Comparative proteomics analysis of degenerative eye lenses of nocturnal rice eel and catfish as compared to diurnal zebrafish

Materials

All zebrafish (Danio rerio) were about 6 months old and supplied from local aquarium stores. Rice eel (Monopterus albus) and walking catfish (Clarias batrachus) were also obtained from a local fish market in Kaohsiung, Taiwan. Rice eel, with the small eyes covered by a layer of skin (Figure 1), can burrow into the humid soil bottom for a period of days in the dry season without exposure to daylight.

Extraction of soluble proteins from fish lenses

Eyeballs were kept and stored at −80 °C in a freezer before dissection. The lenses were removed from the adult fish eyeballs, homogenized, and suspended in the buffer of 20 mM Tris-HCl, pH 6.8 for the extraction of total lens. All animal protocols were approved by the Animal Care and Use Committee at Kaohsiung Medical University. Euthanasia of zebrafish, rice eel and catfish was carried out by treatment with tricaine (Sigma-Aldrich, St. Louis, MO) before protein extraction.

One-dimensional and two-dimensional sodium dodecyl sulfate–polyacrylamide gel electrophoresis

The 1D sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) was prepared in 5% and 15% (gel percentage) for stacking and resolving gels (Hoefer SEM 260 system, Amersham Pharmacia, Piscataway, NJ). Electrophoresis was run for 4.5 h at 80 V after loading about 20 μg protein in each sample lane. The gels were stained with Coomassie blue G-250. For 2D gel, the dried immobilized pH gradient (IPG) strips (Immobiline^TM Drystrip pH 3–10, 13 cm; GE Healthcare, Milwaukee, WI) were rehydrated with 250 μl reswelling solution (7 M urea, 4% NP-40, 2 M thiourea, 1 mg/l bromophenol blue) and 3 ml cover fluid (mineral oil) in a reswelling tray overnight (12–16 h). Lens extracts containing about 200 μg protein each in 0.5% IPG buffer and 1% DTT were then performed by isoelectric focusing (IEF) on IPG strips for the first-dimensional electrophoresis. IEF was performed initially at 300 V for 3 h, then 1,000 V for 3 h, and finally at 8,000 V for 6 h until 40 kV-h was reached at 20 °C on an electrophoresis apparatus (Ettan IPGphor 3). The IPGs of 1D gel analysis were then equilibrated for 20 min in 5 ml of equilibration solution I (0.01% bromophenol blue, 6 M urea, 2% SDS, 30% glycerol, 50 mM Tris-Cl [pH 8.8], and 30 mM DTT), followed by 20 min in 5 mL of equilibration solution II (0.01% bromophenol blue, 6 M urea, 2% SDS, 30% glycerol, 50 mM Tris-Cl [pH 8.8], and 135 mM iodoacetamide). Finally, IPG strips were placed on the top of 15% resolving gels sealed with 0.5% agarose for electrophoresis at the second-dimensional SDS–PAGE. The 2-D gel electrophoresis was carried out at a constant current of 22.5 mA for 6 h, and stained with Coomassie blue G-250.

In-gel digestion

The Coomassie-blue stained 1D SDS–PAGE and 2D gels in triplicate for each sample were prepared for in-gel digestion as described previously [22]; the protein spots (2D gels) or bands (1D gels) were first excised and placed in 0.65 ml siliconized tubes. One hundred microliter buffers (25 mM ammonium bicarbonate [NH₄HCO₃]/50% acetonitrile [ACN]) were added to sample tubes and vortexed for 10 min. The supernatants of vortexed samples were discarded, and then dried with speed-vacuum centrifuge to complete dryness. Digestion was started by adding enzyme solutions containing 20 ng/μl of modified sequence-grade trypsin (Promega, Madison, WI) in 25 mM NH₄HCO₃ (pH 8) buffer to cover sliced and dried gel pieces, rehydrated on ice for 10 min, and then incubated at 37 °C overnight (about 16 h). The digested aqueous solutions were transferred into 0.65 mL siliconized tubes and extracted twice with 30 μl of 50% ACN/5% formic acid (FA). The extracted enzyme-digested samples were dried by centrifugation in a vacuum centrifuge for the removal of ACN and FA, redissolved in 10 μl of 5% ACN/0.1% FA and subjected to analysis by LC-MS/MS.

