Phosphorylation and Glycosylation of Bovine Lens MP20

Abstract

purpose. Membrane protein 20 (MP20) is the second most abundant integral membrane protein in the lens, yet little is known about its function and post-translational modifications. The purpose of this work was the determination of the primary protein structure of MP20 and the types and sites of in vivo modifications.

methods. Bovine MP20 was isolated by anion exchange chromatography or SDS-PAGE followed by digestion with cyanogen bromide (CNBr) or trypsin. The total membrane protein fraction was also digested with trypsin in solution. The CNBr and trypsin peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) and matrix-assisted laser desorption ionization (MALDI) mass spectrometry.

results. Using these approaches to integral membrane protein analysis, complete sequence coverage of MP20 was obtained. The in vivo sites of phosphorylation were identified as Ser-170 and Thr-171, the latter being previously unreported. The most abundant form of MP20 is monophosphorylated at Ser-170, whereas unphosphorylated and diphosphorylated forms were in lower abundance. In addition, two sites of a rare type of glycosylation (C-mannosylation) were identified at tryptophan residues 43 and 61.

conclusions. The functional significance of phosphorylated MP20, the predominant form, remains unknown. Glycosylation of tryptophan residues represents a new lens protein modification that can explain galectin-3 interaction and suggests a topology for MP20 in which these peptides are located in an extracellular domain.

The structural integrity and optical clarity of the lens have been attributed in part to the arrangement of the fiber cells. This arrangement is by specialized interdigitations or ball-and-socket motifs between adjacent cells. Concentrated in these ball-and-socket structures is the second most abundant lens membrane protein (MP), the highly conserved MP20 (also referred to as MP18, MP19, MP21, and LIM2). ^{1 2} Although the function of MP20 remains unknown, it has been shown to relocalize from the cytoplasm to the membrane with fiber cell differentiation ³ and associate with galectin-3, a carbohydrate-binding protein, perhaps providing a basis for the formation of cell-to-matrix or cell-to-cell junctions. ⁴ Clearly, MP20 plays a critical role in lens development, as evidenced by the formation of congenital cataract and microphthalmia in mice with mutations in the gene for MP20. ^{5 6} This conclusion is further substantiated in a recent report of an MP20 knockout mouse, indicating fiber cell disruption and loss of lens optical properties (Shiels A, et al. IOVS 2004;45:ARVO E-Abstract 4595). 

MP20 is a member of the peripheral myelin protein (PMP22)/epithelial membrane protein (EMP)/MP20 protein family, ⁷ a subset of a superfamily of tight junctional proteins, the claudins. ⁸ MP20 shares 20% to 30% sequence identity and 40% to 50% sequence similarity with EMP22, PMP22, and GMP17. ^{7 9} Although the precise physiological and/or structural roles of these proteins are unknown, there is evidence of involvement of PMP22 and EMP in cell growth and proliferation, and findings in one study suggest that these proteins may function in cellular adhesion. ¹⁰ A clearer understanding of the topology and post-translational modifications of MP20 in vivo may give insight into its function and role in lens development. Although the topology of MP20 within the fiber cell membrane is not known, it has been shown to form hexamers within the membrane by cross-linking studies. ¹¹ Oligomeric formation is further supported by the observed dominant negative behavior of an MP20 mutant protein in the MP19^T03 transgenic mouse. ⁶ There is contradictory evidence of the topological structure of MP20. Based on hydropathy analysis, members of the PMP22/EMP family are thought to be tetraspanning integral membrane proteins with two large extracellular loops. ⁷ Kumar et al. ¹² proposed an MP20 structure that is most similar to the proposed PMP22/EMP structure with the N terminus located in the first membrane-spanning region and residues 25-63 in the first extracellular domain. In contrast, Arneson and Louis ¹¹ proposed that a large part of the N-terminal region of MP20 (residues 1-65) is located intracellularly. Last, Chen et al. ¹³ proposed that MP20 has five membrane-spanning regions with residues 1-21 in the first membrane-spanning region and residues 22-64 in a large intracellular loop. 

Because of the lack of protein turnover in mature fiber cells, lens proteins undergo age-related post-translational processing such as truncation, as well as the typical co- and post-translational modifications including glycosylation and phosphorylation. The presence of post-translational modifications on MP20 are consistent with reports of decreased antibody binding in the proposed extracellular loop regions, ¹⁴ with fiber cell age in bovine lenses and with the finding of 19-, 21-, and 23-kDa proteins with N-terminal amino acids identical with the N terminus of MP20. ¹⁵ These authors also suggest that the 19- and 21-kDa bands may be due to truncation of MP20 at the C terminus. The primary sequence of MP20 contains a consensus sequence for N-linked glycosylation at Asn 62 within the first putative extracellular loop. Although previous work has not shown any direct evidence of glycosylation, the interaction between MP20 and galectin-3 has been shown to be disrupted by lactose, which suggests the presence of glycosylation. ⁴ In addition to a putative N-terminal glycosylation site, there are also consensus sequences for phosphorylation by cAMP-dependent protein kinase (PKA), protein kinase C (PKC), cGMP-dependent kinase (PKG), and calmodulin-dependent protein kinase II (CaM kinase II) within the C terminus. ¹⁶ The phosphorylation of MP20 by PKA ^{17 18} and PKC ^{19 20} has been demonstrated by in vitro studies. From the phosphoamino acid analysis and enzymatic digestions presented in these studies, it can be deduced that a serine residue in the C-terminal region is phosphorylated. 

