Sequence Analysis of betaA3, betaB3, and betaA4 Crystallins Completes the Identification of the Major Proteins in Young Human Lens

A combination of Edman sequence analysis and mass spectrometry identified the major proteins of the young human lens as αA, αB, βA1, βA3, βA4, βB1, βB2, βB3, γS, γC, and γD-crystallins and mapped their positions on two-dimensional electrophoretic gels. The primary structures of human βA1, βA3, βA4, and βB3-crystallin subunits were predicted by determining cDNA sequences. Mass spectrometric analyses of each intact protein as well as the peptides from trypsin-digested proteins confirmed the predicted amino acid sequences and detected a partially degraded form of βA3/A1 missing either 22 or 4 amino acid residues from its N-terminal extension. These studies were a prerequisite for future studies to determine how human lens proteins are altered during aging and cataract formation.

Elucidating the structure and function of crystallins, the major proteins of human lens, is important, because alterations in these proteins may contribute to cataract formation. The goal of our laboratories is to determine what changes occur in human crystallins during aging and cataract formation. To accomplish this goal it is first necessary to perform the following in young, normal human lenses: 1) determine which crystallin subunits are present in the young lens; 2) map the positions where these crystallins migrate on two-dimensional electrophoretic gels; and 3) deduce and confirm the amino acid sequences of these proteins.

MATERIALS AND METHODS
Identification of Major Human Crystallins-The posterior poles of human eyes from organ donors 7-months of age or less were obtained from the Lions Eyebank of Oregon within 48 h post-mortem. Following decapsulation, the lenses were homogenized in 1.0 ml of 20 mM phosphate buffer (pH 7.0), and 0.1 mM EGTA. Water-soluble proteins were isolated by centrifugation at 10,000 ϫ g for 30 min. Protein content was assayed by the BCA assay (Pierce) according to the manufacturer's protocol using bovine serum albumin as a standard. Water-soluble fractions were then dried by vacuum centrifugation and stored at Ϫ70°C prior to electrophoretic or chromatographic separation.
Two-dimensional electrophoresis of water-soluble lens proteins, transfer to polyvinylidene difluoride membranes, and direct sequence analysis were carried out as described previously (15). Non-equilibrium pH gradient electrophoresis using pH 3.5-10 ampholine was used in the first dimension and SDS-polyacrylamide electrophoresis in the second dimension. However, most crystallins were blocked on the N terminus and could not be identified by direct sequence analysis. Therefore, electroblots of two-dimensional gels were reversibly stained with Ponceau S, individual proteins digested with trypsin, and peptides isolated from the resulting mixtures for sequencing as described previously (16), except that peptides were separated using a 250 ϫ 2.1-mm C 18 column (Vydac, Hesperia, CA), and linear 100-min 0 -35% acetonitrile gradient containing 0.1% trifluoroacetic acid. The tryptic peptides derived from regions of electroblots thought to contain more than one crystallin species were additionally analyzed by mass spectrometry using either FAB-MS 1 or ESIMS as described below.
The identified crystallins were quantified after staining two-dimensional gels with Coomassie Brilliant Blue R-250, or colloidal Coomassie Brilliant Blue G-250 (Instaview Universal Stain, Gallard-Schlesinger Industries, Inc., Carle Place, NY). Each of the stained gels contained 100 g of total protein. The density of each species was determined from images obtained with a Gel Doc 1000 camera and analysis using Molecular Analyst Software (Bio-Rad).
Cloning and Sequencing of Human ␤A4, ␤B3, and ␤A3/A1 cDNAs-Lenses from organ donors 19-months of age or less were obtained as described above. Total lens RNA was isolated by homogenation of dissected lenses in TRIzol reagent according to the manufacturer's protocol (Life Technologies, Inc.). To amplify the 3Ј end of the human lens ␤A4 cDNA, reverse transcription was performed on total RNA from human lens using an oligo(dT) containing adapter primer as described previously (3). The resulting cDNA was then subjected to 3Ј-RACE PCR according to the manufacturer's protocol (3Ј-RACE System for Rapid Amplification of cDNA Ends, Life Technologies Inc.), using a 3Ј-RACE universal adapter primer, and the gene-specific sense primer AGGCTGACCATCTTCGAGCA matching residues 390 -410 of bovine ␤A4 cDNA (GenBank accession number M60328). Cycling conditions for PCR were as described previously (3), except an annealing temperature of 61°C was used. The amplification resulted in a single PCR product of approximately 550 base pairs.
