doi:10.1006/jmbi.2000.4197 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 304, 253±257

COMMUNICATION

The N-terminal Domain of bbbB2-crystallin Resemblesthe Putative Ancestral Homodimer

Naomi J. Clout1, Ajit Basak1, Karin Wieligmann2, Orval A. Bateman1

Rainer Jaenicke2 and Christine Slingsby1*

1Department ofCrystallography, BirkbeckCollege, Malet Street, LondonWC1E 7HX, UK2Institut fuÈ r Biophysik undPhysikalische BiochemieUniversitaÈt RegensburgD-93040, Regensburg Germany

E-mail address of the correspondC.Slingsby@mail.cryst.bbk.ac.uk

Abbreviations used: bB2-N, N-terbB2-crystallin.

0022-2836/00/030253±5 $35.00/0

bg-crystallins from the eye lens are proteins consisting of two similardomains joined by a short linker. All three-dimensional structures ofnative proteins solved so far reveal similar pseudo-2-fold pairing of thedomains re¯ecting their presumed ancient origin from a single-domainhomodimer. However, studies of engineered single domains of membersof the bg-crystallin superfamily have not revealed a prototype ancestralsolution homodimer. Here we report the 2.35 AÊ X-ray structure of thehomodimer of the N-terminal domain of rat bB2-crystallin (bB2-N). Thetwo identical domains pair in a symmetrical manner very similar to thatobserved in native bg-crystallins, where N and C-terminal domains(which share �35% sequence identity) are related by a pseudo-2-foldaxis. bB2-N thus resembles the ancestral prototype of the bg-crystallinsuperfamily as it self-associates in solution to form a dimer with anessentially identical domain interface as that between the N and Cdomains in bg-crystallins, but without the bene®t of a covalent linker.The structure provides further evidence for the role of two-domain pair-ing in stabilising the protomer fold. These results support the view thatthe bg-crystallin superfamily has evolved by a series of gene duplicationand fusion events from a single-domain ancestor capable of forminghomodimers.

# 2000 Academic Press

Keywords: crystallins; eye lens; protein structure; domain interactions;2-fold symmetry*Corresponding author

The soluble crystallins are the main protein com-ponent in the eye lens, comprising up to 90% of itswet weight, and are derived from two proteinsuperfamilies, the small heat shock-relateda-crystallins and the bg-crystallins (de Jong et al.,1994; Horwitz, 2000; Wistow & Piatigorsky, 1988;Wistow, 1995). Both transparency and refractiondepend on high protein concentration (Benedek,1971; Delaye & Tardieu, 1983), which is providedby concentrated solutions and glasses formed froma diverse array of crystallin monomers and oligo-mers of high solubility (Slingsby et al., 1997). Inlong-lived species such as ourselves these crystallinmolecules need to last a lifetime as they aretrapped inside cells that have no protein syntheticor degradation machinery (Harding & Crabbe,

ing author:

minal domain of rat

1984). The intrinsic stability of protein domains,their potential for higher-order interactions andtheir solubility are important phenotypes for crys-tallin molecules (Jaenicke, 1999). The superfamilyof bg-crystallins is based on a modular design,with each polypeptide chain folding into four simi-lar Greek-key motifs: successive pairs of motifsfold symmetrically to form the similar N andC-terminal domains that are connected by a shortlinker (Slingsby et al., 1997). Half the structuraldiversity required for the appropriate refractiveindex throughout the vertebrate eye lens is thusderived from one kind of building block, thebg-crystallin domain.