Liquid chromatography coupled tandem mass spectrometry analysis from one-dimensional or two-dimensional gels

Electrospray MS was performed using (A) a Waters-Micromass electrospray ionization quadrupole-time of flight (Waters, Manchester, UK) [23,24], or (B) an HCT Ultra ETDIIIon-Trap Mass Spectrometer (Bruker Daltonics, Bremen, Germany) interfaced with DIONEX UltiMate^TM 3000 Nano and Cap system (Dionex, Sunnyvale, CA) capillary high-performance liquid chromatography system. A 100 × 0.075 mm C18 column (3.5 μm particle diameter) with mobile phases of A (0.1% FA in water) and B (0.1% FA in ACN) were used. The peptides were eluted at a flow rate of 0.4 μL/min with an ACN gradient, which consisted of 5%–10% B in 5 min, 10%–50% B in 25 min, and 50%–95% B in 4 min.

The spectra for the eluting fractions were acquired as successive sets of scan modes. The MS scan determines the intensity of the ions in the m/z range of 200 to 2,000, and a specific ion was selected for a tandem MS/MS scan. The former examined the charge number of the selected ion and the latter acquired the spectrum (Collision-Induced Dissociation [CID] spectrum or MS/MS spectrum) for the fragment ions derived by collision-induced dissociation. The centroid MS/MS data of enzyme-digested fragments from protein bands (1D gel) or spots (2D gel) were obtained by using MassLynx 4.0 software (Waters, Manchester, UK), and then submitted them to the bioinformatics search program MASCOT, which was set up to search through the NCBI databases based on zebrafish (D. rerio) with the following parameters: a mass tolerance of 0.2 Da for precursor and fragment ions; one missed cleavage allowed for trypsin digestion; carbamidomethyl cysteine as fixed modification; and oxidized methionine as optional modification. The resultant identification had a statistically significant (p≤0.05) peptide score (based on MS/MS spectra).

Isolation of lens proteins by size-exclusion chromatography

The eye lenses of zebrafish, rice eel, and walking catfish were homogenized in 100 mM NH₄HCO₃ buffer (pH 7.4), containing 5 mM protease inhibitor (Complete Mini Protease Inhibitor Cocktail; Roche Molecular Biochemicals, Indianapolis, IN). The homogenized lens solutions were centrifuged at 27,000g for 1 h at 4 °C. The concentration of the supernatant was determined using the Bio-rad protein assay kit. One hundred microliters of supernatant containing about 200 μg protein was applied to high-performance liquid chromatography TSK-GEL G4000SW_XL (Tosoh Co., Tokyo, Japan) [25,26]. The crystallin fractions were eluted at a flow rate of 0.75 ml/min and monitored for absorbance at ultraviolet 280 nm. Molecular weight standards were used for calibration of molecular sizes of eluted proteins versus their respective elution volume for the chromatography. Eluted crystallin fractions were further analyzed by SDS–PAGE and protein bands identified by silver staining.

Quantitation by ImageMaster analysis

2D gels were digitally imaged with Image Scanner (Amersham Pharmacia) and the quantities of protein spots on 2D gels were analyzed by ImageMaster^TM 2D Platinum Software (Version 5.0). 2D gels were analyzed for each fish lens sample and the relative abundance of selective protein spots was normalized to the total intensity of the entire gel. The proportion of each protein spot was calculated as the mean of determinations for three separate lens samples.

Chaperone-like activity assay

Chaperone-like activities of total lens extracts from the three piscine species were analyzed by measuring the capability to prevent the aggregation of alcohol dehydrogenase denatured by heating treatment or the reduction of disulfide bonds in insulin, as described previously [27,28]. In brief, chaperone-like activity was studied based on the dithiothreitol-induced insulin reductive unfolding and chaperone-assisted refolding. The assay was carried out at 25 °C by recording the turbidity change of OD_360nm within 42 min upon the initiation of dithiothreitol-induced insulin aggregation or until the turbidity curve reaches a plateau.

