Ω-Crystallin of the Scallop Lens

While many of the diverse crystallins of the transparent lens of vertebrates are related or identical to metabolic enzymes, much less is known about the lens crystallins of invertebrates. Here we investigate the complex eye of scallops. Electron microscopic inspection revealed that the anterior, single layered corneal epithelium overlying the cellular lens contains a regular array of microvilli that we propose might contribute to its optical properties. The sole crystallin of the scallop eye lens was found to be homologous to Ω-crystallin, a minor crystallin in cephalopods related to aldehyde dehydrogenase (ALDH) class 1/2. Scallop Ω-crystallin (officially designated ALDH1A9) is 55–56% identical to its cephalopod homologues, while it is 67 and 64% identical to human ALDH 2 and 1, respectively, and 61% identical to retinaldehyde dehydrogenase/η-crystallin of elephant shrews. Like other enzyme-crystallins, scallop Ω-crystallin appears to be present in low amounts in non-ocular tissues. Within the scallop eye, immunofluorescence tests indicated that Ω-crystallin expression is confined to the lens and cornea. Although it has conserved the critical residues required for activity in other ALDHs and appears by homology modeling to have a structure very similar to human ALDH2, scallop Ω-crystallin was enzymatically inactive with diverse substrates and did not bind NAD or NADP. In contrast to mammalian ALDH1 and -2 and other cephalopod Ω-crystallins, which are tetrameric proteins, scallop Ω-crystallin is a dimeric protein. Thus, ALDH is the most diverse lens enzyme-crystallin identified so far, having been used as a lens crystallin in at least two classes of molluscs as well as elephant shrews.

The crystallins comprise approximately 90% of the watersoluble proteins of the cellular lens of vertebrates and are critical for the optical properties of this transparent ocular tissue. Despite their specialized task for vision, the crystallins are a surprisingly diverse group of multifunctional proteins (1)(2)(3). The ␣ (4, 5) and ␤/␥ (6) crystallins are present in the lenses of all vertebrates, while the various enzyme-crystallins are found selectively in discrete species (taxon-specific) (1,(7)(8)(9). The taxon-specific crystallins are related or identical to metabolic enzymes. Both the ubiquitous and taxon-specific crystallins are present at lower concentrations outside of the lens, where they have non-refractive roles.
In many cases, the non-refractive functions of the lens crystallins protect against physiological stress. The most striking example of this is ␣B-crystallin, which is a small heat shock protein (10,11). ␣B-crystallin is constitutively expressed in many tissues (12,13), as well as being stress-inducible and overexpressed in many neurological diseases (5,13,14). The abundant ␣Aand ␣B-crystallins not only influence the optical properties of the vertebrate lens but also act as molecular chaperones to protect against protein aggregation during aging (15). We have called this dual use of a single protein, gene sharing (7,16). The evolutionary strategy of gene sharing and the use of enzyme-crystallins extend to invertebrate lenses (17). The first described example of an invertebrate enzymecrystallin is the glutathione S-transferase-related S-crystallins of cephalopods (17)(18)(19)(20).
Aldehyde dehydrogenase (ALDH) 1 is the only known taxonspecific crystallin that occurs in both vertebrates and invertebrates (17). ⍀-Crystallin is an aldehyde dehydrogenase-derived crystallin in the ocular lens of cephalopods (squid and octopus). In general, it is a minor crystallin of these invertebrate lenses, but is more prevalent in octopus than squid (21). The same protein, however, is the sole crystallin in the transparent lens of the light organ of the squid, Euprymna scolopes, where it is called L-crystallin (22). ⍀/L-crystallin is equally related to cytoplasmic ALDH1 and mitochondrial ALDH2 of mammals as judged by amino acid sequence. Tests for enzyme activity using the most common substrates for the mammalian proteins have proved negative (21,22). In vertebrates, an ALDH1 orthologue with retinaldehyde dehydrogenase activity has been recruited as a lens crystallin in elephant shrews, where it is called -crystallin (23,24). Elephant shrews have another ALDH1 gene called ALDH1-nl (for non-lens), which is expressed in the liver (24).
Although ALDH is not especially abundant in other vertebrate lenses, there are unexpectedly high levels of ALDH3 in the transparent corneal epithelial cells of mammals (25)(26)(27)(28)(29). This abundant corneal protein was originally called BCP 54 (for bovine corneal protein with a molecular mass of 54 kDa) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF148508.
