Identification of 1,1'-bi(4-anilino)naphthalene-5,5'-disulfonic acid binding sequences in alpha-crystallin.

The hydrophobic binding sites in alpha-crystallin were evaluated using fluorescent probes 1,1'-bi(4-anilino)naphthalenesulfonic acid (bis-ANS), 8-anilino-1-naphthalene sulfonate (ANS), and 1-azidonaphthalene 5-sulfonate (1,5-AZNS). The photolysis of bis-ANS-alpha-crystallin complex resulted in incorporation of the probe to both alphaA- and alphaB-subunits. Prior binding of denatured alcohol dehydrogenase to alpha-crystallin significantly decreased the photoincorporation of bis-ANS to alpha-crystallin. Localization of bis-ANS incorporated into alphaA-crystallin resulted in the identification of residues QSLFR and HFSPEDLTVK as the fluorophore binding regions. In alphaB-crystallin, sequences DRFSVNLNVK and VLGDVIEVHGK were found to be the bis-ANS binding regions. Of the bis-ANS binding sequences, HFSPEDLTVK of alphaA-crystallin and DRFSVNLNVK and VLGDVIEVHGK of alphaB-crystallin were earlier identified as part of the sequences involved in their interaction with target proteins during the molecular chaperone-like action. The hydrophobic probe, 1,5-AZNS, also interacted with both subunits of alpha-crystallin. Localization of 1,5-AZNS binding site in alphaB-crystallin lead to the identification of HFSPEEK sequence as the interacting site in this subunit of alpha-crystallin. Glycated alpha-crystallin displayed decreased ANS fluorescence and loss of chaperone-like function, suggesting the involvement of glycation site as well as ANS binding site in chaperone-like activity display.

␣-Crystallin is one of the most predominant eye lens proteins. Its concentration in lens fiber cells is about 40% of the total protein in lens (1). ␣-Crystallin exists as a polydisperse aggregate with an average molecular mass of 800 kDa (2). The two types of subunits, designated ␣A and ␣B, each of which has a molecular mass of 20 kDa, arrange themselves in yet undefined ways to form the aggregate (2). During aging, ␣-crystallin undergoes extensive modifications culminating in the formation of super aggregates and highly cross-linked light-scattering molecules (2). The sequences of the subunits of ␣-crystallin have high homology to small heat shock proteins (3,4) and are highly conserved between species. ␣-Crystallin subunits, once thought to be lens-specific, are now widely known to be present in other tissues as well (5)(6)(7)(8). Despite extensive studies carried out in the past, the quarternary structure or the structurefunction of ␣-crystallin or its subunits has remained an enigma and challenge for researchers.
Recently, the ability of native ␣-crystallin to suppress the aggregation of heat-denatured (9 -20), UV-irradiated (20,21), as well as chemically denatured (22) proteins and enzymes has been demonstrated. It has been proposed that surface hydrophobic sites in the ␣-crystallin aggregate are involved in the binding of ␣-crystallin to target proteins during the display of chaperone-like activity (13,23). A direct correlation between the extent of ␣-crystallin hydrophobicity and chaperone-like activity has been demonstrated by several studies (13,(23)(24)(25)(26)(27). However, the amino acid sequence that contributes to the site responsible for the binding of denatured proteins and hydrophobic site specific probes is not fully understood.
We have shown earlier that both A-and B-subunits in ␣-crystallin interact with bis-ANS 1 in 1:1 stoichiometry at 37°C (26). The number of bis-ANS molecules binding to ␣-crystallin increases if the protein is exposed to higher temperatures or denaturing agents prior to the addition of the fluorophore (26,27). Furthermore, it has been shown that binding of bis-ANS to ␣B-crystallin (25) or ␣-crystallin (26) diminishes the chaperone-like activity of the protein. In the present study we have determined the bis-ANS binding sequences in ␣-crystallin by photocross-linking, peptide mapping and sequencing. The data presented here also show that the bis-ANS binding sequences are also the chaperone sites in ␣-crystallin.

