystallin polypeptidesare modified.

Light scattering experiments [6] demonstratedthat the molecular masses vary from \4 mDa atthe front of the peak to B600 kDa at the rear(Fig. 3c). Rechromatography of the proteins fromthe rear of the peak yielded species close to 500kDa. There was no obvious quantitative relation-


ship between polypeptide modification and molec-ular mass although, clearly, modification wasgreatest in the largest aggregates.

Comparison of the static (RG) and dynamic(RH) scattering parameters yielded useful informa-tion on the shapes of the molecules (Fig. 3d). TheRG/RH ratios were found to vary from \1.20 forthe high molecular weight proteins to B1.0 forthe low. When compared with the theoretical ratioof 0.78 for a spherical particle, it is clear that theproteins cannot be spherical, especially the highmolecular weight species. Theoretical modeling in-dicates the data are more consistent with flexiblecylinders or semi-flexible (worm-like) chains.

Electron microscopy revealed the presence ofsingle particles, similar to those shown in Fig. 1a,as well as chain-like structures of variable lengths(Fig. 1b) in all fractions. Small pieces of sheets(Fig. 1c) were observed in the very early fractions.The diameters of the particles, whether in isolationor assembled into chains, were constant at 13.591.0 nm throughout the peak (Fig. 3e) but theaverage number of particles per chain ranged from1.5 to \4 (Fig. 3f). Short chains were also ob-served in cortical a-crystallins. From the chainlengths and light scattering data, a maximummolecular mass of around 600 kDa can be calcu-lated for a single particle. The actual mass will belower than this since inclusion of the single parti-cles in the calculations significantly reduced theaverage chain lengths. It is probable that singleparticles and short chains observed in the highmolecular weight fractions early in the peak arosefrom chain breakdown during preparation of themicroscope grids. If otherwise, these smaller spe-cies would have been separated from the largechains during the gel filtration.

Similar studies with homopolymers constructedfrom pure aA, aB, aAp and aBp subunits yieldeda maximum molecular mass of 216 kDa for the 9.5nm particle [4]. Again, the true value would belower. Several other observations lead to the sameconclusion. The data have been included in Fig. 2as estimates of the minimum molecular weight.

When lenses are not permitted to cool duringextraction, am-crystallin, with an average MW of320 kDa and a particle diameter of 9.5 nm, isobtained [10]. This is identical in all other regardsto ac-crystallin, the protein isolated in the cold. Itappears that lens cooling induces polymerizationof a-crystallin. These and related observationsstimulated much debate in the 1980s and were

instrumental in disproving the three layer modelsproposed for the structure of a-crystallins.

It is well established that denaturation of a-crystallin with chaotropic agents is reversible andgenerates macromolecules smaller than the origi-nal [1,2]. Work in our laboratory [4,9,15,16] hasdemonstrated that, at low temperatures and lowprotein concentrations, the reconstitution willyield relatively homogeneous species of 3–400kDa. As the temperature or concentration in-creases, so do the average molecular weight andthe heterogeneity, suggesting that hydrophobic in-teractions play a role in the polymerization [9].Sonication, a technique often used nowadays todisrupt the lens, also causes polymerization [31].

Partial dissociation of the a-crystallin polymerscan be achieved under relatively mild conditionsby increasing temperature [12], decreasing ionicstrength [12,17], decreasing pH [18], exposure tolow concentrations of urea [16] or chemical mod-ification [18,19]. These treatments produce stablepopulations of a-crystallins with average molecu-lar weights of 3–400 kDa. By gel filtration of thesepopulations, species as low as 280 kDa can beisolated but these are still polydisperse, indicatingthat the minimum molecular weight is even lower.

The lowest estimate, to date, for the minimummolecular weight species has been obtained fromacid dissociation of the aA homopolymers [20].Exposure to pH 2.5 at low ionic strength gener-ated stable species of 160920 kDa, correspondingto eight subunits and small amounts of an 80 kDamolecule, assumed to be a tetrameric partial disso-ciation product. It is possible that the octamer isthe 9.5 nm particle observed with the electronmicroscope. This might correspond to the‘monomeric’ molecule from which other forms ofthe protein are assembled. It is somewhat soberingto note that similar particles, ranging from 80 to176 kDa, were observed for succinylated a-crys-tallin by Spector and Katz over 30 years ago [18].Their significance was not appreciated at the timeand the observations have been largely forgottensince.