In this work, the primary protein structure of MP20 was determined. The observed modifications provide insight into the potential function, regulation, and interactions of the protein. Previous mass spectrometric analysis of human fiber cell membrane proteins after cyanogen bromide treatment has yielded complete sequence information and modifications of the most abundant lens membrane protein, AQP0, but incomplete coverage of MP20. ²¹ To facilitate the primary structural analysis of MP20, multiple complementary experimental approaches were taken, resulting in complete sequence coverage. Principle sites of phosphorylation and glycosylation were determined. Furthermore, we report a novel lens modification of glycosylation on tryptophan residues. 

Methods

Materials

Calf lenses were obtained from Pel Freez (Rogers, AR). Tris-glycine gels (14%–and 4%–20%), Tris-glycine SDS sample buffer and Tris-glycine SDS running buffer were purchased from Invitrogen (Carlsbad, CA). N-octyl β-d-glucopyranoside, Schiff’s fuschin-sulfite reagent, cyanogen bromide, and other chemicals were purchased from Sigma Aldrich (St. Louis, MO). 

Preparation of Lens Membrane Protein

Isolation of lens fiber cell membranes was conducted according to the procedure of Goodenough. ²² Briefly, lenses were decapsulated and homogenized in cold 10 mM NaHCO₃/10 mM NaF/5 mM EDTA (pH 8.0), followed by centrifugation at 92,000g at 4°C for 20 minutes. The pellet was resuspended in cold Tris buffer (10 mM Tris-HCl, 5 mM EDTA, and 1 mM CaCl₂ [pH 9.1]) and centrifuged as above to remove soluble proteins. The extrinsic membrane proteins were removed by resuspending the pellet in 4 M urea in Tris buffer and incubation at 4°C for 15 minutes followed by centrifugation. This step was repeated with 7 M urea. The remaining membrane proteins in the pellet were either washed with 50 mM NaOH in Tris buffer and centrifuged (followed by a water wash) or subjected to a discontinuous sucrose gradient of 25%/41%/67% (wt/vol) sucrose (prepared in Tris buffer). The sucrose gradient solution was centrifuged at 104,000g at 4°C for 30 minutes. Protein at the 25% to 41% interface was collected and washed three times with 1 mL of deionized water. 

Isolation of MP20 by Anion Exchange Chromatography

Lens membrane proteins isolated from the sucrose gradient as described were separated by anion exchange chromatography after a procedure published by Jarvis and Louis. ²³ The membrane pellet (containing 1 mg of protein as determined by the Bradford assay) was solubilized in 1% (wt/vol) octyl glucoside/25 mM Tris-HCl at pH 7.5 and vortexed intermittently for 30 minutes before injection. The suspension was centrifuged for 10 minutes, and the supernatant was applied to a self-packed anion exchange column (Source 15Q; Amersham Pharmacia Biotech; Piscataway, NJ) with dimensions of 0.6 × 12 cm. The first 10 minutes of the gradient used 100% buffer A (25 mM Tris-HCl/1% octyl glucoside [pH 7.5]). The gradient was stepped up to 150 and 300 mM NaCl after 10 and 20 minutes, respectively, by increasing to 7.5% and 15.0% buffer B (2 M NaCl, 25 mM Tris-HCl, and 1% octyl glucoside [pH 7.5]). Protein was detected by absorbance at 220 nm. The flow rate was 1 mL/min, and fractions were collected. 

Initial fractions containing a 20-kDa protein as determined by SDS-PAGE and were matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry combined. To decrease the detergent concentration before further analysis, the combined fractions were diluted with water and concentrated using filters (Centriprep 10; Amicon, Danvers, MA). Filters were washed with 100 mM NaOH before loading the sample. Dilution with water was performed three times to reduce the octyl glucoside concentration from 1% to <0.1%. On continual dilution, the protein precipitated in the filter reservoir (Centriprep; Ambion) and the precipitate was isolated by centrifugation at 88,000g for 20 minutes. 

