Proteomic and Sequence Analysis of Chicken Lens Crystallins Reveals Alternate Splicing and Translational Forms of βB2 and βA2 Crystallins

Abstract

purpose. To characterize the adult chicken lens proteome using mass spectrometry and two-dimensional gel electrophoresis (2-DE).

methods. Lens proteins from 10-week old chickens were separated by gel filtration and reversed-phase chromatography, and whole protein masses were measured with electrospray mass spectrometry. Water-soluble lens proteins were separated by 2-DE and identified by tandem mass spectrometry of in-gel digests.

results. Whole protein masses were consistent with all major chicken lens crystallin sequences, except for βB2 and βB3. Subsequent cDNA sequencing revealed errors in published sequences translating into 2- and 7-amino-acid differences, respectively, for βB2 and βB3, which were in better agreement with the measured masses. Previously uncharacterized forms of βA2 and βB2 were observed. The novel form of βA2 had four fewer amino acids, was more abundant, and resulted from translation at a second start codon. The novel form of βB2 contained 14 additional amino acids in the interdomain linker and resulted from alternate splicing within intron 4 of the transcript. All examined crystallins, except βA3, for which data could not be obtained, were N-terminally acetylated, and all β-crystallins lacked an initial methionine, except for the smaller βA2 form. In-gel digests identified 29 proteins on the 2-DE map and indicated that truncation occurs within N-terminal extensions of β-crystallins during lens maturation.

conclusions. The complementary techniques 2-DE, mass spectrometry, and DNA sequencing were used to provide the most complete description of the adult chicken lens proteome to date and identified alternate forms of βA2 and βB2.

The ocular lens in vertebrates is composed of highly concentrated soluble proteins known as crystallins that constitute more than 90% of the total lens protein. The α and βγ gene families are present in all vertebrates, and a variety of taxon-specific crystallins are also known. ¹ The α-crystallin family has two members, αA and αB (each approximately 20 kDa), which are related to the small heat shock proteins. ² They form large aggregates in the lens and exhibit chaperone-like activity. The ubiquitous β-crystallins range in size from 20 to 30 kDa, have conserved Greek key motifs, and form dimers or other small oligomers. Six distinct genes for β-crystallins have been described in vertebrates, and they are often the major protein class. Although abundant in mammalian and fish lenses, the monomeric γ-crystallins are apparently absent in birds and reptiles. ¹  

In several nonmammalian species, other taxon-specific crystallins have been characterized and, surprisingly, many of these proteins are very similar to common enzymes. Not only do these usurped enzymes have to satisfy strict optical requirements to be lens structural proteins, they are also expressed at enormous levels in the lens. The dual role of taxon-specific crystallins is discussed in more detail in a review of lens crystallins by Wistow and Piatigorsky. ¹ Chicken lens, the focus of this study, contains δ-crystallin ^{3 4 5} which is closely related to argininosuccinate lyase. It is, by far, the most abundant crystallin in the young chick lens, with two subunits that have molecular masses of approximately 50 kDa each. 

The chicken is a popular animal system for studying developmental changes in crystallin gene expression and regulation, despite its relatively great evolutionary distance from humans. Because γ-crystallins are absent in the chicken lens and δ-crystallin is not present in mammalian lenses, the focus of most research has been on the gene structure, expression, and regulation of α-crystallins ^{6 7 8 9} and β-crystallins. ^{9 10 11 12 13 14 15 16} There are seven different β-crystallins known in the chicken, similar to other species. ¹ Early one-dimensional polyacrylamide gel electrophoresis studies ^{17 18} of chicken lens proteins identified the major crystallin classes, and subsequent two-dimensional (2-D)-gel studies ^{19 20} found a considerably larger number of resolved β-crystallin proteins than known genes, most of which were not identified. 

In this study we used in-gel digests and tandem mass spectrometry to identify the most abundant crystallin isoforms present in 2-D polyacrylamide gel electrophoresis (2-DE) maps of water-soluble lens proteins from 10-week-old chickens. This age was chosen because the abundance of δ-crystallins had decreased to levels similar to those in major β-crystallins, ²⁰ reducing gel saturation effects, and all major β-crystallins are well represented. An alternative form of βA2 was identified and a novel “long-linker” form of βB2 (generally one of the more highly conserved β-crystallins) was observed. Whole-protein molecular weights (MW) of the major crystallin components were also measured by mass spectrometry and compared with weights calculated from β-crystallin sequences. In cases in which there appeared to be MW discrepancies due to possible sequence variations, new cDNA sequencing was performed. Additional evidence for truncation and deamidation of crystallins during lens maturation was also found. 