There are inconsistencies in the hydrodynamicmeasurements which remain to be explained. If theparticles are spherical, the volumes calculatedfrom their dimensions appear large enough toaccommodate almost twice the measured masses.This calculation and the observed high capacity ofthe aggregate for partitioning small molecules


[26,32] suggest that a-crystallin may contain largeopen spaces. It has been postulated that the sitefor the chaperone-like activity is located withinthese spaces [33]. However, such a structure isunlikely since the sedimentation characteristicsand other physicochemical properties appear to betotally consistent with those of a normal proteinwhich has a density of around 1.34 g/ml andcontains up to 25% solvent [6,13,15,21]. It seemsmore likely that the particles are not quite spheri-cal as noted in several previous studies[1,2,6,34,35]. Perhaps, they are disc-like structuresor ellipsoids. A flatter shape would be consistentwith all hydrodynamic measurements to date. Sim-ilar conclusions were reached from a considerationof the rotational properties of the protein in con-junction with electron microscopy [35].

4. Quaternary structure

Despite the lack of a definitive molecular weightfor a-crystallin, a variety of models has beenproposed for the arrangement of subunits. Mostof these models are not consistent with experimen-tal observations such as the equivalence of sub-units. Most are based on a fixed molecular weightand are unable to accommodate or explain thehuge variations observed in the protein’s size. Evi-dence for and against these models has been dis-cussed elsewhere [1,2,6,24] and will not beconsidered here. Only the micelle model [23,24]appears to be consistent with available informa-tion but, as outlined below, this now also seemsunlikely.

In our early studies on the immunological prop-erties of a-crystallins, we had noted that the a-crystallin polypeptide is amphiphilic with adistinct, hydrophobic N-terminal domain of 63–67 amino acids and a more hydrophilic C-terminalsequence [22]. It is probable that the polypeptidesinteract through this hydrophobic N-terminal do-main to form the aggregates. Support for thisproposal is provided by the protein’s behaviorduring denaturation and renaturation.

a-Crystallin remains a large aggregate whilesubstantial unfolding of the C-terminal part of thepolypeptides takes place [16]. Cys-131 [27] andSer-122 [28] in the aA polypeptide and Trp-60 inaB [26] become completely exposed to the solventwith increasing urea, well before the aggregatedissociates directly to subunits at around 4 M

urea. Increasing temperature during slow renatu-ration (removal of urea by dialysis) generates spe-cies of increasing molecular mass [9]. Suchbehavior is consistent with aggregates of subunitsstabilized by strong hydrophobic interactions be-tween their N-terminal domains. These and otherobservations led to our proposal that a-crystallinsmay be comprised of micelle-like assemblies ofsubunits [23,24].

Micellar structures could accommodate variablenumbers of subunits. Their interconversion couldoffer a satisfactory explanation for the variationsin a-crystallins molecular size under different con-ditions. They would also be consistent with otherproperties, notably the observed equivalence ofsubunits, the location of specific amino acids, theimmunological identity of the different forms andthe stoichiometric binding of denatured proteins[1,7,10,15,22,25–27].

The micelle proposal was embraced with enthu-siasm by many and several modified versions havesubsequently been developed [33,36,37]. Recentphysicochemical data have been interpreted assupport for the model [8,29,38]. However, closescrutiny of these data suggests that other interpre-tations may be more appropriate.

An apparent critical micelle concentration(CMC) was obtained from surface tension andconductivity measurements [8,29]. However, thetransitions observed were very small and occurredat protein concentrations (1–10 mg/ml) commonlyused to study the protein. There have been manyprevious studies on the effects of protein concen-tration on the protein’s properties. It is hard toimagine why such transitions were not observedbefore. No change in aggregate size occurs at thisputative CMC, leading to the suggestion that aconformational change may be involved [29]. Thisis highly improbable. Furthermore, studies in ourlaboratory have shown that, over the range 0.01–130 mg/ml, a-crystallins display none of the prop-erties, such as a concentration dependence ofsurface tension, conductivity, aggregation state orconformation, which might reasonably be ex-pected for a micelle [30].

While assembly of subunits and polymers isundoubtedly driven by hydrophobic interactions,it has now become clear that there is no com-pelling evidence to support the micelle hypothesis.Indeed, most observations in the literature aremore consistent with a structure based on a fixed


number of subunits in a defined symmetrical ar-rangement. What that number or arrangementmay be, continues to elude us because of thecomplications arising from polymerization. In or-der to make further progress, it will be necessary toreduce these complications. As will be seen later,molecular biology may provide the solution.

