Molecular Vision 2007; 13:1873-1877 <http://www.molvis.org/molvis/v13/a210/> Received 6 March 2007 | Accepted 2 October 2007 | Published 3 October 2007

Download Reprint

Comparison of two tandem mass spectrometry-based methods for analyzing the proteome of healthy human lens fibers

Chunwei Zhang,¹ Ping Liu,¹ Ningli Wang,² Yaojun Li,³ Linjie Wang³
 
 

¹Department of Ophthalmology, the First Affiliated Hospital, Harbin Medical University, Harbin, China; ²Beijing Tong Ren Eye Center, Beijing Tong Ren Hospital, Capital Medical University, Beijing, China; ³Proteomics Research Center, National Key Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences/Peking Union Medical College, Beijing, China

Correspondence to: Dr. Ping Liu, 23 You Zheng Street, Nan Gang District, Department of Ophthalmology, The First Affiliated Hospital, Harbin Medical University, Harbin, People's Republic of China, 150001; Phone: 86-0451-53643849-3958; FAX: 86-0451-53650320; email: ping_liu53@hotmail.com

Abstract

Purpose: To establish a method for studying the proteome of human lens fibers and to provide the proteome database of lens fibers from healthy male adults.

Methods: We compared two liquid chromatography tandem mass spectrometry (LC-MS/MS)-based methods for studying the proteome of healthy adult human lens fibers. Total proteins were extracted from pooled lens fibers of 12 healthy male adult donors. In one method, the total proteins were digested with trypsin, and the derived peptides were analyzed by strong cation exchange (SCX) coupled with reverse phase liquid chromatography tandem mass spectrometry (RPLC-MS/MS). In the other method, proteins were first resolved by sodium dodecyl sulfate PAGE (SDS-PAGE) and then in-gel digested with trypsin, and the peptides were analyzed by RPLC/MS/MS. The tandem mass spectra of positive results were quality controlled by advanced mass spectrum scanner (AMASS) software. The peptide false positive rate was estimated using the reverse database searching method.

Results: A total of 68 proteins from lens fibers were identified using these two methods based on at least two different peptide matches with reliability of over 97% for each peptide. Among these proteins, 43 were detected by both methods, one was detected only by SCX-RPLC/MS/MS, and 24 were detected only by SDS-PAGE-RPLC-MS/MS.

Conclusions: The data clearly indicated that the SDS-PAGE-RPLC-MS/MS method was more suitable than the SCX-RPLC-MS/MS method for analyzing lens fiber proteome. This work greatly expanded the proteome database of human lens fibers, and the results provided a reference for future studies to detect aging-related and cataract-related changes in human lens fibers proteins.

Introduction

The human lens is a transparent and avascular tissue composed of epithelial and fiber cells. A single layer of epithelial cells covers the anterior surface of the lens, and the fiber cells occupy the remainder of the lens volume. During differentiation, lens epithelial cells undergo significant biochemical and morphological changes that result in the formation of fully differentiated fiber cells. In the lens fibers, virtually all organelles together with their component proteins are removed while a large quantity of crystallins are synthesized. Thus, the lens fibers have the highest protein content in the body, which accounts for more than 35% of the wet weight. About 90% of the lens proteins are high abundant lens fiber structural proteins called crystallins. These proteins play an important role in maintaining the transparency of the lens. Alterations or modifications of these proteins could lead to cataract [1]. Due to the slow turnover rate, many posttranslational modifications that occur during the differentiation and aging processes remain in the lens [2]. Despite the recent progress in proteomic technologies [3,4], the proteome of human lens fibers remains largely unknown. Therefore, a comprehensive profile of human lens fiber proteins will provide a valuable reference for studying aging-related and cataract-related lens protein changes.

The extreme high concentration of crystallins in lens fibers makes it difficult to detect low abundant proteins in the lens fibers because the high abundant crystallins mask the low abundant proteins in both the separation and detection processes. In this study, we compared two liquid chromatography tandem mass spectrometry (LC-MS/MS)-based methods for studying the proteome of healthy adult human lens fibers. One method is the classic shot-gun method in which the protein mixture was digested with trypsin, and the derived peptides were separated by two-dimensional strong cation exchange-reversed phase liquid chromatography (SCX-RPLC), and peptide sequences were determined by tandem MS/MS. In the other method, the lens fiber protein mixture was first separated by sodium dodecyl sulfate PAGE (SDS-PAGE) and then in-gel digested with trypsin, and the peptides were extracted and analyzed by reverse phase liquid chromatography tandem mass spectrometry (RPLC-MS/MS). A total of 68 unique gene products (proteins) were identified in the lens fibers by the combination of the two methods and many of these proteins have not been reported previously. The results indicate that the SDS-PAGE-RPLC-MS/MS method is more sensitive for detection of low abundant proteins in lens fibers, which identified 67 of these proteins whereas the classic shot-gun method only identified 44 of these proteins.

