Deamidation Affects Structural and Functional Properties of Human (cid:1) A-Crystallin and Its Oligomerization with (cid:1) B-Crystallin*

To determine the effects of deamidation on structural and functional properties of (cid:1) A-crystallin, three mutants (N101D, N123D, and N101D/N123D) were generated. Deamidated (cid:1) B-crystallin mutants (N78D, N146D, and N78D/N146D), characterized in a previous study (Gupta, R., and Srivastava, O. P. (2004) Invest . Ophthal-mol . Vis . Sci . 45, 206–214) were also used. The biophysi-cal and chaperone properties were determined in ( a ) homoaggregates of (cid:1) A mutants (N101D, N123D, and N101D/N123D) and ( b ) reconstituted heteroaggregates of (cid:1) -crystallin containing (i) wild type (cid:1) A (WT- (cid:1) A): WT- (cid:1) B crystallins, (ii) individual (cid:1) A-deamidated mu-tants:WT- (cid:1) B crystallins, and (iii) WT- (cid:1) A:individual (cid:1) B-deamidated mutant crystallins. Compared with the WT- (cid:1) A, the three (cid:1) A-deamidated mutants showed reduced levels of chaperone activity, alterations in secondary and tertiary structures, and larger aggregates. These altered properties were relatively more pronounced in the mutant N101D compared sample elution. the dialysis three of guanidine hydrochloride (GdnHCl) GdnHCl determined. Using this method, heteroaggregate formation between the following species at a of 3:1 ( (cid:1) A: (cid:1) B)

␣-Crystallin, the most abundant protein in lens mature fiber cells, constitutes ϳ35% of the total lens protein. In vivo, the ␣A and ␣B subunits at a ratio of 3:1 form an oligomer of 800 kDa. Both ␣Aand ␣B-crystallins are small heat shock proteins (Hsps), 1 and show molecular chaperone activity to protect pro-teins from aggregation in the eye lens (1,2). Because of this property, ␣-crystallin is believed to play a crucial role in maintaining the lens transparency.
Like other small heat shock proteins, ␣-crystallin also contains a highly conserved sequence of 80 -100 residues (residues 62-143 in ␣Aand 66 -147 in ␣B-crystallin) called the ␣-crystallin domain (3,4). Based on similarities with the structure of other Hsps, it is believed that the N-terminal region (residues 1-62 in ␣Aand 1-66 in ␣B-crystallin(s)) of ␣-crystallin forms an independently folded domain, whereas the C terminus (referred as the C-terminal extension; residues 143-173 in ␣Aand 147-175 in ␣B-crystallin) is flexible and unstructured (4). Previous reports show that the removal of N-terminal residues (56 residues) and C-terminal extensions (32-34 residues) of ␣Aand ␣B-crystallins lead to improper folding, reduced chaperone activity, and formation of trimers or tetramers (5)(6)(7)(8)(9). The ␣-crystallin domain is believed to be engaged in the subunitsubunit interactions, but the individual amino acids in subunit interactions and chaperone activity have not been fully identified. Nevertheless, two disease-related point mutations of a highly conserved Arg at equivalent positions in ␣A (R116C) and ␣B (R120G) caused structural changes that lead to hereditary cataracts (10,11).
During cataract development, the conformational changes, unfolding, and subsequent cross-linking of crystallins are believed to lead to accumulation of insoluble cross-linked product of ␣-, ␤-, and ␥-crystallins. Present literature suggests a multifactorial mechanism for the development of cataract-specific cross-linked species, which might be driven by post-translational modifications. These included disulfide bonding (12), glycation (13), oxidation of Trp and His residues (14,15), deamidation (16), and transglutaminase-mediated cross-linking (17). However, despite identification of these modifications in crystallins, their exact roles in the mechanism of crystallin cross-linking remain poorly understood.
Age-related deamidation of crystallins seems to be one of the major post-translational modifications that occurs frequently. In vivo, the deamidation of ␣-, ␤-, and ␥-crystallins has been shown in several studies (18 -21). In ␣A-crystallin, greater deamidation of Gln-50 and Asn-101 has been reported compared with Gln-6 and Asn-123 in high molecular weight and water-insoluble proteins (18,22,23). Likewise, in ␣B-crystallin, deamidation of Gln and Asn residues has been shown (23). 2 Because deamidation introduces a negative charge, it is believed to cause alterations in protein tertiary structure, affecting their structural and functional properties. However, the effects of deamidation on structure and function of ␣Aand ␣B-crystallins have not been investigated until recently. Our recent report showed that the deamidation of the Asn-146 residue and not the Asn-78 residue in human ␣B-crystallin resulted in significant changes in its structural and functional properties (24). Although both Asn-78 and Asn-146 are present in the conserved regions of the ␣B-crystallin, the dramatic effects of deamidation of Asn-146, which is close to the Cterminal region, were surprising. A similar investigation of the effect of deamidation of Asn-101 and Asn-123 (also present in the conserved region) of ␣A-crystallin on its structural and functional properties has been lacking in the literature. Additionally, although interaction between ␣Aand ␣B-crystallins in their oligomeric form is needed for the molecular chaperone activity and for the maintenance of lens transparency (25)(26)(27)(28), the effect of deamidation on subunit interactions and chaperone activity is also presently unknown. However, another intriguing question is how deamidation in ␣A and ␣B subunits affects the formation of the ␣-crystallin heteroaggregate and affects its structural and functional properties.
It has been reported that the ␣Aand ␣B-crystallin subunits constantly undergo rapid exchange, which might be responsible for oligomer stability and suppression of the nonspecific aggregation of crystallins (28). Consistent with this observation was a recent report (29) showing that mutation of R116C in ␣A-crystallin resulted in altered subunit composition with WT-␣B in the heteroaggregates and increased oligomer size compared with the oligomers of WT-␣A and WT-␣B subunits. Similarly, the mutation of Arg to Asp in ␣A-crystallin resulted in a drastic change in protein structure (30). This is consistent with Studer's study, which showed that chaperone activity is coupled to multimerization in bacterium ␣-HSP protein from Bradyrhizobium japonicum (31).
To answer the questions raised above, the objectives of this study were to determine: (a) the effect of deamidation of Asn on structural and functional properties of ␣A-crystallin and (b) the effect of deamidation of ␣Aor ␣B-crystallins on oligomerization with WT-␣A/WT-␣B subunits and their structural and functional properties. The results presented show that the deamidation of both Asn-101 and Asn-123 in ␣A-crystallin altered its structural and functional properties but relatively more so by deamidation of the Asn-101 residue. Similarly, compared with heteroaggregates of WT-␣A-and WT-␣B-crystallins, the heteroaggregates containing deamidated ␣A or ␣B mutants and their counterpart WT proteins showed higher molecular mass, altered tertiary structure, lower exposure of hydrophobic surface, and reduced chaperone activity. Further, heteroaggregate-containing WT␣Acrystallin and deamidated ␣B mutants showed lower chaperone activity, smaller sized oligomers, and a 3-fold lower subunit exchange rate compared with heteroaggregates containing deamidated ␣A mutants and WT␣B-crystallin.