5. Structural predictions

In the absence of an experimentally determined3-dimensional structure, attempts are being madeto predict structures for the polypeptides and forthe aggregates from sequence data. Some of theseare found elsewhere in this volume. Our approachhas been to use a combination of experimentalobservations and conclusions based on analyses ofa-crystallin and heat shock protein sequences. Thishas provided useful information about the poly-merization of a-crystallin.

Exons often code for separate structural units ordomains in proteins. The presence of three exons inits genes [39] suggests a-crystallin polypeptidesmay consist of three domains, comprising around65, 40 and 70 amino acids, respectively. Hydropa-thy analyses support this possibility. The proteinsequences corresponding to the boundaries of theexons are strongly hydrophilic and are predicted tobe antigenic determinants [22]. This is a character-istic of domain interfaces since they are oftenfound on the surface of a protein.

There is little doubt that Domain 1 is a struc-tural entity which is not required for the folding ofthe rest of the polypeptide. Expression of mutantgenes lacking the first exon generates viableproteins [40]. Domain 1, itself, forms large unstruc-tured aggregates, presumably through hydropho-bic interactions. The arrangement of the rest of thepolypeptide is not so clear, with one domain(residues 63–173 [40]) or two (residues 63–104 and105–173 [22,24]) suggested. There is no experimen-tal data to support either.

Domain 1 is of particular interest in the contextof the present discussions. We had suggested thathydrophobic interactions between the N-terminaldomains were the driving forces for assembly ofsubunits and polymerization of the protein [24].Evidence for such a role for Domain 1 has comefrom recent demonstrations that mutant proteinslacking this domain do not form large aggregates

[40]. Therefore, we have undertaken more detailedstudies on residues involved in the polymerization.

Sequence analyses reveal that residues 1–63 con-tain an internal duplication indicating that Do-main 1 has evolved through duplication and fusion

Fig. 4. Alignment of the internal repeat sequences in theN-terminal domain of aA- and aB-crystallin. The sequencesshown are those for the bovine protein but the same relation-ships are observed in all a-crystallins and several small heatshock proteins. Gaps have been inserted for maximum align-ment. Invariant residues are shown in bold and shadingindicates positions in which the hydropathy properties areessentially retained.


of an ancestral gene coding for 30–40 amino acids[41]. It has been proposed that the rest of thea-crystallin polypeptide has also arisen from thisrepeat sequence [42]. However, this is not convinc-ing since the homology is weak.

The repeat in Domain 1 can easily be seen fromthe alignment of the aA- and aB-crystallin se-quences in Fig. 4. Features of the repeats are theinvariant residues—proline in position 8 and 16and aspartic acid in 37—and the retention of sitecharacteristics—75% of the sites, when gaps areignored. All a-crystallin sequences (except parts ofthe dogfish) and some of the small heat shockproteins can be aligned in the same way and showthe same relationships. These similarities suggestthat the two sequences would probably adopt verysimilar conformations.

Our secondary structure predictions using themethod of Rost and Sander [43] suggest thatresidues 14–17 constitute a b-turn. Predictions aresomewhat equivocal for the remaining structurebut b-pleated sheet predominates. It is clear fromimmunological [44] and chemical probing studies[26] that most of the repeat sequences are on thesurface of the subunit as would be expected forresidues involved in intersubunit interactions. Inthe aggregate, some sections, such as residues 1–8are highly accessible to the solvent while others,including amino acids 22–32, are buried.

The most notable differences between the repeatunits are the deletion of the residues in positions5–7, the substantial change in hydrophobicity inposition 15 (Phe/Ser), the charge reversal in posi-tion 26 (Asp/Arg) and the interchange of Gly withcharged residues in positions 32 and 34. Thesewould be expected to modify the properties of thesequences, thereby generating different functions.Information on the specific roles of these repeatsand of the amino acids within them is beingsought with the aid of molecular biology.

6. Site directed mutagenesis

Site directed mutagenesis is being used to ex-plore the role of the two repeat structures in theaggregation of a-crystallins [45]. To date, we havecreated bovine mutant aA polypeptides which lackone or both of the repeat units. The sizes of theaggregates produced were assessed by HPLC gelpermeation chromatography coupled with SDS gelelectrophoresis. The results are shown in Fig. 5.