Methods

Sample preparation

Twelve healthy human lenses from six male donors (age ranged from 35-45 years) were obtained post mortem from subjects intended for autopsy at the Eye Bank of the Department of Ophthalmology of the First Affiliated Hospital, Harbin Medical University. The collection of the donor tissue was approved by the Regional Committee for Scientific Medical Ethics in Heilongjiang Province, China. All of these donors did not have any signs of lens disease or a history of eye surgery. The lenses were collected 5-18 h after death. The capsules and attached epithelia were removed immediately after the extraction of the lenses from the eyes. The epithelium-free lens fibers were frozen in liquid nitrogen and stored at -80 °C until use. All of the procedures for dissecting the eyes and lenses were performed at 4 °C.

Approximately 500 mg frozen pooled lens fibers were grinded under liquid nitrogen and then dissolved in 1.5 ml lysis buffer (7 mol/l urea, 2 mol/l thiourea, 4% CHAPS, 40 mmol/l Tris, 1 mmol/l PMSF, 1 μg/ml aproptinin, and 1 μg/ml leupeptin, pH 7.4). After stirring at 4 °C for 1 h, the lysate was centrifuged at 15,000 g for 15 min. The supernatant was collected and stored at -80 °C until use. The protein concentration was determined by the Bradford method.

SDS-PAGE

The proteins isolated from lens fibers were diluted with SDS-PAGE loading buffer (50 mmol/l Tris-HCl, pH 8.3, 2% SDS, 5% β-mercaptoethanol, and 5% glycerol). After incubation at 95 °C for 3 min, 32 μg of protein/lane was resolved by non-continuous SDS-PAGE. The stacking gel was 5% and the resolving gel was 12%. The ratio of acrylamide to bis-acrylamide was 30:1. The gels were stained with Coomassie brilliant blue after electrophoresis (Figure 1).

Protein digestion

For identification using strong cation exchange (SCX)-RPLC/MS/MS, proteins isolated from lens fibers were reduced with 10 mM dithiothreitol (DTT) at 56 °C for 1 h and alkylated in darkness with 50 mM iodoacetamide at room temperature for 30 min. The proteins were then copolymerized with 6% acrylamide (acrylamide: bis-acrylamide ratio was 30:1) and 0.1% SDS in a 1.5 ml Eppendorf tube. The final concentrations of ammonium persulfate and TEMED were 0.04% and 0.08%, respectively. The protein concentration in the gel was 1 mg/ml. The gel was then cut into approximately three 1 mm pieces and washed three times with 25 mM NH₄HCO₃. About 400 μl washed gel particles were lyophilized and then immersed in 400 μl of 1 ng/μl trypsin (Roche, Penzberg, Germany) solution in 25 mM NH₄HCO₃. After standing at 4 °C for 1 h, 50 μl of 25 mM NH₄HCO₃ without trypsin was added to ensure that the gel particles were immersed in solution during the digestion. The digestion was then performed at 37 °C for 17 h. The resulting peptide mixtures were first extracted with 400 μl 5% trifluoroacetic acid (TFA)/50% acetonitrile (ACN) and then with the same volume of 2.5% TFA/50% ACN. The combined extraction solution was lyophilized and used for LC-MS analysis.

For SDS-PAGE separated samples, the coomassie brilliant blue stained gels were cut into seven slices manually as indicated in Figure 1. The sizes of the SDS-PAGE gels (the resolving gels) were 5x7 cm. The sizes of the gel slices was as follows: slice 1, 0.6x6 cm; slice 2, 0.9x6 cm; slice 3, 0.6x6 cm; slice 4, 0.6x6 cm; slice 5, 0.5x6 cm; slice 6, 1.0x6 cm; slice 7, 0.7x6 cm. The gel was then cut into about 1 mm³ pieces and washed three times with 25 mM NH₄HCO₃. The gel pieces were lyophilized and digested with trypsin according to the method described above.

