Human alphaB-crystallin. Small heat shock protein and molecular chaperone.

The polymerase chain reaction was used to amplify a cDNA sequence encoding the human aB-crystallin. The amplified cDNA fragment was cloned into the bacterial expression vector pMAL-c2 and expressed as a soluble fusion protein coupled to maltose-binding protein (MBP). After maltose affinity chromatography and cleavage from MBP by Factor Xa, the recombinant human aB-crystallin was separated from MBP and Factor Xa by anion exchange chromatography. Recombinant aB-crystallin was characterized by SDS-polyacrylamide electrophoresis (PAGE), Western immunoblot analysis, Edman degradation, circular dichroism spectroscopy, and size exclusion chromatography. The purified crystallin migrated on SDS-PAGE to an apparent molecular weight (Mr ;22,000) that corresponded to total native human a-crystallin and was recognized on Western immunoblots by antiserum raised against human aB-crystallin purified from lens homogenates. Chemical sequencing, circular dichroism spectroscopy, and size exclusion chromatography demonstrated that the recombinant crystallin had properties similar or identical to its native counterpart. Both recombinant aB-crystallin and MBP-aB fusion protein associated to form high molecular weight complexes that displayed chaperonelike function by inhibiting the aggregation of alcohol dehydrogenase at 37 °C and demonstrated the importance of the C-terminal domain of aB-crystallin for chaperone-like activity.

The cytoplasm of human lens cells is composed predominantly of a group of soluble globular proteins known as the crystallins. The crystallin proteins in the human lens are categorized as the ␣and ␤/␥-crystallin families (1). The structure, stability, and short range order of the crystallins are thought to contribute to the transparency of the vertebrate ocular lens (2). ␣-Crystallin is the most abundant of the crystallins in the vertebrate lens and is composed of two M r 20,000 subunits, ␣A and ␣B, which associate to form high molecular weight oligomers ranging from approximately 3 ϫ 10 5 to 1.2 ϫ 10 6 daltons (3). The two proteins are derived from single copy genes that share approximately 55% identity (1). The ␣-crystallins also share sequence similarity with the small heat shock proteins of numerous species (4). The tertiary structures of the ␣-crystallins have not been elucidated, but it is possible that they share the two-domain structure found in other crystallins (5,6). In vitro a chaperone-like activity has been described for bovine ␣-crystallins in suppressing the aggregation of proteins denatured at high temperature (7)(8)(9)(10)(11)(12)(13). Human ␣B-crystallin expression is found under normal conditions in many nonlens cells and tissues, including heart, brain, skeletal muscle, kidney, placenta, and lung (14 -22) and, like the ubiquitous small heat shock proteins, is dramatically up-regulated in response to stress and under pathological conditions (23)(24)(25)(26)(27)(28)(29)(30). ␣B-Crystallin has biochemical properties that result in its copurification with mammalian heat shock protein 28 from human skeletal muscle (18). The ␣B-crystallin gene has been shown to contain a heat shock element in its promoter that may be subject to a heat-regulated control mechanism (31).
The chemical nature of the interactions between human ␣-crystallins and other proteins is poorly understood because of the difficulty with isolation of sufficient quantities of unmodified protein from human lenses. For this reason, recombinant techniques have been used to characterize structure-function relationships of the individual ␣-crystallin subunits. The conformational properties of substrate proteins bound to recombinant human ␣A-crystallin was recently reported (32); however, unlike human ␣B-crystallin, human ␣A-crystallin lacks a heat shock element in its promoter, is not induced by stress or pathological conditions, and has a limited expression pattern in the body. In this article we report expression and characterization of the small heat shock protein human ␣B-crystallin cloned from a fetal lens cDNA library. To the best of our knowledge this is the first demonstration that human ␣B-crystallin displays molecular chaperone activity. The recombinant expression system described here provided an excellent source of unmodified human ␣B-crystallin that assembled into a high molecular weight oligomer, and displayed chaperone-like activity against protein aggregation.

EXPERIMENTAL PROCEDURES
Construction of Human Fetal Lens cDNA Library-RNA was isolated from 9 -12-week human fetal lenses as described previously (33). Oligo(dT)-primed cDNA was prepared from 2 g of total RNA using the Boehringer Mannheim cDNA synthesis kit according to the manufacturer's instructions. The cDNA was EcoRI-linkered, digested with EcoRI, and ligated to EcoRI gt11 arms (Promega, Leiden, The Netherlands). After in vitro packaging, the phages were plated on Escherichia coli Y1090 cells.