Materials
The restriction endonucleases NcoI and EcoRV, the molecular weight protein markers and DNA markers were purchased from Amersham Biosciences and Promega (Madison WI), respectively. The T7 promoter, T7 terminator, and other primers used in the study were obtained either from Invitrogen or from the University of Alabama at Birmingham Oligo Synthesizing Core Facility. Molecular biology grade chemicals were purchased from Sigma, unless stated otherwise. All chemicals for two-dimensional gel electrophoresis were from Amersham Biosciences or Bio-Rad (South San Francisco, CA). DEAE-Sephacel, agarose gel A1.5, and butyl-Sepharose were obtained from Amersham Biosciences. Two fluorescent probes, lucifer yellow iodoacetamide (LYI) and 4-acetamido-4Ј-[(iodoacetyl) amino]stilbene-2,2Ј-disulfonic, disodium salt (AIAS) were purchased from Molecular Probes, Inc. (Eugene, OR). Unless indicated otherwise, all other chemicals used in this study were purchased from Fisher or Sigma.

Bacterial Strains and Plasmids
E. coli BL21 (DE3) pLysS bacterial strain was obtained from Promega. The human ␣B-crystallin cDNA cloned on a plasmid pDIRECT was received from Dr. Mark Petrash (Washington University, St. Louis, MO). Cells were propagated in Luria broth, and recombinant bacteria were selected using ampicillin.

Site-specific Mutagenesis
Deamidation of Asn to Asp at position 101 or 123 or both in ␣A-cDNA was introduced using the QuikChange site-directed mutagenesis kit and following the manufacturer's instructions (Stratagene, La Jolla, CA). Recombinant human ␣A-crystallin coding sequence was cloned in pDIRECT as described earlier (32). These constructs were used as templates with the mutated primers (where Asn was replaced by Asp) ( Table I) to generate site-specific mutations. Briefly, 25 ng of template was used, and the PCR conditions were as follows: predenaturing at 95°C for 30 s, followed by 16 cycles of denaturing at 95°C for 30 s, annealing at 55°C for 1 min, and extensions at 68°C for 5 min. After digestion with DpnI, 2 l of the PCR product was used to transform XL1-Blue supercompetent cells provided with the above Stratagene kit. The positive clones were confirmed by DNA sequencing.

Expression and Purification of Wild-type and Deamidated Mutant Proteins
Escherichia coli BL21 (DE3) pLysS was transformed with mutant amplicons using a standard E. coli transformation procedure. The crystallin constructs were grown in LB medium to a cell density of ϳ0.5-0.6 at 600 nm and induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at 37°C for 3 h. The purification method for the WT and three ␣A mutants was similar to one used previously to purify the WT and deamidated ␣B mutants (24). Cells were harvested and, after a freezethaw cycle, were lysed in 50 mM Tris-HCl buffer, pH 7.5, containing 0.5 mM EDTA and 0.3 mM NaCl (TEN buffer). Following centrifugation at 17,000 ϫ g for 20 min, the supernatant fraction was collected and dialyzed against 50 mM Tris-HCl buffer, pH 7.9, containing 0.5 mM EDTA and 1 mM DTT (TED buffer), and subjected to DEAE-Sephacel ion exchange chromatography using a 3 ϫ 30-cm column. The bound proteins were eluted with a gradient of 0 to 0.5 M NaCl in TED buffer. The ␣A-crystallin species-containing fractions, identified by SDS-PAGE using a 15% polyacrylamide gel (33), were pooled and dialyzed against 50 mM phosphate buffer, pH 7.0, at 5°C for 24 h. Next, the preparation was subjected to hydrophobic interaction chromatography using a butyl-Sepharose column (3 ϫ 20 cm). The column equilibration buffer was 50 mM phosphate buffer, pH 7.0, containing 0.5 M (NH 4 ) 2 SO 4 , and the bound proteins were eluted with a decreasing (NH 4 ) 2 SO 4 concentration (i.e. 0.5 M to 0 M). The crystallin-containing fractions were identified as above by SDS-PAGE, pooled, and concentrated by ultrafiltration using an Amicon stirred cell (Millipore Corp., Billerica, MA). The purity of WT and mutant ␣A-crystallin species was examined by SDS-PAGE and two-dimensional gel electrophoresis as described below. The protein concentrations were determined with either a Pierce protein determination kit or absorbance at 280 nm. The total protein yields of different ␣A species were as follows: WT-␣A-crystallin, 1 mg/ml; mutant N101D, 1.6 mg/ml; mutant N123D, 1.3 mg/ml; and mutant N101D/N123D, 1 mg/ml.

Two-dimensional Gel Electrophoresis
The protein samples were mixed with resolubilization buffer (5 M urea, 2 M thiourea, 2% CHAPS, 2% caprylylsulfobetaine 3-10, 2 mM tributyl phosphine, 40 mM Tris, pH 8.0) in the ratio 2:1, respectively (34). Each preparation was subjected to two-dimensional gel electrophoresis (isoelectric focusing in the first dimension and SDS-PAGE in the second dimension). Isoelectric focusing separation was carried out using Immobiline dry strips (pH range of 3-10) by following the manufacturer's instructions (Amersham Biosciences). SDS-PAGE in the second dimension was performed using a 15% polyacrylamide gel by Laemmli's method (33).