Liquid chromatography tandem mass spectrometry

For SDS-PAGE-RPLC-MS/MS, the lyophilized tryptic peptides from each slice were dissolved in 80 μl buffer A (0.1% formic acid). Fused silica tubing (150x170 μm id) packed with C-18, 5 μm spherical particles with pore diameter 300 Å (MERCK Superspher 100 RP-18, Merck, San Diego, CA) was used to separate the peptides. The peptides were eluted with buffer B (0.1% formic acid, 99.9% ACN) on a gradient from 5%-30% for 6 h (flow rate: 2 μl/min).

For SCX-RPLC/MS/MS, 200 μg of sample was separated by SCX capillary column (150x320 μm id) packed with poly (2-sulfethyl aspartamide)-based SCX resins (PolyLC INC, Columbia, MD) of 5 μm spherical particles with a pore diameter of 300 Å and by the coupled reversed phase (RP) capillary column (MERCK Superspher 100 RP-18). Peptides were sequentially eluted from the SCX column with 0, 25, 50, 75, 100, 125, 150, 250, 500, and 1000 mM ammonium acetate, respectively. The peptides from each SCX elution step were eluted from the RP column with 5%-30% buffer B gradient for 4 h (flow rate: 2 μl/min).

Eluted peptides were analyzed by an LCQ-DECA XP^plus electrospray ion trap mass spectrometer (ThermoFinnigan, San Jose, CA). Ions were detected in a survey scan from 400-2000 atomic mass units (amu) followed by five data-dependent MS/MS scans (5 microscans each, isolation width 3 amu, 35% normalized collision energy, dynamic exclusion for three min) in a completely automated fashion.

All of these experiments were repeated three times.

Data processing

All MS/MS spectra were searched using Bioworks 3.1 against the database ipi.human.v3.11 [5] with enzyme constraints with a static modification of +57 Da on cysteine residues and a differential modification of 16 Da on methionine residues. The precursor-ion mass tolerance was 1.40 Da and the fragment-ion mass tolerance was 1.50 Da. We used SEQUEST criteria as follows: ΔCn is greater than or equal to 0.1; Rsp=1; Xcorr is greater than or equal to 1.9 for singly charged peptides; Xcorr is greater than or equal to 2.2 for doubly charged peptide; Xcorr is greater than or equal to 3.75 for triply charged peptides. Then AMASSv1.13 software (a free copy of AMASSv1.13 software can be obtained from the author) was used to filter the SEQUEST results with the following three parameters: MatchPct (percentage of matched high-abundant ions) is greater than or equal to 95, Cont (continuity of matched fragment ions in b, y series) is greater than or equal to 30, Rscore (randomicity score) is less than 2.7 [6,7]. Proteins with two or more peptides approved by advanced mass spectrum scanner (AMASS) were accepted as positive identifications. The overlapping rate equals the number of identified proteins in common in at least two runs divided by the number of total identified proteins and multiplied by 100%.

All of the raw data were also searched with the same parameters against the reversed database to estimate the false positive rate. If a set of peptides match multiple proteins in the database, the protein with highest peptide coverage was reported. Theoretical molecular weights and the functions of the identified proteins were analyzed manually using the Swiss-Prot database.

Results

Identification of the proteins of lens fibers with SDS-PAGE-RPLC-MS/MS or SCX-RPLC-MS/MS

Lens fiber proteins were analyzed using the SDS-PAGE-RPLC-MS/MS method or the SCX-RPLC-MS/MS method. Each method was repeated three times. Proteins that were identified in any of the three experiments of each method were reported. A total of 68 unique proteins were identified from lens fibers based on two or more positive peptides with a reliability >97% for each peptide (Table 1 in Appendix 1). The false positive rate was calculated as peptide number in reverse database/peptide number in forward database x100%. The average of false positive rates was 2.25% for EQUEST/AMASSES-filtered positive peptides. Among the 68 proteins identified with at least two different peptides match criteria, 67 were identified by SDS-PAGE-RPLC-MS/MS method and 44 were identified by the SCX-RPLC/MS/MS method. Forty-three of these proteins were identified by both methods. The proteins that were identified by the SCX-RPLC/MS/MS method are relatively highly abundant proteins in the lens fibers. Only one low abundant protein was identified by the SCX-RPLC/MS/MS method but not by the SDS-PAGE-RPLC-MS/MS method. In contrast, 23 low abundant proteins were identified by the SDS-PAGE-RPLC-MS/MS method but not by the SCX-RPLC/MS/MS method. These data indicate that the SDS-PAGE-RPLC-MS/MS method works better than the SCX-RPLC/MS/MS method for identifying low abundant lens fiber proteins. One of the reasons for the relative high sensitivity of the SDS-PAGE-RPLC-MS/MS method is that highly abundant crystallins were partially separated from the low abundant proteins by SDS-PAGE before in-gel tryptic digestion. This reduced the possibility of high abundant peptides masking low abundant peptides during the LC-MS/MS analysis.