Shotgun strategy for gel-free proteomics analysis

The detailed protocol was modified from that of a previous report [19] and is included in Appendix 1.

Comparison of morphology and the lens sizes of the three piscine lenses

All of the three piscine species are freshwater fishes, with rice eel and catfish generally living in the muddy fields during daytime, and preying food at night. Regarding the body size, the zebrafish is smaller than the rice eel and catfish. For the lens sizes, the catfish is bigger than the other two (Figure 1). Although the rice eel is larger than the zebrafish in body size, its lens is much smaller than that of the zebrafish due to its degenerative eye lenses. Ratios of lens size versus body length for the three species we studied are about 0.1 cm/4.5 cm for zebrafish, 0.09 cm/50 cm for rice eel, and 0.35 cm/55 cm for catfish. The ratios appeared to follow the order of zebrafish > catfish > rice eel. It is to be noted that catfish, rice eel, and zebrafish of the teleostean class were found to possess lenses that are harder than those of evolutionarily higher classes of vertebrates such as mammals. Even though the increased hardness of lenses was found to be reflected by the respective decrease in water content inside human lenses [31], biochemical factors influencing the hardness of animal eye lenses are not clear and warrant further study. Catfish lenses were also found to be harder than those of rice eel and zebrafish, with the latter being the softest among the three species.

Fractionation of lens proteins by size-exclusion chromatography

Figure 2A shows a typical elution pattern of lens extracts from three different species belonging to two nocturnal fish and one diurnal fish. Three well-defined peaks were obtained for the zebrafish lens, in contrast to four for the rice eel and five for the catfish. It is of interest that, in contrast to higher vertebrates, the first peak of the zebrafish elution pattern contains α-crystallin, as well as the aggregated form of β-crystallin (Figure 2B); this is also the case for the fractionation of catfish lenses, but is not common for higher classes of vertebrates, including reptiles, birds, and mammals [25,26,32,33]. It is also of interest that in contrast to zebrafish, catfish and rice eel contain very little α-crystallin, as revealed by the absence of α-crystallin subunit bands of about 20 kDa in SDS–PAGE (Figure 2B). In Figure 2C,D we show a comparative separation of porcine lens proteins into five well-defined fractions of HMα, α, βH, βL and γ crystallin classes similar to our previous characterization of bovine lens extract [32]. The proportions for each piscine crystallin fraction (Figure 3) are calculated based on the area of chromatographic peaks of each crystallin class as compared with the well-defined fractions of HMα (6.6%), α (31.4%), βH (24.6%), βL (23.0%), and γ (14.3%) for mammalian lenses. One salient feature of the crystallin contents in bony fishes of the piscine class is that the γ-crystallin contents of fishes are much higher than those of mammalian species such as the porcine γ-crystallin reported in this study (Table 1 and Figure 3). Our superior and reproducible resolution of the crystallins allowed us to make a systematic comparison between the crystallins of different classes. The well-defined and characteristic distribution of subunit compositions for each isolated fraction (Figure 2A-D) justified the use of size-exclusion gel in the general characterization and classification of lens proteins from various vertebrate and invertebrate species [33].

Gel-based proteomic analyses of piscine lens extracts

The global protein-expression profiles of piscine lenses were analyzed using 1D (Figure 4) or 2D (Figure 5) gel electrophoresis. In this study, we first performed proteomics analysis on conventional 2D gel. The isoelectric point ( pI) range for the first-dimension isoelectric focusing (IEF) in 2D gel was 3–10, and the second dimension SDS–PAGE was run at 15% polyacrylamide gel (Figure 5). The proteomics analysis showed that more than half of lens proteins located on the weakly basic and low molecular weight regions (6.5 < pI <7.5 and 18,000 < molecular weight <32,000), corresponding mostly to β- and γ- crystallins. The proteomic analyses for zebrafish, rice eel, and catfish showed that we had positively identified about 49, 28, and 33 protein spots (No. 1–49, No. 1–28, No. 1–33 for lens proteins of zebrafish, rice eel, and catfish, respectively), as confirmed and verified by LC-MS/MS (Table 2).

