Domain Swapping in Human αA and αB Crystallins Affects Oligomerization and Enhances Chaperone-like Activity*

αA and αB crystallins, members of the small heat shock protein family, prevent aggregation of proteins by their chaperone-like activity. These two proteins, although very homologous, particularly in the C-terminal region, which contains the highly conserved “α-crystallin domain,” show differences in their protective ability toward aggregation-prone target proteins. In order to investigate the differences between αA and αB crystallins, we engineered two chimeric proteins, αANBC and αBNAC, by swapping the N-terminal domains of αA and αB crystallins. The chimeras were cloned and expressed in Escherichia coli. The purified recombinant wild-type and chimeric proteins were characterized by fluorescence and circular dichroism spectroscopy and gel permeation chromatography to study the changes in secondary, tertiary, and quaternary structure. Circular dichroism studies show structural changes in the chimeric proteins. αBNAC binds more 8-anilinonaphthalene-1-sulfonic acid than the αANBC and the wild-type proteins, indicating increased accessible hydrophobic regions. The oligomeric state of αANBC is comparable to wild-type αB homoaggregate. However, there is a large increase in the oligomer size of the αBNAC chimera. Interestingly, swapping domains results in complete loss of chaperone-like activity of αANBC, whereas αBNAC shows severalfold increase in its protective ability. Our findings show the importance of the N- and C-terminal domains of αA and αB crystallins in subunit oligomerization and chaperone-like activity. Domain swapping results in an engineered protein with significantly enhanced chaperone-like activity.

␣A and ␣B crystallins, members of the small heat shock protein family, prevent aggregation of proteins by their chaperone-like activity. These two proteins, although very homologous, particularly in the C-terminal region, which contains the highly conserved "␣-crystallin domain," show differences in their protective ability toward aggregation-prone target proteins. In order to investigate the differences between ␣A and ␣B crystallins, we engineered two chimeric proteins, ␣ANBC and ␣BNAC, by swapping the N-terminal domains of ␣A and ␣B crystallins. The chimeras were cloned and expressed in Escherichia coli. The purified recombinant wild-type and chimeric proteins were characterized by fluorescence and circular dichroism spectroscopy and gel permeation chromatography to study the changes in secondary, tertiary, and quaternary structure. Circular dichroism studies show structural changes in the chimeric proteins. ␣BNAC binds more 8-anilinonaphthalene-1-sulfonic acid than the ␣ANBC and the wild-type proteins, indicating increased accessible hydrophobic regions. The oligomeric state of ␣ANBC is comparable to wild-type ␣B homoaggregate. However, there is a large increase in the oligomer size of the ␣BNAC chimera. Interestingly, swapping domains results in complete loss of chaperone-like activity of ␣ANBC, whereas ␣BNAC shows severalfold increase in its protective ability. Our findings show the importance of the N-and C-terminal domains of ␣A and ␣B crystallins in subunit oligomerization and chaperone-like activity. Domain swapping results in an engineered protein with significantly enhanced chaperone-like activity.
␣-Crystallin, a major lens protein having homology with small heat shock proteins (1)(2)(3), prevents aggregation of other proteins like a molecular chaperone (4). We had earlier shown that ␣-crystallin can prevent photo-aggregation of ␥-crystallin, which may have relevance in cataractogenesis (5). By using various non-thermal modes of aggregation, it was shown that chaperone-like activity of ␣-crystallin is temperature-dependent. A structural perturbation above 30°C enhances this activity severalfold (6,7). In order to probe the molecular mechanism of the chaperone-like activity and its enhancement upon structural perturbation, we have been studying ␣-crystallin and its constituent subunits. Our recent study on the ␣A and ␣B homoaggregates showed that, despite high sequence homology, these proteins differ in their stability, chaperone-like activity, and the temperature dependence of this activity (8). This study also indicated different roles for the two proteins in the ␣-crystallin heteroaggregate in the eye lens and as separate proteins in non-lenticular tissues. Several investigators have introduced mutations in ␣A and ␣B crystallins to gain an insight into the structure-function relation (9 -12). Derham and Harding in their recent review (13) list about 30 sitedirected mutations from different laboratories. These mutations either result in some decrease or no change in the protective ability. It is interesting to note that point mutations in both ␣A and ␣B crystallin, R116C and R120G, respectively, result in significant loss of activity and are associated with human diseases (14 -19).