EXPERIMENTAL PROCEDURES
Materials-bis-ANS, ANS, and 1,5-AZNS were obtained from Molecular Probes, Inc. (Junction City, OR). The stock solutions of bis-ANS were prepared in 95% alcohol, and the concentration was determined by absorbance at 385 nm using an extinction coefficient, ⑀ 385 ϭ 16,790 cm Ϫ1 M Ϫ1 (28). Lysyl endopeptidase was purchased from Wako Bioproducts. Sequence grade trypsin was obtained from Sigma. ␤L-crystallin (29) was isolated from bovine lenses (15). All other chemicals were of the highest grade commercially available.
Preparation of ␣-Crystallin-␣-Crystallin was isolated from young bovine lens cortex by gel filtration on Sephadex G-200 and ion-exchange chromatography on trimethylaminoethyl-fractogel column (EM-Separations) as described earlier (26,30). The ␣-crystallin thus obtained was Ͼ99% pure as judged by SDS-PAGE, and this preparation was used in this study.
Photoincorporation of Bis-ANS into ␣-Crystallin-Photoincorporation of bis-ANS to ␣-crystallin was carried out as described earlier (26), with slight modification of the original procedure described by Seale et al. (31). Following photolysis, the sample was analyzed by HPLC and SDS-PAGE, and the fluorescence associated with protein bands was documented by photography using TMAX 100 film (Eastman Kodak Co.) under UV light (360 nm). The gel was later stained with Coomassie Blue. The efficiency of bis-ANS incorporation to ␣-crystallin during 15-min photolysis was determined by quantitative densitometry function of Image-1 system (Universal Imaging Corp.). * This work was supported in part by National Institutes of Health Grant EY 11981 and a grant-in aid from Research to Prevent Blindness, Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To investigate whether prior binding of denatured proteins to ␣-crystallin prevents bis-ANS binding and photoincorporation, ␣-crystallin and alcohol dehydrogenase (ADH) (1:6 ratio) were incubated at 48°C for 1 h. Following incubation, the reaction mixture was cooled to 25°C, and bis-ANS was added. The final bis-ANS concentration was 12.5 M. Photolysis of the sample was carried out as above for 15 min, and the aliquots were subjected to SDS-PAGE under reducing conditions. The fluorescent bands were photographed as above and the gel was stained with Coomassie Blue.
Separation of Bis-ANS-labeled ␣Aand ␣B-crystallin-The photolyzed ␣-crystallin-bis-ANS complex was treated with 5 mM dithiothreitol for 2 h and filtered. The ␣Aand ␣B-subunits were separated from one another by HPLC using a C18 column (218TP1010 from The Separation Group, Hesparia, CA) and linear gradient (0 -60% over a period of 1 h) formed between 0.065% trifluoroacetic acid in water and 0.065% trifluoroacetic acid in acetonitrile. The flow rate was 1 ml/min. The elution was monitored by absorbance (280 nm) and fluorescence (390 nm excitation and 490 nm emission). bis-ANS-labeled ␣Aand ␣Bcrystallins were further purified by SDS-PAGE and recovered by electroelution, and the SDS was removed by ether precipitation (32).
Identification of Bis-ANS-labeled Peptides-The bis-ANS-labeled ␣Aand ␣B-crystallins were digested with lysyl endopeptidase (1:30, enzyme/protein) for 4 h at 37°C. The peptides were separated by reverse phase HPLC on a Vydac C18 column (218TP54) equilibrated with 20 mM sodium phosphate buffer, pH 6.5 ϩ 5% acetonitrile (solvent A). The elution of bound peptides was carried out with a linear gradient (0 -60%) formed by solvent A and solvent B (20 mM phosphate buffer, pH 6.5, in 95% acetonitrile). A flow rate of 1 ml/min over 120 min was maintained, and 1-ml fractions were collected. The elution was monitored at 220 nm for absorption and fluorescence (390 nm excitation and 490 nm emission). The amino-terminal sequences of bis-ANS-labeled peptides were determined by Edman degradation on an Applied Biosystems PROCISE CLC protein sequencing system.