Periodic Acid-Schiff Staining

Glycosylated proteins were detected by periodic acid-Schiff (PAS) staining according to a procedure described by Thornton et al. ²⁴ Bovine fetuin and bovine serum albumin (BSA) were used, respectively, as positive and negative controls for glycosylated protein. Fractions from the anion exchange separation were collected, and the detergent was removed. Aliquots from the fractions were spotted onto nitrocellulose membrane, allowed to air dry, and stained by PAS. After the membranes were stained, a scanner was used to process the images. 

CNBr Cleavage

MP20 isolated by anion exchange chromatography was digested with cyanogen bromide according to the procedure of Schey et al. ²¹ Protein was resuspended in a ratio of 1:1 (vol/vol) n-propanol:1.5 M Tris [pH 8.7]. Cysteine residues were reduced by treatment with 1000 molar excess tributyl phosphine followed by alkylation with 2000 M excess 4-vinyl pyridine. The reaction mixture was incubated under argon for one hour on ice. After incubation, the reaction mixture was centrifuged at 92,000g for 20 minutes at 4°C, and the pellet was washed with 1.5 M Tris to remove unreacted reagents. Delipidation was performed by overnight incubation in 95% ethanol at −20°C. After delipidation, the sample was centrifuged as described earlier, the supernatant was removed, and acetone was added to the membrane protein pellet. After centrifugation and removal of the supernatant, the membrane proteins were washed twice more after addition of water. 

The protein pellet was dissolved in 75% trifluoroacetic acid (TFA) and cleaved with a 500 molar excess of cyanogen bromide (CNBr) to methionine. The reaction was performed under argon, in the dark for 18 hours, at which time the reaction was quenched with water and dried completely in a vacuum. For analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS), the cyanogen bromide cleavage products were dissolved in 10 μL TFA and 2 μL buffer B, followed by 88 μL of water. The sample was injected onto a reversed phase column (C3 Zorbax) with dimensions of 1.0 × 150 mm. Buffer A contained 0.05% (vol/vol) TFA in water and buffer B was composed of 2:1 isopropanol-acetonitrile and 0.05% (vol/vol) TFA. ²⁵ HPLC (model 1100; Agilent) was used, with the gradient set at 2% B for 2 minutes, 2% to 25% B in 20 minutes, 25% to 55% B in 50 minutes, 55% to 70% B in 15 minutes, 70% to 98% B in 10 minutes. The flow rate was 25 μL/min, and the column effluent was split directing 5 μL/min into an ion-trap mass spectrometer (ThermoFinnigan LCQ Classic) set to scan between m/z 400 to 2000. Fractions were collected from the remaining eluate. Selected ion chromatograms were generated for each peptide and phosphopeptide by using the predicted m/z ratios. Relative amounts of phosphorylation were calculated (Xcalibur software; ThermoFinnigan), from the areas of each peak compared with the total area of peptide 164-173 in the unphosphorylated and the mono- and diphosphorylated states. 

Tryptic Digestion

The total membrane protein fraction from the cortical region of a lens was isolated and rinsed with 50 mM NaOH, as described earlier. Reduction and alkylation of cysteine residues and overnight delipidation were performed as described earlier. Digestions were performed with Promega (Madison, WI) sequencing-grade trypsin (an enzyme to protein ratio of 1:100). Membranes were suspended in 80% 10 mM ammonium bicarbonate (pH 8.0)/20% acetonitrile and digested for 22 hours at 37°C. Water (500 μL) was added to each sample and the samples were centrifuged at 92,000g (4°C). The supernatant containing the soluble peptides was removed and dried under vacuum. 

The soluble peptides were analyzed by LC-MS/MS. Samples (100 μg) were dissolved in 40 μL of mobile phase A and injected onto a C18 column (Microtech Scientific; Vista, CA) with dimensions of 1.0 × 150 mm. Mobile phase A was 0.02% TFA/0.1% acetic acid/2% acetonitrile; buffer B was 0.02% TFA, 0.1% acetic acid, and 90% acetonitrile. The gradient was 0% to 65% B over 65 minutes and 65% to 98% B over 15 minutes. HPLC was used with a flow rate of 50 μL/min. The effluent was directed into the ion-trap mass spectrometer set to scan between m/z 350 to 2000. 

SDS-PAGE and In-Gel Digestion

The total membrane protein fraction was isolated and rinsed with 50 mM NaOH, as described earlier. Lens membrane samples were prepared without boiling in SDS sample buffer (Invitrogen) and 2.5% β-mercaptoethanol according to the manufacturer’s instructions. Samples were loaded onto 4% to 20% Tris-glycine gels, and electrophoresis was conducted (Xcell Mini Cell; Invitrogen). Proteins were visualized by Coomassie brilliant blue staining. Bands of interest were excised and destained with 60% 25 mM ammonium bicarbonate (pH 8.0) and 40% ethanol at 37°C three times for 20 minutes. Gel pieces were dehydrated with acetonitrile and dried in a speed vac. The protein was reduced and alkylated under Ar by adding 50 μL of 5 mM tributylphosphine and 93 mM 4-vinylpyridine prepared in 100 mM ammonium bicarbonate (pH 8.0) and vortex mixing for 1 hour. Gel pieces were rinsed three times with 100 mM ammonium bicarbonate (pH 8.0) and twice with 50% 25 mM ammonium bicarbonate (pH 8.0) and 50% acetonitrile. Gel pieces were rinsed with acetonitrile and dried in a speed vac. 