Human ␣A and ␣B crystallins are coded by three exons (20,21) and are thought to have arisen due to gene duplication. They share high sequence homology with the small heat shock proteins, which are found in all organisms, from prokaryotes to humans (22). ␣A and ␣B crystallins are constitutively expressed during normal growth and development. ␣A crystallin is expressed predominantly in the eye lens with small amounts being present in spleen and thymus (23), whereas ␣B crystallin is expressed not only in the eye lens, but also in several other tissues such as heart, skeletal muscle, placenta, lung, and kidney (24,25). The main function of these proteins in the lens appears to provide transparency and prevent precipitation by binding to other aggregation-prone proteins. In the lens, ␣A and ␣B crystallins exist as heteroaggregates of approximately 800 kDa. Both the recombinant ␣A and ␣B crystallins exist as high molecular mass oligomeric proteins of approximately 640 and 620 kDa, respectively (26). The size of these proteins can vary a little depending on the pH and ionic strength, and they differ in their structure, function, tissue expression, and abnormal deposition in disease.
␣B crystallin has a heat shock element upstream to the gene and is induced during stress (3,28). Apart from maintaining lens transparency, its in vivo functions include interaction with intermediate filaments (29) and regulation of cytomorphological rearrangements during development (30). ␣B crystallin is hyperexpressed in neurological disorders such as Alzheimer's' disease, Creutzfeldt-Jacob disease, and Parkinson's disease (31)(32)(33).
The charged C-terminal domain is conserved in all the members of the small heat shock protein family, whereas the hydrophobic N-terminal domain is variable in length and sequence similarity (34). The N-and C-terminal domains are thought to form two structural domains with an exposed Cterminal extension (35). To investigate the role of the N-terminal domains in the differential structural and functional properties of human ␣A and ␣B crystallins, we have swapped their N-terminal domains coded by exon 1. A unique XmnI restriction site at the beginning of the ␣-crystallin domain in a 20-nucleotide stretch in exon 2, with 100% sequence identity in human ␣A and ␣B crystallin genes, has been used to create chimeric proteins ␣ANBC and ␣BNAC. We have used biophysical methods to study the structural and functional properties of wild-type ␣A and ␣B crystallins as well as the chimeras in order to get an insight into the effect of swapping and the role of the N-terminal domain in oligomerization and chaperonelike activity.

Sequencing of Human Chimeric ␣ANBC and ␣BNAC Crystallins
Sequencing was done with T7 promoter primer using the dye terminator cycle sequencing kit (Perkin-Elmer) in an 3700 ABI automated DNA sequencer. The coding regions of both the ␣ANBC and ␣BNAC chimeras were found to be mutationless with no change in the reading frame.

Fluorescence Measurements
Intrinsic Fluorescence-Intrinsic fluorescence spectra of wild-type and chimeric proteins were recorded using a Hitachi F-4000 fluores-cence spectrophotometer with the excitation wavelength of 295 nm. The excitation and emission band passes were set at 5 and 3 nm, respectively. Intrinsic fluorescence spectra were recorded using 0.2 mg/ml protein in 10 mM phosphate buffer, which was incubated at 37°C for 10 min.
8-Anilino-1-naphthalenesulfonic Acid (ANS) Binding-Wild-type and chimeric proteins (0.2 mg/ml) in 10 mM phosphate buffer, pH 7.4, containing 100 mM NaCl were equilibrated at 37°C in the sample holder of Hitachi F-4000 fluorescence spectrophotometer using a Julabo thermostated water bath for 10 min. To these protein samples, 20 l of 10 mM ANS was added. Fluorescence spectra were recorded with an excitation wavelength of 365 nm. The excitation and emission band passes were 5 and 3 nm, respectively.

Circular Dichroism Studies
Circular dichroism spectra were recorded using a Jasco J-715 spectropolarimeter. All spectra reported are the average of 5 accumulations. Far-and near-UV CD spectra were recorded using 0.05-and 1-cm pathlength cuvettes, respectively.