To amplify the 5Ј end of the human lens ␤A4 cDNA, reverse transcription was performed on total RNA from human lens using the gene-specific antisense primer GTCATCGCTCAGCTCTCCTT complementary to residues 392-411 of the final human ␤A4 cDNA sequence.
The cDNA was then homopolymer tailed according to the manufacturer's protocol (5Ј-RACE System for Rapid Amplification of cDNA Ends, Life Technologies Inc.). The resulting cDNA was then PCR amplified using a second gene-specific antisense primer, TCTTGCCCAGGAAGT-TCTCT, complementary to residues 371-391 of the final human ␤A4 cDNA and a 5Ј-RACE anchor primer supplied by the manufacturer (Life Technologies, Inc.). Cycling conditions for PCR were as described previously (3), except an annealing temperature of 55°C was used. The amplification resulted in a single PCR product of approximately 450 base pairs.
To amplify the unknown 5Ј end of human ␤B3 cDNA, the same procedure described above for 5Ј-RACE of human ␤A4 cDNA was followed. The gene-specific antisense primer, CAGACCACAAGCTG-CATCTGT, complementary to nucleotides 8 -28 in exon 5 of human ␤B3 (GenBank accession number X15145), was used for reverse transcription of total lens RNA. A second gene-specific antisense primer, CCTC-CGGCCTCTGAATATT, complementary to nucleotides 115-133 in exon 4 of human ␤B3 (GenBank accession number X15144), was used to perform 5Ј-RACE of human ␤B3 cDNA. Cycling conditions were as described previously (3), except an annealing temperature of 60°C was used. A PCR product of approximately 400 base pairs was obtained.
Preliminary mass spectrometric analysis of human ␤A3 protein suggested that a portion of the reported genomic sequence of human ␤A3 contained several errors in nucleotide identification in exons 4 -6 (Gen-Bank accession numbers M14304, M14305, and M14306). Therefore, a corresponding region of human ␤A3 cDNA was amplified and sequenced for comparison with the earlier genomic sequence. Total human lens RNA was reversed transcribed using the gene-specific antisense primer, GCAAGGTCTCATGCTTGAGG, complementary to residues 185-204 of exon 6 of human ␤A3 (GenBank accession number M14306). After treating with RNase H, the cDNA was PCR amplified using the gene-specific sense primer, TGATCAGGAGAACTTTCAGG, matching residues 366 -385 of exon 3 of human ␤A3 (GenBank accession number M14303) and the antisense primer used above in the reverse transcription reaction. Cycling conditions for PCR were as described previously (3), except an annealing temperature of 55°C was used. A PCR product near the expected size of 589 base pairs was produced.
PCR products were cloned with either the Prime PCR Cloner Cloning System (5 Prime 3 3 Prime, Inc., Boulder, CO) or with the Original TA Cloning® Kit (Invitrogen, San Diego, Ca). Plasmid DNA was isolated using either a FlexiPrep Kit (Pharmacia-Biotech, Inc., Piscataway, NJ), or Quantum Prep kit (Bio-Rad). Following screening to confirm the presence of the correct sized insert, sequencing was performed using either the AutoRead sequencing kit and automated laser fluorescence DNA analysis system (Pharmacia Biotech), or a Cycle Sequencing Kit and Model 373A DNA Sequencer (Applied Biosystems, Inc., Foster City, CA). Depending on the plasmid utilized, sequencing was performed using either M13/pUC forward and reverse primers or T7 and Sp6 primers. Three clones of each of the 3Ј-and 5Ј-RACE products of the ␤A4 cDNA, and 2 clones of the PCR product of the ␤A3 cDNA were sequenced in both sense and antisense directions. The sequence from 4 antisense strands of the 5Ј-RACE product of ␤B3 cDNA were sequenced. DNA sequences were edited and theoretical pI values of proteins calculated using Geneworks 2.5 software (IntelliGenetics, Mountain View, CA).