Methods

Measurement of Chicken Crystallin Masses

All experiments using chickens were approved by the University of Delaware and Oregon Health and Science University (OHSU) institutional review boards and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Lenses from 10-week old white leghorn chickens (SkyLane Farms, Hubbard, OR) were homogenized without protease inhibitors, and the water-soluble fraction was isolated by centrifugation and the protein content assayed as previously described. ²¹ Crystallin oligomers and monomers were separated by gel filtration on a 2.5 × 95-cm Sephacryl S-300 HR column (Amersham Biosciences, Piscataway, NJ) at a flow rate of 25 mL/hour, using a 20 mM Tris (pH 7.5) and 100 mM NaCl mobile phase. α-Crystallin aggregates and six fractions containing decreasing sizes of β-crystallin aggregates and δ-crystallin were isolated. Masses of individual crystallin subunits were determined by separation of 5 to 20 μg of protein from the gel filtration–purified fractions on a reversed-phase C4 column with online electrospray ionization mass spectrometry, as previously described, ²² except a 19-μL/min flow rate and 0.2% acetic acid were used in a mobile phase containing no trifluoroacetic acid (TFA). Protein fractions eluting from the C4 column were not collected and no additional protein identification data other than the whole mass measurements were obtained. The mass spectra of proteins were deconvoluted on computer (Biomass software, Bioworks 3.1; Thermo Finnigan, San Jose, CA), calibrated with horse myoglobin and could be determined to approximately 0.01% accuracy on an ion trap mass spectrometer (LCQ classic; Thermo Finnigan). Masses of amino acid sequences were calculated on computer (PAWS: freeware edition for Windows 95/98/NT/2000, ProteoMetrics, LLC, New York, NY; available at http://65.219.84.5/PAWSDL.html). 

2-D Gel Electrophoresis

Separation of 400 μg of 10-week-old chicken water-soluble lens protein was performed using 2-DE. The first-dimension isoelectric focusing (IEF) was performed using 18-cm nonlinear pH 3 to 10 immobilized pH gradient (IPG) strips (Immobiline DryStrip; Amersham Biosciences), with a programmed voltage gradient on a commercial system (Multiphor II; Amersham Biosciences), as previously described. ²³ IPG strips were rehydrated in the first-dimension buffer (8 M urea, 2% 3-([3-cholamidopropyl]dimethylammonio-2-hydroxy-1-propanesulfonate [CHAPS], 2% 3-10 NL IPG buffer, 50 mM dithiothreitol [DTT], and a trace amount of bromophenol blue), which also contained the sample (400 μg). Second-dimension separation was performed on a 24 × 18.5-cm 12% SDS polyacrylamide gels with Coomassie G-250 staining, as previously described. ²³ Gel images were acquired with a 12-bit gray-scale commercial optical scanner (Expression 1600; Epson America, Inc., Long Beach, CA), and image analysis was performed with the 2-DE gel analysis software (Melanie 3.09g; Swiss Institute of Bioinformatics, Geneva, Switzerland). The pH scale of the 2-DE gel was calibrated assuming that the most abundant spots of βB1, βB3, βA3, βA2 (long form), βB2 (short form), and βB2 (long form) had isoelectric points (pI) of 4.7, 5.1, 5.6, 6.0, 6.5, and 8.4, respectively, calculated using the Compute pI/MW tool (http:/us.ExPASy.org/tools/pi_tool.html). The amino acid sequences used were from this work and published sequences. N-terminal acetylation was approximated by removing a lysine residue from the amino acid sequence. The approximate MW of proteins was extrapolated from a reference lane of low MW standards (Bio-Rad, Hercules, CA). 

Protein and Peptide Identifications

Proteins resolved on Coomassie-stained 2-DE gels were excised and digested within gel pieces with trypsin, as previously described, ²⁴ except the destaining step used for silver-stained gels was replaced by two washes in 50 mM ammonium bicarbonate, 50% acetonitrile before drying. Dried in-gel digests were dissolved in 10 μL 5% formic acid and analyzed by liquid chromatography electrospray ionization tandem mass spectrometry (LC-MS) using a ion trap mass spectrometer (LCQ classic; Thermo Finnigan). Analysis by LC-MS was performed using a 250 × 0.5-mm stable bond C18 column (Zorbax; Agilent Technologies, Palo Alto, CA). Samples were applied to the column through a microtrapping cartridge (Michrom Bioresources Inc, Auburn, CA) and the peptides separated using a 10-μL/min flow rate, mobile phase containing 0.2% acetic acid, 0.005% heptafluorobutyric acid, and 60 minutes of 0% to 30% acetonitrile gradient. Data-dependent tandem mass spectra on major peptide ions were automatically collected by using a dynamic exclusion feature to extend analysis to less-abundant peptides. 