In-gel tryptic digestions were performed following the procedure used by van Montfort et al. ²⁶ Porcine trypsin (200 ng; Promega) was added, and gel pieces swelled at room temperature for 20 minutes. Thirty microliters of 0.08% octyl-glucoside and 80% 25 mM ammonium bicarbonate (pH 8.0), 20% acetonitrile was added, the samples were overlaid with Ar, and incubations were performed at 37°C for approximately 20 hours. After removing the overlay buffer, peptides were extracted from the gel with two additions of 1% TFA and 70% acetonitrile at 37°C and intermittent vortexing for 20 minutes. The extracted peptides were dried to near completeness and desalted with C₁₈ pipette tips (Zip Tips; Millipore, Bedford, MA) for MALDI-TOF analysis. 

MALDI-TOF Mass Spectrometric Analysis

For analysis of intact membrane proteins by MALDI mass spectrometry, samples were resuspended in 6 μL of 77% vol/vol formic acid 23% vol/vol hexafluoroisopropanol (or 7:3 formic acid; HFIP) followed by 2 μL water. After the solubilization of protein, the samples were analyzed immediately to avoid potential formylation of residues. Sinapinic acid matrix (50 mM 3,5-dimethoxy-4-hydroxy-cinnamic acid) in 70% formic acid was spotted onto the plate. Internal calibration was used with insulin, thioredoxin, and apomyoglobin as standards. For analysis of tryptic peptides by MALDI-TOF, samples were desalted with C₁₈ (Zip Tips; Millipore) and eluted in 70% acetonitrile and 0.1% TFA. The samples were directly eluted onto 0.5 μL spots of α-cyano-4-hydroxycinnamic acid (α-CHCA) on the plate. A MALDI mass spectrometer (Voyager-DE STR Biospectrometry Workstation; Applied Biosystems, Foster City, CA) was used, and calibration was performed with angiotensin I, ACTH, and bovine insulin as standards. For MALDI-TOF/TOF analysis, peptides isolated from the in-gel digestion were desalted with C₁₈ pipette tips (Zip Tips; Millipore) and mixed with α-CHCA (1:1) and spotted onto the sample plate. A proteomics analyzer (model 4700; Applied Biosystems) was used in positive ion mode with a laser intensity between 4200 and 4500 nm. Precursor ions were selected with a window of 10 and 1000 to 5000 shots were averaged for each spectrum. 

Results

Purification of MP20

To facilitate the primary structural analysis of bovine MP20, the protein was separated from the more abundant lens membrane protein, AQP0, under mild conditions. Lens membrane proteins were solubilized in 1% octyl glucoside and separated according to pI by anion exchange chromatography at pH 7.5 (Fig. 1A) . Due to the basic nature of MP20, this protein was not retained by the column and was eluted at 3.5 minutes, as previously reported by Jarvis and Louis. ²³ A step gradient was performed and proteins eluting without NaCl, at 150 mM NaCl, and at 300 mM NaCl were collected. The fractions were analyzed by SDS-PAGE and MALDI-TOF to confirm the purity of the MP20 fractions. By MALDI-TOF analysis, the enriched MP20 fractions isolated between 2 and 4 minutes, as shown in Figure 1A , were found to have a broad signal centered at m/z 19,853, 168 Da higher than predicted. The AQP0 peak was found to have a signal at m/z 28,231, 7 Da higher than predicted (data not shown). 

Anion exchange fractions were pooled for each eluting peak and equal volumes were spotted onto nitrocellulose membrane. Staining by PAS (Fig. 1B)shows a positive reaction for the first and second pooled fractions, indicating the presence of glycosylated protein. All other later eluting fractions, which contained proteins including AQP0, MP70, and connexin 46, as previously reported by Jarvis and Louis, ²³ did not stain with this method. 

Complete Map of the Bovine MP20 Sequence

To confirm the sequence and to identify the sites and nature of post-translational modifications, MP20 fractions isolated by anion exchange chromatography were pooled and cleaved with CNBr. The peptide fragments were analyzed by LC-MS/MS, and the sequence of each peptide was confirmed. All the peptides predicted from the amino acid sequence of MP20 were observed, with exception of the first single methionine residue (Table 1) . By selecting the predicted m/z values of each MP20 CNBr peptide ion, selected ion chromatograms were generated illustrating the presence and chromatographic separation of the cyanogen bromide peptides (Fig. 2) . Peptide 105-149 which contains three tryptophan residues and peptides 6-29 and 30-68, each of which contain two tryptophan residues, were also detected, with multiple sites of tryptophan oxidization identified by tandem mass spectrometry. Oxidation of the tryptophan residues most likely occurred during CNBr digestion. 