Confirmation of Deduced ␤A3/A1, ␤B3, and ␤A4 Protein Sequences using Mass Spectrometry-The deduced amino acid sequences of ␤A3/ A1, ␤B3, and ␤A4 crystallins were confirmed by: 1) determining the molecular mass of intact proteins; 2) determining the molecular masses of tryptic peptides; and 3) confirming the sequences of tryptic peptides. The types of mass spectrometry employed were ESIMS, FAB-MS, and tandem mass spectrometry (MS/MS).
Soluble lens protein was isolated from 32-week gestation, 3-7-dayold, and 42-year-old human donors as described above. Proteins were dissolved in gel filtration buffer containing 50 mM phosphate (pH 7.0), 150 mM NaCl, and 5-mg portions injected onto a Superose 6 HR 10/30 gel filtration column (Pharmacia Biotech) at a 0.2 ml/min flow rate. Protein elution was monitored at 280 nm and ␤-crystallin aggregates of approximately 160,000 and 50,000 molecular weight (␤ H -and ␤ L -crystallin fractions, respectively) were collected. These aggregates were then concentrated and desalted using Centricon 10 microconcentrators (Amicon, Inc., Beverly, MA) in preparation for mass analysis.
␤-Crystallin subunits were partially purified from ␤ H and ␤ L aggregates using a Vydac 4.6 ϫ 150-mm C 4 reversed phase column and 25-60% acetonitrile gradient containing 0.1% trifluoroacetic acid over 35 min and 1 ml/min flow rate. Reversed phase HPLC purified ␤-crystallins were then injected into the mass spectrometer with a 50:50 solution of acetonitrile and water at a flow rate of 5 l/min. The masses of the isolated ␤-crystallins were determined using a Micromass Platform II electrospray ionization mass spectrometer with a quadrupole analyzer and Mass Lynx software (Micromass, Manchester, UK). For protein analysis, the instrument was calibrated with horse skeletal muscle myoglobin over the range of 700-1600 Da. Accuracy of protein molecular mass determinations was Ϯ3 Da.
As an alternative to reversed phase HPLC, two-dimensional electrophoresis was used as a preparative tool to isolate the truncated ␤A3/A1 species. Total soluble proteins from a 4-day-old human donor were separated by two-dimensional electrophoresis, proteins visualized by precipitation of SDS-protein complexes with ice-cold 0.25 M KCl (17), and the trucated ␤A3 species excised and electroeluted from 12 gels. Proteins were electroeluted into a Centricon 10 microconcentrator using the method recommended by the manufacturer (Amicon, Inc.). Electroelution into this device facilitated the concentration and desalting of the sample for analysis by ESIMS as described above.
The deduced amino acid sequences of ␤A3/A1, ␤B3, and ␤A4 were confirmed from the masses of peptides in tryptic digests of the proteins (50:1 substrate:trypsin, pH 8.2, 4 h). Lenses of 32-week-gestation, 4-day-old, and 42-year-old donors were used to isolate ␤B3, ␤A3, and ␤A4 for tryptic digestion, respectively. Tryptic peptides of ␤A4 and ␤B3 were prepared from proteins isolated by gel filtration and reversedphase HPLC. Peptides of ␤A3 were prepared from protein separated by two-dimensional electrophoresis, transfer to polyvinylidene difluoride membrane, and digestion from the membrane surface with trypsin. The peptides were fractionated by C 18 reversed-phase HPLC with a gradient of 2-50% acetonitrile in water, containing 0.1% trifluoroacetic acid. For peptide analysis, the mass spectrometers used included a Kratos MS-50 fast atom bombardment mass spectrometer (Kratos Analytical, Manchester, UK), a Micromass Platform II electrospray ionization mass spectrometer and a Micromass Autospec mass spectrometer with an orthogonal TOF analyzer for the MS/MS analyses. Some analyses employed a microbore column on-line to the electrospray ionization mass spectrometer (with a flow rate of 50 l/min), with 5 l/min entering the mass spectrometer and 45 l/min monitored by a UV detector and collected for further analysis. The instruments were calibrated over the range of 200-3000 with NaI; the accuracy of the determinations was Ϯ0.3 Da.