In addition, another 50 proteins were identified based on a single positive peptide that was detected at least twice per run (Table 2 in Appendix 1). Although the lower coverage rate reduced the confidence of the identities of these proteins, these data provide valuable clues for future studies to investigate the functions of these proteins in the lens.

Reproducibility of the two protein identification methods in different experiments

Based on at least two different peptide match criteria, proteins identified by the SCX-RPLC/MS/MS method had an overlapping rate of 55% in the three parallel experiments. The reproducibility of the proteins identified by the SDS-PAGE-RPLC-MS/MS method was above 91%. The difference in protein detections between different experiments was due to the absence of some low abundant proteins in some experiments. This occurred because the low abundant peptides have little opportunity to be selected as precursor ions compared to highly abundant ones during the data-dependent acquisition [8]. The frequency for peptides derived from some low abundant proteins to be picked and identified could be much less than one per run. Thus, some low abundant proteins were not identified in every experiment. Because the majority of high abundant proteins in lens fibers are crystallins and the molecular weights of these proteins are 20-30 kDa, SDS-PAGE can effectively separate these high abundant crystallins from other proteins. Thus, the SDS-PAGE-RPLC-MS/MS method has higher sensitivity and reproducibility than those of the SCX-RPLC/MS/MS method for identifying low abundant proteins in lens fibers.

Analysis of the identified proteins

The distribution of theoretical molecular weights of the identified proteins was analyzed, and the results are shown in Figure 1B. Most of these proteins have theoretical molecular weights between 20 kDa and 90 kDa and pI values between 4 and 10. Two proteins had theoretical molecular weights less than 20 kDa.

Many of the identified proteins were shown as either posttranslational modification or degradation products. For example, α-, β-, and γ-crystallins as well as phakinin, filensin, retinal dehydrogenase, and glyceraldehyde-3-phosphate dehydrogenase were identified in the seventh slice of the SDS-PAGE gel where the apparent molecular weights were less than 20 kDa (Figure 1). However, these proteins have molecular weights of 20-74 kDa in their intact forms. These data indicate that there were significant fragmentations of these proteins in the normal lenses. All of the three classes of crystallins were also identified in the first and second slices of the SDS-PAGE gel where the apparent molecular weights were above 50 kDa. Because the monomers of these proteins have molecular weights less than 30 kDa, the slower migration of these proteins in SDS-PAGE suggests they are cross-linked products.

The identified proteins were categorized based on their subcellular localizations. For the known proteins, their subcellular localizations were assigned according to the Swiss-Prot databases. For all the others, localizations were predicted based on their similarities to known proteins using the PSORT II program. Among the identified proteins, 60 are cytoplasmic or cytoplasmic membrane proteins (88%) and eight are nuclear proteins (12%). The detection of nuclear proteins in the differentiated lens fibers may reflect the presence of some nuclei in the outer cortex or the residues of nuclear proteins that were not degraded during lens differentiation.

Discussion

The lens fiber is a terminally differentiated tissue. The fully differentiated fibers are devoid of most cellular organelles such as nuclei, mitochondria, endoplasmic reticulum, and Golgi apparatus. Whereas most epithelial proteins were removed during the differentiation process, a larger quantity of lens fiber specific proteins such as crystallins were synthesized. The extreme high concentrations of crystallins in lens fibers make it challenging to determine composition and activities of other proteins. Therefore, other than the highly abundant crystallins, the proteome of the lens fibers remains largely unknown. In this study, we compared two LC-MS/MS-based methods for identifying proteins in human lens fibers. We found that the SDS-PAGE-RPLC-MS/MS method is more sensitive than the classic SCX-RPLC/MS/MS method [9,10] for identifying low abundant proteins in lens fibers. The number of proteins identified by the SDS-PAGE-RPLC-MS/MS method was significantly more than that identified by the SCX-RPLC/MS/MS method (67 proteins versus 44 proteins). The highly abundant proteins in lens fibers were identified by both methods, and the majority of the proteins that were only identified by the SDS-PAGE-RPLC-MS/MS method are the low abundant proteins. One reason for the difference is that some of the low abundant proteins were separated from the highly abundant crystallins by SDS-PAGE. This separation reduced the suppression of the data-dependent acquisition of the low abundant peptides by high abundant peptides during the LC-MS/MS analysis. The SDS-PAGE-RPLC-MS/MS method takes advantage of the fact that subunits of all crystallins in the lens have molecular weights of 20-30 kDa and they can be separated from most of the low abundant proteins in the lens fibers by SDS-PAGE. Therefore, the SDS-PAGE-RPLC-MS/MS method is more suitable than the classic shot-gun method for identifying low abundant proteins in lens fibers although this method is a time-consuming and labor-consuming procedure.