Higher assembly in bg-crystallins involvesdomain pairing about an approximate 2-fold axis.In the monomeric g-crystallins the N and Cdomains pair intra-molecularly (Blundell et al.,1981). bB2-crystallin, the predominant b-crystallinsubunit, forms four-domain dimers by pairingdomains in an equivalent manner to monomeric

# 2000 Academic Press

Table 1. Crystallographic data for bB2-N

Crystal parametersSpace group P65

Number of molecules per asymmetric unit 2

Cell dimensions (AÊ )a = b = 110.96,

c�29.88Angles (deg.) a = b = 90, g = 120Solvent content (v/v) (%) 45

Data collectionMax. resolution (AÊ ) 2.35No. of unique reflections 9079Completeness (%) 99.8I/s(I)>20 (%) 79.9Rmerge (%) 5

RefinementR-factor (%) 20.49R-free (%) 24.62No. of water molecules 61RMSD bond lengths (AÊ ) 0.007RMSD bond angles (deg.) 1.26

The rat bB2-N construct (residues ÿ15 to 88) was expressedin Escherichia coli and puri®ed (Wieligmann et al., 1999). It wascrystallised from hanging drops whereby 1 ml of protein solu-tion (6 mg/ml) and 1 ml of well solution were suspended at4�C over a well solution containing 0.2 M magnesium acetatetetrahydrate, 0.1 M sodium cacodylate (pH 6.5) and 15% (w/v)PEG 8000. Diffraction data were collected from a frozen crystal.Glycerol was added to the crystallisation buffer (20% (v/v)) asa cryoprotectant. Data to 2.35 AÊ were collected, using the oscil-lation method, on a MAR research 30 cm image plate systemusing synchrotron radiation at the CLRC Daresbury Laboratory(l = 0.96 AÊ ). The point group of the data was determined usingautoindexing in DENZO (Otwinowski, 1993). Intensity mea-surement and pro®le ®tting were carried out using MOSFLM,and the data were scaled and merged using scala (CCP4, 1994).The N-terminal domain (residues 1-82) of bB2-crystallin (Baxet al., 1990) was used as a search model to solve bB2-N bymolecular replacement using the program AMoRe (Navaza,1994). The structure was re®ned using CNS (BruÈ nger et al.,1998).

254 The N-terminal Domain of �B2-crystallin

gB-crystallin, but inter-molecularly (Bax et al.,1990), by domain swapping (Figure 1; Schluneggeret al., 1997). It is generally accepted that the b andg-crystallins have evolved from a one-domain com-mon ancestor that subsequently duplicated anddiverged, followed by gene fusion events yieldingthe modern two-domain polypeptide chains of band g-crystallins (Wistow & Piatigorsky, 1988;Lubsen et al., 1988; Norledge et al., 1996). What isless certain is the timing of the gene fusion eventwith respect to the association behaviour of indi-vidual domains. An early gene fusion event wouldbe consistent with the hypothesis that an ancestralsingle bg-crystallin domain did not evolve an inter-domain interface until covalent linkage allowedoptimisation of a weakly interacting interface. Sucha mechanism has been proposed for the RosettaStone model for the evolution of protein-proteininteractions between unrelated domains (Marcotteet al., 1999), although this is unlikely to be the casefor a symmetrical interface where the domains areidentical or very similar. A late gene fusion is con-sistent with the ancestral single-domain capable offorming homodimers of domains with subsequentduplication and divergence of the single-domainancestor leading to the formation of heterodimersof prototype N and C domains.

Biophysical and crystallographic experiments onengineered single g-crystallin domains have notresolved the issue of whether single domains canself-associate in solution to form a non-covalentdimer that mimics the native domain. Mixingthe recombinant N and C-terminal domains ofgB-crystallin in equal amounts in solution does notyield a g-crystallin-like heterodimer, and this hasled to the proposal that a covalent linker isrequired to raise the local concentration of domainsfor dimer formation (Mayr et al., 1994). However,there is evidence that the N and C-terminaldomains of bovine gS-crystallin can associate insolution (Wenk et al., 2000). g-crystallin singledomains are monomeric in dilute solution,although 2-fold domain pairing of identicaldomains has been observed at the high proteinconcentration of crystal lattices (Norledge et al.,1996; Basak et al., 1998; Palme et al., 1998). How-ever, the geometry of the native pseudo-symmetri-cal interface of N and C-terminal domains was notalways recovered with just one kind of domain. Agood candidate domain to provide support for thelate fusion of ancestral single domains is one thatself-associates in dilute solution.