Proteins present in each digested spot were then identified (Sequest ²⁵ software; Thermo Finnigan) to correlate experimental MS/MS spectra with theoretical MS/MS spectra calculated from peptide sequences in a chicken subset of the Swiss-Prot protein database (Swiss-Prot, Release 42.0, October 2003; Swiss Institute of Bioinformatics, Geneva, Switzerland). SEQUEST searches were repeated with the database modified to contain the new sequences for βB2 and βB3 produced in this study and compared with the previous search results. Peptides identified by SEQUEST analysis were considered correct if the cross-correlation scores (Xcorr) were greater than 1.8, 2.5, and 3.5 for 1+, 2+, and 3+ ions, respectively. Two different unique peptides matched to the same protein were required for positive protein identification and sequence coverage was usually greater than 50%. N-terminal acetylation was specified as a differential (variable) modification in the search and the MS/MS spectra of any modified peptides were manually validated. The criteria for validation were as follows: Xcorr greater than the values just listed, parent ion mass and calculated mass in close agreement, fragment ions well above noise, continuous y- or b-ion series longer than 50% of the peptide amino acid sequence, the most intense fragment ions had to be assigned, and the occurrence of enhanced cleavage N-terminal to proline residues. 

Sequencing of βB2 cDNA, βB2 Intron IV, and βB3 cDNA

A βB2-crystallin cDNA clone (number pgp2n.pk007.k9) from a chicken pituitary expressed sequence tag (EST) library (Chick EST Project, University of Delaware, Newark, DE) was obtained and sequenced by using the T7 and SP6 primers in the cloning plasmid pCMVSPORT6 (Invitrogen, Carlsbad, CA). A partial chicken βB3-crystallin cDNA previously cloned into a vector (pBlueScript; Stratagene, La Jolla, CA) and two chicken βB3-crystallin 5′ rapid amplification of cDNA ends (RACE) previously cloned in pPCRScript ¹³ (Stratagene) were sequenced using the T3 and T7 primers within each plasmid. 

The sequence of the primers designed to amplify across intron 4 of the chicken βB2-crystallin gene were based on the published cDNA sequence ⁹ (GenBank S52930; http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) and the predicted location of the intron splice site. ¹³ The forward primer was 5′-ACA GCA TCA CCT CCC TGA GA, and the reverse primer was 5′-GGC ACG ATT TTG TGC TCC TGT. These primers were used in a PCR amplification of chicken genomic DNA (BD Biosciences-Clontech, Palo Alto, CA), using PCR Master Mix (Qiagen, Valencia, CA). The resultant product was isolated and directly sequenced with the same primers. All DNA sequencing was performed at the DNA Sequencing Core Facility at the Charles C. Allen Jr. Biotechnology Laboratory, University of Delaware. 

Results

Masses of Chicken Lens Crystallins and Additional Sequences for Chicken βB2 and βB3

Chicken crystallins were separated by gel filtration and reversed-phase chromatography, and whole protein masses were measured for comparison with masses calculated from the reported amino acid sequences and to detect the presence or absence of posttranslational processing. Agreement between a measured protein mass and the calculated mass of the protein sequence does not prove that the sequence is accurate; however, disagreement often indicates that there is a sequence discrepancy. Masses corresponding to the predicted masses of αA, αB, phosphorylated αB, βA1, βA2, βB1, and δ1 crystallin were observed and are listed in Table 1 . Due to its low abundance, a species corresponding to βA3 could not be detected. The 9-mass-unit difference for δ1 is outside of the expected ±5-Da uncertainty and could indicate a possible sequence variation, but no attempt to resequence δ1 was made. The observed mass of βB3 differed significantly from the mass calculated from the previously published sequence. ¹³ The negative 3-mass-unit difference for βB2 is close to the ±2-Da uncertainty, but it is the only negative difference (many of the mass differences are between +1 and +2 Da), and also suggests a possible sequence discrepancy. Sequencing of additional βB2 and βB3 cDNAs and βB2 genomic DNA was performed to reconcile these discrepancies. 

Sequence of the 5′ end from a newly isolated chicken βB2 cDNA is shown in Figure 1A . This region differed from that of an earlier βB2 cDNA sequence (GenBank S52930) at two nucleotides, resulting in a change of P16 and N17 to S16 and K17. The calculated mass of acetylated βB2, missing its N-terminal methionine, and containing these two amino acid substitutions was 24,975. Unlike the mass calculated from the earlier sequence, this mass fell within the 0.01% mass error of the experimentally determined βB2 mass (Table 1) . This corrected sequence has been submitted (GenBank AY539830). A previously reported chicken βB3 cDNA was resequenced, and corrections from the previously reported sequence are shown in Figure 1B . The new sequence differed from the previous sequence ¹³ at eight nucleotides, resulting in 7 amino acid changes and a calculated mass of 24,361, assuming methionine removal and N-terminal acetylation. This calculated mass was within the experimental error of the measured mass of βB3 (Table 1) . The previous GenBank submission (U28146) was corrected to reflect these new data. 