Labeling of Recombinant Crystallins with Different Fluorescence Probes
The cysteine residue in recombinant ␣A-crystallin species was covalently labeled with LYI (8.4 mM) by following the method of Bova et al. (28) and the manufacturer's instructions (Molecular Probes). The desired crystallin species (1 mg/ml) was dissolved in 20 mM MOPS buffer, pH 7.9, containing 100 mM NaCl. The disulfide bonds in proteins were reduced using a 10-fold molar excess of Tris-(2-carboxyethyl)phosphine) (15 l of 1 M solution) to liberate free thiol for maximum labeling followed by incubation first at room temperature for 14 h and then at 37°C for 6 h. Upon completion of the reaction, a 5-fold excess of ␤-mercaptoethanol was added to consume excess thiol-reactive reagent. Similarly, using exactly the above conditions, the recombinant ␣Bcrystallin species were labeled with AIAS (8.4 mM) by incubation at room temperature for 12 h. The extent of labeling was determined spectrophotometrically using molar extinction coefficients of 47,000 mol Ϫ1 cm Ϫ1 at 336 nm for AIAS and 11,000 mol Ϫ1 cm Ϫ1 at 426 nm for LYI (values given by Molecular Probes), and corrected protein concentration for the contribution of the dye at 280 nm was determined. The unbound (AIAS) dye was then separated from fluorescence-labeled protein by a Sephadex G-25 column chromatography, using 50 mM phosphate buffer, pH 7.5, containing 100 mM NaCl, and 2 mM EDTA for column equilibration and sample elution. Similarly, the unbound LYI was separated from the protein preparations by dialysis against 50 mM phosphate buffer, pH 7.5, for 24 h at 4°C with three changes of the buffer.

Reconstitution of ␣-Crystallin Heteroaggregates
The purified WT-␣A was mixed with purified WT-␣B or the individual purified ␣B-deamidated mutants (i.e. N78D, N146D, or N78D/ N146D) at a 3:1 ratio. Similarly, purified WT-␣B was mixed with purified individual ␣A-deamidated mutants (i.e. N101D, N123D, or N101D/N123D) at a 1:3 ratio. The molecular weights of the resulting heteroaggregates were determined using the static light scattering (SLS) method as described below. The chaperone activities, surface hydrophobicity, and intrinsic tryptophan fluorescence of the heteroaggregates were also determined using the methods described below. The rate of subunit exchange in the heteroaggregates (WT-␣A:WT-␣B, WT-␣A:␣B-N146D, ␣A-N101D:WT-␣B, ␣A-N123D:WT-␣B, and ␣A-N101D/ N123D:WT-␣B) was determined by mixing the AIAS-labeled WT-␣B or individual ␣B mutants with LYI-labeled WT-␣A or individual ␣A mutants and using the fluorescence resonance energy transfer (FRET) method as described below.
In a separate experiment, the formation of heteroaggregates following denaturation of ␣A or ␣B species by guanidine hydrochloride (GdnHCl) and renaturation after the removal of GdnHCl by dialysis was determined. Using this method, the heteroaggregate formation between the following species at a ratio of 3:1 (␣A:␣B) was examined: The purified WT-␣A/deamidated ␣A mutants were mixed with their purified counterparts in a 3:1 ratio (␣A:␣B), followed by denaturation in 4 M GdnHCl and renaturation as described by the Bera et al. method (30). Briefly, the individual heteroaggregates were mixed to 4 M GdnHCl (final concentration) and incubated at 4°C for 6 h. This was followed by dialysis against 50 mM Tris-HCl, pH 7.9, at 5°C for 48 h with four changes of the buffer. To ascertain whether the GdnHCl denaturation and renaturation treatment provided a functionally active heteroaggregate, their chaperone activity against insulin as a target protein was determined.

Determination of Structural and Functional Properties of WT-␣A-crystallin and Its Deamidated Mutants
Chaperone Activity Assays-The chaperone activity was determined by using three target proteins (i.e. insulin, citrate synthase (CS), and recombinant human ␥D-crystallin) essentially by the methods previously described (24). The aggregation of insulin by 20 mM DTT at 25°C, citrate synthase at 43°C, and ␥D-crystallin at 63°C either in the absence or at varying concentrations of different ␣A-crystallin species (i.e. WT-␣A, N101D, N123D, and N101D/N123D mutants) was determined. The aggregation was monitored using light scattering at 360 nm with time (60 min) using a Shimadzu UV-VIS scanning spectrophotom-eter (model UV2101 PC) equipped with a six-cell holder (Shimadzu model CPS-260) and a temperature controller (Shimadzu model CPS 260).
Fluorescence Studies-All fluorescence spectra were recorded in corrected spectrum mode using a Shimadzu RF-5301PC spectrofluorometer with excitation and emission band passes set at 5 and 3 nm, respectively. The total intrinsic Trp fluorescence intensities of the WT-␣A, the three deamidated ␣A mutants, and the ␣-crystallin heteroaggregates (0.15 mg/ml each), dissolved in 10 mM phosphate buffer, pH 7.4, containing 100 mM NaCl, were recorded with an excitation at 295 nm and emission between 300 and 400 nm.
The binding of a hydrophobic probe, 8-anilino-1-naphthalenesulfate (ANS) to WT-␣A, deamidated ␣A mutants, and the ␣-crystallin heteroaggregates was determined by recording fluorescence spectra with excitation at 390 nm and emission between 400 and 600 nm. In these experiments, 15 l of 0.8 mM ANS (dissolved in methanol, 12 M) was added to the protein preparations (0.15 mg/ml, dissolved in 10 mM phosphate buffer, pH 7.4, containing 100 mM NaCl) and incubated for 10 min at 37°C. In another experiment, the WT-␣A and the three ␣A mutants were heated with the above described ANS concentration at 43°C for 10 min, and fluorescence spectra were determined after cooling the preparations to room temperature.
Circular Dichroism Studies-To investigate the conformational changes in the WT-␣A and the three deamidated ␣A mutants, the CD spectra were determined using a JasCo spectropolarimeter model 62DS at room temperature. The ␣A-crystallin preparations at 0.1 mg/ml or 1 mg/ml (dissolved in 50 mM potassium phosphate buffer, pH 7.4) were used for recording the far-and near-UV CD spectra, respectively. The path length was 0.1 and 1 cm during the far-and near-UV CD spectra determination, respectively. The spectra reported are the average of five scans, corrected for buffer blank and were smoothed. Secondary structures were estimated using the PROSEC program (35).
Static Light Scattering-Determination of molecular weights of WT-␣A, the three deamidated ␣A mutants, and the ␣-crystallin heteroaggregates was carried out using an SLS instrument (model 202, Precision Detectors), which exploits gel permeation chromatography, coupled to low and high angle laser light scattering, and a differential refractive index detector. All proteins were dissolved in 50 mM Tris-HCl, pH 7.9, and preparations were filtered through a 0.2-m filter prior to their analysis. Results utilized both 90 and 15°light scattering detection. To reconstitute ␣-crystallin heteroaggregates (1 mg/ml), the purified preparations of deamidated ␣A and deamidated ␣B species were mixed at a ratio of 3:1 (␣A/␣B).
Measurement of Subunit Exchange Rate-FRET was carried out to determine the subunit exchange rate in ␣-crystallin heteroaggregates containing deamidated ␣A/␣B and their counterpart WT proteins as described by Bova et al. (28). Briefly, the exchange reaction was initiated by mixing 0.4 mg/ml LYI-labeled WT-␣A or LYI-labeled individual ␣A mutants with 0.4 mg/ml AIAS-labeled WT-␣B crystallin. At the desired time intervals (described in the figure legends), 20 l of the reaction mixture was withdrawn and diluted 100-fold with buffer A (50 mM sodium phosphate, pH 7.5, containing 100 mM NaCl and 2 mM DTT). The emission spectrum of the aliquots (excitation at 426 nm and emission at 525 nm) was recorded using a Shimadzu spectrofluorometer (model RF 5301PC). The rate constant was obtained by fitting the data to the exponential function F(t) ϭ C 1 ϩ C 2 e Ϫkt , where F(t) represents the fluorescence intensity at 525 nm, and k is the rate constant for subunit exchange. The constants C 1 ϩ C 2 ϭ 1 at time 0, and C 1 is the fluorescence intensity at time ϰ.

