α-Crystallin prevents protein aggregation under various stress conditions through its chaperone-like properties. Previously, we demonstrated that MGO (methylglyoxal) modification of αA-crystallin enhances its chaperone function and thus may affect transparency of the lens. During aging of the lens, not only αA-crystallin, but its client proteins are also likely to be modified by MGO. We have investigated the role of MGO modification of four model client proteins (insulin, α-lactalbumin, alcohol dehydrogenase and γ-crystallin) in their aggregation and structure and the ability of human αA-crystallin to chaperone them. We found that MGO modification (10–1000 μM) decreased the chemical aggregation of insulin and α-lactalbumin and thermal aggregation of alcohol dehydrogenase and γ-crystallin. Surface hydrophobicity in MGO-modified proteins decreased slightly relative to unmodified proteins. HPLC and MS analyses revealed argpyrimidine and hydroimidazolone in MGO-modified client proteins. The degree of chaperoning by αA-crystallin towards MGO-modified and unmodified client proteins was similar. Co-modification of client proteins and αA-crystallin by MGO completely inhibited stress-induced aggregation of client proteins. Our results indicate that minor modifications of client proteins and αA-crystallin by MGO might prevent protein aggregation and thus help maintain transparency of the aging lens.
- client protein aggregation
- human lens
The main constituents of the mammalian lens are crystallins, namely α-, β- and γ-crystallins. αA-Crystallin is the major crystallin, accounting for nearly 50% of proteins in the lens. It consists of two subunits, αA- and αB-crystallins, which typically exist in a molar ratio of approx. 3:1 in the mammalian lens . Although it was believed for many years that α-crystallin was strictly a lens-specific protein, both αA- and αB-crystallins have now been found in many non-lenticular tissues [2,3]. α-Crystallin is a key member of the sHSP (small heat-shock protein) family, and its structural and functional similarities are conserved from bacteria to humans [4,5]. Like other sHSPs, α-crystallin acts as a molecular chaperone in vitro by preventing aggregation of other proteins under various stress conditions [5–8].
During aging and cataract formation, lens proteins may be damaged by various factors, including oxidative stress and UV radiation [9,10]. Damaged proteins thus formed tend to aggregate, forming large insoluble complexes that compromise lens transparency. The chaperone activity of α-crystallin is thought to inhibit aggregation of damaged proteins, thus helping in maintaining lens transparency.
It is believed that α-crystallin binds to client proteins that have undergone mild structural perturbation. Previously, several investigators found that it binds to client proteins that are in their ‘ladder state’ [11–13]. Hydrophobic pockets on the surface of α-crystallin are thought to interact with client proteins during its chaperone function [5,7,14].
Many post-translational modifications such as phosphorylation , de-amidation , glycation [17,18] and oxidation  are detrimental to the chaperone function of α-crystallin. Glycation is a major post-translational modification; it produces stable adducts on proteins, which are collectively known as AGEs (advanced glycation end-products) [17,18]. Previous studies suggest that a reactive α-dicarbonyl compound, MGO (methylglyoxal), formed from triose phosphate intermediates of glycolysis, is a major source of AGEs in the human lens [20,21]. Several years ago, we established that reaction of MGO with αA-crystallin enhances its chaperone function . Other groups have studied this reaction and report similar or opposite findings [23–25]. We further showed that replacement of selected arginine residues with alanine mimicked the effects of MGO . Based on those findings, we proposed that MGO modification of αA-crystallin might be beneficial to the lens in maintaining transparency during aging. We believe that in addition to modifying αA-crystallin during aging, MGO can modify other lens proteins as well. Other such proteins could be client proteins for αA-crystallin's chaperone function. To determine if client protein modification affected their chaperoning by αA-crystallin, we first established the effect of MGO modification on chemically or thermally induced aggregation of four model client proteins, and we then measured the effect on their binding to MGO-modified and native αA-crystallins. We also examined the consequences of MGO modification on the structural characteristics of client proteins.
Bovine insulin, CS (citrate synthase), α-lactalbumin, CA (carbonic anhydrase) and DTT (dithiothreitol) were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). TNS [2-(p-toluidino)naphthalene-6-sulfonic acid, sodium salt] was obtained from Molecular Probes (Invitrogen, Carlsbad, CA, U.S.A.). All other chemicals were of analytical grade.