§ To whom correspondence should be addressed: Laboratory of Molecular and Developmental Biology, National Eye Institute, National Institutes of Health, Bldg. 6 (30,31). ALDH1 is also found at high concentrations in the transparent stromal keratocytes of the rabbit cornea (32). The surprisingly high levels of ALDH, as well as of several other enzymes, has suggested the possibility that these proteins play a structural role in cornea as do the enzyme-crystallins in the lens (33).
In the present investigation we have examined the protein composition of the ocular lens of the scallop. Scallops have numerous complex eyes protruding from the mantle lining the outer edge of each shell. The highly developed scallop eyes have a single-layered corneal epithelium covering the cellular lens (34 -36). These eyes have a double retina, and an argenteum, which reflects focussed images on the distal photoreceptor cells. The distal retina reacts to negative light stimuli generated by the images, while the proximal retina reacts to positive light stimuli (37). We show here by cDNA cloning that the predominant water-soluble protein of the cellular lens of the scallop belongs to the ALDH1/2 family, as does ⍀-crystallin of cephalopods. Our immunofluorescence data also suggest that ⍀-crystallin is present in the scallop cornea.

MATERIALS AND METHODS
Biological Materials-The adult sea scallops (Placopecten magellanicus) and bay scallops (Argopecten irradians) were obtained from the Marine Biological Laboratory, Woods Hole, MA. Juvenile bay scallops were obtained from Mook Sea Farms, Inc., Walpole, ME.
Protein Analyses-Cutting the adductor muscle with scissors separated the two shells of fresh scallops. The eyes were removed from the mantle along the edges of the separated shells with surgical scissors, and the lenses were dissected from the eyes under the microscope. Homogenates were made in 20 mM Tris-HCl (pH 7.9) and 0.1 M NaCl, and centrifuged at 10,000 ϫ g for 10 min. The supernatant fractions were taken as the water-soluble protein. These were subjected to electrophoresis in an SDS-12.5% polyacrylamide gel and stained with Coomassie Blue. The tryptic peptides derived from the major band of protein were microsequenced (Harvard Microchemical Facility, Cambridge, MA).
Native lens proteins were examined by Superose HR-12 and HR-6 column chromatography in phosphate-buffered saline (PBS). Amino acid sequences were analyzed and the evolutionary tree was made by using the Phylip program (38) by means of the values calculated by the Clustral W program (39).
The SegMod algorithm (40) implemented in the program GeneMine was used to build a homology model of scallop ALDH, based on the crystal structure of human mitochondrial ALDH2 (41) and the considerable sequence homology of the two proteins. The program CHARMM (42) was used to compare the model and template structures and to calculate the fractional accessible surface area (fASA) of all residues in the model. The fASA for each residue was obtained as the ratio of its accessible surface area (43) in the modeled protein structure to that in an extended Gly-X-Gly tripeptide. A probe radius of 1.4 Å was used.
Western Immunoblots and Immunofluorescence-The ALDH2 antibody was a generous gift of Dr. Henry Weiner (Biochemistry Department, Purdue University, West Lafayette, IN). It was a polyclonal antibody, prepared in rabbits against Escherichia coli recombinantly expressed human liver ALDH2. The anti-⍀/L-crystallin polyclonal antibody was a gift of Dr. Margaret J. McFall-Ngai (Pacific Biomedical Research Center, Kewalo Marine Laboratory, University of Hawaii, Honolulu, HI). It was made in rabbits against purified L-crystallin from the lens of the light organ of the squid, E. scolopes. For Western immunoblots, the proteins were transferred from the gel onto filters; the latter were developed using Pierce immunopure ABC staining kit (Pierce, Rockford, IL). Alternatively, Western blotting was performed by the enhanced chemiluminescence procedure using the Amersham Pharmacia Biotech ECL RPN 2106 Kit.