Assay for Protein Aggregation
Chaperone-like activity of the wild-type and chimeric proteins was studied by the insulin aggregation assay (6,36). The extent of protection by the wild-type ␣A and ␣B crystallins and the chimeric proteins was studied by incubating insulin (0.2 mg/ml) with various concentrations of the wild-type and chimeric proteins for 10 min at 37°C. Aggregation was initiated by the addition of 20 l of 1 M dithiothreitol (DTT) after the incubation.

RESULTS AND DISCUSSION
Construction and Expression of the Chimeric Human ␣A and ␣B Crystallins-Human ␣A and ␣B crystallin genes have a unique site for the restriction enzyme XmnI at the beginning of exon 2. A 20 nucleotide stretch at the XmnI site in both ␣A and ␣B crystallins has 100% sequence identity. Swapping of the domains does not disturb the reading frame (Fig. 1). Since XmnI site is slightly into the exon II, the excised N-terminal fragment has additional 15 amino acids. Of the 15 amino acids, 8 are identical and the rest are chemically conserved. Ligation of the N-terminal domain of ␣A crystallin with the C-terminal region of ␣B crystallin results in the chimeric polypeptide ␣ANBC crystallin, which is 171 amino acids long. Similarly, the ligation of the N-terminal region of ␣B crystallin with C-terminal domain of ␣A crystallin creates polypeptide ␣BNAC crystallin that is 177 amino acids long. Henceforth, the chimeras are referred to as ␣ANBC and ␣BNAC. Overexpression and purification of the chimeric proteins was carried out as described earlier for the wild-type proteins. The wild-type and chimeric proteins were purified to greater than 95% homoge-neity, as judged by SDS-polyacrylamide gel electrophoresis (data not shown), and moved as ϳ20-kDa proteins as expected. Interestingly, when ␣ANBC is eluted from a Mono Q ion exchange column with a 0 -2 M NaCl gradient, it elutes at ϳ100 mM NaCl like the wild-type ␣B crystallin. On the other hand, ␣BNAC elutes at ϳ350 mM NaCl, similar to wild-type ␣A crystallin. The number of positively and negatively charged amino acids are identical in wild-type ␣A crystallin and ␣BNAC (ArgϩLys ϭ 20; AspϩGlu ϭ 25) and in wild-type ␣B crystallin and ␣ANBC (ArgϩLys ϭ 24; AspϩGlu ϭ 25). A recently proposed model for ␣-crystallin suggests that the hydrophobic N-terminal domain is mostly buried in the oligomer (37). Thus, the C-terminal domain may largely determine the surface charge distribution of the proteins. This could be one of the reasons for the similarity in Mono Q elution profiles of wild-type proteins and chimeras that contain C-terminal regions identical to those of the wild-type proteins.
Superose-6 Gel Permeation Chromatography-To investigate the consequences of domain swapping on the molecular masses, chimeric and wild-type proteins were chromatographed on a FPLC Superose-6 gel filtration column (Fig. 2). The average molecular masses of wild-type ␣A and ␣B crystallins were observed to be ϳ640 and ϳ620 kDa, respectively. These sizes are consistent with earlier reports (16,26). The chimera ␣ANBC elutes at the same elution volume as that of wild-type ␣B with an apparent molecular mass of ϳ620 kDa. However the ␣BNAC chimera oligomerizes into large polydisperse aggregates, with species exceeding 2000 kDa. This finding shows an important difference in ␣A and ␣B crystallins. The ␣ANBC chimera consisting of the N-terminal domain of ␣A crystallin and the C-terminal domain of ␣B crystallin still possesses the oligomer size of wild-type ␣A and ␣B crystallins. Thus, it appears that the N-terminal domain of ␣B crystallin can be replaced by the N-terminal domain of ␣A crystallin with no alteration in the oligomeric status. However, the N-terminal domain of ␣B crystallin in fusion with the C-terminal domain of ␣A crystallin forms very large aggregates, probably due to altered packing of the subunits with an increase in intersub-unit interaction. This kind of increase in the oligomer size was earlier observed in the R116C mutant of ␣A crystallin (15). The monomer sizes of the proteins of the small heat shock protein family range from 12 to 43 kDa. Almost all members of this family multimerize to form large aggregates, ranging in size from 400 to 800 kDa with only one exception till date; sHSP 12.6 of Caenorhabditis elegans, which has the shortest N-and C-terminal domains, is monomeric (38). The N-terminal domain is variable in both length and sequence in the sHSP superfamily, which might be responsible for the varying multimeric sizes. Bova et al. (27) showed that sequential truncation from the N terminus of ␣A crystallin reduces oligomeric size. In the present study, the sequence length of the swapped N-terminal domain between ␣A and ␣B crystallin is similar, so the variation in sequence of this domain is likely to be responsible for the differential multimerization of the chimeric proteins.