A single N-terminal domain of rat bB2-crystallin(bB2-N) has been shown to exist as a dimer indilute solution (>50 mg/ml), and is the ®rst singlebg-crystallin domain which does so (Wieligmannet al., 1999). b-crystallins differ from g-crystallins inhaving N-terminal sequence extensions, and thisbB2-N construct includes the entire 15 residue N-terminal extension plus the linker peptide (residues83-88) which acts as an ``arti®cial'' C-terminalextension. It was possible that the N-terminal

extension was involved in stabilising the solutiondimer of domains.

The structure of bB2-N was solved at 2.35 AÊ res-olution using molecular replacement (Table 1). Thespace group is P65 with two molecules per asym-metric unit (molecules A and B). The numbering ofthe residues of the rat bB2-N polypeptide followsBax et al. (1990), where the residue numbering ofbovine bB2-crystallin was based on a structuralalignment with bovine gB-crystallin so that theN-terminal extension is numbered ÿ15 to ÿ1.Comparison of the rat bB2-N domain to the nativebovine bB2-crystallin N-terminal domain (Bax et al.,1990) reveals very little difference in the structures:there are three sequence differences and the struc-tures have an RMSD of 0.99 AÊ for Ca-backboneresidues 1-84. Mass spectrometric analysis con-®rmed that the domain comprises residues ÿ15to 88 (Mcalc�11,896.00 Da, Mobs�11,896.17 Da).However, it was not possible to model the ®rst 13residues of the N-terminal extension of bB2-N,nor linker residues 85 to 88, in either molecule ofbB2-N, suggesting a disordered conformation forthe N-terminal extension and residues 85 to 88.This is further supported by an increase in

Figure 1. The homodimer of bB2-N is very similar todomain pairing of the N and C domains in native bB2.(a) bB2-crystallin is shown in light blue (Bax et al., 1990).The bB2-N homodimer is superimposed on the bB2-crystallin structure and molecule A of bB2-N is shownin purple whilst molecule B is shown in yellow. Thesuperposition was performed using the program O(Jones & Kjeldgaard, 1994) and the Figure was drawnusing the program Setor (Evans, 1993). The conservedhydrophobic side-chains of the interface region areshown in red for both bB2-crystallin and bB2-N, and itcan be seen that they occupy equivalent positions. Thepolar residues that surround the hydrophobic interfaceresidues are shown in blue for both structures. Thetopologically equivalent polar residues from the bB2-crystallin homodimer and the bB2-N homodimer occupyequivalent positions in the domain interfaces of the twostructures. (b) Detail of the hydrophobic and polar resi-due interactions involved in domain pairing lookingdown the 2-fold axis relating molecule A and molecule

The N-terminal Domain of �B2-crystallin 255

B-factors of the de®ned residues towards the Nand C termini.

The disordered N-terminal sequence extensioncan be assumed not to be stabilising the solutiondimer. However, the two molecules in the asym-metric unit are related to each other by a non-crystallographic approximate 2-fold of 179.4�analogous to the pseudo-2-fold pairing of the Nand C-terminal domains in native bg-crystallins.When the inter-molecularly paired N and C-term-inal domains of the native bB2-crystallin dimer aresuperimposed onto the bB2-N domain pair, thestructures have an RMSD of 1.46 AÊ for the Ca

backbone residues. The domain interface betweenthe two domains in the asymmetric unit is remark-ably similar to that between the N and C-terminaldomains of native bB2-crystallin, with the equival-ent residues from the second molecule of bB2-Nperforming the role of those from the C-terminaldomain of native bB2-crystallin (Figure 1(a)). Thetotal solvent-accessible surface area buried in thedimer interface is 765 AÊ per molecule. The core ofthe interface is made up from the hydrophobicpatch residues Val43, Val56 and Ile81 from eachbB2-N domain. In the native bB2-crystallin thesame patch residues, plus the topologically equiv-alent residues from the C-terminal domain (Val132,Leu145 and Ile170), form an analogous core. Innative bB2-crystallin the central hydrophobic patchis encircled by polar residues that make speci®cinteractions across the interface and hence aredependent on the speci®c orientation of the twodomains related by the 2-fold axis. Glu58 andArg79 form an ion pair with Arg168 (topologicallyequivalent to Arg79) and Glu147 (topologicallyequivalent to Glu58), respectively. The side-chainof Gln54 hydrogen bonds with the backbone ofresidue 145 from the C-terminal domain, while thetopological equivalent, Gln143, hydrogen bondswith residue 56 from the N-terminal domain. Inthe crystal structure of bB2-N the two equivalention pairs between Glu58 and Arg79 are formed, aswell as the two hydrogen bonds involving Gln54(Figure 1(b)). Thus the pairing between bB2-Ndomains and the native bB2-crystallin N and Cdomains is essentially the same, and these inter-actions are highlighted in Figure 1(b). Additionalhydrogen bonds in the bB2-N dimer interfaceare found between Gly40 and Asp84 (Figure 2),and between Gly52, Glu58 with Ser70, Lys82respectively.