For immunofluorescence, segments of the mantle containing 3-5 eyes were cut free from freshly opened scallops. The tissue samples were immersed for 6 h in 4% paraformaldehyde in PBS and washed for 24 h in PBS. The eyes were subsequently removed from the mantle, embedded in Tissue-Tek O.C.T. compound (Miles Laboratories, Inc., Elkhart, IN) and 15-20-m sections were cut on a cryostat. The sections were preincubated with SuperBlock (Pierce) for 30 min and incubated overnight at room temperature with anti-⍀/L-crystallin at a dilution of 1:200 for bay scallop eyes and 1:500 for sea scallop eyes. After 3 washes in PBS, biotinylated secondary antibodies (Vector Laboratories, Burlingame, CA) were added for 1 h. The slides were washed again, covered with streptavidin-fluorescein isothiocyanate (Vector Laboratories) for 1 h, mounted in Dako fluorescent mounting medium (Dako, Carpinteria, CA), and viewed with a Zeiss (Oberkochen, Germany) microscope. Control experiments were performed using either PBS or preimmune serum instead of the primary antibody.
Northern Blot-Total RNA from adult scallop tissues was extracted with NucleoSpin RNA II Kit (CLONTECH, Palo Alto, CA). The RNA species were separated by electrophoresis on a 1% agarose, 3% formaldehyde gel and were blotted onto nylon filters (Hybond-N). The blot was hybridized with a 677-bp probe that contains a region of cDNA from nucleotides 51 to 728l. This was obtained by digesting pGEM-T-ALD5Ј plasmid with ApaI and PstI. The hybridization was carried out at 65°C for 1 h in QuikHyb solution (Stratagene, Jolla, CA). The blot was washed once with 2 ϫ SSC ϩ 1% SDS at room temperature for 15 min, once with 1 ϫ SSC ϩ 1% SDS at room temperature for 15 min, and twice with 0.5 ϫ SSC ϩ 1% SDS at 65°C for 15 min. The hybridized blot was exposed to x-ray film (Kodak) at Ϫ80°C with two intensifying screens overnight for ⍀-crystallin and for 6 h for actin. After the removal of the first probe, the filter was hybridized under the same conditions with a probe against P. magellanicus actin (GenBank accession number U55046). The actin probe was obtained by RT-PCR and contained the region from nucleotide 305 to 624.
RNA Isolation and RT-PCR-RNA was isolated from adult scallop eyes and from juvenile animals using the guanidinium isothiocyanate method with an RNA isolation kit (Stratagene, La Jolla, CA). In order to reduce a brown contaminating pigment that inhibited the reverse transcriptase reaction, the following modification was necessary. After ethanol precipitation, the RNA pellet was resuspended in proteinase K digestion buffer (20 mM Tris-HCl, 20 mM EDTA, 0.5% SDS, 2 mg/ml proteinase K), incubated for 1 h at 42°C and re-extracted with phenol/ chloroform. For RT-PCR analysis, 1 g of total RNA was reverse transcribed using random oligodeoxynucleotide hexamers as primers and Superscript II (Life Technologies), followed by PCR using degenerate primers encoding conserved regions of ALDH proteins: forward primer AA(AG)CCNGCNGA(AG)CA(AG)ACNCC (corresponding to amino acids KPAEQTP) and reverse primer GGNCC(AG)AA(TGA)AT(CT)TC-(CT)TC(TC)TT (corresponding to amino acids KEEIFGP). PCR products were cloned into the T-vector and sequenced. The three scallop ALDH cDNA sequences were identified among the PCR products and the corresponding clones designated pALD-eye (for ALDH-crystallin), pALD1, and pALD3.
cDNA Library Construction-The scallop eye-specific cDNA library in the phage vector was constructed using the SMART technology, which favors the production of full-length clones, essentially as recommended by the manufacturer (CLONTECH, Palo, CA). Briefly, 2 g of the total eye RNA were used for the reverse transcription using Superscript II enzyme and primer ATTCTAGAGGCCGAGGCGGCCGACAT-G(T)30(AGC)N. The reverse transcription was carried out in the presence of the SMART oligonucleotide AAGCAGTGGTATCAACGCAGAG-TGGCCATTATGGCCGGG. Double stranded cDNA was cut with SfiI and cloned into SfilI cut TriplEx2 vector. Approximately 1.5 ϫ 10 7 primary phage clones were obtained. The library was amplified by plating the primary library on 15, 20 ϫ 20-cm LB agar plates; the phage were harvested by washing with 100 mM NaCl, 10 mM MgSO 4 , 35 mM Tris-HCl (pH 7.5), and 0.01% gelatin.