This study also advanced our knowledge about the proteome of human lens fibers. A total of 68 proteins were identified from lens fibers with a stringent criterion (at least two different peptide match). Another 50 proteins were identified with a less stringent criterion (at least two hits of a single-peptide match). This work greatly expanded the proteome database of human lens fibers because only 22 proteins were identified as lens proteins in the Swiss-Prot databases in previous human lens proteome work [2,11].

The high concentration and proper arrangement of crystallins in the lens fibers play an important role in maintaining the transparency of the eye lens [12]. Mutation or modification of these proteins may contribute to cataract formation [1]. Furthermore, the turnover rates of lens proteins are very low. Many posttranslational modifications that occurred during differentiation and maturation of lens fibers may remain in the lens. Posttranslational modifications also occur during aging and cataractogenesis. Thus, detection of such modifications of lens protein could likely reveal the molecular mechanisms for cataract development [2]. Phosphorylation and deamidation are the most common modification of lens proteins [13,14]. COOH-terminal truncation of α-crystallins and NH₂-terminal truncation of β-crystallins could also lead to cataract formation [15,16]. The partial degradation of α- and β-crystallins and the increased acidity of γ-crystallins have been documented by two-dimensional electrophoresis proteome maps of the mouse lens [17]. Although the SDS-PAGE-RPLC-MS/MS method is not suitable for detecting modifications that cause changes in isoelectric points, it can readily detect fragmentation and cross-links. We identified the degradation (fragmentation) products of many lens proteins. These include α-, β-, and γ-crystallins as well as phakinin, filensin, retinal dehydrogenase, and glyceraldehyde-3-phosphate dehydrogenase. This study also detected cross-linked products of all three classes of crystallins. Since the lens fibers were obtained from clear lenses, these data indicate that these modifications occurred during either differentiation or the aging process.

In addition to well-characterized lens fiber specific and cytoskeleton proteins such as α-, β-, and γ-crystallins, phakinin, filensin, actin, and tubulin, many enzymes involved in sugar metabolism are among the highly abundant proteins in lens fibers. These enzymes include glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase, pyruvate kinase, fructose-bisphosphate aldolase A, sorbitol dehydrogenase, phosphoglycerate mutase, aldose reductase, malate dehydrogenase, α-enolase, lactate dehydrogenase A, and glycogen phosphorylase. The presence of these enzymes in lens fibers is consistent with the active sugar metabolism in this tissue. The active sugar metabolism in lens fibers not only provides the energy but also plays a role in maintaining the redox status in the lens fibers.

Retinal dehydrogenase, which catalyzes the conversion from retinaldehyde to retinoic acid, is another high abundant protein in the lens fibers. Retinoic acid plays an important role in regulating cell growth and gene expression. The high concentration of retinal dehydrogenase in lens fibers suggests that the lens may have an active retinoid metabolism. However, it remains to be determined how retinoic acid executes its regulating functions in lens fibers.

The 14-3-3 family of proteins was originally identified as an abundant brain protein. In mammals, the 14-3-3 protein family comprises seven homologous isoforms. The 14-3-3 family members are expressed in all eukaryotes, and their ability to bind other proteins is their crucial characteristic. The 14-3-3 proteins interact with several signaling and proto-oncogene proteins by their discrete phosphoserine/threonine-binding motifs. To date, more than 300 binding partners have been identified, and most of which are phosphoproteins. It has become clear that 14-3-3 proteins are involved in the regulation of most cellular processes including several metabolic pathways, redox-regulation, transcription, RNA processing, protein synthesis, protein folding and degradation, cell cycle, cytoskeletal organization, and cellular trafficking. The detection of several isoforms of 14-3-3 proteins in the lens fibers suggests that 14-3-3 proteins-regulated signaling pathways also exist in differentiated lens fibers. Further characterization of the 14-3-3 proteins in lens fibers will be essential for elucidating the signaling pathways that are regulated by these proteins.