Sequences of tryptic peptides were also confirmed by MS/MS. This technique consists of one mass spectrometer to isolate the peptide of interest, a chamber where the peptide is fragmented by collision with xenon, and a second mass analyzer which determines the masses of the resulting fragments (18). The tandem mass spectrometer used in this investigation (Micromass Autospec oa-TOF) consisted of a conventional magnetic sector instrument as the first mass spectrometer and an orthogonal acceleration time-of-flight analyzer as the second mass spectrometer (19). These analyses were performed in the fast atom bombardment mode of ionization.

Identification of Crystallin Subunits of Young Lens Separated by Two-dimensional Electrophoresis-Two-dimensional
FIG. 1. Two-dimensional electrophoresis of soluble lens protein from 3-day-old human donor. Identification of major crystallin species was performed by both Edman sequencing and mass spectrometric analysis as summarized in Table I and Figs. 2 and 3. The identification of ␤B1-crystallin and its degradation product ␤B1 (16-251) was previously reported (3). Note that ␥S and ␤A1, as well as ␤A3and ␤B3-crystallins co-migrate. The left side of the gel is basic and the right side is acidic. Only the region of the two-dimensional gel containing crystallins is shown. electrophoresis of the constituent proteins in the water-soluble fraction from lenses of young human donors revealed 11 major proteins (Fig. 1). The identification of ␤B1 and ␤B1 missing 15 residues from its N terminus (␤B1, residues 16 -251) was previously reported (3). Due to blockage of the N termini, the identification of ␤B2, ␤A3, ␤A4, ␥S, ␣A, and ␣B subunits was performed by trypsinization of electroblotted proteins, separation of tryptic fragments, and 6 -15 cycles of Edman sequencing of isolated tryptic fragments (Table I). These results confirm the previously reported positions of human ␤B1, ␤B2, ␥S, ␣A, and ␣B on two-dimensional gels (12)(13)(14) and demonstrate the presence of ␤A3 and ␤A4 in human lens and their positions on two-dimensional gels.
Due to their lack of acetylated N termini, the positions of ␥C and ␥D, as well as a truncated ␤-crystallin, could be determined by direct sequence analysis of electroblotted proteins (Fig. 1, Table I). The truncated subunit contained a sequence matching that of ␤A3/A1 and could have arisen from truncation of the first 22 amino acids of ␤A3 or the first 4 amino acids of ␤A1. ␤A1 and ␤A3 are identical, except ␤A3 contains a longer N-terminal extension due to the use of an alternate start codon in the single ␤A3/A1 transcript (20). For simplicity, the truncated ␤-crystallin is referred to as ␤A3 (23-215). The presence of ␤A3 (23-215) in human lens was recently reported in another laboratory (21).
Although mRNAs coding for ␥A, ␥B, ␥C, and ␥D have been reported in fetal and neonatal human lenses, transcripts for ␥C and ␥D were more abundant (7). In the present study, ␥C and ␥D crystallins were detected, but no ␥A or ␥B crystallins were observed (Fig. 1, Table I). Earlier mass spectrometric and chromatographic analysis also detected ␥C and ␥D in human lenses, and found either small or undetectable amounts of ␥A and ␥B (10, 11). This suggested that little translation of either   ϭ 4), resulting from the densitometric analysis of individual two-dimensional electrophoretic gels of soluble protein of newborn, 3-, 4-, and 7-day old human donors.
b The individual amounts of ␥S, ␤A1, ␤A3, and ␤B3 could not be estimated because these subunits migrated to similar positions during two-dimensional electrophoresis.
c The remaining 2.3% of the total crystallin was due to minor, unidentified species migrating to the acidic sides of ␤A3/␤B3 and ␤A4.   b Several analyzed fractions of the reverse-phase purified tryptic fragments contained more than one peptide.
␥A or ␥B mRNA may occur in human lens. The lower migration position of ␥D was not expected (Fig. 1), since its calculated molecular weight was nearly identical to ␥C. This is likely explained by anomalous migration of ␥-crystallins during SDS-PAGE and not by post-translational modification.