Cloning, expression and purification of recombinant human αA-crystallin
Human αA-crystallin with an N-terminal His tag and an entero-kinase cleavage site was amplified by PCR and cloned into pET23d vector . The recombinant protein was expressed in Escherichia coli BL21(DE3)pLysS cells. The protein was purified using an Ni–agarose column. The purity was checked by SDS/PAGE and Western blotting.
Preparation of bovine γ-crystallin
Bovine lenses were homogenized on ice in 10 mM Tris/HCl (pH 7.2) containing 100 mM NaCl, 1 mM EDTA, 0.02% sodium azide and 0.2 mM PMSF and then centrifuged at 18000 g for 45 min at 4 °C. The supernatant fraction was loaded on to a Sephacryl S300-HR column (95 cm×1.5 cm). γ-Crystallin corresponded to the fifth protein peak of the Sephacryl S300 column; the protein was collected, extensively dialysed against PBS and stored at –20 °C. SDS/PAGE separation showed a single band at ∼20 kDa.
Modification of client proteins by MGO
Client proteins (1.0 mg/ml insulin, α-lactalbumin, CA or γ-crystallin) in 50 mM sodium phosphate buffer (pH 7.4) were incubated with 0–1000 μM MGO for 2 days at 37 °C. We then dialysed all incubation mixtures against 50 mM sodium phosphate buffer (pH 7.4) to eliminate unreacted MGO.
Co-modification of CA and αA-crystallin by MGO
MGO (100 μM) in 50 mM sodium phosphate buffer (pH 7.4) was added to a solution containing CA and αA-crystallin (1.0 mg/ml each). This mixture was then incubated for 2 days at 37 °C. After incubation, the solution was dialysed against 50 mM sodium phosphate buffer (pH 7.4) to eliminate excess MGO.
Client protein aggregation assay
DTT-induced aggregation of insulin and α-lactalbumin
Freshly prepared DTT (final concentration 20 mM) was added to 80 μg of unmodified and MGO-modified insulin. Light scattering at 400 nm measured the aggregation for 1 h (at 25 °C) . Assays were performed in 96-microwell plates and light scattering was monitored with a microplate reader (model 190; Molecular Devices, Sunnyvale, CA, U.S.A.).
Aggregation of α-lactalbumin by DTT was measured in a similar manner with a microplate reader. The incubation mixture contained α-lactalbumin (100 μg) in 50 mM sodium phosphate buffer (pH 7.4) with 100 mM NaCl and 2 mM EDTA. The plates were incubated at 37 °C, and light scattering was monitored at 360 nm . Final reaction volume for both assays was 250 μl.
Thermal aggregation of CA and γ-crystallin
Thermal aggregation assays of unmodified and MGO-modified CA (150 μg each) were performed in a Beckman DU spectrophotometer (Beckman, Fullerton, CA, U.S.A.). The proteins were heated to 60 °C in 50 mM sodium phosphate buffer containing 100 mM NaCl (pH 7.4), and light scattering at 400 nm was monitored for 1 h in the kinetic mode. Thermal aggregation of unmodified and MGO-modified γ-crystallin samples (0.25 mg/ml) was initiated by incubation at 65 °C, and light scattering at 400 nm was monitored for 1 h.
We repeated these assays in the presence of αA-crystallin. The ratio of α-crystallin to insulin, α-lactalbumin, CA and γ-crystallin was 1:20, 1:15, 1:1 and 1:20 (w/w) respectively. For the experiment involving co-incubation of MGO-modified CA and MGO-modified αA-crystallin, we used an αA-crystallin/CA ratio of 1:1 (w/w).