Confirmation of Specific Mutations at Desired Sites in ␣A-crystallin
In two individual ␣A-crystallin mutants, the Asn-101 or Asn-123 residues were changed to Asp, and in another double mutant, both Asn-101 and Asn-123 were changed to Asp. The resulting three deamidated proteins are referred as N101D, N123D, and N101D/N123D mutants throughout.
DNA sequencing results confirmed the mutations of Asn to Asp at the desired positions in the three mutants. To confirm this further, the expressed WT-␣A-crystallin and the three mutant proteins were analyzed by the MALDI-TOF mass spectrometric method after trypsin digestion. The isotopic distribution of tryptic fragments further confirmed mutation(s) of Asn to Asp at the desired sites in the three mutant proteins. The tryptic peptides with a mass of 1627.4 (residues 100 -112; HNERQDDHGYISR; Fig. 1B) and 2785.7 (residues 120 -145; LPSNVDQSALSCSLSADGMLTFCC; Fig. 1C) were detected in the N101D and N123D mutants, respectively. The corresponding comparative peaks showed mass of 1626.4 and 2784.9 in WT-␣A-crystallin (Fig. 1A). A gain of one mass unit in the mutant proteins compared with WT-␣A suggested mutation of Asn-101 and Asn-123 to Asp in the above two mutants. Similar results (i.e. gain of 1 mass unit) were observed in the two tryptic peptides with masses of 1627.4 and 2785.7 from the N101D/N123D mutant compared with WT-␣A, confirming double mutations of Asn-101 and Asn-123 to Asp in the protein.

Expression and Purification of Human Recombinant WT-␣A and Three Deamidated ␣A Mutants
The WT-␣A crystallin and the three mutated ␣A species were expressed in E. coli. The SDS-PAGE and MALDI-TOF mass spectrometric analyses showed the expression of full-length recombinant ␣A-crystallin. The expressed proteins were purified to homogeneity by a combination of methods as described under "Experimental Procedures," and each species on SDS-PAGE analysis showed a single ϳ20-kDa protein band (Fig. 2).