Photoincorporation of 1,5-AZNS to ␣-Crystallin and Identification of Binding Site in ␣B-crystallin-The photoincorporation of 1,5-AZNS to ␣-crystallin was accomplished using a procedure described by Dockter and Koseki (33). 1 mM 1,5-AZNS and 1.25 M ␣-crystallin were used in this study. Following photoincorporation, A-and B-subunits of ␣-crystallin were separated by HPLC as above using a C18 column. Although the fluorophore was incorporated to both the subunits, only ␣B-subunit was further analyzed to determine the 1,5-AZNS incorporation site. Labeled ␣B-crystallin was digested with sequencing grade trypsin (1:50 ratio), and the resulting peptides were separated as described earlier (30). The elution profile was monitored at 220 nm. All fractions were tested for fluorescence in a Perkin-Elmer Spectrophotometer model 650-40 (excitation and emission maxima of 334 and 440 nm, respectively). The major fluorescent peptide eluting at 45 min from the HPLC column was subjected to amino acid sequencing in an Applied Biosystems 470A sequencer.
Glycation of ␣-Crystallin, Chaperone Assay, and ANS Binding-Glycation of ␣-crystallin was carried out in 0.1 M phosphate buffer, pH 7.0, using 10 mg/ml protein and 20 mM L-ascorbic acid (34). After incubation at 37°C for 4 weeks, the reaction mixture was dialyzed, and the glycated protein was tested with ␤L-crystallin for chaperone-like activity (15). The interaction of glycated ␣-crystallin (0.25 M) with ANS was examined by fluorescence measurement in a Perkin-Elmer Spectrofluorimeter model 650 -40. The samples with ANS were excited at 390 nm, and the emission was measured at 490 nm in a cuvette with 1-cm path length and slit width of 5 nm. The ratio of protein to probe was approximately 1:50. ␣-Crystallin incubated without ascorbic acid and processed similarly was used as the control.

RESULTS
Interaction of Bis-ANS with ␣-Crystallin-The binding of bis-ANS, the environment-sensitive probe, to ␣-crystallin results in severalfold increases in fluorescence intensity of the probe, and the emission maxima is blue shifted to ϳ490 nm from its emission maximum of 533 nm in aqueous medium (27). We have shown recently that bis-ANS interacts strongly with ␣-crystallin, and the bound fluorophore cannot be removed by dialysis (26). Covalent bonds are formed between proteinbound bis-ANS and the amino acids forming the binding pocket when the complexes are exposed to long UV light (26,31). The photoincorporation of bis-ANS to ␣-crystallin is directly proportional to the duration of photolysis. By image analysis we estimated that about 15% of the bound bis-ANS covalently cross-links to ␣-crystallin in the 15 min of photolysis used in our study (Fig. 1, lane 1). Although photolysis for longer durations results in higher amount of photoincorporation, there is increased subunit cross-linking and generation of high molecular weight species. Therefore the studies described here were limited to 15 min of photolysis. As ␣-crystallin-bis-ANS was dialyzed to remove the free bis-ANS prior to photolysis, it is unlikely that nonspecific incorporation of the fluorophore to ␣-crystallin occurred during our experiments with bis-ANS and ␣-crystallin.
Effect of Denaturing Protein Bound to ␣-Crystallin on the Photoincorporation of Bis-ANS-Earlier we showed that prior binding of bis-ANS to ␣-crystallin diminishes the chaperonelike activity of the protein (26). To determine whether the binding of denaturing proteins to ␣-crystallin at the chaperone site can affect subsequent bis-ANS photoincorporation, ␣-crystallin and ADH (1: 6 ratio) were incubated at 48°C for 1 h prior to the addition of bis-ANS and photolysis. SDS-PAGE of such an experiment is shown in Fig. 1. The result shows a significant decrease in photoincorporation of bis-ANS to ␣-crystallin when ADH was heat-denatured and allowed to bind to ␣-crystallin prior to the addition of the fluorophore (compare lanes 1 and 2 in Fig. 1).