Phosphorylation of Bovine MP20 Determined by LC-MS/MS

Mass spectrometric analysis of CNBr digests indicated single and double phosphorylation of the C-terminal peptide; however, the sites of phosphorylation were difficult to identify in the tandem mass spectra because of the dominating losses of vinyl pyridine and phosphoric acid. Tryptic digests of lens membrane proteins rinsed with 50 mM NaOH were analyzed by LC-MS/MS to identify the phosphorylation sites. The unphosphorylated C-terminal tryptic peptide 169-173 and the singly and doubly phosphorylated peptides with one missed tryptic cleavage, residues 168-173, were observed in the data (see Table 1for the predicted and observed m/z ratios). The tandem mass spectra shown in Figures 3A and 3Bwere obtained on fragmentation of the doubly charged ions ([M+2H]²⁺) of the mono- and diphosphorylated peptides, with respective m/z ratios of 405.3 and 445.3. Fragmentation of the peptides in the ion trap results in cleavage of amide bonds and the formation of the complementary N terminus containing b-ions and C terminus containing y-ions. Labeling of the b and y ions was performed according to nomenclature by Biemann. ²⁷ The most intense peak in each spectrum corresponds to a loss of phosphoric acid from the doubly charged molecular ion ([M+2H*]²⁺). Serine-170 was found to be the site of in vivo phosphorylation in the monophosphorylated peptide corresponding to residues 168-173. Product ions corresponding to a loss of phosphoric acid from Ser-170 are observed for b₃ and y₄. Figure 3Bshows that the second site of phosphorylation is the adjacent Thr-171 residue. Product ions corresponding to the expected m/z for phosphorylated residues Thr-171 (y₃) and Ser-170 (y₄) are observed at m/z 453.2 and 620.1, respectively. Peak areas for the unphosphorylated, mono- and diphosphorylated peptides were calculated from selected ion chromatograms of the expected m/z ratios for each peptide and averaged from three anion exchange separations. The areas for the mono- and diphosphorylated peptides were found to be approximately 69.4% and 14.3%, respectively, of the total area measured for the unphosphorylated and mono- and diphosphorylated peptides. These values may be used as rough estimates of the relative amounts of phosphorylation and most likely underestimate the values, since the ionization efficiency of a peptide in the phosphorylated state is generally lower when compared with the unphosphorylated peptide. ²⁸ Our interpretation of these data is that MP20 exists predominantly in the monophosphorylated state. 

Glycosylation of Bovine MP20 Determined after In-Gel Digestion

The cyanogen bromide peptide corresponding to residues 30-68 contains a potential site for N-glycosylation at residue 62. The reported interaction with galectin-3 suggests glycosylation of MP20; however, no glycosylated peptides were observed after CNBr cleavage and no shifts in peptide masses were observed after deglycosylation attempts with PNGase F. Therefore in-gel tryptic digestion was attempted, to avoid the highly acidic conditions required for CNBr cleavage. The MP20 bands from the gel were excised, destained, treated with 4-vinyl pyridine, and digested with trypsin. The peptides were then extracted and analyzed by MALDI-TOF, revealing all the expected tryptic peptides from the extramembranous regions (between residues 33-65, 131-135, and 163-173) and two sites of glycosylation. Peaks corresponding to the unmodified tryptic peptides 33-44 and 51-65 were observed at m/z 1343.8 and 1925.5, respectively. In four different lenses (approximately 27% of the samples analyzed), signals with masses corresponding to the addition of a hexose residue (162-Da increase) for residues 33-44 and 51-65 were observed in addition to the unmodified peptides. A representative MALDI-TOF spectrum is given in Figure 4A . Peaks for the peptides containing the putative glycosylation were observed at m/z 1505.9, 1668.0, and 2087.8 and sequence data were obtained by MALDI-TOF/TOF (Figs. 4B 4C) . When peptides from the same sample were extracted from the gel without the reduction and alkylation steps, the glycosylated peptides were not observed. Also, the glycosylated peptides were not observed with common gel extraction protocols, including sonication in the presence of solutions containing 5% formic acid and acetonitrile. 