HPLC assay for argpyrimidine
Protein samples (300 μg each) were hydrolysed with 6 M HCl for 20 h at 110 °C. The acid was evaporated in a Speed Vac system, and the pellet was suspended in 200 μl of water and filtered through a 0.45-μm centrifugal filter. The amino acid content of each hydrolysate was estimated with ninhydrin as described in . The samples were then injected into a C18 reversed-phase column (218TP54; Grace Vydac) and separated with a gradient system consisting of water and acetonitrile. Solvent A was 0.01 M HFBA (heptafluorobutyric acid) in water, and solvent B was 70% acetonitrile in water with 0.01 M HFBA. The solvent programme was as follows: 0–39 min, 16% B; 40–50 min, 20% B; 50–60 min, 22% B; 60–62 min, 28% B; 62–71 min, 100% B; 71–80 min, 16% B. The column eluate was monitored with an online fluorescence detector set at λex and λem of 335 and 385 nm respectively. Under these conditions, argpyrimidine had an Rt (retention time) of ∼28 min. Argpyrimidine in the protein samples was quantified by comparison with peak areas of known synthetic standards.
HPLC–ESI (electrospray ionization)-MS
Protein reduction, S-alkylation and digestion
MGO-modified insulin (20 μg) was dissolved in 20 μl of 100 mM Tris/HCl (pH 8.0) containing 8 M urea. The protein was then reduced with 20 mM DTT for 2 h and treated with 50 mM iodoacetamide for 30 min at 25 °C in the dark. The alkylated proteins were diluted 10-fold in water and digested with Asp-N (Roche) for 16 h at 25 °C. After heating at 95 °C for 5 min to inactivate Asp-N, the sample was then digested further with staphylococcal (Staphylococcus aureus) V8 protease (Pierce) overnight at 25 °C. The ratio of protease to protein was 1:50 for both incubations. The resulting peptide solution was vacuum-evaporated and reconstituted in 0.1% formic acid.
Identification of hydroimidazolone
LC-MS/MS (liquid chromatography tandem MS) analyses of the digests were performed in a Thermo Finnigan linear ion-trap mass spectrometer (model LTQ-XL) coupled with an Ettan MDLC system (GE Healthcare). The digests were chromatographed using a gradient of acetonitrile from 0 to 60% in aq. 0.1% formic acid for 30 min at a flow rate of 300 nl/min. The mass spectrometer was operated in a data-dependent MS to MS/MS switching mode, with the four most intense precursor ions in each MS scan subjected to MS/MS analysis. Data thus obtained were searched with Mascot Daemon software for modification of arginine residues. Further interpretation of the MS/MS spectrum of the modified peptide R22*GFFYTPKA30 (the asterisk represents modification of the residue) was performed with MS-Product software (http://www.prospector.ucsf.edu/ucsfhtml4.0/msprod.htm).
TNS fluorescence measurements
Unmodified and MGO-modified client proteins (100 μg/ml each) were incubated separately with 20-fold molar excess of TNS [in DMF (dimethylformamide)] for 2 h at 25 °C. Fluorescence emission spectra of TNS-bound samples were recorded between 350 and 520 nm with an LS-55 PerkinElmer spectrofluorimeter at 25 °C with excitation and emission band passes of 10 and 20 nm. TNS emission was measured after excitation at 320 nm.
We previously showed that modification by MGO improves αA-crystallin's chaperone function , and we thought that this phenomenon might be important for maintaining lens transparency. Presumably, along with αA-crystallin in the aging lens, client proteins are also modified by MGO. To determine how MGO modification affects aggregation of client proteins, we first examined the aggregation profiles in the presence or absence of MGO (10–1000 μM) for four proteins: insulin, α-lactalbumin, CA and γ-crystallin. We chose these particular client proteins to compare the impact of MGO modification on both chemically and thermally induced aggregation.
Surprisingly, we found that MGO modification inhibited both chemically and thermally induced aggregation of the proteins, and the extent of inhibition depended on MGO concentration (Figure 1). When the proteins were modified with 10–25 μM MGO, aggregation was inhibited by only 10–20%. However, the inhibition increased substantially with 100 μM MGO. While the DTT-induced aggregation was completely suppressed when insulin was modified with 1000 μM MGO (Figure 1A), aggregation of α-lactalbumin was completely suppressed at only 100 μM MGO (Figure 1B). Thermal aggregation of γ-crystallin was completely suppressed when modified with 1000 μM MGO. These observations led us to conclude that modification by MGO renders client proteins resistant to chemical and thermal aggregation.