During reversed-phase separation and mass measurements of chicken βB2 and βA2, an additional MW form of each protein was observed (Fig. 2) . The first reported sequence of βB2 contained 218 residues. ⁹ A later study reported a 204 residue βB2 sequence, which suggested the first cDNA sequenced was an alternate form, a product of incomplete splicing. ¹³ The shorter form of βB2 was otherwise identical with the earlier 218-residue protein, except it was missing 14 residues from V102 to K115. After substitution of the two different amino acids reported above, the 204-residue form of βB2 (new GenBank entry AY539829) had a calculated mass of 23,345 assuming N-terminal acetylation and methionine removal. This is in excellent agreement with the experimentally determined mass of 23,344 (Fig. 2A) . In addition, the smaller form of chicken βA2 with a mass of 22,523 (Fig. 2B) agreed with the calculated mass of βA2 (22,526) if the second in-frame ATG within the reported βA2 c-DNA (GenBank U28145) was used as an alternate start codon and the N terminus of the protein was acetylated. 

2-DE and MS/MS Analysis of Individual Spot Digests

Water-soluble lens proteins from 10-week-old chickens were resolved using 2-DE, as shown in Figure 3A . The gel had roughly 70 spots within the given pH and MW ranges. The most abundant 29 spots were excised, and in-gel trypsin digested and the peptides identified using MS/MS. Individual protein spots that were identified are labeled on an identical gel image in Figure 3B . When more than one form of the same protein was present, lowercase letters were used to designate different forms in order of decreasing MW, and by increasing pI for spots at similar weights. Assignment of the unmodified form of each crystallin was based on similarity to crystallins identified on 2-DE gels from 14-day embryonic lenses (data not shown). 

βA3/A1 Crystallin

The gene encoding these proteins in chicken has been studied, ^{10 11} and it was found that both proteins are translated from a single transcript. The alternate start codon for βA1 results in a protein that is 17 amino acids shorter than βA3, in both chickens and mammals. ^{21 22 23 26 27 28 29} MS/MS results unambiguously assigned βA3/A1 protein to spots βA3, βA3/A1(a), βA3/A1(b), βA3/A1(c) and βA1 (Fig. 3B) . Digests from the spot designated βA1 contained the N-acetylated peptide AQTNPLPVPMGPWK from residues 1-14 of βA1 and identified this protein as the unmodified form of βA1 (MS/MS results not shown). This confirmed the result of the whole mass measurement of βA1, indicating that the N-terminal methionine is removed and that N-terminal acetylation occurs. Although a peptide from the unique 17 amino acids of the N terminus of βA3 was not recovered from the spot designated βA3, this protein was assigned to unmodified βA3 based on its apparent mass and pI. It remains unknown whether βA3 retains it N-terminal methionine or is acetylated, because its whole mass could not be measured. The proteins labeled βA3/A1(a–c) were probably modified forms derived from either βA3 and/or βA1. The two spots βA3/A1(a) and βA3/A1(b), to the acidic side of βA1, could be due to deamidations in βA1 which shift the pI to more acidic values. The spot labeled βA3/A1(c) was blotted from a 2-DE gel and subjected to Edman sequencing. The N-terminal sequence PLPVPMGP confirmed that this protein resulted from removal of either 22 amino acids from βA3, assuming the protein retains its N-terminal methionine, or four amino acids from βA1. 

βA2 Crystallin

Two prominent spots denoted βA2(a) and βA2(b) in Figure 3B were identified as βA2 protein. MS/MS results from the higher MW βA2(a) spot identified the N-terminally acetylated peptide TSEAMDTLGQYK corresponding to residues 1-12 of βA2. Similarly, MS/MS results from the lower MW βA2(b) spot identified the N-terminally acetylated peptide MDTLGQYK corresponding to residues 5-12 of βA2 (Fig. 4) . Because the N-terminal peptide from βA2(b) was acetylated, this lower MW form of βA2 was probably the result of alternate start codon usage during translation, and not proteolytic processing. In support of this interpretation, the lower MW βA2(b) was also observed in greater abundance than the higher MW βA2(a) form in lenses of chicken embryos (data not shown). These results confirm the tentative identification of the lower MW form of βA2, observed during whole mass measurements in Figure 2 , as an alternate translation product from the second in-frame start codon of chicken βA2 mRNA (GenBank U28145). 

βB1 Crystallin

The β-crystallin naming convention was adopted from bovine crystallins, where βB1 is the β-crystallin with the most basic pI. In contrast, βB1 is the most acidic crystallin in the chicken lens (Fig. 3B) . Two truncated forms of βB1, spots βB1(b) and βB1(c), similar in MW to βB1(a), and a third more significantly truncated form, spot βB1(d), were observed. MS/MS analysis from digests of βB1(b) and βB1(c) indicated that the C terminus was intact, but identified no peptides from the N terminus of the truncated proteins. Therefore, proteins from βB1(b) and βB1(c) were blotted to PVDF membranes, and the combined protein spots were subjected to Edman degradation, yielding the sequence KTAAPGQAAE, corresponding to residues 5-14 of chicken βB1. βB1(b) may result from deamidation of βB1(c). MS/MS analysis of βB1(d) digest did not identify either N- or C-terminal cleavage sites, and Edman sequencing was not attempted. 