Effect of Deamidation of Asn-101 and Asn-123 on Structural and Functional Properties of ␣A-crystallin
Comparison of Effect of the Deamidation on the Chaperone Activity of WT-␣A and Deamidated ␣A Mutants-The chaperone activity of WT-␣A and ␣A mutant species was determined using three target proteins (insulin, citrate synthase, and ␥Dcrystallin) at three varying ratios of chaperone: target protein (i.e. 0.5:1, 1:1, and 2:1). With insulin as a target protein, no chaperone activity in the N101D mutant was observed at the ratio of 0.5:1 crystallin/target protein (Fig. 3A), but the other two mutants (N123D and N101D/N123D) showed relatively higher activity, although the levels were significantly lower than the WT-␣A-crystallin. At the higher ratios of chaperone/insulin (1:1), the N101D mutant again lacked any chaperone activity, but the activity was observed at the 2:1 ratio, although it was lowest compared with the other two mutants (N123D and N101D/ N123D). Together, the data of chaperone activity toward insulin suggested that the mutation of N101D resulted in relatively greater loss (50 -100%) of chaperone activity compared with the other two mutations (N123D and N101D/N123D).
With CS and recombinant ␥D-crystallin as target proteins, the N101D mutant again showed relatively lower chaperone activity (ϳ50%) compared with the WT-␣A crystallin and the other two mutants (i.e. N123D and N101D/N123D) (Fig. 3, B and C). Further, the N123D mutant also showed a lower chaperone activity compared with WT, but it was relatively greater than the N101D mutant. Additionally, although the N101D/ N123D mutant showed a lower chaperone activity than the N123D mutant with insulin and CS as target proteins, its activity was almost at the same level as in the other two mutants with ␥D-crystallin as the target protein. Together, the data suggested a significant loss of chaperone activity in ␣A following deamidation of Asn-101 compared with deamidation of Asn-123, and a progressive decrease in chaperone activity Surface Hydrophobicity of WT-␣A and ␣A-deamidated Mutants-Because past studies have suggested that the interaction between the chaperone molecule (␣-crystallin) and target protein largely involves hydrophobic residues, the fluorescence spectra of ANS bound to WT-␣A crystallin and the three deamidated mutants were determined (Fig. 4). ANS fluoresces at about 512 nm, but upon binding to WT-␣A, it showed fluorescence at 482 nm (Fig. 4A). The three ␣A mutants showed a decrease in ANS fluorescence intensity compared with WT and also a shift from 482 (seen in WT-␣A) to 396 -480 nm. Furthermore, compared with WT-␣A, the decrease in the fluorescence intensity was highest in the N101D mutant (32%) followed by N123D (19%) and N101D/N123D (10%) mutants. Together, the results suggested a deamidation-induced decrease in available hydrophobic surface area in the following order: WT-␣A Ͼ ␣A-N101D/N123D Ͼ ␣A-N123D Ͼ ␣A-N101D. The decrease in surface hydrophobicity was in parallel with a decrease in chaperone activities of the ␣A species (see above).
Because ␣A-crystallin is shown to be a better chaperone at higher temperatures, the ANS binding of WT-␣A and the three mutants was determined at 43°C (Fig. 4B). The fluorescence intensity was 20 -30% higher in all protein species compared with the levels observed at 37°C. Although the relative hydrophobicity of WT-␣A and the mutants showed temperature dependence, the order of the decrease in fluorescence intensities at 43°C was different than seen at 37°C and was as follows: Analysis of Secondary and Tertiary Structures of WT-␣A and ␣A Deamidated Mutants-Compared with the highest fluorescence intensity of Trp at 336 nm of WT-␣A crystallin, the N101D mutant showed an emission at 332 nm. The N123D and N101D/N123D mutants also showed similar shifts in the emission maxima with a slight difference in fluorescence intensity (Fig. 5). Together, the results suggested that deamidation of Asn-101 and Asn-123 residues caused altered microenvironment of the Trp residue in the ␣A mutants compared with WT-␣A.
The far-and near-UV CD spectra of WT-␣A and the three ␣A mutants are shown in Fig. 6, A and B, respectively. The far-UV CD spectrum of WT-␣A crystallin was similar as reported previously (36) with minimum around 218 -220 nm, suggesting that the protein is in ␤-sheet conformation and is properly folded. However, the mutants exhibited varied profiles (Fig.  6A). The levels of secondary structures were estimated using the PROSEC program and are shown in Table II. The content of ␣-helix (11%), ␤-sheet (44%), and ␤-turns (17%) and random coil (28%) in WT-␣A were consistent with the previous data (36). Compared with WT-␣A, the ␤-sheet and ␤-turn contents were reduced to 50% in the N123D mutant and 70 -80% in the N101D and N101D/N123D mutants, with significant increase in levels of ␣-helix and random coil conformation.
The near-UV CD spectrum of a protein is mainly a representation of the local environments of the aromatic amino acid residues (Trp, Phe, and Tyr). The near-UV spectra of the three mutants did not overlap with the WT-␣A (Fig. 6B). Like previous data (36), the near-UV spectra of the WT-␣A also showed five distinct maxima around 259, 265, 273, 279, and 287 nm and five distinct minima around 262, 268, 275, 284, and 292 nm. The peak signal caused by Phe in the 250 -270 region showed little alteration in the three mutants compared with WT-␣A. However, the peaks beyond 270 nm in all the three mutants were markedly different in their intensity compared with the WT-␣A (Fig. 6B). Indeed, the spectra beyond 270 nm of the three deamidated ␣A mutants were very different from the spectrum of WT-␣A, which confirmed differences in the microenvironment of Tyr and/or Trp following deamidation.
Determination of Quaternary Structures of WT-␣A and the Deamidated Mutants-To compare the quaternary structures of the homoaggregates of WT-␣A and the deamidated ␣A mutants, their molecular masses were determined by the static light scattering method using a TSK G-4000 PW XL column (TOSO Haas, Montgomeryville, PA). The estimated molecular mass is shown in Table III. The three ␣A mutants showed significant difference in oligomer sizes compared with WT-␣Acrystallin (i.e. N101D/N123D formed the largest oligomers of 823 kDa followed by N123D oligomers of 809 kDa, N101D of 608 kDa, and WT-␣A of 670 kDa).  (Table III). Whereas the mass of the oligomers of WT-␣A:WT-␣B was 745 kDa, the reconstituted oligomers with deamidated proteins showed higher mass (i.e. the mass of the oligomers containing WT-␣B and deamidated ␣A mutants ranged between 878 to 1867 kDa, and that of WT-␣A:deamidated ␣B mutants ranged between 728 and 1121 kDa). Additionally, the results showed that compared with ␣A-N101D mutant, the ␣A-N123D mutant produced heteroaggregates of larger sizes with the WT-␣B subunit, suggesting that deamidation of Asn affected the oligomerization process. The fact that the deamidated N101D/N123D mutant produced even larger oligomers with the WT-␣B subunit suggested that both Asn residues might be involved in maintaining the quaternary structure and also in the subunit interaction with ␣B-crystallin. However, whereas the size of the heteroaggregate of ␣B N78D mutants:WT-␣A was almost the same as that of WT-␣A: WT-␣B, the oligomers produced by ␣B N146D mutants:WT-␣A were of larger sizes, suggesting that Asn-146 plays a role in subunit interaction with the ␣A-crystallin. Taken together, the deamidation in ␣A and ␣B subunits resulted generally in loosely organized oligomer structures.