ALDH Cloning and Sequencing-Full-length ALDH-crystallin cDNA was obtained by screening approximately 5 ϫ 10 6 phage clones of the amplified eye library with the 650-bp insert from pALD-eye. Three of the positive phage cDNA inserts were subcloned into the plasmid vector and sequenced on both strands using an automated DNA sequencer ALF (Amersham Pharmacia Biotech) and primer walking strategy. The resulting DNA sequence has been deposited into the GenBank (accession number AF148508).
The deduced amino acid sequence was compared with those contained in the GenBank data base by using the TFASTA program from the Genetics Computer Group (GCG) package (44). The alignment of the sequences was performed with the Pileup Program of GCG (45).
Protein Expression and Purification-The restriction sites BclI and EcoRI were added to the 5Ј and 3Ј ends of ALDH coding sequence, respectively, using primers GCAGTTGATCACATATGAGTACGC-CTATCAAAAAC (5Ј BclI) and GGAATTCGTTATGTTTGGTAAG-GCACTTGCG (3Ј EcoRI). The modified ALDH cDNA was then introduced into E. coli expression vector pETH2␣ carrying the N-terminal 6 ϫ histidine tag. The restriction sites NdeI and EcoRI were added to the 5Ј and 3Ј ends of ALDH coding sequence, respectively, using primers GCAGTTGATCACATATGAGTACGCCTATCAAAAAC (5Ј NdeI) and GGAATTCGACTTTGGTTGGAGTTTTGAT (3Ј EcoRI) in order to introduce the cDNA into the E. coli expression vector pETH2␣HIS carrying the C-terminal 6 ϫ histidine tag. The recombinant proteins were expressed in the E. coli strain BL21(DE3, pLysS) (46) and purified on Ni-NTA-agarose (Qiagen, Valencia, CA) under denaturation conditions in the presence of 10 mM ␤-mercaptoethanol. The proteins were renatured by dialysis against a stepwise urea gradient to lower the chance of irreversible precipitation of the recombinant proteins. Final dialysis buffer was either 20 mM Tris-HCl (pH 7.0), 100 mM KCl, or 50 mM phosphate (pH 8.0), and 100 mM NaCl containing 0.1% Triton X-100 or 0.1% Tween 20. All dialysis buffers contained 1 mM dithiothreitol.
Microscopy-Light and electron microscopy were performed on eyes that were removed and fixed by immersion at room temperature for at least 24 h. For light microscopy, the eyes were fixed in 4% paraformaldehyde in 50 mM Na-potassium phosphate buffer (pH 7.2) with 8% sucrose, embedded in glycol methacrylate, sectioned longitudinally along the central axis at a thickness of 1 to 2 m, and stained with toluidine blue, hematoxylin, and eosin, or by the periodic-acid Schiff reaction. For electron microscopy the eyes were fixed in 2.5% glutaraldehyde in 50 mM Na-cacodylate buffer (pH 7.2) with 6% sucrose and embedded in epoxy resin. Ultra-thin sections were stained with uranyl acetate and lead citrate, micrographs were taken with a JEM-100CX electron microscope (JEOL USA Inc., Peabody, MA) and several low magnification montages were prepared to preserve orientation and compare with micrographs taken by light microscopy while obtaining higher resolution.
ALDH Activity Assays-ALDH activity was assayed by the increase in the absorbance at 340 nm caused by the reduction of NADϩ to NADH, according to the procedure of Bostian and Betts (47). Baker yeast ALDH was obtained commercially (Sigma).

RESULTS
Scallop Eyes-The cornea, comprising a single layer of epithelial cells, cellular lens, and double-layered retina can be seen in the micrographs of transected eyes of juvenile bay (Fig.  1, A and B) and adult sea (Fig. 1C) scallops. The cornea is contiguous with the pigmented epithelial cells extending along both sides of the eye. The electron micrograph in Fig. 1B shows lens, corneal, and pigmented cells of the juvenile bay scallop eye from the region demarcated by the bracket in Fig. 1A. The lens cells of the juvenile eye are nucleated. In contrast to the organelle-rich cytoplasm of the corneal cells, the cytoplasm of the lens cells lacks organelles and appears homogeneous, as is typical of lens cells of other species, from jellyfish (48) to humans (49). Regularly arranged microvilli are present along the anterior surface of the corneal epithelial cells (Fig. 1B).