βB2 Crystallin

Two prominent spots in Figure 3B (βB2(a) and βB2(d)) and four other lower-abundance spots (βB2(b), βB2(c), βB2(e), and βB2(f)) were identified as βB2. The N-terminal acetylated tryptic peptide ASEHQMPASK was observed in the MS/MS data from both prominent spots (data not shown). Consequently, alternate start codons, such as observed in βA3/A1 and βA2, are not responsible for the two major forms of βB2. Therefore, the higher MW βB2(a) was hypothesized to be the alternately spiced form of βB2 containing 14 extra amino acids within the connecting peptide consistent with the whole protein masses observed in Figure 2A . This hypothesis was confirmed by MS/MS data of digests from βB2(a) and βB2(d). In spot βB2(a), the unique peptide QPLPTR contained within the extra 14 amino acid region was observed and the fragmentation spectra is shown in Figure 5 . Prominent loss of ammonia and water from Q107, and loss of water from T111 ³⁰ increase the confidence of the peptide assignment. In contrast, the MS/MS data from spot βB2(d) contained the peptide SDSITSLRPIKVDSQEHK that could only result from the lower MW form of βB2 without the additional 14 amino acids in the connecting peptide. βB2(a) containing the extra 14 residues is the most basic crystallin in the chicken lens, because of the presence of four additional basic residues not found in βB2(d). The predicted shift in pI from 6.5 to 8.4 is in excellent agreement with the 2-DE migration positions observed in Figure 3 . 

The MS/MS data from the digests of the remaining lower-abundance βB2(b), βB2(c), βB2(e), and βB2(f) did not identify any peptides to allow assignments of modifications, including either N- or C-terminal truncation sites within these proteins. However, the most abundant truncation product βB2(f) was blotted, subjected to Edman sequencing, and yielded the sequence SKQQPAS. This indicated that βB2(f) resulted from the loss of 8 amino acids from the N terminus of βB2. While not confirmed, βB2(b) likely resulted from a similar N-terminal truncation of the larger, alternately spliced βB2(a). 

To more closely examine the cause of the alternate splicing of βB2 in chicken lens, a region of chicken βB2 genomic DNA (and the translated amino acids) around the intron 4 region was determined (Fig. 6) . The consensus GT sequence ³¹ at the 5′ splice site and the alternate 5′ splice site are highlighted. Use of this alternate downstream splice site would result in the addition of the 14 extra amino acids in the translated protein, as observed earlier. 

βB3 Crystallin

Chicken βB3 had considerable heterogeneity by 10 weeks of age. There were two major apparently full-length forms marked βB3(a) and βB3(b) and three truncated forms marked βB3(c-e) (Fig. 3B) . MS/MS analysis of the peptides from the major full-length βB3(b) spot confirmed the results of the whole mass measurement of βB3 shown in Table 1 , indicating that the N-terminal methionine is removed from chicken βB3, and that the protein is acetylated. The ensemble of MS/MS data from all five βB3 spots confirmed the 7 amino acid changes reported in Figure 1B . 

Attempts to identify the likely truncation site responsible for spots βB3(c–e) from MS/MS data and Edman sequencing were inconclusive. However, the truncation sites are probably within the N-terminal extension, since the pIs of the truncation products were progressively increased, and there are three glutamic acids between residues 8 and 18. Phosphorylation of βB3 both in vitro and in vivo has been reported, ³² but no evidence of phosphorylation from the whole protein mass measurements or in any MS/MS data from 2-DE gels was observed. 

α Crystallins

We assigned two spots in Figure 3B to αA [αA(a) and αA(b)], and three spots to αB [αB(a-c)]. Retention of the N-terminal methionine and N-terminal acetylation was confirmed by MS/MS analysis of spots αA(a), αA(b), and αB(c). A whole mass matching a phosphorylated form of αB was observed (Table 1) but the acidic αB(a) and αB(b) spots were rather faint and we could not assign the site of phosphorylation from MS/MS data. No phosphorylated αA peptides were observed in MS/MS analysis of spot αA(a). This is consistent with the whole mass measurement (Table 1) where no phosphorylated form of αA was observed. Spot αA(a) is more likely due to deamidation, which would not be observed in this study due to the single mass unit increase occurring with deamidation. The lack of αA phosphorylation observed in chicken lens was in contrast to mammals and consistent with earlier results. ³³ Previous studies found that the more acidic form of αA in chicken lens was due to deamidation at N149 ³⁴ and that the more acidic form of chicken αA increases in abundance with age. ⁷  

Taxon-Specific Crystallins

δ1-Crystallin (similar to arginosuccinate lyase) and τ-crystallin (α enolase) were both observed in 10-week-old chicken lenses (Fig. 3B) . MS/MS analysis of digests from the spot labeled δ1 found that the N-terminal methionine is removed from chicken δ1 and that the protein is acetylated, confirming the results shown in Table 1 . There are two genes encoding δ-crystallins in chicken lens called δ1 ^{3 5} and δ2 ⁴ ; however, δ1 is more highly expressed in chicken lenses. ¹ The results of the 2-DE mapping of spots from 10-week-old chicken confirm this finding, because no δ2-specific peptides were observed in the gel digests. Although the presence of τ-crystallin was confirmed by MS/MS analysis, information regarding the presence or absence of the N-terminal methionine or acetyl group was not obtained. 