Fig. 5. Determination of aggregate size for mutant a-crys-tallins. All proteins were expressed in the baculovirus systemand fractionated by HPLC-GPC on Zorbax G250 and 450columns connected in series. The location of the mutantproteins was determined using SDS-PAGE of the fractions.Lanes are identified according to the elution volume of theproteins: a, a1–173-crystallin; b, D1

2N (a36–173-crystallin); and c,DN (a64–173-crystallin).

The mutant lacking the first of the repeatunits, aA36–173, produced aggregates which co-elute with native and recombinant a-crystallinsat about 17 min (Fig. 5a,b) indicating that thethree have similar sizes, around 600 kDa. Weconclude that residues 1–34 of the a-crystallinpolypeptide are not involved in the interactionswhich generate either quaternary structure orpolymers. This is consistent with observationsthat the closely related small heat shock proteinsalso form aggregates and 9.5 nm particles [5]even though they lack much of the N-terminaldomain.


However, the mutant protein lacking both re-peat units, i.e. aA64–173, is clearly smaller, as indi-cated by its greater elution time of 24 min (Fig.5c). Comparison with standards indicates it maybe a molecule of 20–30 kDa which would proba-bly be a dimer of the 12.8 kDa polypeptide. Asimilar protein generated by Merck et al. [40] wasreported to be a tetramer. Crosslinking data pre-sented in the same study suggested this was adimer of dimers. A tetrameric molecule was alsoobserved in our acid dissociation studies [20] andby Spector and Katz in the succinylated protein[18]. It seems likely that the tetramer is either the‘monomeric’ molecule from which a-crystallins areassembled or a substructure of an octamericmolecule as suggested earlier. Thus, we may con-clude that residues 36–63 are responsible for poly-merization of a-crystallins but not for thegeneration of its quaternary structure.

It is probable that aA residues 46–63 (repeat 2,positions 16–37) are actually involved in the inter-actions. Eight of the amino acids invariant in alla-crystallin polypeptides (P46, Y48, R49, R54, D58,G60, S62, E63) are found within this segment.Residues important in the polymerization wouldbe retained through evolution. Furthermore, anti-bodies raised against a peptide corresponding toresidues 49–63 of the aA polypeptide (repeat posi-tions 20–37) do not react with a-crystallin [44]suggesting that at least part of this sequence isburied. We are now producing mutants with singleamino acid changes order to identify the specificresidues involved.

7. Conclusion

What, then, is our current view of the structureof a-crystallin? We are now certain that it is not amicelle but still believe that the polypeptide isfolded into three independent structural domainswhich all have different roles in the structuralorganization and functions of the protein. Recentobservations suggest that a-crystallins may be as-sembled from a ‘monomeric’ unit comprising eightsubunits arranged in two tetramers. It is thoughtthat these tetramers are stabilized through interac-tions in the second domain (residues 64–104)while assembly into an octamer or higher polymerrequires hydrophobic interactions involvingmainly the last 17 residues of the N-terminal do-main. The octamer particle is proposed to be

disc-like with a diameter of 9.5 nm and a height of3 nm. The larger particles, chains and sheet-likestructures could be assembled from the octamers.

Much remains to be learnt about this importantprotein. Undoubtedly, with the enormous interestgenerated by the relationship between a-crystallinand the heat shock proteins, as well as the putativechaperone-like activity, elucidation of the com-plete three-dimensional structure of the proteincannot be too far away.

Acknowledgements

The work from our laboratory was supported,in part, by grants from the National Health andMedical Research Council of Australia, the RERoss Trust, The Australian Research Council, TheVictorian Lions Foundation and The VictorianDepartment of Human Services.

References

[1] Stevens, A., Augusteyn, R.C. Prog. Polymer Chem. 1997;in press.

[2] Bloemendal H. Molecular and Cellular Biology of theEye Lens. New York: Wiley, 1981.

[3] Koretz JF, Augusteyn RC. Curr. Eye Res. 1988;7:25–30.[4] Augusteyn RC, Koretz JF, Schurtenberger P. Biochim.

Biophys. Acta 1989;999:293–9.[5] Arrigo AP, Suhan JP, Welch W. J. Mol. Cell Biol.

1988;8:5059–71.[6] Schurtenberger P, Augusteyn RC. Biopolymers

1991;31:1229–40.[7] Stevens A, Walsh R, Augusteyn RC. Curr. Eye Res.