B. The conserved hydrophobic residues (Val43, Val56and Ile81) are shown in red. Glu58 and Arg79, whichform an ion pair, are shown in dark blue and light blue,respectively. Gln54, which makes a hydrogen bond withthe backbone oxygen atom of Val56, is shown in green,and the hydrogen bond is also shown in green. TheFigure was drawn using the program Setor (Evans,1993).

Figure 2. Detail of the differing contributions of theC-terminal extensions to domain pairing. View of theC-terminal extension of bB2-crystallin showing how itcrosses the homodimer interface to interact with theN-terminal domain of the partner bB2-crystallin poly-peptide. The polypeptides of the bB2-crystallin homodi-mer are shown in blue. The bB2-N homodimer issuperimposed on the bB2-crystallin homodimer, withmolecule A shown in purple and molecule B shown inyellow. The superposition was performed using the pro-gram O (Jones & Kjeldgaard, 1994) and the Figure wasdrawn using the program Setor (Evans, 1993). Residuesinvolved in contacts between the partner domains of thetwo homodimers are labelled. Hydrogen bonds areshown in green.

256 The N-terminal Domain of �B2-crystallin

In native bB2-crystallin the ordered region of theC-terminal extension (residues 171-175) crosses thedomain interface and interacts with the N-terminaldomain, thus acting like a non-covalent linker.When the bB2-N domain pair is superimposedonto the native bB2-crystallin domain pair, resi-dues Lys82, Val83 and Asp84 of bB2-N, domain B,follow the direction of the C-terminal extension ofnative bB2-crystallin, equivalent to Arg171, Asp172and Met173. Asp84 interacts with domain A bymaking a hydrogen bond via its backbone nitrogenatom to the oxygen atom of Gly40 (Figure 2). Innative bB2 the backbone nitrogen atom of Gln174makes the equivalent hydrogen bond, but is out ofregister by one residue. A key additional interfacehydrophobic interaction in native bB2 is betweenTrp175 of the C-terminal extension and Pro1 of theN-terminal domain (Figure 2). In bB2-N the residuethat is topologically equivalent to Trp175 is Gln86(unde®ned by the electron density in bB2-N), andso this hydrophobic interaction is abolished. Pro1

is part of an additional surface hydrophobic patchthat is conserved in all b-crystallins, and whichincludes Leuÿ2, Gly40 and Pro41. Thus, the arti®-cial C-terminal extension of the B domain of bB2-Npartially compensates for the C-terminal extensionof native bB2-crystallin, facilitating domain pairingby contributing a backbone hydrogen bond, butthe Trp175-Pro1 hydrophobic contact is absent.The main driving force for dimerisation of bB2-Ndomains must thus derive from the interfacehydrophobic surface patch and the surroundingspeci®c polar contacts (Figure 1), and is strongenough to enable domain pairing from dilute sol-ution without requiring a covalent linker to raisethe local protein concentration. This implies thatthe interface between domains evolved beforedomain fusion.