Many of the crystallins described above were identified from the sequence of one peptide in each tryptic digest. This did not exclude the possibility that some spots observed on two-dimensional gels actually contained more than one protein species. Therefore, mass spectrometry was used to determine the masses of all peptides in tryptic digests from each of the two regions previously thought to contain only ␥S or ␤A3. The region containing ␥S (as determined by Edman sequencing) yielded peptides with masses matching peptides expected from both ␥S and ␤A1. Representative mass spectra showing two peptides with masses matching residues 84 -94 and 148 -153 of ␥S ( Fig. 2A), and a peptide with a mass matching the acetylated N terminus of ␤A1 (residues 1-14) (Fig. 2B) are shown. The presence of the acetylated N-terminal tryptic peptide (Fig.  2B) indicated that the species comigrating with ␥S was actually ␤A1, and not a post-translationally derived form of ␤A3.
A protein corresponding to ␤B3 was not identified during limited Edman sequencing of the major protein species of young human lens (Table I). However, the ESIMS determined masses of the tryptic peptides derived from the region containing ␤A3 included masses corresponding to peptides from both ␤A3 and ␤B3. For example, masses matching the acetylated N-terminal peptide (residues 1-17) and peptide 96 -109 of ␤A3 (Fig. 3A), as well as peptide 110 -122 of ␤A3 and peptide 153-167 of ␤B3 (Fig. 3B) were found.
Densitometric Analysis of Two-dimensional Gels-Once the major crystallin subunits in young human lenses were identified, the relative amount of each subunit was determined by densitometric analysis of two-dimensional electrophoretic gels of the total soluble lens protein from newborn, 3-, 4-, and 7-day old human donors (Table II). The relative amounts of ␤A1 and ␥S, as well as ␤A3 and ␤B3, could not be determined from two-dimensional gels of whole soluble protein, because these proteins did not resolve from one another. However, two-dimensional electrophoresis of isolated ␤ H -and ␤ L -crystallin aggregates from 3-and 7-day-old human lenses removed ␥S and allowed estimation of the relative amount of ␤A1. ␤A1 and ␤A4 were found in nearly equal quantities in the young human lens (gels not shown).
Human ␤A4, ␤B3, ␤A3/A1-crystallin cDNAs-A proposed full-length human ␤A4-crystallin cDNA was obtained by 3Јand 5Ј-RACE PCR because there was no previously published sequence for human ␤A4. Cloning and sequencing of these PCR products resulted in the 809-base pair sequence shown in Fig.  4A (GenBank accession number U59057). An ATG start codon was found at nucleotide 37 and a stop codon at nucleotide 625. A putative polyadenylation signal was also found at nucleo- . The predicted pI was 5.63. Excluding the N-terminal extensions, the amino acid sequence of human ␤A4 shared 92-94% sequence identity with rat 2 and bovine (22) ␤A4, and 65% identity with chicken ␤A4 (23). The N-terminal extensions of human, rat, and chicken ␤A4 were all 10 residues in length, supporting the previous suggestion that the N-terminal extension of bovine ␤A4 is also 10 residues in length (23).
Since only exons 4 -6 of human ␤B3 were previously reported (24), the 5Ј end of human ␤B3 cDNA was amplified by 5Ј-RACE PCR and sequenced (Fig. 4B). The resulting sequence was 379 nucleotides in length, contained a 71-nucleotide untranslated region at its 5Ј end, and coded for amino acids 1-102 of human ␤B3 (GenBank accession number U71216). This sequence overlapped with the first 114 base pairs of the previously reported sequence for human ␤B3 exon 4 (Genbank accession number X15144). Combining this sequence with the previously reported sequences of human ␤B3 exons 4 -6 (Gen-Bank accession numbers X15144, X15145, and X15146) resulted in a deduced sequence for a 211-amino acid protein with a calculated mass of 24,224, including the acetylated N-terminal methionine. The calculated pI was 5.77.