In summary, this study compared two different methods for studying the proteome of human lens fibers and established the preliminary proteome database for the matured lens fibers. This information could be a valuable reference for future studies to investigate aging-related and cataract-related alterations of lens proteins. The SDS-PAGE-RPLC-MS/MS method is not only useful for identifying proteins in lens fibers; it may also be used for determining the proteome of other types of terminal differentiated tissues such as muscles and red blood cells.

Acknowledgements

This work was supported in part by grants from the Science and Technology Foundation of the Department of Education, Heilong Jiang Province (No:11521162); Foundation for Key Problems in Science and Technology, Heilong Jiang province (No:GB03C602-1); and Foundation for the Innovation of Graduate Students, Heilong Jiang Province (No:SCX2005038).

References

1. MacCoss MJ, McDonald WH, Saraf A, Sadygov R, Clark JM, Tasto JJ, Gould KL, Wolters D, Washburn M, Weiss A, Clark JI, Yates JR 3rd. Shotgun identification of protein modifications from protein complexes and lens tissue. Proc Natl Acad Sci U S A 2002; 99:7900-5.

2. Paron I, D'Elia A, D'Ambrosio C, Scaloni A, D'Aurizio F, Prescott A, Damante G, Tell G. A proteomic approach to identify early molecular targets of oxidative stress in human epithelial lens cells. Biochem J 2004; 378:929-37.

3. Klose J. Large-gel 2-D electrophoresis. Methods Mol Biol 1999; 112:147-72.

4. Wolters DA, Washburn MP, Yates JR 3rd. An automated multidimensional protein identification technology for shotgun proteomics. Anal Chem 2001; 73:5683-90.

5. Kersey PJ, Duarte J, Williams A, Karavidopoulou Y, Birney E, Apweiler R. The International Protein Index: an integrated database for proteomics experiments. Proteomics 2004; 4:1985-8.

6. Sun W, Li F, Wang J, Zheng D, Gao Y. AMASS: software for automatically validating the quality of MS/MS spectrum from SEQUEST results. Mol Cell Proteomics 2004; 3:1194-9.

7. Li F, Sun W, Gao Y, Wang J. RScore: a peptide randomicity score for evaluating tandem mass spectra. Rapid Commun Mass Spectrom 2004; 18:1655-9.

8. Liu H, Sadygov RG, Yates JR 3rd. A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal Chem 2004; 76:4193-201.

9. Link AJ, Eng J, Schieltz DM, Carmack E, Mize GJ, Morris DR, Garvik BM, Yates JR 3rd. Direct analysis of protein complexes using mass spectrometry. Nat Biotechnol 1999; 17:676-82.

10. Washburn MP, Wolters D, Yates JR 3rd. Large-scale analysis of the yeast proteome by multidimensional protein identification technology. Nat Biotechnol 2001; 19:242-7.

11. Lampi KJ, Ma Z, Shih M, Shearer TR, Smith JB, Smith DL, David LL. Sequence analysis of betaA3, betaB3, and betaA4 crystallins completes the identification of the major proteins in young human lens. J Biol Chem 1997; 272:2268-75.

12. Benedek G. Why the eye lens is transparent. Nature 1983; 302:383-4.

13. Lampi KJ, Amyx KK, Ahmann P, Steel EA. Deamidation in human lens betaB2-crystallin destabilizes the dimer. Biochemistry 2006; 45:3146-53.

14. Zhang Z, Smith DL, Smith JB. Human beta-crystallins modified by backbone cleavage, deamidation and oxidation are prone to associate. Exp Eye Res 2003; 77:259-72.

15. Kamei A, Takamura S, Nagai M, Takeuchi N. Phosphoproteome analysis of hereditary cataractous rat lens alpha-crystallin. Biol Pharm Bull 2004; 27:1923-31.

16. Lampi KJ, Shih M, Ueda Y, Shearer TR, David LL. Lens proteomics: analysis of rat crystallin sequences and two-dimensional electrophoresis map. Invest Ophthalmol Vis Sci 2002; 43:216-24.

17. Ueda Y, Duncan MK, David LL. Lens proteomics: the accumulation of crystallin modifications in the mouse lens with age. Invest Ophthalmol Vis Sci 2002; 43:205-15.