Construction of Human ␣B-Crystallin Expression Vector-The coding region for human ␣B-crystallin was isolated from the fetal lens cDNA library by polymerase chain reaction and inserted into the cloning vector pCR TM II (Invitrogen, San Diego, CA). The following primers, which correspond to the 5Ј-and 3Ј-ends of the coding region for human ␣B-crystallin (16,36), were synthesized and used in the polymerase chain reactions: 5Ј-CCAGAATTCATGGACATCGCCATCCACCAC-3Ј (forward) and 5Ј-CCATCTAGATCATTTCTTGGGGGCTGCGGT-3Ј (reverse). EcoRI and XbaI sites were attached to the 5Ј-ends of the forward * This work was supported by National Institutes of Health Grant EY04542 and Vision Training Grant Research Training in Biotechnology Grant T32 GM 08437, EY07031. 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.
ʈ To whom correspondence should be addressed: Dept. of Biological Structure, Box 357420, Seattle, WA 98195-7420. Tel.: 206-685-0950; Fax: 206-543-1524; E-mail: clarkji@u.washington.edu. and reverse primers, respectively. The coding region for the human ␣B-crystallin was then removed from the pCR TM II vector at its flanking restriction sites and ligated into EcoRI-XbaI-digested pMAL TM -C2 (New England Biolabs, Beverly, MA) to produce pMAL-C2-␣B3. The coding region was inserted downstream of the malE gene of E. coli, which encodes MBP, 1 resulting in the expression of a MBP-␣B fusion protein. The coding region of this expression construct was confirmed by DNA sequence analysis using the dideoxy chain termination method (34).
Expression and Purification of Human MBP-␣B Fusion Protein-The pMAL-C2-␣B3 expression plasmid was used to transform competent E. coli JM109 cells (Stratagene, San Diego, CA). One liter of L broth that contained 100 g/ml ampicillin was inoculated with 10 ml of an overnight culture of the transformed E. coli cells, and cells were grown with vigorous shaking at 37°C. The cells were incubated until the culture reached an optical density of ϳ0.5 at A ϭ 600 nm, at which point protein expression was induced by addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.3 mM (Sigma). Four hours after induction, cells were harvested by sedimentation and resuspended in 50 ml of column buffer (20 mM Tris-HCl, pH 7.4, 200 mM NaCl, and 1 mM EDTA). Cells were stored overnight at Ϫ20°C. After thawing, the cells were disrupted by sonication on ice by eight 30-s cycles at 70 watts on a Branson (Plainview, NY) Ultrasonics power sonifier. Insoluble cellular debris was removed by sedimentation at 9,000 ϫ g for 30 min at 4°C. Soluble fusion protein present in the supernatant was purified by adsorption to a 10-ml amylose resin affinity column (New England Biolabs) for 1 h at 25°C. After washing with 10 column volumes of column buffer, bound fusion protein was eluted using column buffer that contained 10 mM maltose. Preparations of fusion protein were concentrated using Centriplus-10 microconcentrators (Amicon, Beverly, MA). Preparations of fusion protein contained Ͻ5% contaminating proteins as assessed by SDS-PAGE and Coomassie Blue staining. Concentrations of purified fusion protein were determined by a protein assay (Bio-Rad).
Purification of Recombinant Human ␣B-Crystallin-MBP was cleaved from human ␣B using the protease Factor Xa, the recognition sequence of which is encoded in the pMAL-c2-␣B3 vector within the fourth and fifth codons 5Ј from the coding region for ␣B-crystallin. Recombinant human ␣B-crystallin was purified by anion exchange chromatography in the presence of 8 M urea. Briefly, cleaved fusion protein was dialyzed against ion exchange buffer (20 mM Tris-HCl, pH 8.0, 25 mM NaCl, 1 mM EDTA, and 8 M urea). MBP and Factor Xa were separated from human ␣B-crystallin by absorption to a 10-ml column of Q-Sepharose anion exchange resin (Pharmacia Biotech Inc.). In the presence of ion exchange buffer, recombinant human ␣B-crystallin had no affinity for the Q-Sepharose anion exchange resin and hence could easily be separated from MBP and Factor Xa by this method. Preparations of recombinant human ␣B-crystallin were dialyzed overnight against ion exchange buffer that lacked 8 M urea to promote reoligomerization and then concentrated on Centriplus-10 microconcentrators. Preparations of recombinant human ␣B-crystallin contained Ͻ5% contaminating proteins as assessed by SDS-PAGE and Coomassie Blue staining. Concentrations of purified recombinant human ␣B-crystallin were determined by a protein assay (Bio-Rad).