1996;15:215–8.[8] Radlick LW, Koretz JF. Biochim. Biophys. Acta

1992;1120:193–200.[9] Clauwaert J, Ellerton HD, Koretz JF, Thomson K,

Augusteyn RC. Curr. Eye Res. 1989;8:397–403.[10] Thomson JA, Augusteyn RC. Exp. Eye Res.

1983;37:367–77.[11] Spector A, Li L-K, Augusteyn RC, Schneider A, Freund

T. Biochem. J. 1971;124:337–43.[12] Bindels JG, Siezen RJ, Hoenders H. J. Eur. J. Biochem.

1980;111:435–44.[13] Thomson JA, Augusteyn RC. Curr. Eye Res.

1988;7:563–9.[14] Augusteyn RC, Parkhill E, Stevens A. Exp. Eye Res.

1992;54:219–28.[15] Thomson JA, Augusteyn RC. Biochim. Biophys. Acta

1989;994:246–52.[16] Thomson JA, Augusteyn RC. J. Biol. Chem.

–45.[17] Augusteyn RC, Ellerton HD, Putilina T, Stevens A.

Biochim. Biophys. Acta 1988;957:192–201.


[18] Spector A, Katz E. J. Biol. Chem. 1965;240:1979–85.[19] Ortworth BJ, Olesen PR, Sharma K, Prabhakaram M.

Exp. Eye Res. 1993;56:107–14.[20] Stevens A, Augusteyn RC. Biophys. J. 1993;65:1648–55.[21] Siezen RJ, Berger H. Eur. J. Biochem. 1978;91:397–405.[22] Puri N, Augusteyn RC, Owen EA, Siezen RJ. Eur. J.

Biochem. 1983;134:321–6.[23] Thomson, J.A. a-crystallin; a model for multisubunit

protein assemblies. University of Melbourne, 1985, The-sis.

[24] Augusteyn RC, Koretz JF. FEBS Lett. 1987;222:1–5.[25] Winzor D, Stevens A, Augusteyn RC. Arch. Biochem.

Biophys. 1989;268:221–6.[26] Augusteyn RC, Ghiggino KP, Putilina TP. Biochim.

Biophys. Acta 1993;1162:61–71.[27] Augusteyn RC, Hum TP, Putilin T, Thomson JA.

Biochim. Biophys. Acta 1987;915:132–9.[28] Carver J, Aquilina JA, Truscott RJ. Biochim. Biophys.

Acta 1994;1164:22–8.[29] Carver J, Aquilina JA, Cooper P, Williams GA, Truscott

RJ. Biochim. Biophys. Acta 1994;1204:195–206.[30] Loutas J, Stevens A, Howlett GJ, Augusteyn RC. Exp.

Eye Res. 1995;62:613–20.[31] Putilina T, Zhang WZ, Augusteyn RC. Curr. Eye Res.

1991;10:113–20.

[32] Stevens A, Augusteyn RC. Eur. J. Biochem.1997;243:792–7.

[33] Groth-Vaselli B, Kumosinksi TF, Farnsworth PN. Exp.Eye Res. 1995;61:249–53.

[34] Siezen RJ, Berger H. Eur. J. Biochem. 1978;91:397–405.[35] van Haeringen B, Eden D, van den Bogaerde MR, van

Grondelle R, Bloemendal M. Eur. J. Biochem.1992;210:211–6.

[36] Carver JA, Aquilina JA, Truscott RJW. Exp. Eye Res.1994;59:231–4.

[37] Walsh MT, Sens AC, Chakrabarti B. J. Biol. Chem.1991;266:20079–84.

[38] Doss EW, Ward KA, Koretz JF. Exp. Eye Res.1997;65:255–66.

[39] Van den Heuvel R, Hendricks W, Quax W, BloemendalH. J. Mol. Biol. 1985;185:273–84.

[40] Merck B, De Haard-Boekman WA, Oude-Essink BB,Bloemendal H, de Jong WW. Biochim. Biophys. Acta1992;1130:267–76.

[41] Barker C, Ketcham LK, Dayhoff MO. J. Mol. E1978;10:265–81.

[42] Wistow GJ. FEBS Lett. 1985;181:1–6.[43] Rost BR, Sander C. Proteins 1994;19:55–72.[44] Takemoto LJ, Emmons T, Granstrom D, Emmons T,

Jacobs S. Biochim. Biophys. Acta 1989;995:259–63.[45] Cannon, J.F., Fang, Z-P, Maroudas, P., Augusteyn,

R.C. in preparation.

.

alpha-crystallin polymers and polymerization

Documents