We considered that MGO may induce cross-linking of client proteins, which would thus attain a rigid structure that would resist unfolding. To test this possibility, we incubated client proteins with various concentrations (10–1000 μM) of MGO for 48 h. In samples incubated with MGO up to 100 μM, we found no apparent cross-linking of client proteins, but at 1000 μM MGO it became apparent (Figure 2). Because protein aggregation was inhibited at 100 μM MGO (Figure 1) in the absence of protein cross-linking, we assume that other changes are more likely to be the cause.
Our previous study had established that MGO-modified αA-crystallin binds more client proteins than unmodified αA-crystallin when compared on a molar basis . Because of this finding, we wanted to investigate how MGO modification affected the chaperone function of αA-crystallin. Because modification by 100 μM MGO either partially or almost completely inhibited client protein aggregation, we used proteins modified with less than 100 μM MGO for these experiments. Figure 3 shows that inhibition of aggregation (ΔP) increased in unmodified and MGO-modified client proteins after addition of human αA-crystallin. The plots indicate that αA-crystallin's chaperoning efficiency remained unaltered, even though MGO modification of client proteins inhibited their aggregation. For example, αA-crystallin, whether MGO-modified or not, inhibited insulin aggregation by ∼45%. Likewise, γ-crystallin aggregation was inhibited by ∼90% for both unmodified and MGO-modified protein. From these results, we concluded that modification of client proteins with low concentrations of MGO (<100 μM) did not affect chaperoning by αA-crystallin during either thermal or chemical aggregation.
MGO presumably modifies both αA-crystallin and its client proteins simultaneously within the aging lens. Accordingly, we measured the effect of modification of both αA-crystallin and CA (chosen as a representative client protein) by 100 μM MGO. We found that MGO-modified αA-crystallin inhibits CA thermal aggregation by ∼50% (Figure 4). Similarly, modification of CA with MGO inhibited its aggregation by ∼66%. When αA-crystallin and CA were both modified by MGO, inhibition was nearly complete (92%). Taken together, these findings indicate that MGO modification of αA-crystallin and its client proteins helps to keep client proteins soluble. Because protein turnover in the lens is extremely low, this phenomenon could be important for maintaining lens transparency.
In two earlier studies, we found that MGO modification of Arg21 and Arg103 to argpyrimidine improved the chaperone function of human αA-crystallin [22,26]. We now wanted to determine if these modifications correlated with resistance to thermal and chemical aggregation of client proteins. We noted a higher concentration of argpyrimidine in proteins incubated with 100 μM MGO than in those incubated with 10 μM MGO. We found that the extent of resistance to aggregation with MGO treatment related directly to the amount of argpyrimidine in client proteins: the higher the argpyrimidine concentration (Figure 5), the greater the resistance to aggregation (Figure 1).
Hydroimidazolone is another MGO-derived arginine modification on proteins. We could not measure this modification directly because of limitations in our HPLC instrumentation. However, we were able to detect it in MGO-modified insulin by MS. We sequentially digested 100 μM MGO-modified insulin with Asp-N and V8 and analysed the product by LC-MS/MS. The precursor ion mass increment of 54 Da between the modified and unmodified peptides suggests the modification adduct of hydroimidazolone. Figure 6 shows the tandem mass spectra of the arginine-residue-containing peptide, R22GFFYTPKA30, from unmodified and MGO-modified insulin B-chains. The results indicate that MGO modification of hydroimidazolone occurs at Arg22. We did not detect argpyrimidine, although it was identified by HPLC in another experiment (Figure 5). The low abundance (∼6 pmol/100 μmol of amino acid) of this modification probably precluded its detection. We also noted a loss of 44 Da, which is consistent with decarboxylation during the collision-induced dissociation. Because insulin has only one arginine residue, modification of this single residue to hydroimidazolone (and possibly argpyrimidine as well) appears to prevent aggregation by DTT. Formation of hydroimidazolone as well as argpyrimidine likely prevents aggregation in other proteins as well.