Discussion

A previous 2-DE study of chicken crystallins ²⁰ with a pH range from 3 to 9.5 and molecular mass range of approximately 15 to 90 kDa had migration patterns very similar to Figure 3 . A total of 24 spots were attributed to β-crystallins in that study but positive spot identifications were not performed. This study used complementary techniques of 2-DE, mass spectrometric analysis of proteins and peptides, and DNA sequencing to provide the most complete description of the adult chicken lens proteome to date. We observed approximately 40 2-DE spots that were probably β-crystallins, 22 of which were positively identified by MS/MS. The utility of Edman sequencing to identify N-terminal truncations and the importance of accurate sequence information in protein databases was also demonstrated. These results allowed discovery of both a new form of βA2 produced from an alternate start codon and a βB2 arising from alternately spliced mRNA. 

Leaky ribosomal scanning can lead to alternate forms of a protein from a single mRNA using alternate start codons. ^{35 36} Although this is very rare in eukaryotic genes, the chicken has two such genes, namely, βA3/A1 and βA2. Alternate translation of the βA3/A1 gene leading to two forms of the protein occurs in other vertebrates, but this is the first time that a βA2 gene with an alternate start codon has been reported. In mammals, βA1 crystallin is more abundant than βA3. This could be the result of two factors: a “weak” Kozak consensus sequence for the first start codon and an extremely short (5–12 bp) 5′ untranslated region. ^{37 38} “Strong” Kozak consensus sequences consist of an A or G 3 bp upstream of the start codon and a preferred G immediately after the start codon. The 5′ cDNA sequences for βA3/A1 for mouse, human, bovine, and chicken are compared in Figure 7A , where the two start codons and flanking Kozak sequences are highlighted. A strong Kozak consensus sequence is present in all four species for the second start codon, whereas only chicken βA3/A1 cDNA has a strong Kozak consensus sequence for the first ATG. Apparently, the short 5′ untranslated region (5 bp) negates any benefit of the Kozak consensus sequence, because we still observed approximately 10 times more βA1 than βA3 in chicken lens. Similar observations led Werten et al. ³⁹ to conclude that the short 5′ untranslated region is the dominant factor leading to a greater abundance of βA1. 

Figure 7B shows the 5′ region of βA2 cDNA sequences for mouse, human, bovine, and chicken where start codons are again in bold and Kozak consensus sequences underscored. The second start codon is only present in chicken βA2, explaining the absence of the shorter form of βA2 in these other species. Furthermore, unlike the first ATG in the chicken βA2 cDNA, the second ATG is flanked by a strong Kozak consensus sequence. The weak context for the first start codon probably increases the probability that ribosomal scanning continues until the strong Kozak consensus sequence is encountered at the second start codon. This could explain why the lower-MW βA2(b) spot in Figure 3 is approximately 1.3 times more intense than the higher-MW βA2(a) spot. However, unlike βA3/A1 cDNA, the greater length of the 5′ untranslated region in the chicken βA2 cDNA leads to more equal translation from the first and second start codons. Although the function of this alternately translated form of chicken βA2 is unknown, we hypothesize that similar forms of βA2 may exist in other avian species. 

The detection of alternately spliced forms of βB2 in chicken was an unexpected result and to date has not been observed in any other vertebrate lens. Although, cDNAs for both forms of βB2 have been previously reported, ^{9 13} the longer form was postulated to be poorly represented in the pool of βB2 transcripts. However, based on quantification from 2-DE gels in the present study, the larger form of βB2 comprises approximately 20% of the total βB2 in chicken lens. The extra 14 amino acids in this alternate form of βB2 are in the linker region that connects the N- and C-terminal domains of the protein. The presence of four basic residues in these additional amino acids results in a crystallin subunit with the highest pI in chicken lens. Mutations in residues in the linker region of human βB1 ⁴⁰ were found to have significant effects on the associative properties of βB1 dimers. The presence of an elongated highly basic extension would be expected to also significantly alter the associative properties of βB2. ⁴¹ Future studies should examine what, if any, effect the extra 14 amino acids have on the structure of βB2. Perhaps the unusual properties of crystallins in chicken lens partially accounts for the low percentage of solidlike protein determined by ¹³C nuclear magnetic resonance (NMR) in this species, ⁴² and the excellent accommodative ability of birds. ⁴³  

The anomalous migration pattern of βB1, where the apparent molecular mass is approximately 35 kDa versus an expected mass of 27 kDa, is similar to patterns observed in other avian species. ⁴³ In that study, βB1 from several avian species reacted strongly in a glycan detection procedure and glycosylation was proposed as an explanation for the slow migration. ⁴³ However, our measurement of the mass of chicken βB1 in the present study suggests that chicken βB1 is nonglycosylated. This is consistent with the lack of a signal peptide or N- or O-linked glycosylation sites within the βB1 protein. We speculate that the slow gel migration of βB1 is due to its N-terminal extension or some other intrinsic feature of βB1 and not due to glycosylation. 