Figure 4Bshows the MALDI-TOF/TOF spectrum of the selected precursor ion at m/z 1505.9 containing residues 33-44 with an additional 162 Da. The observed fragmentation pattern is consistent with a hexose residue on Trp-43. In addition, there is a large peak at m/z 163.1, due to ionized hexose, and a loss of 120 Da (C₄H₈O₄) was observed for the glycosylated y₅ ion. The loss of 120 Da is consistent with data obtained by electrospray ionization (ESI)-MS/MS for a protein containing a C-glycosidic linkage to tryptophan as found by others, ^{29 30} and indicates that the hexose residue may be attached by a carbon–carbon bond. Note that the specific hexose involved cannot be directly determined by mass spectrometry. Another smaller peak in the MALDI-TOF spectrum at m/z 1668.0 was sequenced and the fragmentation pattern indicates that there is the addition of two hexose residues at Trp-43 (data not shown). Sequence analysis by MALDI-TOF/TOF required a greater number of laser shots to be averaged and resulted in a weak spectrum, but the data show losses of 282 Da (due to one hexose residue and 120 Da) from the molecular ion and several y ions (including y₂) that contain Trp-43. Figure 4Cshows the MALDI-TOF/TOF analysis of the precursor ion at m/z 2087.8 containing residues 51-65 and an additional 162 Da. Again, the fragmentation pattern indicated the addition of a hexose residue to Trp-61. The y₅ ion containing the hexose residue is observed at m/z 809.4. 

Discussion

The MP20 sequence has been cloned and the protein partially characterized by amino acid analysis, peptide mapping, and partial Edman degradation ^{12 15 31} ; however, to date, the entire protein sequence has not been confirmed nor have the sites of glycosylation and in vivo phosphorylation been determined. In this study, several experimental strategies were used to confirm the sequence and determine the types and sites of post-translational modifications in MP20. Use of anion exchange chromatography to purify bovine MP20 followed by LC-MS/MS analysis of CNBr peptides obtained 100% sequence coverage. 

MP20 contains consensus sequences for phosphorylation by PKA, PKC, PKG, and CaM kinase II at the C terminus, ¹⁶ and in vitro studies with PKA and PKC revealed phosphorylation of a C-terminal serine residue. ^{17 18 19 20} Tandem mass spectrometric analysis of the lens membrane products from CNBr cleavage and tryptic digestion allowed detection of the MP20 peptides phosphorylated in vivo. Phosphorylation of Ser-170 was confirmed by tandem mass spectrometry in the singly phosphorylated peptide and a second phosphorylation site at Thr-171 was identified in the doubly phosphorylated peptide. The areas of the peaks calculated from the selected ion chromatograms for the monophosphorylated and diphosphorylated peptides were found to be approximately 69.4% and 14.3%, respectively, of the total area measured for the unphosphorylated, mono- and diphosphorylated peptides. Depending on the ionization efficiencies of the phosphorylated peptides, these data indicate that the C-terminal region of MP20 is predominantly in the monophosphorylated form. One possible role of MP20 phosphorylation is in protein trafficking; however, because the unphosphorylated form of MP20 appears to undergo changes in localization, ^{2 3 32} this suggests that there is another signal, such as glycosylation, that is involved in trafficking. Clearly, further studies are needed to determine the precise role of the phosphorylation of MP20 in the lens. 

MP20 has recently been shown to be a putative ligand for galectin-3, a carbohydrate-binding protein, also found in the lens. ^{4 32} The change in localization of MP20 from the cytoplasm to the membrane of the fiber cells appears to correlate with the change in localization of galectin-3 in the lens. ⁴ Because the MP20 CNBr peptide containing residues 30-68 contains a consensus sequence for N-linked glycosylation and residue 62 is the only asparagine in the protein, the glycosylation state of this peptide was considered. Although the anion exchange fractions eluting before AQP0 stained positively by PAS, indicating the presence of a glycosylated protein, glycosylation at Asn-62 was not found in the tryptic digests or in the CNBr digests of MP20. Previous work with an N-linked high-mannose glycated membrane protein, demonstrated that this type of glycosylation was stable to cysteine alkylation and cyanogen bromide cleavage in acidic conditions. ²⁵ These negative results for bovine MP20 indicate that (1) a small undetectable percentage of protein is glycosylated, (2) there is another type of glycosylation present, or (3) a co-eluting protein is glycosylated. MALDI-TOF analysis of MP20 anion exchange fractions shows one broad peak 168 Da higher than predicted, indicating that the protein is relatively pure and that extensive glycosylation is not present. 

MS/MS analysis of in-gel digestions of MP20 revealed glycosylation at Trp-43 and -61, representing the first observation of tryptophan glycosylation in the lens. Variable amounts of glycosylation at these two tryptophan residues was found in approximately 27% of the bovine lenses analyzed. Sequence analysis by MALDI-TOF/TOF also indicates that there is a small amount of peptide with two hexose residues present at Trp-43, a previously unreported modification. Although MALDI-TOF does not allow for absolute quantitative comparison of data and the effect of glycosylation on the ionization efficiencies of these peptides is not known, Trp-43 appeared to be more extensively glycosylated than Trp-61. Ranges of glycosylation, determined from the relative intensities of the peaks at the observed m/z ratios, are at approximately 20% to 70% and 5% to 56% for Trp-43 and -61, respectively. The variance in the amount of glycosylation may be due to individual differences of each lens or possibly to instability of the glycosylated protein during storage, to the pH extremes used during isolation of the membrane proteins, or to the presence of glycosidases and proteases in the samples. Another peptide in a proposed extracellular loop that contained Trp-134 was not found to be glycosylated. This suggests that glycosylation of Trp-43 and -61 is specific and is not an artifact of sample preparation. 