SDS-PAGE and Western Immunoblot Analyses of Recombinant
Human ␣B-Crystallin-Proteins were analyzed on 4 -20% polyacrylamide electrophoretic gels in the presence of 0.1% SDS (Novex, San Diego, CA) and were stained with Coomassie Blue R-350 (Pharmacia). Proteins were electrophoretically transferred to a polyvinylidene difluoride membrane using a Blot Module II (Novex). Antiserum raised against ␣B-crystallin purified directly from human lens homogenates was used as a primary antibody (35). For immunodetection, alkaline phosphatase conjugated to goat anti-rabbit IgG antibody and 5-bromo-4-chloro-3Јindolylphosphate and p-nitro blue tetrazolium chloride (Bio-Rad) were used.
Edman Degradation of Human ␣B-Crystallin-N-terminal sequencing of the purified human ␣B-crystallin was performed by sequential Edman degradation (15 cycles) using an Applied Biosystems 477A liquid phase protein sequencer with an on-line 120A phenylthiohydantoin analyzer after immobilization on a polyvinylidene difluoride membrane.
Circular Dichroism Spectroscopy of Human ␣B-Crystallin-Circular dichroism spectra were obtained using a Jasco 720 spectropolarimeter. A 0.5-mm path length cell was used. Sixteen scans were averaged per sample, and spectra of the buffers were subtracted from the spectra of the protein samples.
Reassociation of Human MBP-␣B and ␣B-Crystallin-Recombinant human MBP-␣B and ␣B-crystallin were fractionated by size exclusion chromatography on a Macrosphere GPC 300-Å, 7 m, 250 ϫ 4.6-mm size exclusion column (Alltech, Deerfield, IL) using a Hewlett-Packard (Palo Alto, CA) 1090 high performance liquid chromatography with the following mobile phase: 0.1 M KH 2 PO 4 (pH 7.0) and 0.2 M NaCl, at a flow rate of 0.1 ml min Ϫ1 . High molecular weight protein standards (Pharmacia) were used to calibrate the column.
Aggregation Measurements-The effect of human ␣B-crystallin on protein aggregation was measured as described previously (7). Briefly, the aggregation of ADH at 37°C was measured as the apparent optical density at A ϭ 360 nm using a Shimadzu (Kyoto, Japan) UV-160 UV-visible recording spectrophotometer equipped with a temperaturecontrolled cuvette holder. In a total reaction volume of 400 l, 5 M equine liver ADH (Sigma) was incubated at 37°C with varying amounts of purified human MBP-␣B or ␣B-crystallin. All reagents were diluted into the following reaction buffer: 50 mM sodium phosphate buffer (pH 7.0), 0.1 M NaCl, and 2 mM EDTA. Stock solutions were stored on ice until they were mixed at room temperature and quickly placed in the temperature-controlled sample chamber of the spectrophotometer. The temperature in the cuvette was monitored using a bead thermistor installed in a cuvette within the sample chamber. The optical density in each cell was recorded every 60 s.

RESULTS
Characterization of Human ␣B-Crystallin cDNA Clone-The cDNA-coding region of ␣B-crystallin, isolated by polymerase chain reaction from a human fetal lens cDNA library, was ligated into the plasmid pMAL TM -c2 to produce pMAL-c2-␣B3. Double-stranded dideoxy sequencing of both strands encoding pMAL-c2-␣B3 demonstrated that the ␣B sequence was 100% identical to the corresponding exon sequence of the ␣B-crystallin gene and to the coding regions of partial-and full-length ␣B-crystallin cDNA clones (16,36,37). The N terminus of the recombinant human ␣B-crystallin contained 4 additional amino acids: isoleucine, serine, glutamic acid, and phenylalanine, all of which are derived from the insertion of the human ␣B-coding region into the EcoRI and XbaI sites in the polylinker of pMAL TM -c2. Hence, pMAL-c2-␣B3 contained an open reading frame of 537 base pairs, predicted to encode a polypeptide of 179 amino acids after cleavage and separation from MBP.