Four protein spots in the 2D gel of zebrafish were identified as αA-crystallin (protein spots #1–3) and αB-crystallin (protein spot No. 4), respectively. We also detected 27 protein spots that belong to the class of β-crystallins (protein spots No. 5–31), and 12 protein spots that belong to the class of γ-crystallins (protein spots No. 32–43). It is noteworthy that no protein spots corresponding to α-crystallins for rice eel were detected. We found 14 protein spots that belong to the class of β-crystallins (protein spots No. 1–14), and 5 protein spots in the class of γ-crystallins (protein spots No. 15–19). On the other hand, two protein spots located on the acidic region on the 2D gel of nocturnal catfish were identified as αA-crystallin (protein spots No. 1–2). We also detected 21 protein spots that belong to the class of β-crystallins (protein spots No. 3–23), and 7 protein spots to the class of γ-crystallins (protein spots No. 24–30). In Table 2, we have also listed 6, 9, and 3 protein spots (denoted as others) as anomalous proteins shown in 2D gels of zebrafish, rice eel, and catfish, respectively, because their identities cannot be revealed through searching and comparison with those identified crystallins deposited in the database of zebrafish.

Comparison of gel-based one-dimensional or two-dimensional gel proteomics

In our proteomic study of porcine lens proteins [34], we encountered poor solubility of some proteins in pre-LC-MS/MS 2D gel separation. To improve the detection sensitivity for low-abundance proteins, fractionation of lens proteins was thus performed directly on the total extracts of three piscine lenses by 1D SDS–PAGE gels (Figure 4) instead of prerunning first-dimensional IEF in 2D gel without adding the strong protein denaturing and solubilizing agent of SDS (Figure 5). The proteins were separated into more than 10 different protein bands or zones from the total protein mixture each of three piscine lenses (Figure 4). In comparison with zebrafish lens-proteome, α-crystallin was found to be almost absent in the rice eel lens, similar to that observed in the 2D gel analysis described above. Loss of specific crystallins analogous to the missing α-crystallin from degenerative eyes for rice eel was also reported for lenses of some species under specific or pathological conditions, such as mutation-induced congenital cataract formation [35-37]. This pointed to the fact that differential crystallin expression under environmental stresses or pathological conditions may result in degenerative or deteriorating crystallin formation in the lenses of nocturnal rice eels.

It should be noted that 1D gel is methodologically less tedious and more time saving than 2D gels, being able to afford a respectable and extensive protein separation suitable for protein identification analysis after LC-MS/MS (Figure 4). The unambiguous identification of some major crystallin species of β- and γ-crystallin classes were confirmed and verified in addition to α-crystallins reported previously.

Although 2D gel electrophoresis coupled with tandem MS has been considered as the method of choice in conventional proteomics study [38], only up to about 2,000 individual polypeptide chains at most can be resolved on a single 2D gel [39,40]. The number of detected proteins is still relatively small as compared to the whole proteome corresponding to the human genome, which encodes about 20,000-30,000 proteins. The 2D gel analysis is especially under-representative of some special classes of proteins, such as low-abundance transcription factors and membrane proteins [40-42], because of the low solubility of these classes of proteins in the first dimensional IEF protein separation of 2D gel electrophoresis in the absence of protein-solubilizing SDS detergent.