Biophysical measurements have shown that theisolated N-terminal domain of bB2-crystallin ismuch less stable than other bg-crystallin singledomains, including the bB2 C-terminal domain(Wieligmann et al., 1999). The bB2-N domain pro-vides a clear example of where higher order inter-actions provide mutual stabilisation of theprotomer fold (Jaenicke, 1999). Pinpointing thestructural basis for this lower stability is dif®cult,considering pair-wise comparisons of the variousN and C-terminal domains of b and g-crystallinsshow only around 35% sequence identity. How-ever, in the native bB2-crystallin protein, the N-terminal domain is stabilised at the expense of adecrease in stability of the C-terminal domain(Wieligmann et al., 1999). A contribution to thisstability re-distribution may reside in the asym-metric structural feature of the native interface,namely the covering of a surface hydrophobicpatch on the N-terminal domain by a constrainedsequence extension from the C-terminal domain(Figure 2). This mutual stabilisation between non-identical domains probably contributes to theobserved heterologous N:C domain pairing asopposed to N:N and C:C pairings in native oligo-meric b-crystallins. The C-terminal domain onlyforms monomers in solution (Wieligmann et al.,1999), whereas the N-terminal domain provides amore favourable partner.

It is intriguing that at some stage during theevolution of the two-domain bB2-crystallin poly-peptide the N-terminal domain has effectivelyevolved a structural dependence on the C-terminaldomain. The bB2-crystallin polypeptide is involvedin domain swapping (Bax et al., 1990; Schluneggeret al., 1997) and it has been invoked to have arole in solubilising b-crystallin hetero-oligomers(Bateman & Slingsby, 1992) in the eye lens by sub-unit exchange (Slingsby & Bateman, 1990). Theseprocesses require the N-terminal domain tobecome unpaired from the C-terminal domain. It ispossible that the solubilising/exchanging role ofbB2-crystallin is aided by the ease of unfolding ofthe N-terminal domain.

The co-ordinates have been deposited in thePDB with accession code 1e7n.

The N-terminal Domain of �B2-crystallin 257

Acknowledgements

The ®nancial support of the Medical Research Councilis gratefully acknowledged. The work has also been sup-ported by an EU BioMed Grant (BMH4-CT98-3895).

References

Basak, A. K., Kroone, R. C., Lubsen, N. H., Naylor, C. E.,Jaenicke, R. & Slingsby, C. (1998). The C-terminaldomains of gS-crystallin pair about a distorted two-fold axis. Protein Eng. 5, 337-344.

Bateman, O. A. & Slingsby, C. (1992). Structural studieson bH-crystallin from bovine eye lens. Expt. EyeRes. 55, 127-133.

Bax, B., Lapatto, R., Nalini, V., Driessen, H., Lindley,P. F., Mahadevan, D., Blundell, T. L. & Slingsby, C.(1990). X-ray analysis of beta-B2-crystallin andevolution of oligomeric lens proteins. Nature, 347,776-780.

Benedek, G. B. (1971). Theory of transparency of theeye. Appl. Optics. 10, 459-473.

Blundell, T., Lindley, P., Miller, L., Moss, D., Slingsby,C., Tickle, I., Turnell, B. & Wistow, G. (1981). Themolecular structure and stability of the eye lens:X-ray analysis of g-crystallin II. Nature, 289, 771-777.

BruÈ nger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S.,Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J.,Rice, L. M., Simonson, T. & Warren, G. L. (1998).Crystallography & NMR system: a new softwaresuite for macromolecular structure determination.Acta Crystallog. sect. D, 54, 905-921.

Collaborative Computational Project Number 4 (1994).The CCP4 suite - programs for protein crystallogra-phy. Acta Crystallog. sect. D, 50, 760-763.

de Jong, W. W., Lubsen, N. H. & Kraft, H. J. (1994).Molecular evolution of the eye lens. Prog. Retin. EyeRes. 13, 391-442.

Delaye, M. & Tardieu, A. (1983). Short-range order ofcrystallin proteins accounts for eye lens transpar-ency. Nature, 302, 415-417.

Evans, S. V. (1993). Setor: hardware lighted three-dimensional solid model representations of macro-molecules. J. Mol. Graphics, 11, 134-138.