Expression and Purification of Human MBP-␣B and ␣B-Crystallin- Fig. 1 contains SDS-PAGE (A) and Western immunoblot analysis (B) of the expression and purification of recombinant human ␣B-crystallin. Fig. 1A is a Coomassie Blue stain of a polyacrylamide electrophoretic gel run in the presence of 0.1% SDS. Control of induction of MBP-␣B fusion protein expression was apparent in crude cell lysates of bacterial cultures transformed with the human ␣B-crystallin expression construct before and after the addition of isopropyl-␤-D-thiogalactopyranoside (Fig. 1A, lanes 1 and 2, respectively). Treatment of the affinity-purified fusion protein (Fig. 1A, lane 3) with the serine protease Factor Xa demonstrated that this fusion protein was cleaved into two distinct polypeptides (Fig. 1A, lane 4) that migrate to molecular weights corresponding to native MBP (ϳ43,000) and native human ␣B-crystallin (ϳ22,000). The recombinant human ␣B-crystallin was isolated by anion exchange chromatography in the presence of 8 M urea and was found to contain Ͻ5% contaminating proteins as assessed by SDS-PAGE (Fig. 1A, lane 5). Further confirmation of the expression and purification of recombinant human ␣B-crystallin was demonstrated by Western immunoblot analysis using antihuman ␣B-crystallin antiserum (Fig. 1B). One predominant immunoreactive band was observed in the purified uncleaved fusion protein (Fig. 1B, lane 1), in the mixture of cleaved fusion protein (Fig. 1B, lane 2), and with purified human ␣B-crystallin (Fig. 1B, lane 3).
Chemical Sequencing Results of Human ␣B-Crystallin-To verify that the M r ϳ22,000 band corresponded to human ␣Bcrystallin, the cleaved fusion protein mixture was immobilized on a polyvinylidene difluoride membrane, and the M r ϳ22,000 band was subjected to sequential Edman degradation for 15 cycles. The chemical sequencing results demonstrated that after the first four vector-derived N-terminal amino acids (ISEF) the next 11 residues of this protein (MDIAIHHPWIR) were 100% identical to the deduced amino acid sequence predicted from the exon sequence of the ␣B gene (36).
Secondary Structure of Human ␣B-Crystallin-In Fig. 2 the far UV circular dichroism spectra of recombinant human ␣Bcrystallin in the absence and presence of 1% SDS are presented. The spectrum of recombinant human ␣B-crystallin in the presence of buffer alone (Fig. 2, solid line) resembles published spectra for total native bovine ␣-crystallin as well as recombinant and native bovine ␣A-crystallin and is typical of a spectrum for a predominantly ␤-pleated structure (12, 38 -40). On addition of SDS to a final concentration of 1%, an increase in molar ellipticity was observed at all wavelengths, indicative of major structural changes (Fig. 2, dashed line) in the spectrum, as reported previously for total native bovine ␣-crystallin (38).
Reassociation of Human MBP-␣B and ␣B-Crystallin-Fractionation of the recombinant MBP-␣B fusion protein and the ␣B-crystallin alone by chromatography on a size exclusion column under nondenaturing conditions demonstrated that they formed high molecular weight oligomers between 2.32 ϫ 10 5 and 4.40 ϫ 10 5 in size (Fig. 3).
Complete Inhibition of the Thermally Induced Aggregation of ADH by Human MBP-␣B and ␣B-Crystallin-The apparent optical density was a direct measure of the aggregation of ADH at 37°C over 60 min. Human ␣B-crystallin and MBP-␣B suppressed the aggregation of ADH over a range of concentrations from 1 to 100 nM (Fig. 4, A and B, respectively). We assumed molecular weights of 8.0 ϫ 10 4 for ADH, 1.3 ϫ 10 6 for MBP-␣B-crystallin, and 4.4 ϫ 10 5 for ␣B-crystallin. At a molar ratio of 200:1 (ADH:human MBP-␣B/␣B-crystallin), partial inhibition of ADH aggregation was observed. At a molar ratio of 50:1 (ADH:human MBP-␣B/␣B-crystallin), complete suppression of ADH aggregation was observed. At a molar ratio of 5,000:1 (ADH:human MBP-␣B/␣B-crystallin), no effect on the aggregation of ADH was observed. As a control, MBP alone was tested at a molar ratio of 1:1 or lower (ADH:MBP), and no effect on the aggregation of ADH was observed (data not shown).