Major truncation products of βB1, βB2, βB3, and βA3/A1 were observed in chicken lens. The major truncation products of βB1, βB2, and βA3/A1 resulted in the removal of 5, 8, and 22 or 4 residues from the N-terminal extension of the proteins, respectively. The N-terminal extensions of mammalian β-crystallins are also shortened during maturation and aging of the lens. ^{44 45 46 47} We hypothesize that the N-terminal extensions of β-crystallins function to assist the assembly of ordered protein complexes in the lens cytosol necessary for transparency. After proper assembly, the N-terminal extensions may no longer be necessary, and their removal may assist in the dehydration of lens fibers occurring with further maturation. Because unscheduled removal of these extensions has been associated with cataract formation in rodents, ⁴⁵ determining which proteases cause N-terminal extension removal from β-crystallins is of great interest. Although α- and β-crystallins in rodent lenses are predominately cleaved by calpains, ^{45 48} the present results suggest that a yet unknown protease is responsible for crystallin degradation in chicken lenses. Chicken α-crystallins are excellent substrates for purified m-calpain in vitro, ⁴⁹ yet α-crystallins remained intact in 10-week-old chicken lenses. In addition, although chicken β-crystallins were partially degraded by purified m-calpain in vitro, the cleavage sites in the N-terminal extensions determined by Edman sequencing (David L, unpublished results, 1996) were at different positions than the cleavage sites observed in vivo in the present study. These results are similar to human lenses, where α-crystallins also remain largely intact, ⁵⁰ and the cleavage sites caused by m-calpain in vitro ⁵¹ are unlike cleavage sites observed in vivo. ^{21 52}  

Thus, although calpains may not be responsible for crystallin truncation in all species, a specific β-crystallin truncation product not produced by m-calpain was generally observed. This was the truncation product of βA3/A1 missing either 22 or 4 residues from its N terminus. Besides chicken, this truncation product was also observed in human and bovine lenses. ^{21 46 47} In all three species, cleavage of βA3/A1 occurred at the C-terminal side of asparagine in the conserved consensus sequence NPXP. Future studies should examine which proteolytic activity is responsible for this cleavage, because this enzyme is probably active in the lenses of all vertebrates and may be responsible for a major fraction of β-crystallin truncation in nonrodent species. 

In conclusion, these studies provide a framework for future studies of chicken lenses using modern tools of proteomics to characterize developmental, maturational, and pathologic alterations in chicken crystallins. Chickens will continue to be an important species in improving understanding of the biology of the human lens. 

 Supported by National Eye Institute Grants EY007755 and EY10572 (LLD) and EY12221 (MKD). JRT was a Fight for Sight Summer Fellow.

 Submitted for publication February 9, 2004; revised March 22, 2004; accepted March 31, 2004.

 Disclosure: P.A. Wilmarth, None; J.R. Taube, None; M.A. Riviere, None; M.K. Duncan, None; L.L. David, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Comparison between Expected and Measured Whole Protein Masses of 10-Week Chicken Lens Crystallins Measured by LC-MS

Protein	SwissProt Accession Number	Calculated Mass (Da)*	Measured Mass (Da), †	Mass Difference (Da)	New Calculated Mass (Da), ‡
αA	P02504	19829	19829	0
αB	Q05713	20076	20074	2
αB(+PO4)	Q05713	20156	20155	1
βA1	P10042	22934, §	22931	3
βA2	P55164	22915, §	22913	2
βA2, ∥		22526	22523	3
βB1	P07530	27178, §	27176	2
βB2, ¶	Q05714	24971, §	24974	−3	24975, §
βB2, ∥		23341, §	23344	−3	23345, §
βB3	P55165	24421, §	24360	61	24361, §
δl	P02521	50707, §	50698	9

*  Average masses of N-acetylated proteins calculated using the listed SwissProt sequences. Acetylation was confirmed by MS/MS analysis of in-gel digests for all listed crystallins except βB1.

†  Masses determined by ESI-MS with an uncertainty of ±0.01%.

‡  Masses calculated with new βB2 and βB3 amino acid sequences from this work.

§  Masses calculated without initial methionine residue. Presence or absence of methionine was confirmed for all listed crystallins, except βB1, by MS/MS analysis of in-gel digests.

∥  Sequences for the additional forms of βA2 (alternate downstream start codon) and βB2 (more common form missing the extra 14 amino acids) were derived from SwissProt sequences P55164 and Q05714, respectively.