Because the mass of ␤A3 determined by ESIMS (25,192) did not agree with the mass calculated from the published sequence (20), a portion of the human ␤A3 cDNA containing the 3Ј coding region was amplified by PCR and sequenced. The 549-base pair sequence with a stop codon at nucleotide 517 is shown in Fig. 4C (GenBank accession number U59058). This sequence differs from the earlier sequence at two positions within exon four (Genbank accession number M14304), and 4 positions within exon six (Genbank accession number M14306). Each of these nucleotide differences resulted in a change in the deduced amino acid sequence. The calculated masses for the new sequences of ␤A3 and ␤A1 were 25,193 and 23,102, respectively, including the acetylation of their N termini. The pI values of ␤A3 and ␤A1 were 5.58 and 6.17, respectively.
Confirmation of the Deduced ␤A4, ␤B3, and ␤A3/A1 Amino Acid Sequences-The molecular weights of the subunits in lens ␤ H -and ␤ L -crystallin aggregates from a 3-day-old donor were determined by ESIMS (Table III). Two of the components had masses corresponding to ␤B1 and ␤B2. Two other components had masses corresponding to previously identified partially degraded ␤B1-crystallins (3). A spot corresponding to ␤B1 (40-251) was not identified during Edman sequencing of proteins separated by two-dimensional electrophoresis (Fig. 1). This protein, as well as other partially degraded forms of ␤B1 (3), may not have been detected by electrophoresis because they may co-migrate with other crystallins.
Further confirmation of the deduced sequences of human ␤A4, ␤B3, and ␤A3 was obtained from the molecular weights of the peptides in tryptic digests of the proteins. Peptides with masses corresponding to all portions of the sequences of ␤A4, ␤B3, and ␤A3 were found (Fig. 6). Analysis of the tryptic peptides also confirmed that the N termini of ␤A4, ␤B3, ␤A3 (Fig. 6), and ␤A1 (Fig. 2B) were all acetylated, and showed that the N-terminal methionine was removed from ␤A4 and ␤A1, but retained on ␤B3 and ␤A3. The peptides marked with an asterisk in Fig. 6 were additionally analyzed by MS/MS to confirm the proposed sequence. An example of an MS/MS spectrum of a peptide from ␤A4 is shown in Fig. 7. The spectrum shows the fragments formed by collisional activation of ␤A4 peptide 106 -117, which has the sequence, LTIFEQENFLGK. Collisional activation causes the peptide to fragment, primarily along the backbone. Fragments formed with the charge remaining with the C terminus are called the a, b, and c series; fragments formed with the charge remaining with the N terminus are called x, y, and z series (26). Masses of expected fragments due to the b and yЉ series (the Љ indicates an additional H on the fragment) are shown in the diagram at the top. The presence of many peaks in the spectrum with masses corresponding to the expected fragments confirmed that the peptide analyzed has the given sequence.

DISCUSSION
This investigation: 1) identified and quantified the major protein species in the young human lenses following separation by two-dimensional electrophoresis; 2) determined and confirmed the sequences of human ␤A4, ␤B3, and ␤A3/A1 crystallins; and 3) demonstrated the presence of a truncated ␤A3/A1 crystallin. The present report completes the mapping of the major human crystallin species separated by two-dimensional electrophoresis first initiated by De Vries et al. (12) and Datiles et al. (13). Furthermore, the work demonstrates how an analysis of human crystallins combining two-dimensional electrophoresis, Edman sequencing, and mass spectrometry can rapidly identify and characterize the primary structure of crystallins and detect their post-translational modifications. The information in this report will be used as a reference point  a For the acidic ␤-crystallins and ␤B3 the molecular masses were calculated based on the deduced amino acid sequences in Fig. 4. For ␤B1 and ␤B2 the calculated molecular masses were based on previously confirmed sequences (3,11).
for a thorough characterization of modifications occurring during lens maturation, aging, and opacification.