Harding, J. J. & Crabbe, M. J. C. (1984). The lens: devel-opment, proteins, metabolism, and cataract. In TheEye (Davson, H., ed.), 3rd edit., vol. 1B, pp. 207-492,Academic Press, London.

Horwitz, J. (2000). The function of alpha-crystallin invision. Semin. Cell Dev. Biol. 11, 53-60.

Jaenicke, R. (1999). Stability and folding of domain pro-teins. Prog. Biophys. Mol. Biol. 71, 155-241.

Jones, T. A. & Kjeldgaard, M. (1994). Making the ®rsttrace with O. In From First Map to Final Model(Sawyer, L., Isaacs, N. & Bailey, S., eds),

Proceedings of the CCP4 study weekend, 6-7January 1994.

Lubsen, N. H., Aarts, H. J. M. & Schoenmakers, J. G. G.(1988). The evolution of lenticular proteins: theb- and g-crystallin super gene family. Prog. Biophys.Mol. Biol. 51, 47-76.

Marcotte, E. M., Pellegrini, M., Ng, H-L., Rice, D. W.,Yeates, T. O. & Eisenberg, D. (1999). Detecting pro-tein functions and protein-protein interactions fromgenome sequences. Science, 285, 751-753.

Mayr, E.-M., Jaenicke, R. & Glockshuber, R. (1994).Domain interactions and connecting peptides inlens crystallins. J. Mol. Biol. 235, 84-88.

Navaza, J. (1994). AMoRe: an automated package formolecular replacement. Acta Crystallog. sect. A, 50,157-163.

Norledge, B. V., Mayr, E.-M., Glockshuber, R., Bateman,O. A., Slingsby, C., Jaenicke, R. & Driessen, H. P. C.(1996). The X-ray structures of two mutant crystal-lin domains shed light on the evolution of multi-domain proteins. Nature Struct. Biol. 3, 267-274.

Otwinowski, Z. (1993). Oscillation data reductionprogram. In Data Collection and Processing, Proceed-ings of the CCP4 Study, Weekend, 29-30 January 1993,(Sawyer, L., Isaacs, N. & Bailey, S., eds) pp. 56-62,SERC, Daresbury Laboratory, UK.

Palme, S., Jaenicke, R. & Slingsby, C. (1998). Unusualdomain pairing in a mutant of bovine lens gB-crystallin. J. Mol. Biol. 279, 1053-1059.

Schlunegger, M. P., Bennett, M. J. & Eisenberg, D.(1997). Oligomer formation by 3D domainswapping. A model for protein assembly andmissassembly. Advan. Protein Chem. 50, 61-122.

Slingsby, C. & Bateman, O. A. (1990). Quaternary inter-actions in eye lens b-crystallins: basic and acidicsubunits of b-crystallins favor heterologous associ-ation. Biochemistry, 29, 6592-6599.

Slingsby, C., Norledge, B., Simpson, A., Bateman, O. A.,Wright, G., Driessen, H. P. C., Lindley, P. F., Moss,D. S. & Bax, B. (1997). X-ray diffraction and struc-ture of crystallins. Prog. Retin. Eye Res. 16, 3-29.

Wenk, M., Herbst, R., Hocger, D., Kretschmar, M.,Lubsen, N. H. & Jaenicke, R. (2000). Gamma-S-crystallin of bovine and human eye lens: solutionstructure, stability and folding of the intact two-domain protein and its separate domains. Biophys.Chem. 86, 2-3.

Wieligmann, K., Mayr, E.-M. & Jaenicke, R. (1999). Fold-ing and self-assembly of the domains of bB2-crystal-lin from the rat eye lens. J. Mol. Biol. 286, 989-994.

Wistow, G. (1995). Molecular Biology and Evolution ofcrystallins, pp. 37-82, Springer-Verlag, R. G LandesCompany, Heidelberg.

Wistow, G. & Piatigorsky, J. (1988). Lens crystallins - theevolution and expression of proteins for a highlyspecialized tissue. Annu. Rev. Biochem. 57, 479-504.

Edited by A. R. Fersht

(Received 17 July 2000; received in revised form 3 October 2000; accepted 3 October 2000)