The Lens Crystallin of Scallops Is an ALDH Family Member-The soluble proteins of the whole eye, excised lens, and other scallop tissues were analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 2). The Coomassie Blue-stained left-hand panel of Fig. 2 shows that there were relatively few watersoluble proteins in the whole eye extracts derived from the bay  7), mantle (lane 8) and adductor muscle (data not shown) were very heterogeneous in size. There was relatively little staining at the region in the gel corresponding to a molecular mass of 50 kDa in the lanes used for these non-ocular tissues.
Since only the 50-kDa protein was detectable in the isolated lens, we performed Western blots to test whether it was related to the similar sized ALDH/⍀-crystallin of cephalopod (squid and octopus) lenses, inasmuch as both scallops and cephalopods are molluscs. Cephalopod ⍀-crystallin is closely related to mammalian ALDH1 and -2 (17,21). The right-hand panel of Fig. 2 shows that the 50-kDa protein of the eyes (lanes 3 and 4) and excised lens (lane 2) reacted strongly by Western immunoblotting with a polyclonal anti-ALDH2 antibody made against recombinant human ALDH2 (generously provided by Dr. Henry Weiner, Purdue University, West Lafayette, IN). Note that the intensity of the band obtained in the Western blot using only 0.05 g of protein from the isolated lens (lane 2) was similar to that obtained with 5 g of the total eye proteins (lanes 3 and 4). Commercially obtained Baker yeast ALDH reacted with this antibody as well (lane 1). The 50-kDa eye protein of the bay scallop also reacted with a polyclonal antibody made against purified ⍀/L-crystallin of the octopus (E. scolopes) light organ lens (22)  The Western blots in the right-hand panel of Fig. 2 showed that a 50-kDa protein, as well as varying amounts of more rapidly migrating polypeptides, from the digestive gland (lane 6), stomach (lane 7), and mantle (lane 8) also reacted with the anti-ALDH2 antibody. Although immunoreactive 50-kDa proteins were present in these non-ocular tissues, it should be noted that there was 100 -200 times more protein analyzed from the non-ocular tissues than from the eye. In contrast to the other tissues, the gill (lane 5) had neither a 50-kDa polypeptide nor smaller polypeptides that reacted with the human anti-ALDH2 antibody. Without additional data it is uncertain whether the larger reactive proteins in the nonocular extracts were cross-reacting proteins or ALDH-related proteins that were modified or were not reduced into subunits under the present conditions. We next investigated the native size of the major watersoluble protein of the isolated bay scallop lens by Superose HR-12 column chromatography (Fig. 3). The results gave a single peak of protein with a molecular mass of approximately 100 kDa. SDS-polyacrylamide gel electrophoresis of the 100-kDa protein after reduction with dithiothreitol gave a single band of protein with a molecular mass of approximately 50-kDa protein as expected (data not shown). The isolated 100-kDa protein re-chromatographed on a Superose HR-6 column as a 100-kDa protein (data not shown). No magnesium was present in the elution buffer since this ion is known to reduce tetrameric horse liver ALDH to dimers (50). A similar result was obtained with the sea scallop eye proteins (data not shown). Finally, commercially obtained yeast ALDH and recombinant human ALDH2 (generously provided by Dr. Henry Weiner, Purdue University, West Lafayette, IN)) chromatographed as a 200-kDa protein on the Superose HR-6 and HR-12 columns (data not shown).
The 50-kDa protein band of the sea scallop eye was isolated from the SDS-polyacrylamide gel and three tryptic peptides were microsequenced. These peptide sequences were homologous to ALDHs in the data base and are identified in the deduced protein sequence presented in Fig. 4 (see below). Together, these data indicate that the scallop lens crystallin is a dimer comprising two 50-kDa subunits that are related to the ALDH1/2 family of proteins.