The relative amounts of the total ␣-, ␤-, and ␥-crystallin subunits determined from two-dimensional gels were 27.7, 42.5, and 27.5%, respectively. This corresponds to previously published amounts of ␣-, ␤-, and ␥-crystallin aggregates isolated by gel filtration from lenses of similar age to the newborn lenses used in this study (27,28). The percent contribution of individual ␣and ␥-crystallin subunits to the total lens protein was also consistent with the literature (8, 10). The present Electrospray ionization mass spectrum of N-terminally degraded ␤A3 (23-215) from a 4-day-old lens electroeluted from two-dimensional electrophoretic gels. The mass includes an acrylamide adduct (ϩ71 Da) at each of the five cysteine residues. The mass of ␤A3 (23-215) calculated from the sequence is 22,646; the calculated mass including the acrylamide adducts is 23,001. report for the first time estimates the relative amounts of various ␤-crystallin subunits in human lens. The relative amounts of the various ␤-crystallins apparently differ greatly between species. No ␤A2 subunits were detected in newborn human lens, or in rat lens (16), while bovine (29) and chicken lenses 3 contained significant amounts of ␤A2. ␤B3 is also a major protein of rat lens (30), but its relative amount in human lens is much lower. Changes in the relative concentrations of various crystallins in lens may alter its properties. Therefore, future studies will further characterize the relative amounts of crystallins and how this changes during lens maturation.
While two-dimensional electrophoresis was capable of resolving the majority of human crystallins, several crystallins did not resolve completely from one another. Initially Edman sequencing identified just one of the crystallins because only a single tryptic peptide from a complex mixture was analyzed. In these cases, the presence of co-migrating crystallins was detected by mass spectrometric analysis. The co-migration of human ␤A3 and ␤B3 was unexpected, since these two proteins migrate to different positions during two-dimensional electrophoresis of rat or bovine crystallins (16,29), and have theoretical pI values which differ by 0.2 pH units.
Independent confirmation of the cDNA determined sequences of human ␤A4, ␤B3, and ␤A3/A1 crystallins by mass spectrometric analysis can detect errors in the deduced sequences and characterize post-translational modifications. A major finding of the present paper was the truncation of ␤A3/ A1-crystallin. N-terminal degradation of ␤B1 has previously been reported (3). Even in the lenses from donors of less than 1-week-old, partially degraded forms of both ␤B1 and ␤A3/A1 were present. ␤B1 and ␤A3/A1 may be the most proteolytically labile crystallin subunits in the human lens. The major truncated forms of ␤A3/A1 and ␤B1 result from cleavage between an Asn 22 -Pro 23 of ␤A3 and Asn 15 -Pro 16 of ␤B1. A similar site is also present in ␤B2 at Asn 15 -Pro 16 . However, no cleavage of ␤B2 was seen in the lenses examined in this study. The asparagine-proline sequence in ␤B2 occurs at the interface between the N-terminal extension and the first "Greek key" containing motif (22). 1 H NMR spectroscopy has confirmed that the Nterminal extensions of ␤-crystallins have relatively greater flexibility than the Greek key containing motifs (31). Therefore, the proteolytically resistant Asn 15 -Pro 16 region of ␤B2 may be relatively inaccessible compared to the Asn-Pro regions of ␤B1 and ␤A3 which are cleaved. Of interest is the additional observation that ␤B1 and ␤A3 have longer N-terminal extensions than the proteolytically resistant ␤B2 and ␤A4 (57 and 30 residues versus 15 and 10 residues, respectively). The protease(s) responsible for these cleavages in newborn human lens remain unknown.
In conclusion, two-dimensional electrophoresis combined with Edman sequencing and mass spectrometry successfully identified the major crystallins in the young human lens and confirmed their predicted sequences and masses. This combination of techniques permitted separation and identification of proteins and modifications not otherwise possible. Knowledge of the sequences of all the human ␤-crystallins and their exact molecular weights obtained in this investigation will facilitate identification of the various modified forms which become more numerous with age. The ability to electroelute a modified crystallin species from two-dimensional gels and carry out mass spectrometric measurements on the recovered protein was noteworthy (Fig. 5). This technique should speed progress in identifying the numerous protein modifications occurring in human lenses with age. Post-translational modifications of crystallins may alter the way they interact with one other and affect the highly ordered protein structures required to maintain lens clarity. Determining the modifications of crystallins in human lenses from donors of different ages and stages of cataract will contribute to our understanding of cataractogenesis. Such information is important, because cataract is still a major cause of blindness. When the charge on the peptide stayed with the N terminus, fragments from the b series were seen; when the charge stayed with the C terminus, fragments from the yЉ series were seen (26). Fragments labeled w were from cleavage of the side chains of the amino acid residues. Horizontal arrows labeled x2.00 and x150.00 indicate the magnification of the mass spectral response in these regions.