In addition to formation of argpyrimidine and hydroimidazolone, MGO modification could influence aggregation of proteins through structural perturbations. Two reports assert that MGO modification perturbs the structure of α-crystallin so as to significantly increase surface hydrophobicity, although these studies failed to correlate surface hydrophobicity with chaperone function [22,24]. We used far- and near-UV CD spectroscopy to examine secondary and tertiary structural perturbation by MGO (10 and 100 μM) in all four client proteins. We found no apparent change in either the secondary or tertiary structure in any of these proteins (results not shown), suggesting that the resistance to thermal and chemical aggregation is not due to structural perturbation.
We used the hydrophobic probe, TNS, to define how MGO modification influences the surface hydrophobicity of the four client proteins. TNS-bound insulin, α-lactalbumin, CA and γ-crystallin displayed fluorescence with maximum emission (λmax) at 441, 459, 425 and 461 nm respectively. The λmax did not change when these four proteins were modified by 10 and 100 μM MGO. However, we noted a slight reduction in fluorescence (from ∼6% to 23%), indicating a decrease in surface hydrophobicity. Argpyrimidine fluoresces at 380 nm when excited at 335 nm. We believe that the amount of argpyrimidine formed was insufficient to produce a shoulder second peak relative to the TNS fluorescence at 380 nm. From these findings, we assume that even a slight decrease in surface hydrophobicity together with the formation of the MGO adducts, argpyrimidine and hydroimidazolone, conveys resistance to protein aggregation. However, at this point, we are unable to determine which of these factors is the most important.
The presence of argpyrimidine (Figure 5) and mild reduction of surface hydrophobicity suggest that the TNS- and MGO-binding sites are located in different regions in the four client proteins. Therefore we determined whether the TNS- and MGO-binding sites are mutually exclusive. We found argpyrimidine in CA and γ-crystallin (not treated with TNS, but incubated with MGO) to be ∼450 and ∼1150 nmol/μmol of amino acid. Surprisingly, we found similar concentrations in proteins treated first with TNS and then modified with MGO. This would suggest mutually exclusive binding sites for TNS and MGO on these two proteins.
We considered the possibility that MGO could have displaced TNS and modified arginine to the same extent as in the MGO-modified protein not treated with TNS. Accordingly, we determined the fluorescence of TNS-bound MGO-modified CA (with or without prior TNS treatment) (Figure 7). If MGO displaced TNS, it would result in lower fluorescence. However, the fluorescence intensity of TNS-treated and MGO-modified protein (λmax=404 nm) was higher than that of TNS-treated CA (λmax=420 nm); therefore we do not believe that MGO displaced TNS during incubation. These results also confirm that the MGO- and TNS-binding sites are mutually exclusive.
Our objective was to determine the impact of MGO modification of client proteins on chaperoning by αA-crystallin. Because MGO is a rapidly reacting molecule and it induces structural changes in proteins through formation of chemical adducts, we expected MGO modification to enhance thermally and chemically induced aggregation of those proteins and to decrease the chaperone function of αA-crystallin. Contrary to these expectations, we found that MGO modification inhibited in a concentration-dependent manner the chemically and thermally induced aggregation of four different proteins while having no effect on chaperoning by αA-crystallin. Our findings are similar to those of Wang and Spector , who found that oxidation of client proteins failed to influence chaperoning by α-crystallin. Even more intriguing is the fact that co-modification of both αA-crystallin and client protein with MGO affords greater protection against aggregation of a client protein than modification of either protein alone.
α-Crystallin binds to proteins that have undergone mild structural perturbation (molten globule state) and prevents their further denaturation and aggregation [31–34]. By this mechanism, α-crystallin helps to maintain lens transparency during aging. This protein protective mechanism is particularly important in the lens where protein turnover is negligible. However, lens proteins can be altered by post-translational modifications such as oxidation, glycation, truncation and de-amidation, all of which contribute to protein cross-linking and aggregation . The loss of α-crystallin chaperone function as a consequence of such modifications would diminish the protection that it affords against protein aggregation. In this context, modification of αA-crystallin by MGO is unique, because it enhances the chaperone function. Our research suggests that this enhancement is likely due to formation of hydroimidazolone and argpyrimidine adducts on selected arginine residues of αA-crystallin.