¶  Alternate splice variant of βB2 protein having 14 additional amino acids.

Figure 1.

 

cDNA and deduced amino acid sequences for (A) the 5′ region of chicken βB2 crystallin (GenBank AY539829 and AY539830) and (B) the complete coding region of chicken βB3 crystallin (GenBank U28146). The amino acid changes (bold) were the result of single nucleotide substitutions, except for the four consecutive amino acid changes in βB3 that were caused by a frameshift due to insertion of 2 bases, followed 10 bases later by a 2-base deletion.

Figure 2.

 

Deconvoluted spectra showing the masses of (A) chicken βB2 and (B) βA2 crystallin. Proteins from a high-molecular-weight β-crystallin gel filtration fraction were separated using reversed-phase chromatography, and masses were directly measured using ESI-MS. The βB2 fraction, eluting at 26.7 minutes, contained two proteins. The 24,974-mass species agreed with the expected mass of βB2 containing the 14 additional residues in its connecting peptide, whereas the 23,344 species agreed with the mass of the βB2 without the elongated connecting peptide. The βA2 fraction, eluting at 30 minutes, also contained two species. The mass of 22,913 was in agreement with the form of βA2 translated from the first start codon, whereas the 22,523 mass form corresponded to βA2 translated from the second start codon.

Figure 3.

 

2-DE separation of water-soluble proteins from whole lens of 10-week-old chickens. (A) Coomassie-stained image from 400 μg of protein. (B) Identical image to (A) with the proteins that were excised and analyzed by mass spectrometry circled and labeled by crystallin subclass. Left: MW markers. The pH calibration was derived from several β-crystallins spanning the range of interest.

Figure 4.

 

MS/MS spectrum of the N-terminal tryptic peptide (MDTLGQYK) of the smaller form of βA2 translated from the second start codon showing the presence of a +42-mass increase at the N terminus due to acetylation. The y- and b-series fragment ion peaks are labeled and superscripts “o” and “*” indicate neutral loss of water (−18) and ammonia (−17), respectively. Masses for b-series ions in parenthesis are the masses without acetylation. The singly charged parent ion m/z was 997.5 and matched the predicted acetylated peptide mass. The MDTLGQYK peptide was the top-scoring result from a Sequest search with Xcorr = 2.37 and deltaCN = 0.11, using a chicken-only subset of the SwissProt database.

Figure 5.

 

MS/MS spectrum of the tryptic peptide (QPLPTR) derived from the extra 14-amino-acid sequence contained in the connecting peptide of the alternatively spliced form of βB2. The singly charged parent ion had an m/z of 711.6. The y- and b-series fragment ion peaks are labeled, and superscripts of “o” and “*” indicate neutral loss of water (−18) and ammonia (−17), respectively. The prominent loss of water and ammonia indicates the presence of glutamine and threonine residues and supported the identification of this peptide. The QPLPTR peptide was the top-ranked Sequest search result with Xcorr = 1.34 and deltaCN = 0.30, using a chicken-only subset SwissProt database.

Figure 6.

 

Partial genomic DNA sequence of the chicken βB2 gene illustrating the alternative 5′ splice site for intron 4 (GenBank AY539831). Regions corresponding to exon 4, exon 5, and the intervening intron 4 are labeled above the DNA sequence, whereas the translated amino acid sequence is shown below. Horizontal bar: intron borders including the additional one at the alternate 5′ splice site. Bold sequences: consensus nucleotides typically present at the beginning and ends of intron sequences. Within the genomic sequence, the nonitalic capital letters represent the portions of chicken βB2 exons 4 and 5 present in both cDNAs. Italic capital letters in the genomic sequence indicate nucleotides present only in the cDNA sequence (GenBank S52930) that results in a longer exon 4 and 14 additional amino acids in the connecting peptide of βB2 (also italic). The four additional basic amino acids unique to the longer form of βB2 are underscored.

Figure 7.

 

A comparison of the 5′ cDNA sequences of (A) βA3/A1 in mouse (GenBank NM_009965), human (NM_005208.3), bovine (NM_174523.2), and chicken (M15658) and of (B) βA2 in mouse (GenBank AY160973), human (NM_057094), bovine (NM_174524), and chicken (U28145). The complete 5′ untranslated region is shown for all βA3/A1 sequences. The numbers in parenthesis preceding the sequences for βA2 indicate the length of the 5′ untranslated region not shown. Start codons are in bold, and the corresponding amino acid sequences are listed below each cDNA sequence. The guanine or adenine bases that represent strong Kozak consensus sequences flanking the start codons are underscored.

The authors thank Marjorie Shih for technical assistance with the Edman sequencing, Thomas R. Shearer for providing the 10-week-old chicken lenses, Kirsten Lampi for assistance in early 2-DE experiments, and the University of Delaware’s Chick EST Project for providing expressed sequence tag clones.