Characterization of Reconstituted Heteroaggregates
Chaperone Activity of Reconstituted Heteroaggregates-The chaperone activities of the reconstituted ␣-crystallin heteroaggregates were determined using two target proteins (i.e. insulin and recombinant ␥D-crystallin). Compared with the chaperone activity of the oligomer of the WT-␣A:WT-␣B, the oligomers containing either deamidated ␣A or deamidated ␣B mutants and their counterpart WT proteins showed about 50% lower chaperone activity (Fig. 7, A and B). The only exception was the heteroaggregates of WT-␣A:␣B-N78D mutant, which showed a higher chaperone activity than all of the other heteroaggregates. In general, the decrease in chaperone activity in the heteroaggregates showed the following order: Together, the results suggested that deamidation of Asn residues in both ␣Aand ␣B-crystallins affected the chaperone activity of the heteroaggregate, and this could lead to higher molecular weight aggregates seen during aging or cataract development.
Hydrophobicity of the Reconstituted Heteroaggregates-As stated above, because chaperone activity is believed to be partly due to the interactions of hydrophobic patches of ␣-crystallin with a target protein (42), a hydrophobic probe such as ANS could provide information as it fluoresces upon binding to apolar surfaces. As shown in Fig. 8, the fluorescence intensities of ANS bound to oligomers containing deamidated ␣A or ␣B mutants and their counterpart WT proteins differed significantly. Compared with the heteroaggregates of WT-␣A:WT-␣B, the exposed hydrophobic surface area was reduced in all the oligomers. The ANS-binding results showed a progressive decrease in the available hydrophobic surface area in the follow- Intrinsic Trp Fluorescence of Reconstituted Heteroaggregates-The Trp fluorescence of different reconstituted ␣-crystallin heteroaggregates was determined with excitation at 295 nm and emission between 300 and 400 nm. The heteroaggregates containing WT-␣A:WT-␣B showed the emission maxima at 336 nm, whereas the heteroaggregate containing WT-␣A: deamidated ␣B mutants or WT-␣B:deamidated ␣A mutants showed differences in fluorescence spectra. Compared with ␣-crystallin heteroaggregate containing WT-␣A:WT-␣B, the heteroaggregates containing WT-␣A:␣B-N78D/N146D mutant and WT-␣B:␣A-N101D/N123D mutant showed maximum quenching in fluorescence intensity with a blue shift to 331-333 nm (Fig. 9). The heteroaggregates containing WT-␣A:␣B-N78D mutant or WT-␣A:␣A-N123D mutant exhibited greater fluorescence intensity with the same max. The heteroaggregates containing WT-␣:␣A-N101D mutant or WT-␣A:␣B-N146D mutant showed intermediate fluorescence intensities with a shift to 337 nm. Together, the results suggested that the microenvironment of the Trp residues of the reconstituted crystallin heteroaggregates containing deamidated mutant proteins was altered.
Rate of Subunit Exchange in Heteroaggreagtes-The WT-␣A, WT-␣B, ␣A mutants (N101D, N123D, and N101D/N123D), and ␣B mutant (N146D) were labeled with fluorescent probes, and their subunit exchange rates were determined in the heteroaggregates by the FRET method. The cysteine residue of WT-␣A and ␣A-deamidated mutants were labeled with LYI (a sulfhydryl-specific fluorophore), and the WT-␣B and the ␣B-deamidated mutant were labeled with AIAS fluorophore as described under "Experimental Procedures." The emission maxima of LYI-labeled WT-␣A/␣A mutants (excitation at 426 nm) and AIAS-labeled WT-␣B/␣B mutants (excitation at 336 nm) were at 525 and 415 nm, respectively. The overlap of the emission spectrum of AIAS fluorophore with the absorption spectrum of the LYI fluorophore suggested that energy was transferred between the donor and acceptor fluorophores during FRET (data not shown). Following the mixing of the AIAS-labeled WT-␣B or N146D mutant (donor) or the LYI-labeled WT-␣A or deamidated ␣A mutants (acceptor) in a 3:1 ratio (␣A/␣B), the subunit exchange rate was determined. A quenching of donor fluorescence at 426 nm concomitantly occurred with an increase in acceptor fluorescence at 525 nm in all of the heteroaggregates. Emission spectra of one such example is shown in Fig. 10A and other heteroaggregates also showed similar spectrum. The heteroaggregates containing WT-␣A and WT-␣B showed changes in fluorescence intensity until 120 min, but the quenching of donor fluorescence continued until the 240 min in the heteroaggregates containing WT-␣A and deamidated ␣B mutants and also in the heteroaggregates containing deamidated ␣A and WT-␣B subunits. The acceptor fluorescence intensities at 525 nm for all the heteroaggregates are plotted against time in Fig. 10B. The subunit exchange rate from the increase in acceptor fluorescence intensity was calculated by fitting the data to the exponential function F(t)/F(0) ϭ A 1 ϩ A 2 e Ϫkt , where k is the subunit exchange rate constant. The subunit exchange rate was found to be 0.051 min Ϫ1 for heteroaggregate containing WT-␣A and WT-␣B, which is in agreement with the previously published results (5,26). Compared with heteroaggregates containing WT-␣A:WT-␣B, the heteroaggregates containing deamidated ␣A mutants:WT-␣B showed a relatively lower subunit exchange rate (i.e. WT-␣B:␣A-N101D, k ϭ 0.041 min Ϫ1 ; WT-␣B:␣A-N123D, k ϭ 0.0393 min Ϫ1 ; WT-␣B:␣A-N101D/N123D, k ϭ 0.0417 min Ϫ1 . The het- FIG. 4. Fluorescence spectra of WT-␣A and three deamidated ␣A mutants following ANS binding. Fluorescence spectra of ANS bound to WT-␣A and to the three deamidated mutants at 37°C (A) and 43°C (B) are shown. The concentration of each ␣A species was 0.15 mg/ml, and that of ANS was 12 M. All of the protein preparations were heated with ANS at 43°C for 10 min, and fluorescence spectra were determined after cooling the preparations to room temperature. Fluorescence spectra of the preparations were recorded with excitation at 390 nm and emission between 400 and 600 nm. eroaggregates containing WT-␣A and deamidated ␣B mutants showed a nearly 3-fold reduced subunit exchange rate (WT-␣A: ␣B-N146D, k ϭ 0.016 min Ϫ1 ), suggesting that deamidation of Asn residues in ␣A-/␣B-crystallins altered the electrostatic interactions, more so by Asn-146 in ␣B-crystallin. Hence, it appears that electrostatic interaction, which is the first step in subunit-subunit interactions, followed by noncovalent interac-tions, occurred during the heteroaggregate multimeric assembly.
The heteroaggregate containing deamidated ␣A mutants: WT-␣B also formed larger oligomers and exhibited relatively greater chaperone activity and more surface hydrophobicity compared with heteroaggregates containing WT-␣A and deamidated ␣B mutants. The data suggested that the presence of WT-␣B in the ␣-crystallin heteroaggregate increased the surface area, which most likely favors subunit exchange and ex-  6. A, far-UV CD spectra of WT-␣A and three deamidated ␣A mutants. The spectra were recorded at the concentration of 0.1 mg/ml of each species with a cell path length of 0.1 cm. The spectra reported are the average of five determinations. B, near-UV CD spectra of WT-␣A and three ␣A-deamidated mutants. The spectra were determined at a 1 mg/ml concentration of each preparation with a cell with a 1-cm path length. The reported spectra are the average of five determinations of each sample and are corrected for the cell blank and smoothed.