Comparison of major lens-crystallin families of the three piscine species

The lens crystallins of vertebrates form a complex group of highly conserved structural proteins with distant evolutionary relationships [43]. To date, most physicochemical studies on the characterization of crystallins have placed more emphasis on species of higher vertebrates, with relatively fewer reports on lenses from the lower aquatic vertebrates, i.e., various classes of fishes [44]. The piscine lenses are usually spherical and hard as compared to the more flexible and soft lenses found in the avian class. The hardness of lenses has been shown to be related to the state and content of water, i.e., degree of hydration, in different lenses of vertebrate species [45]. The poor solubility and susceptibility of lens proteins from most fishes to denaturation have hampered detailed biochemical characterization of piscine crystallins under nondenaturing conditions [46,47].

In this report, we have adopted a systematic and general approach to isolate and characterize piscine crystallins of three different teleostean species by the proteomics approach to shed some light on the development and evolution of distinct crystallin families in piscine lenses. Herein, we focus on the analysis of the three prominent crystallin families, i.e., α-, β-, and γ-crystallins.

Quantitative analysis of α-crystallin

α-Crystallin constitutes a major class of lens proteins present in all vertebrate eye lenses [48,49]. Native α-crystallin from mammalian lenses is commonly isolated as a large water-soluble aggregate with a molecular mass of about 600–800 kDa. It consists of two homologous subunits, αA and αB, of about 55%–60% sequence identity, each with a molecular mass of 20 kDa and in a binding ratio (αA/αB) of about 3:1 for most mammalian lenses [50]. Recently, α-crystallin—and especially its αB subunit—has received a lot of interest and attention because it was shown to possess structural and functional similarities to small heat-shock proteins [51,52]. Moreover, in vitro studies of α-crystallin indicated a chaperone-like activity associated with this lens protein [53,54]. Fewer studies on α-crystallin of piscine lenses have been reported due to its relative lower abundance as compared with β and γ crystallins in fish lenses. Therefore, it is deemed important to reevaluate the proportion of each major crystallin family in fish lenses of these three teleostean species (Figure 3). As described above, we found three protein spots of αA and one spot of αB for zebrafish; two spots of αA for catfish; and the absence of αA and αB for rice eel (Table 2 and Figure 4, Figure 5). Protein spot #4 in 2D gel of zebrafish lens was identified through search in NCBI database and listed as α B chain without specifying its being α Ba or α Bb. Quantitation of protein spot #4 in 2D gel (Table 2) of zebrafish lens is therefore either αBa or αBb. We do not know why we detect only one αB chain. However, we did detect two αB chains by 1D gel and shotgun proteomics (see Tables in Supplemental Materials on 1D gel and shotgun proteomics).

It is to be noted that the presence of more than one spot per crystallin member of the α-, β-, or γ-crystallin family in 2-D gel usually reflects truncated crystallin fragments or posttranslational modifications. However, a recent genomic study [55] indicated the existence of two αB-crystallins in the zebrafish, one lens specific (αBa-crystallin) and one ubiquitous (αBb crystallin). Determining the function of αA- or αB-crystallin in the piscine [33] and amphibian [56] lenses has been difficult because pure αA- or αB-crystallin is difficult to obtain from the above-mentioned size-exclusion chromatography. α-Crystallin and aggregated β-crystallins are always eluted together in the void volume of high-molecular-weight fraction. This also makes the accurate estimation of percent abundance of α-crystallin in fish lenses more difficult than that for γ-crystallin (Figure 3). In the literature, the reported percent abundance for piscine lenses also varies. For instance, the percent abundances of α-crystallin estimated previously by conventional protein analysis [57] were higher than that (7.8%) of a recent report employing a proteomic approach coupled with the densitometry of Coomassie stained 2D gels [58].

In our previous characterization of catfish α-crystallin from the same walking catfish (C. batrachus) as the species used in this study [27], it was found that in contrast to mammalian lenses with a subunit association ratio (αA-crystallin/αB-crystallin) of about 3:1, α-crystallin from catfish lens showed a ratio of about 19:1. This could also account for the difficulty of detecting catfish αB-crystallin in the current proteomics analysis. On the other hand, the possibility that the concentration of α-crystallin is too low or posttranslationally modified in the degenerated eye lenses of rice eel cannot be excluded. We have therefore investigated the chaperone-like activity of α-crystallin based on total lens extracts of three fish lenses. It is as expected that the chaperone activity of zebrafish is higher than that of catfish, which in turn is also much higher than that of rice eel, which exhibits no detectable activity (unpublished results).