The in vivo enzymatic glycosylation of tryptophan resides was recently identified in RNase 2 from human urine ³³ and also in several other proteins. ²⁹ Nuclear magnetic resonance (NMR) analysis revealed that this type of glycosylation, C-mannosylation, involves the attachment of a mannose residue to C-2 in tryptophan through a carbon-carbon bond. ³⁴ In the MALDI-TOF/TOF analysis of bovine MP20, the type of hexose and the position of attachment to the indole could not be determined. It is possible that the glycosylation of MP20 involves an N-linkage to the indole, as has been observed to occur through an enzymatic pathway in fruit. ³⁵ A consensus sequence has been proposed, but proteins containing C-mannosylation show variability in this sequence, suggesting that more than one C-mannosyltransferase may be present or that other local features of the protein may be involved. ³⁰ The in vivo function of C-mannosylation is not yet known, but site-directed mutagenesis of tryptophan residues that are C-mannosylated results in a loss of protein trafficking from the endoplasmic reticulum. ³⁶ Besides a trafficking role, glycosylation of MP20 may result in the interaction of this protein with galectin-3. Galectin-3 has been shown to bind to a variety of oligosaccharides containing β-galactoside ³⁷ and oligomannosides. ³⁸ Binding of galectin-3 to the sites of glycosylation in MP20 is further supported by the observation that antibodies to residues 26-44 and 50-65 had increased binding to MP20 after a short exposure to trypsin, possibly due to the removal of a protein that blocked access to these regions. ¹¹  

From the data presented in this study, several hypotheses may be made about the structure and function of MP20 in the lens. The proposed structure of the MP20 domains that fits with the predicted extracellular locations of the glycosylated residues and the intracellular location of phosphorylated residues (Fig. 5)is similar to those proposed by Kumar et al. ¹² and Taylor et al. ⁷ In contrast, the glycosylated tryptophan residues would not be present in extracellular domains in models proposed by Arneson and Louis ¹¹ and Chen et al. ¹³ In work by Gray et al., ³ MP20 localization changed from the cytoplasm in the outer cortex to the membranes of the fiber cells on differentiation. Furthermore, work by Tenbroek et al. ² showed that between 0.5 and 2.0 mm into the lens, MP20 localization changed from patterns dispersed on the broad sides of fiber cells to the narrow interdigitated sides. That both studies were performed with antibodies against a synthesized C-terminal MP20 peptide (unphosphorylated) may suggest that there is another signal besides phosphorylation, such as glycosylation, that induces trafficking of MP20. 

In summary, the deduced sequence of bovine MP20 has been confirmed with the primary phosphorylation site identified as Ser-170. A second site of phosphorylation has also been identified in the diphosphorylated protein at Thr-171. Glycosylation at tryptophan residues 43 and 61 indicate that this region of the protein is located in an extracellular domain. The impact of the covalent modifications of MP20 will be determined as the function of the protein is further elucidated. The structural modifications of MP20 may be important in the regulation of protein trafficking, function, and/or the development of protein–protein interactions. 

²  Contributed equally to the work and therefore should be considered equivalent authors.

 Supported by Grants EY13462 (KLS) and R24 EY14793 from the National Eye Institute and by Grant 5 T32-HL07260 from the National Heart, Lung, and Blood Institute.

 Submitted for publication July 28, 2004; revised September 29, 2004; accepted October 12, 2004.

 Disclosure: L.A. Ervin, None; L.E. Ball, None; R.K. Crouch, None; K.L. Schey, 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.

 

Figure 1.

 

(A) Anion exchange chromatogram monitored at 220 nm showing separation of calf lens proteins solubilized in 1% octylglucoside. (B) Periodic acid-Schiff–stained dot blot of anion exchange fractions. (F) Positive control, fetuin; (B) negative control, BSA. Fractions with retention times of (1) 2 to 4 minutes, (2) 4 to 6 minutes, (3) 6 to 10 minutes, (4) 14 to 16.5 minutes, (5) 16 to 16.5 minutes, (6) 16.5 to 22 minutes, (7) 24 to 27 minutes, and (8) 33 to 36 minutes.

Table 1.