TABLE II
The levels (percentages) of secondary structure contents in WT-␣A and the three ␣A-deamidated mutants The secondary structure contents of the protein species were determined from the from the far UV-CD spectra as shown in Fig. 6A.   posed the chaperone sites, eventually regulating both the structural and functional properties.

DISCUSSION
Both ␣Aand ␣B-crystallins, either as homoaggregates or a heteroaggregate, show chaperone activity, and this property is believed to prevent aggregation of unfolded proteins and thus prevent cataract development (37). A previous study has correlated the significance of oligomeric state, different regions, and specific amino acids in both crystallins to their functional, structural, and oligomerization properties (38). However, this structure-function relationship is not yet fully understood. Similarly, the effect of in vivo post-translational modifications of specific amino acids on the structure-function relationship of both ␣Aand ␣Bcrystallins is also poorly understood. Although deamidation of ␣Aand ␣B-crystallins is one of the most frequent in vivo modifications during aging and cataract development (see Introduction), the effects of deamidation on its structural and functional properties have not been examined except in one of our recent reports (24). This study showed that the deamidation of Asn-146 and not Asn-78 (the two deamidation sites) in ␣B-crystallin caused significant changes in structural and functional properties of the crystallin. Because both Asn-78 and Asn-146 reside in the conserved region (␣crystallin domain, residues 66 -147) of the ␣B-crystallin (3,4), the vicinity of the Asn-146 residue toward the conserved Cterminal region (residues 147-175) might be responsible for the above effects. The ␣A-crystallin also contains two potential deamidation sites (i.e. Asn-101 and Asn-123) in the conserved region (␣-crystallin domain, residues 62-143) and are also in the vicinity of the conserved C-terminal region (i.e. residues 143-173) (3,4). Because no information exists in the literature regarding effects of deamidation of these two residues on the structural and functional properties of ␣A-crystallin, the present study was undertaken. First, we examined the effects of deamidation of Asn-101 or Asn-123 residues or both on the structural and functional properties of ␣A-homoaggregates. Next, we reconstituted heteroaggregates using deamidated ␣A/␣B mutants and their counterpart WT proteins and determined their structural and functional properties.
The major findings during the study of ␣A-homoaggregates were as follows. (a) Deamidation of either Asn-101 or Asn-123 or both residues in ␣A resulted in altered secondary and tertiary structures, oligomerization properties, and reduced levels of the chaperone activity with three target proteins (insulin, citrate synthase, and human recombinant ␥D-crystallin). (b) The N101D mutant exhibited the lowest chaperone activity, and maximum change is secondary, tertiary, and quaternary structures, compared with the other two deamidated ␣A mutants. Therefore, the Asn-101 residue is of relatively greater importance compared with Asn-123 in maintaining the structural and functional integrity of ␣A-crystallin.
The major findings during the investigation of reconstituted heteroaggregates were as follows. (a) Relative to reconstituted heterooligomers of WT-␣A and WT-␣B-crystallins, the oli- gomers of larger sizes were generated in heteroaggregates of deamidated ␣A or ␣B mutants and their counterpart WT proteins. However, heteroaggregates containing deamidated ␣A mutants and WT-␣B were of larger sizes than heteroaggregates of deamidated ␣B mutants and WT-␣A crystallin. (b) Compared with the heteroaggregates containing WT-␣A and WT-␣B, the heteroaggregates containing deamidated ␣Amutants and WT-␣B exhibited higher chaperone activity, altered tertiary structures, and lower subunit exchange rate. The heteroaggregate containing WT-␣A and deamidated ␣B mutants showed a 3-fold lower subunit exchange rate, suggesting that WT-␣B in the ␣-crystallin heteroaggregate modulates its dynamic structural and functional properties.
Together, the above findings suggest that deamidation of Asn in ␣Aand ␣B-crystallins have a major impact on structural and functional properties of their homoaggregates and heteroaggregates. The results also suggest the relative impact of deamidation of Asn-101 in ␣A-crystallin, and Asn-146 in ␣B-crystallin is far greater than the deamidation of Asn-123 in ␣A-crystallin, and Asn-78 in ␣B-crystallin. Further, the negative charges generated due to the conversion of Asn to Asp at the above sites apparently changed the conformation of crystallins and in turn their chaperone activity. The two deamidation sites (Asn-101 and Asn-123) in the ␣A-crystallin, like other small heat shock proteins, fall within the highly conserved sequence of the ␣-crystallin domain (i.e. residues 62-143) (3, 4). Therefore, the major question is why the deamidation of these two residues in ␣A-crystallin causes the above changes. It is FIG. 8. Fluorescence spectra of reconstituted ␣-crystallin heteroaggregates following ANS binding. Fluorescence spectra of different heteroaggregates (0.15 mg/ml) bound to ANS 37°C. Fluorescence spectra of the samples were recorded following excitation at 390 nm and emission between 400 and 600 nm. The concentration of ANS was 12 M.
FIG. 9. Intrinsic Trp fluorescence spectra of reconstituted ␣-crystallin heteroaggregates. The preparations were excited at 295 nm, and fluorescence spectra were recorded with emission between 300 and 400 nm. The concentration of each preparation was 0.15 mg/ml. more intriguing by the fact that between the two Asn residues of ␣A-crystallin, only the Asn-123 and not the Asn-101 has been conserved in the mammalian species (39). Because a frequent in vivo deamidation of Asn-101 and not of the Asn-123 residue has been reported in past studies (18, 40), our above findings suggest that deamidation of Asn-101 would compromise the structural and functional properties of the crystallin in vivo during aging.
Our studies show that both chaperone activity and exposed hydrophobic regions were affected following deamidation at both sites in ␣A-crystallin. This finding is important, because previous studies have shown that the hydrophobic regions of ␣-crystallin are involved in binding with substrate proteins during chaperone activity (41)(42)(43). Residues 71-78 of ␣A-crys-tallin represent the most hydrophobic region of the molecule (44), which is also an ANS-binding site. However, such binding caused a decrease in chaperone activity (42). Recent studies in ␣A-crystallin have suggested that factors other than hydrophobicity might also play important roles in the chaperone activity (45)(46)(47). The observation that the hydrophobicity is not the sole determinant of chaperone activity was also supported by our findings. The ␣A-N101D mutant did not show parallel coherence between chaperone activity and hydrophobicity with an increase in temperature. Although this mutant exhibited increased surface hydrophobicity at elevated temperature (43°C), it did not exhibit an equivalent increase in chaperone activity compared with the WT-␣A. An explanation for this anomaly might be that ANS (a hydrophobic probe), binds to all of the exposed hydrophobic patches in the N101D mutant at elevated temperature because of a change in conformation, but all of these binding sites might not be large enough to bind to the target protein (CS and recombinant ␥D-crystallin), which resulted in reduced chaperone activity.