It has been reported that αA-crystallin of zebrafish is a chaperone protein that can keeps the γ-crystallin of mutant zebrafish lenses soluble [7]. In contrast, the evolutionary and functional basis for the lower abundance of αB crystallin and the apparent absence of αA and αB crystallins in the degenerative lenses of catfish and rice eel, respectively, are especially intriguing; this topic remains elusive and deserves a further detailed study in the future employing genomics approaches.

Quantitative analysis of β- and γ-crystallins

By proteomic analysis, the most abundant β- and γ-crystallins were found to comprise a diverse group of heterogeneous classes of βA1a, βA1–2, βA2a, βA2–2, βA4, βB1, βB2, and βB3 subunit crystallins, and γB, γD, γM2, γM3, γM5, γM7, γN-A, γN-B, γS1, and γS2 monomeric crystallins of the γ-crystallin class (Table 2). As we reported previously using a conventional protein chemistry methodology [33] and gene cloning [59], the class of carp γ-crystallin with high methionine content (γM-crystallin) was also found in most teleostean fishes, similar to those found for the diverse γ- and γS-crystallins of catfish [60,61] and γMs-crystallins of zebrafish [58], which is the most abundant crystallin family and comprises more than 30% of the lens.

In zebrafish, five βA-crystallins and three βB-crystallins comprised about 36% of total proteins in the zebrafish lens [58]. Many of the β-crystallins were extensively modified; for example, βB1 and βB2 were identified as 11 and 8 spots, respectively. Similar to the α-crystallin family, modified β-crystallins with different masses are probably due to truncations and other modifications with similar mass and the different isoelectric points may be due to phosphorylation or other modifications. In this pilot study, we did not pursue protein modifications of these nocturnal lenses due to scarcity of the lens samples for the top-down proteomic strategy [12]. There is always an issue for proteomics analysis of novel tissues lacking the complete genomic or proteomic databases such as rice eel and catfish regarding the possible false-negative results and discrepancies of protein components reported by different groups [58]. Therefore, for βB3 crystallin, which was found to be missing or reduced in rice eel or catfish, this may be due to the false negative identification (for rice eel) or reduced expression of existing crystallins with functional degeneration in adaptive dark environments (for catfish). However, we are confident that the positive identification of γ-B and γ-D is real and not artificially false identification, because they were revealed through widely accepted standard Mascot software analysis with high scores (Appendix 1). The analytical methodology for the estimation of the accuracy in peptides and their corresponding protein identifications made by MS/MS and database search has always been a big issue in proteomics analysis and identification. Other factors that may contribute to the ambiguity and uncertainty in protein identification could arise from the pre-MS/MS methodologies or protocols for sample treatment before protein separation, such as 1D gel, 2D gel, or shotgun proteomics strategies. Current state-of-the-art shotgun proteomics techniques may allow the sensitive identification of parent proteins from individual peptides in many complex protein mixtures of cell or tissue extracts. Specifically, shotgun proteomics is more sensitive than 2D gel electrophoresis for the separation and detection of proteins with low abundance.

The β- and γ-crystallins are evolutionarily related families of proteins that make up a large part of the refractive structure of the vertebrate eye lens [62]. Each family has a distinctive gene structure that reflects a history of successive gene duplications. Recently, Wistow et al. [63] made a survey of γ-crystallins expressed in mammal, reptile, bird, and fish species, which has resulted in the important discovery of γN-crystallin, an evolutionary bridge between the β and γ families. In all species examined, γN-crystallins have a hybrid gene structure, half β and half γ, leading to the supposition that they may be the “missing link” between β and γ crystallin lineages.