Photoincorporation of Bis-ANS to ␣-Crystallin Subunits and Localization of Incorporated Bis-ANS-To determine the bis-ANS binding sites in ␣-crystallin, the fluorophore was initially allowed to bind to purified ␣-crystallin by the addition of saturating amounts of the probe and removal of the excess by dialysis to minimize nonspecific photoincorporation of free bis-ANS activated during photolysis. Photolysis of the ␣-crystallinbis-ANS complex by UV-A light (366 nm) resulted in covalent incorporation of the fluorophore to both ␣Aand ␣B-subunits as we reported earlier (26).
In order to identify the sites of bis-ANS incorporation into ␣Aand ␣B-crystallins, ␣-crystallin was modified with bis-ANS, and the subunits were separated by HPLC and purified by SDS-PAGE. The purified bis-ANS-modified proteins were later digested with lysyl endopeptidase. The resulting fluorescent peptides were separated from other peptides by reverse phase HPLC. The HPLC profile of ␣A-crystallin peptides and the fluorescence is shown in Fig. 2. Edman degradation of fluorescent peptides eluting at 37 and 40 min revealed the same NH 2 -terminal sequence HFSPE. These peptides probably have the sequence HFSPEDLTVK and HFSPEDLTVKVQED-FVEIHGK, corresponding to residues 79 -88 and 79 -99 of ␣Acrystallin (Fig. 3), since we used lysyl endopeptidase to digest ␣A-crystallin. Incomplete digestion at Lys-88 due to the bis-ANS incorporation at/or near Lys-88 may have generated the latter peptide. Furthermore, the peptide eluting at 40 min yielded low levels of Phe (10 pmol) in the second cycle compared with 20 pmol of Ser in the third sequencing cycle, indicating that Phe-80 was modified by bis-ANS during photoincorpora- tion. We could not determine the location of bis-ANS insertion site in peptide eluting at 37 min by examining the results of five sequence cycles used to determine the identity of the peptide.
The fluorescent peptide eluting at 89 min showed the NH 2terminal sequence RTLGPF. The peptide showed a mobility equivalent to 6.5 kDa on SDS-PAGE (data not shown). Therefore, the peptide eluting at 89 min is likely to represent the residues 12-70 of ␣A-crystallin. Analysis of the fluorescent material eluting at 72 min failed to show any amino acid during sequencing cycles. Since the same peak also did not contain appreciable 220 nm absorption, it is likely that the fluorescence at this region may be due to the peptide-bound unstable bis-ANS or its derivative. To localize the bis-ANS-bound amino acid in peptide eluting at 89 min (Fig. 2), the peptide was subjected to trypsin digestion and HPLC analysis. Fig. 2, inset, shows the HPLC profile of the trypsin digest. The single fluorescent peptide eluting at 66 min (Fig. 2, inset) was sequenced as described earlier. The observed sequence for the 66-min peptide, QSLFR, corresponds to residues 50 -54 of ␣A-crystallin (Fig. 3). During the sequencing of this peptide the yield of Phe was low (82 pmol) compared with other amino acids (which were in the range of 110 -160 pmol), suggesting a possible modification of Phe-53 in ␣A-crystallin by bis-ANS.
The two fluorescent peptides of ␣B-crystallin, generated by lysyl endopeptidase digestion, eluted from the HPLC column at 38 min as a doublet (Fig. 4). The same peaks also showed maximal fluorescence emission at 490 nm when excited at 390 nm and revealed NH 2 -terminal sequence DRFSV and VLGDV. These two ␣B-peptides can only be from the sequence DRFS-VNLDVK and VLGDVIEVHGK (Fig. 3), since we have used lysyl endopeptidase for digestion. We were unable to conclude which of the amino acid in ␣B-crystallin formed a cross-link with bis-ANS by examining the results from five cycles of sequencing reaction used to determine the identity of the fluorescence peptide. Although the fluorescent material eluting at 72 min was the largest fluorescent peak, it gave no amino acids during sequencing cycles. Since the same peak also did not contain measurable 220 nm absorption, it is likely that the fluorescence at this region may be due to the peptide-bound unstable bis-ANS or its derivative. Since the chemistry of the reaction is not known at the present time, further studies are required to see whether the 72-in fluorescent peak was due to the bis-ANS that was originally bound to the two peptides we have identified or to a different peptide.