 

Peptides Predicted from the Amino Acid Sequences of MP20

Residues	Predicted Molecular Weight	Observed Molecular Weight
Cyanogen bromide fragments*
1	101.1	ND
2–5	498.2	498.2
6–29, †	2576.0	2575.0
30–68, † , ‡	4768.5	4767.6
69–81	1364.8	1364.6
82–104	2447.8	2447.2
105–149, †	5179.1	5179.1
150–158	1047.5	1047.4
159–163	699.3	699.4
164–173, §	1358.7, 1438.7, 1518.7	1359.2, 1439.4, 1519.2
Tryptic peptides, ∥
33–44 , ¶	1342.5, 1504.5, 1666.5	1342.2, 1504.9, 1667.0
45–50 , #	789.0	789.0
51–65 , # , **	1924.2, 2086.2	1924.8, 2086.8
131–135	679.7	679.8
163–167 , #	779.9	779.8
168–173 , ††	808.9, 888.9	808.6, 888.6
169–173	572.7	572.8

 Bold indicates sequence was confirmed. ND, not detected.

*  All sequences were confirmed by tandem mass spectrometry.

†  Fragment contains oxidization at Trp residues.

‡  N-terminal pyroglutamic acid detected.

§  Unphosphorylated, mono- and diphosphorylated.

∥  Peptides were detected by LC-MS/MS and MALDI-TOF.

¶  Peptide without glycosylation and with the addition of one or two hexose residues.

#  Peptide contains alkylated Cys.

**  Peptide without glycosylation and with the addition of one hexose residue.

††  Mono- or diphosphorylated.

Figure 2.

 

Selected ion chromatograms of RP-HPLC separated cyanogen bromide fragments of bovine MP20. The selected m/z of each peptide and the retention times are shown. Residues 6-29, 30-68, and 105-149 were found to be unmodified and also to contain oxidation of multiple tryptophan residues. The selected ion chromatogram for residues 105-149 shows the elution of the peptide with the addition of three oxygen atoms (48 Da).

Figure 3.

 

ESI-MS/MS analysis of the mono- and diphosphorylated MP20 tryptic peptide 168-173. Phosphorylation of this peptide results in a missed cleavage by trypsin. (A) Sequence analysis of the doubly charged ion at m/z 405.3 corresponding to the monophosphorylated peptide. Inset: predicted m/z for the y and b ion series, including phosphorylation at Ser-170. *Observed losses of 98 Da due to losses of phosphoric acid. [M+2H*]²⁺: loss of phosphoric acid from the doubly charged unfragmented peptide. Losses of 17 and 18 Da are due to losses of NH₃ and H₂O, respectively. (B) Sequence analysis of the doubly charged ion at m/z 445.3 corresponding to the diphosphorylated peptide. Inset: predicted m/z for the y and b ion series, including phosphorylation at Ser-170 and Thr-171. Observed losses of 98 (*) and 196 (**) Da due to losses of one and two phosphoric acid groups. [M+2H*]²⁺: loss of phosphoric acid from the doubly charged unfragmented peptide.

Figure 4.

 

MALDI-TOF spectra of MP20 tryptic peptides isolated from an in-gel digestion. (A) Peaks at m/z 1343.8 and 1925.5 correspond to [M+H]⁺ for the unmodified peptides 33-44 and 51-65, respectively. The peaks at m/z 1505.9 and 1668.0 correspond to increases in mass due to the addition of one or two hexose residues to 33-44. The peak at m/z 2087.8 corresponds to an increase in mass due to the addition of one hexose residue to 51-65. Sequences of these peaks were confirmed from the expected m/z ratios for the y and b ions formed by fragmentation of the peptide during MALDI-TOF/TOF analysis. (B) MALDI-TOF/TOF analysis of the peak at m/z 1505.9. Inset: predicted m/z values for the y and b ion series (shown in the inset) are based on a hexose addition to Trp-43. Observed losses of 162 and 120 Da are due to the respective losses of hexose and C₄H₈O₄. A loss of 120 Da is characteristic of C-mannosylation. ^{29 30} (C) MALDI-TOF/TOF analysis of the peak at m/z 2087.8. Predicted m/z values for the y and b ions are based on addition of a hexose at Trp-61 and 4-vinyl pyridine at Cys-51. Observed losses of 105 and 162 Da are due to losses of vinyl pyridine and a hexose residue, respectively.

Figure 5.

 

Proposed structure of bovine MP20 (from TMpred at www.ch.EMBnet.org; provided in the public domain by the Swiss Institute of Bioinformatics, Basel, Switzerland). Filled background: Tryptic and CNBr cleavage sites. Sites of identified in vivo glycosylation (Hex) and phosphorylation (P) are shown.

The authors thank Susana Comte-Walters (Medical University of South Carolina) for operation of the proteomics analyzer (model 4700; Applied Biosystems) and the Mass Spectrometry Facility, Medical University of South Carolina.