The γ-crystallin, unlike other lens crystallin, is monomeric in solution, and has the highest sulfhydryl content of all crystallins. X-ray crystallographic investigation by Summers et al. [64] has shown that the protein has a symmetric structure of two globular domains packed together with a single connection. It is suggested that the stability of γ-crystallin may be due to the interaction of polarizable amino acid groups and sulfur-containing residues such as methionine present in the core of each domain. Prediction of secondary and tertiary structure of carp high-methionine γM-crystallin by computer graphic simulation [65] has shown that carp γ-M1 crystallin comprises 22 methionine residues (12.4%), with 6 buried inside and 16 exposed on the surface. Carp γ-M2 crystallin with 24 methionine residues (14%) showed an essentially similar distribution pattern of methionine residues on the protein surface. Interestingly, most hydrophobic methionine residues are located on the protein surface with only a few buried inside the protein surface or in the interface between two motifs of each domain. The exposed hydrophobic and polarizable methionine cluster on the protein surface may have a bearing on the crystallin stability and dense packing in the piscine species, and probably also provides a malleable nonpolar surface for the interaction with other crystallin components for the maintenance of a clear and transparent lens.

There has been speculation [66] about the role of methionine in the regulation of the surface polarity of proteins as judged by the unique properties of the thioether sulfur atom in this generally hydrophobic amino acid. Incorporation of multiple methionine residues into carp γ-crystallins may therefore represent one efficient way of optimizing various crystallin interactions and recognitions in the intact lens. The existence of unusually high levels of methionine residues on the crystallin surface suggests that these hydrophobic amino acids with polarizable sulfur atoms might play a significant role in the recognition and interaction of crystallin molecules in the piscine lenses. Previous studies concerning the cold-adapted Antarctic toothfish [67,68] also described several methionine-rich γM-crystallin isoforms, of which the high methionine residues may have predisposed the toothfish lens to biochemically attenuate γ-crystallin hydrophobicity, thereby allowing for cold adaptation. Reduced structural constraints upon γM crystallins due to the presence of high methionine residues could have allowed for greater evolutionary plasticity, resulting in increased polydispersity of γ-crystallins contributing to the cold stability of the Antarctic toothfish lens. The complete lens proteome of zebrafish has indeed provided good model system to study investigations of vertebrate lens development, function, and diseases [58]. The similarity and differences of lens crystallins present in the three teleostean species studied herein will certainly provide some insights into the molecular basis for the structural and functional evolution of different crystallin families in fishes and other vertebrates. Future studies will be needed to further elucidate the posttranslational modification of these normal and degenerative piscine lens systems. The preliminary proteome map provided here lays a firm foundation for these investigations.

Conclusions

The eye lens is an extraordinary tissue in terms of its development and evolution. In cave or nocturnal animals [13,14], the eye is sometimes reduced or eliminated as a consequence of adaptation to life in visual darkness. How the cavefish and rice eel adapted to develop degenerative eye lenses under dim or totally dark environments has remained an interesting evolutionary issue since Darwin’s time. In The Origin of Species, Darwin found no reason to invoke natural selection to explain the loss of eyes in cave animals [17]. As a result, the evolutionary mechanisms responsible for eye degeneration in cave-adapted animals have remained elusive. Opposing hypotheses invoking neural mutation [69] or natural selection [70], each with certain genetic and developmental expectations, have been advanced to explain eye regression, although little or no experimental evidence has been provided to support or refute either hypothesis. In this study, we have performed a comparative proteomics analysis on eye lenses of nocturnal rice eel and catfish as compared to diurnal zebrafish. The results of this analysis form a molecular basis to investigate further the evolution of piscine lenses in the future. The total α-, β-, and γ-crystallins in the three fish species analyzed by current proteomics methodology clearly indicate the complexity and diversity of crystallin species present in the piscine class of vertebrates. It is noteworthy that the unexpected finding that α-crystallin is absent in the degenerative eye lenses of rice eel points to the fact that α-crystallin acting as a chaperone protein may be essential in diurnal vertebrate species to protect lens proteins from aggregation and to maintain functional transparency of the lens under varied environmental conditions. A detailed study on the correlation of the chaperone function and eye degeneration in rice eel certainly warrants future investigation.