The 1,5-AZNS binding site in ␣B-crystallin was determined by peptide mapping and sequencing of the fluorophore-containing peptide. A peptide eluting from C18 column at 45 min (Fig.  5) with a sequence HFSPEELK was found to have incorporated 1,5-AZNS. This sequence corresponds to residues 83-90 in bovine ␣B-crystallin (Fig. 3). Examination of the chromatographic profiles obtained during the sequencing of the fluorescent peptide showed that Asp in the peptide was modified by 1,5-AZNS during photolysis. Although ␣A-crystallin was also labeled with 1,5-AZNS, we did not analyze the sample further.
Effect of Glycation of ␣-Crystallin on Its Chaperone-like Function and Interaction with ANS-Earlier studies have shown that glycation of ␣-crystallin reduces its chaperone-like activity (36). To determine whether this was due to an alteration of the hydrophobic chaperone site in ␣-crystallin, we measured the interaction of glycated ␣-crystallin with hydrophobic probe ANS. ANS, like bis-ANS is a polarity-sensitive reagent. ␣-Crystallin glycated with ascorbate for 4 weeks showed a 25% decrease in its ability to increase ANS fluorescence compared with the controls. When the same glycated ␣-crystallin was tested with ␤L-crystallin in a heat denaturation assay (15), a marked decrease in chaperone-like activity was observed (Fig. 6). DISCUSSION The presence of surface hydrophobic sites on ␣-crystallin has been known for a number of years (2). Since the demonstration of chaperone-like activity with ␣-crystallin (9), considerable interest has been shown in the hydrophobic sites within ␣-crystallins as these sites have been implicated in the chaperonelike function of the protein (13,(23)(24)(25)(26)(27). The hydrophobic sites in ␣-crystallin and its subunits have been studied during recent years using probes such as ANS (37), bis-ANS (26,27), and pyrene (13). We have shown recently that UV photolysis of the ␣-crystallin-bound bis-ANS leads to photoincorporation of the fluorophore to the protein subunits similar to that seen with chaperone GroEL (31) and HSP18.1 (38) and ␣B-crystallin (25). Prior binding of denatured proteins to ␣-crystallin resulted in diminished photoincorporation of the fluorophore bis-ANS (Fig.  1, lane 2) compared with ␣-crystallin alone. Earlier we showed that prior binding of bis-ANS to ␣-crystallin partially suppresses its chaperone-like activity (26). Taken together these data suggest that both ADH and bis-ANS share common binding sites in ␣-crystallin. The sequence analysis of binding sites discussed below confirms this view. Similar sharing of bis-ANS binding site and denaturing protein binding site in heat shock protein 18.1 has been reported recently (38).
The role of hydrophobic sites and the amino acid residues that contribute to their makeup within the multimeric chaperone GroEL has been confirmed (31). The available data show that the bis-ANS binding sequences are part of the chaperone sites in GroEL (31,39). The bis-ANS binding sites in small heat shock proteins have also been identified (38), but the chaperone sites in those proteins are yet to be determined. The two bis-ANS binding sequences in ␣A-crystallin, HFSPEDLTVK and HFSPEDLTVKVQEDFVEIHGK (Fig. 3), in part represent the chaperone site we identified earlier (40). On the basis of deuterium exchange studies Smith et al. (41) have also proposed that residues 72-75, in ␣A-crystallin, are a potential chaperone site. It should be noted that under the experimental conditions described here to determine bis-ANS binding sequences, other hydrophobic sequences in ␣A-crystallin, namely the residues 3-10, 27-37, or 130 -145, did not label with bis-ANS. It is possible that those sequences may be buried inside the protein molecule. The only other sequence that was labeled with bis-ANS was QSLFR peptide. Although it is not a hydrophobic sequence by itself, the sequence can be interpreted as part of an extended hydrophobic region between residues 44 and 57 (Fig.  3). It should also be noted that none of the peptides arising from the COOH terminus of ␣A-crystallin were labeled with bis-ANS. Since a loss in chaperone activity of ␣A-crystallin has been correlated with COOH-terminal truncation (20,42), and no bis-ANS binding site has been identified in that region, further studies are needed to determine the role of this region in chaperoning.