Because buried charged amino acids generally pair with amino acids of opposite charges and often have functional importance (48), an unpaired negatively charged residue following deamidation at Asn-101 would perturb the tertiary structure. This might have caused the formation of homoaggregates of a larger size in the deamidated mutants. Additional reports of loss of chaperone activity in both ␣A and ␣B on mutation of charged residues (10, 49 -51) further suggest that for the ␣Aand ␣B-crystallins conserve their net charge through evolution (52). Our finding that an additional negatively charged residue due to deamidation resulted in changed tertiary structure and relatively larger oligomers compared with WT-␣A crystallin was supported by additional results. The results of far-UV CD spectra clearly showed that the homoaggregates of deamidated ␣A-crystallin adopted more of an ␣-helical conformation compared with WT-␣A crystallin. The near-UV spectra, which provide information regarding the aromatic environment in the tertiary structure of a protein and therefore also represent packing of a polypeptide to form tertiary structures, confirmed changed conformation of the deamidated ␣A-crystallin homoaggregates compared with that of WT-␣A crystallin. The three mutants showed peaks beyond 270 nm that were markedly different in their intensity compared the WT-␣A crystallin, suggesting differences in their tyrosine and/or tryptophan microenvironments.
In mammalian lens, ␣-crystallin exists as a heteroaggregate of ␣Aand ␣Bsubunits and exists as a molar ratio of 3:1, although previous reports indicated that various ratios can be formed in vitro (53,54). The interaction between ␣Aand ␣B-crystallins has been investigated in several past studies. It has been shown that the ␣Aand ␣B-crystallin complex has greater thermal stability than either proteins alone (55), suggesting the greater potential of the complex to protect proteins under stress. In the quaternary structure of ␣A-crystallin, the small multimers of ␣A-subunits remain in dynamic equilibrium with the oligomeric complex (5). This study further showed that ␣B-crystallin readily exchanges with the subunit of ␣A-crystallin but not with other proteins unrelated to Hsp.
It is believed that the interactions between the subunits involve ionic as well as hydrophobic interaction, because reduced subunit exchange rates were observed under either very high or very low salt concentrations (28,56). Because deamidated forms of ␣A or ␣B or both might exist in vivo, it is logical to study structural and functional properties of the heteroaggregates containing deamidated ␣Aor ␣B mutants with their counterpart WT proteins. As stated under "Results," ␣-crystallin was reconstituted using deamidated subunits ␣A/␣B and their counterpart WT proteins at a ratio of 3:1. Our results Increase in relative fluorescence intensity at 525 nm due to FRET from the AIAS-labeled crystallin to the LYI-labeled WT-␣A-crystallin or deamidated ␣A-crystallin species. Each curve represents the best statistical fit of the data to the exponential function F t /F 0 ϭ C 1 ϩ C 2 e Ϫkt , where k represents the subunit exchange rate. The subunit exchange rate of the heteroaggregates containing deamidated ␣Aand WT-␣B was higher than heteroaggregates containing WT-␣A-crystallin and deamidated ␣B-crystallin.
show relatively reduced levels of the subunit exchange and chaperone activity in heteroaggregates containing deamidated ␣A-/␣B-crystallins with their counterpart WT proteins compared with similarly reconstituted aggregates with WT-␣A and WT-␣B. The reduced chaperone activity in these heteroaggregates could be due to change in tertiary structure and heterologous subunit packing upon deamidation. This argument was supported by the observed blue shift to 331-333 nm during intrinsic Trp fluorescence measurement of heteroaggregate of deamidated ␣A/␣B and their counterpart WT protein compared with 336-nm peaks in the oligomers of WT-␣A and WT-␣B crystallin. The larger size oligomers in the above heteroaggregates containing either deamidated ␣A or ␣B with their counterpart WT protein further support the above argument. Compared with the heteroaggregates of WT-␣A and WT-␣B, a 3-fold decrease in the subunit exchange rate in the heteroaggregates containing deamidated ␣B and WT-␣A was observed, but the relative rate was only slightly lower in the aggregates of deamidated ␣A and WT-␣B. Similar differences were observed when oligomer sizes and chaperone activity were compared.
Together, the above data suggested that mutation of Asn in ␣Aor ␣B-crystallins in the conserved region resulted in modified structure and function, possibly due to a different mode of subunit assembly. This effect was greater in the heteroaggregates containing deamidated ␣B compared with those containing deamidated ␣A subunit. The ␣-crystallin heteroaggregates containing deamidated ␣A mutant and WT-␣B oligomers (i.e. ␣A-N123D:WT-␣B and ␣A-N101D/N123D:␣B-WT) also showed high oligomer sizes and slight increase in chaperone activity compared with other mutants, suggesting that the presence of WT-␣B modulates the chaperone activity in these oligomers. Apparently, the WT-␣B in the ␣-crystallin heteroaggregate modulates the formation of oligomeric complex, the flexible building units, extended surface area, and the chaperone activity in these oligomers either directly or by inducing packing changes in the heteroaggregate.
In our study, the presence of ␣B-crystallin in the multimeric ␣-crystallin complex (containing deamidated ␣A mutant and WT-␣B) affects the subunit exchange and increases the chaperone activity and oligomer sizes. How the oligomeric structure favors free and rapid subunit exchange and relative increase in chaperone activity remains largely unexplored. Probably, optimum surface area facilitates free subunit exchange and leads to the formation of stable quaternary structure. The time-dependent structural and functional changes with these oligomers may elucidate a possible mechanism for chaperone activity. We are currently attempting to answer these questions.