Intrinsic and ANS Fluorescence-The emission maximum of tryptophan is highly sensitive to solvent polarity and depends on the accessibility of tryptophan residues to the aqueous phase. Fig. 3 shows the intrinsic fluorescence spectra of wildtype and chimeric proteins. The intrinsic fluorescence spectra of the wild-type ␣B crystallin and ␣BNAC are similar. Both the tryptophans are present in the N-terminal domain, which are likely to be in a similar environment even after domain swapping. A slight blue shift, noticeable in the red region of the emission profile of ␣BNAC, compared with the wild-type ␣B crystallin suggests that the tryptophans in the chimera are marginally less solvent accessible. The intrinsic fluorescence spectra of the lone tryptophan of wild-type ␣A crystallin, which is present in the N-terminal domain, and ␣ANBC are similar, indicating no alteration of the tryptophan environment in the chimeric ␣ANBC protein with respect to the wild-type ␣A crystallin. Fig. 4 shows the spectra of ANS in the presence of wild-type and chimeric proteins. ANS fluorescence spectra show marked differences in emission intensity with no apparent change in emission maxima. The ␣ANBC chimera binds the least amount of ANS among all the proteins compared. The ␣BNAC chimera, on the other hand, binds ANS several times more when compared with wild-type ␣B crystallin, wild-type ␣A crystallin, and ␣ANBC chimera. This finding suggests that there are more hydrophobic regions accessible to ANS in the ␣BNAC chimera than in ␣ANBC chimera. The molecular basis for this finding is not yet clear. However, the gel permeation chromatography data together with ANS fluorescence suggest that ␣BNAC might be forming a large porous oligomer.
Circular Dichroism Measurements of Chimeric ␣ANBC and ␣BNAC Crystallins- Fig. 5 shows far-UV circular dichroism spectra of wild-type and chimeric proteins. CD spectra of wildtype ␣A and ␣B crystallins, shown in panel A, are comparable with the CD spectra of recombinant human ␣A and ␣B crys- tallins reported earlier (15,16,26). Both the spectra show characteristic ␤-sheet protein profile as expected. Chimeric proteins also show ␤-sheet CD profiles. The CD spectrum of ␣ANBC is comparable to the spectra of wild-type ␣A and ␣B crystallins. However, ␣BNAC shows increased ellipticity.
Near-UV CD spectra (Fig. 6) also show a similar trend. Spectra of wild-type ␣A and ␣B are comparable to earlier reported spectra for recombinant human ␣A and ␣B crystallins (15). The CD spectrum of the chimeric ␣ANBC is comparable to that of ␣B crystallin with increased chirality for ␣ANBC. The CD spectrum of ␣BNAC on the other hand is comparable to that of wild-type ␣A crystallin.
Domain swapping results in some change in secondary and tertiary structure of ␣ANBC with observable change only in the secondary structure for ␣BNAC.