Both of these important MGO-induced modifications [36,37] are found in the human lens, where they constitute two of the major modifications to lens proteins. Although exactly which specific human lens proteins are modified is not yet clear, the major protein of the lens, i.e. α-crystallin, is likely included.
It is puzzling how MGO modification can afford increased protection against thermal and chemical denaturation of client proteins in the absence of any observable structural perturbation. A similar observation was made with βL-crystallin; it had increased resistance to thermal denaturation after H2O2-induced oxidation . We considered that MGO modification might decrease surface hydrophobicity. In fact, we noted a slight, but not significant, decrease in TNS fluorescence in MGO-modified client proteins. Whether this slight decrease can account for the resistance to aggregation remains to be verified.
The fact that MGO modification occurred even after TNS binding and that MGO failed to displace bound TNS on CA suggests that these two molecules bind to different regions of this protein. Although MGO- and TNS-binding sites are mutually exclusive, the TNS fluorescence intensity decreased slightly in MGO-modified client proteins compared with unmodified client proteins. Because MGO-binding sites influence the overall surface hydrophobicity of a protein, such a subtle change may make these client proteins resistant to thermal and chemical stress.
Importantly, modification of the single arginine in insulin was sufficient to increase its resistance to DTT-induced denaturation. One possibility is that arginine modification might prevent reduction of the disulfide bond by DTT, thus improving protection. A more likely scenario is that modification of positively charged arginine residues by MGO and conversion of these into neutral adducts decreases hydrophobic interaction during protein aggregation. This phenomenon may also occur in γ-crystallin and α-lactalbumin, both of which had increased thermal resistance after MGO modification. Taken together, it is evident that resistance to thermal and chemical stress by MGO modification is not restricted to a single protein or one type of stress.
We considered that MGO modification might mildly perturb client proteins sufficiently to enhance their binding to αA-crystallin, thus improving resistance to aggregation, as we found in the co-incubation experiments. However, our UV-CD spectroscopy data showed no change in either the secondary or the tertiary structure in client proteins that had been modified by MGO (<100 μM). MGO modification of αA-crystallin increases its surface hydrophobicity, which we believe underlies its enhanced chaperone function. Our observation that co-modification of αA-crystallin and client proteins further enhances resistance to thermal and chemical denaturation beyond that of either modified protein alone suggests a complex protective system. Such a protective system may be beneficial in the aging eye.
Finally, whether MGO in the lens is beneficial or harmful is difficult to predict. On the one hand, MGO-mediated lysine–lysine and lysine–arginine cross-linking (at high MGO concentrations) is likely to contribute to protein aggregation during aging. On the other hand, MGO modification of arginine residues on αA-crystallin and its chaperoning client proteins (at low MGO concentrations) might prevent aggregation of proteins during lens aging. An additional observation that MGO modification inhibits glycation-mediated loss in chaperone function and synthesis of pentosidine in α-crystallin  further suggests a beneficial role. Therefore, based on these observations, it is tempting to speculate that MGO modifications in the aging lens might benefit the lens by enhancing the chaperone function of αA-crystallin along with improving the resistance of client proteins against stress-induced aggregation.
This study was supported by NIH (National Institutes of Health) grants R01EY-016219 and R01EY-09912 (R. H. N.) and P30EY-11373 (Visual Sciences Research Center of Case Western Reserve University), Carl F. Asseff, M. D. Professorship to R. H. N., RPB (Research to Prevent Blindness; New York, NY, U.S.A.) and Ohio Lions Eye Research Foundation. We thank Michael Zagorski and Krzysztof Palczewski of Case Western Reserve University for use of the CD spectropolarimeter and the fluorescence spectrofluorimeter.
Abbreviations: AGE, advanced glycation end-product; CA, carbonic anhydrase; DTT, dithiothreitol; HFBA, heptafluorobutyric acid; LC-MS/MS, liquid chromatography tandem MS; MGO, methylglyoxal; sHSP, small heat-shock protein; TNS, 2-(p-toluidino)naphthalene-6-sulfonic acid, sodium salt
- © The Authors Journal compilation © 2008 Biochemical Society