DISCUSSION
Human lens crystallins undergo extensive posttranslational modifications during the aging process (41)(42)(43)(44). These modifications lead to protein heterogeneity, which has precluded the successful purification of large amounts of homogenous human crystallins to be used for biophysical, functional, and structural analyses. Here we report for the first time expression and characterization of functional human ␣B-crystallin in E. coli.
DNA primers designed against the 5Ј-and 3Ј-ends of the ␣B coding region successfully amplified a ϳ525-base pair cDNA from a human fetal lens cDNA library, confirming that this gene is transcriptionally active in the human lens (data not shown). The amplified coding region for this crystallin was subcloned into a bacterial expression plasmid and expressed in the bacterial cytoplasm as a soluble fusion protein coupled to MBP. After cleavage and separation from MBP, SDS-PAGE and Western immunoblot analysis confirmed that the recombinant protein migrated to its predicted molecular weight (M r ϳ22,000). This polypeptide was recognized by antiserum raised against human ␣B-crystallin purified directly from lens homogenates. N-terminal chemical sequencing demonstrated that after the first four vector-derived amino acids (ISEF) the recombinant human ␣B-crystallin was 100% identical to the deduced amino acid sequence predicted from the exon sequence of the human ␣B-crystallin gene (36). Although the N-terminal methionine of the recombinant ␣B-crystallin is preceded by four additional residues (isoleucine, serine, glutamic acid, and phenylalanine), circular dichroism spectroscopic analysis indicated that the secondary structure of the reassociated ␣B resembled the secondary structures previously reported for total purified bovine ␣-crystallin as well as recombinant and purified bovine ␣A-crystallin (12, 38 -40).
On a size exclusion chromatography column recombinant human MBP-␣B and ␣B-crystallin eluted as high molecular weight oligomers similar in size (between 2.3 ϫ 10 5 and 4.4 ϫ 10 5 ) to total ␣-crystallin purified directly from human eye lenses (45). Interestingly, MBP-␣B fusion protein was able to associate into a high molecular weight oligomer despite the presence of the M r 42,700 N-terminal fusion partner MBP. Assuming molecular weights of 1.3 ϫ 10 6 for MBP-␣B and 4.4 ϫ 10 5 for ␣B-crystallin, both the human MBP-␣B and ␣B-crystallin displayed molecular chaperone activity, as demonstrated by their complete suppression of the aggregation of ADH at a molar ratio of approximately 50:1 (ADH:crystallin). When the molecular weights of the subunits of MBP-␣B and ␣B-crystallin were used (6.5 ϫ 10 4 and 2.2 ϫ 10 4 , respectively), complete suppression of the aggregation of ADH was observed at a molar ratio of approximately 2.5:1 (ADH:crystallin). A systematic evaluation of the precise stoichiometry involved in the suppression of ADH aggregation by MBP-␣B and ␣B-crystallin will need to be addressed in the future.
It was recently demonstrated using site-directed mutagenesis that specific residues at the C terminus of recombinant bovine ␣A-crystallin influenced its reassociation and chaperone activity (46). To date, no mutants of human ␣B-crystallin have been characterized; however, in the studies presented here the presence of four extra N-terminal residues had no observable effect on the secondary structure, reassociation, and chaperone activity of recombinant human ␣B-crystallin. Strikingly, human MBP-␣B formed a large functional structure and was able to completely suppress the aggregation of ADH at concentrations identical to those used with recombinant human ␣Bcrystallin alone. Taken together, these results contribute to the evidence that the well conserved C-terminal domain of the ␣-crystallins is responsible for the proper assembly into large functional oligomers (5, 6) .
At the amino acid level ␣B-crystallin is remarkably well conserved among mammalian species. This high degree of conservation among species may indicate a critical function for ␣B-crystallin in nonlens cells, in which expression occurs in response to environmental and pathological stress. The successful expression of functional recombinant human ␣B-crystallin creates the first opportunity to characterize the chemical basis of the interactions between ␣B and other proteins in lens and nonlens cells under normal and pathological conditions.