The two regions in ␣B-crystallin identified as bis-ANS binding sequences are at the COOH-terminal domain (Fig. 3). This is in contrast with the recent report published by Smulders and de Jong (25) on the photoincorporation of bis-ANS to recombinant rat ␣B-crystallin, where the authors observed incorporation of bis-ANS to the NH 2 -terminal domain of the protein.
Since prior exposure of ␣-crystallin to urea can affect the bis-ANS binding (26), it is yet to be determined whether the bis-ANS binding to rat ␣B-crystallin was influenced by urea used during the isolation of the recombinant protein. Of the two bis-ANS binding sequences identified in ␣B-crystallin during the present study (Fig. 3), the FSVNLDVK portion of the DRFSVNLDVK sequence is the same as the mellitin binding sequence we determined by cross-linking studies, 2 and the VLGDVIEVHGK sequence is one of the alcohol dehydrogenase binding sites determined earlier (30). The DRFSVNLDVK sequence follows the alcohol dehydrogenase interacting site in ␣B-crystallin reported by us earlier (30). In a separate experiment we have determined that another hydrophobic site-specific probe, 1,5-AZNS, binds to ␣B-crystallin sequence 83-90. The 1,5-AZNS-labeled peptide is the sequence between the two bis-ANS binding sequences in ␣B-crystallin. The structural differences between bis-ANS and 1,5-AZNS may have contrib-2 K. Krishna Sharma and G. S. Kumar, manuscript in preparation.  6. Effect of glycation on chaperone-like activity of ␣-crystallin. The assays were done as described under "Experimental Procedures" using ␤L-crystallin and glycated ␣-crystallin. Squares, ␤L-crystallin; circles, ␤L-crystallin ϩ ␣-crystallin; triangles, ␤L-crystallin ϩ glycated ␣-crystallin. 200 g of ␤L-crystallin and 30 g of ␣-crystallin or glycated ␣-crystallin were used in this study. uted to the difference in the site of incorporation of these two probes to ␣B-crystallin. Nevertheless, both bind at the highly conserved region of the protein (3,4).
The two bis-ANS binding sequences in ␣B-crystallin are separated by 10 amino acids, which have a major in vitro glycation site (43). Glycation of ␣-crystallin decreases ANS binding. Glycation also reduces chaperone-like activity of ␣-crystallin (Fig.  6). The data from this study suggest that the glycation-induced loss of chaperone-like activity reported earlier (36) may be due to the modification of Lys residues 90 and 92 of ␣B-crystallin that may be part of the hydrophobic/chaperone site.
The most hydrophobic region of ␣B-crystallin, residues 28 -34, proposed as a potential chaperone site by deuterium exchange studies (39) was not labeled by bis-ANS during our study. The two bis-ANS binding sequences in ␣B-crystallin identified during the present study belong to a region of high homology between HSP18.1 and ␣B-crystallin (38). Earlier studies have shown that this region in HSP18.1 is the primary bis-ANS binding region (38). On the basis of these data it can be stated that the entire third exon sequence of ␣B-crystallin, which has a high degree of homology to other heat shock proteins (3), is responsible for its chaperone-like function.
Although we estimated the binding of one bis-ANS molecule per subunit of ␣crystallin (26), in this study we see two sequences in each subunit as bis-ANS binding sites. The occurrence of two UV-sensitive anilinonaphthalene centers in bis-ANS and the activation of either one of the centers and insertion to the adjacent amino acid in the binding pocket may be the cause for the observation of two peptides as binding sites. Alternately, multiple conformation of ␣-crystallin subunits and their interaction with bis-ANS may result in labeling of more than one peptide sequence as a binding site.