Chaperone-like Activity-Insulin B-chain aggregates in the presence of DTT. At 37°C a 1:1 (w/w) ratio of wild-type ␣A and ␣B crystallin to insulin prevented this aggregation completely. At ratios of 1:2 and 1:4, aggregation was prevented to lesser extents, as shown in Fig. 7 (panels A and B). Interestingly, the chimera ␣BNAC showed enhanced chaperone-like activity. The initial scatter value for ␣BNAC chimera without insulin was very high. The large molecular size of ␣BNAC could be respon-sible for the high scatter. We had earlier observed a similar high initial scatter value for the R116C mutant of ␣A crystallin, which also forms a large aggregate (Ͼ2000 kDa) (16). The data were normalized to determine the protective ability of the ␣BNAC protein. At 37°C complete protection was observed at a 1:6 w/w ratio of ␣BNAC to insulin. Significant protection was observed even at 1:8, 1:12, and 1:16 ratios of ␣BNAC to insulin (Fig. 7D). The ␣BNAC chimera shows 3-4-fold increase in the DTT-induced aggregation of 0.2 mg/ml insulin alone (Ins) and in the presence of 1:1, 1:2, and 1:4 w/w wild-type ␣A crystallin:insulin, respectively. B, effect of wild-type ␣B crystallin. Panel shows aggregation of 0.2 mg/ml insulin alone (Ins) and in the presence of 1:1, 1:2, and 1:4 w/w wild-type ␣B crystallin:insulin, respectively. C, effect of ␣ANBC chimera. Panel shows aggregation of 0.2 mg/ml insulin alone (Ins) and in the presence of 1:2 and 1:1 w/w ␣ANBC:insulin, respectively. D, effect of ␣BNAC chimera. Panel shows aggregation of 0.2 mg/ml insulin alone (Ins) and in the presence of 1:6 1:8, 1:12, and 1:16 w/w ␣BNAC:insulin, respectively. chaperone-like activity compared with the wild-type proteins. ␣ANBC, in contrast, shows complete loss of chaperone-like activity. A 1:2 (w/w) ratio of ␣ANBC to insulin does not show any protective ability toward DTT-induced aggregation of insulin. Increasing the ␣ANBC ratios to 1:1 and 2:1 w/w with respect to insulin does not show any increase in protection (Fig.  7C). In fact, ␣ANBC promotes the aggregation process as observed by increased light scattering.
The swapped N-terminal domain (exon 1 encoded) is comparable in length between human ␣A and ␣B crystallins. There are some differences in the sequences in this region. One of the prominent differences is the increase in the number of proline residues. The N-terminal domain of ␣A crystallin contains 5 proline residues, whereas the same region for ␣B crystallin has 9 proline residues (two prolines in tandem). The swapping alters the number of proline residues in the chimeric proteins. ␣BNAC contains 9 prolines in its N-terminal domain, a gain of 4 prolines in comparison to the same region of wild-type ␣A crystallin. Far-UV CD spectrum shows some enhancement in the secondary structure. Whether the local secondary structural changes can alter the subunit topology and consequently intersubunit interactions remains to be investigated. Although we point out differences in the number of proline residues, there are other sequence variations, and marginal changes in predicted pI and the total length of the chimeric proteins. Clearly discernible changes are oligomeric status, accessible hydrophobic surfaces, and chaperone-like activity.
It is interesting to note that, despite being similar to wildtype ␣B crystallin in the aggregate molecular mass and circular dichroism spectra, the chimeric ␣ANBC possesses no chaperone-like activity. The most important difference between the two chimeric proteins is the accessible hydrophobicity. ANS, a hydrophobicity probe, very clearly distinguishes the two chimeric proteins. We believe that the lack of accessible surface hydrophobicity, probably due to altered subunit packing in ␣ANBC chimera, results in its loss of chaperone-like activity.
The enhanced chaperone-like activity of ␣BNAC chimera could be because of the exposure and availability of more hydrophobic surfaces when compared with the wild-type proteins. Increased ANS binding of the ␣BNAC chimera supports this possibility. We observed an increase in oligomeric size and chaperone-like activity in the case of the ␣BNAC chimera. However, the increase in size and enhancement of chaperonelike activity may not be necessarily correlated. The point mutation R116C in ␣A crystallin leads to increased oligomer size but results in significant loss of chaperone-like activity. Swapping the N-terminal domain between human ␣A and ␣B crystallins makes a more effective chaperone in the case of ␣BNAC chimera, whereas ␣ANBC chimera loses its protective abilities completely. To the best of our knowledge, this is the first report where a 3-4-fold increase in chaperone-like activity is observed. This phenomenon may have a therapeutic significance in diseases occurring due to protein misfolding.