cDNA Cloning of the Scallop ALDH/Crystallin-Degenerate PCR primers were designed against conserved ALDH amino acid sequences (KPAEQTP and DEEIFGP) and used for RT-PCR on sea scallop eye RNA (see "Materials and Methods"). A 650-bp partial cDNA clone was obtained that was unequivocally assigned to the 50-kDa protein since it encoded one of the tryptic peptides from that protein identified above by microsequencing (HGNVGYFVQPTVFS(?)V(?)EDM). This partial cDNA clone was used to screen a phage cDNA library made from total RNA isolated from purified eyes of sea scallops. Approximately 5 ϫ 10 6 phage clones were screened resulting in many positively hybridizing clones. Three clones were isolated and sequenced. A 2486-bp cDNA encoding a 53.6-kDa protein with the 3 tryptic peptides identified above from the ϳ50-kDa protein isolated from the SDS-polyacrylamide gel was obtained (Fig. 4, peptides underlined). A polyadenylation signal (boxed in Fig. 4) preceded the poly(A) tail. The 5Ј 47 nucleotides shown in Fig. 4 were obtained by 5Ј-rapid amplification of cDNA ends (data not shown) and were not in the cloned cDNA. Two closely related partial sequences were also obtained by RT-PCR cloning from total RNA derived from juvenile bay scallops (data not shown).
Sequence Comparisons of the Scallop ALDH/Crystallin with Other ALDHs-The scallop ALDH/crystallin sequence was aligned with ALDH/⍀-crystallins of the octopus and squid lens (21), with ALDH1/-crystallin of the elephant shrew lens (24), and cytoplasmic ALDH1, mitochondrial ALDH2, and ALDH3 of the human (Fig. 5). Included in this alignment was 2-hydroxymuconic semialdehyde dehydrogenase, a microbial ALDH which clusters with the ALDH1/2 class of proteins (51, 52). 2-Hydroxymuconic semialdehyde dehydrogenase was included since it is a dimer like the scallop ALDH/crystallin, unlike the other tetrameric native ALDH1/2 proteins.
The alignments show the homology among these family members. The amino acid sequence identities between the scallop ALDH/crystallin and the other proteins are 67% for ALDH2, 64% for ALDH1, 61% for ALDH1/-crystallin of the elephant shrew, 56 and 55% for ALDH/⍀-crystallin of the octopus and squid, respectively, 38% for 2-hydroxymuconic semialdehyde dehydrogenase, and 30% for ALDH3. We conclude that the scallop crystallin is orthologous to ⍀-crystallin of cephalopods and belongs to the ALDH1/2 class of proteins (52,53). The sea scallop ⍀-crystallin sequence was compared with all other recorded ALDHs by the Human Gene Nomenclature Committee and officially called ALDH1A9 (54).
We have denoted several amino acids of interest in Fig. 5 (see the figure legend for index numbers that correspond to the comprehensive analysis performed by Perozich et al. (52)). Arginine 101 (denoted by # in Fig. 5) is conserved in all eukaryotic ALDHs examined (52), including scallop ⍀-crystallin, except for squid ⍀-crystallin, where it has a leucine (21). The boxed amino acids denoted 1-4 are also conserved in scallop ⍀-crystallin; these residues are conserved in all 145 ALDHs that have been compared to date (52). Scallop ⍀-crystallin has also conserved two critical residues (boxed and indicated with * in Fig. 5) that are required for enzymatic activity of mammalian ALDHs. These are cysteine 331, which acts as a nucleophile, and glutamate 297, which serves as a general base that activates the essential cysteine (55)(56)(57). Glutamate 523 (indicated by q in Fig.  5), required for full enzymatic activity of mammalian ALDH2 (58,59), is conserved in scallop ⍀-crystallin. Finally, a series of 14 arrows denote amino acids that are known to be involved in either hydrogen bonding or van der Waal interactions with NAD(P) (60). Glycine 254, valine 278, and especially, valine 282 (marked by arrows and ૺ in Fig. 5) may form an altered NAD pocket in scallop ⍀-crystallin limiting the binding of NAD required for enzymatic activity (see below and Fig. 8). It is not surprising that there are other numerous differences among these ALDHs since only about 10% of all the residues are conserved in more than 80% of the different family members (52). Finally, we constructed an evolutionary tree using the Phylip program by means of the values calculated by the Clustral W program (Fig. 6). Consistent with the sequence comparisons above, scallop ⍀-crystallin belongs to the ALDH1/2 branch of proteins (52,53). Scallop ⍀-crystallin is closer to human ALDH2 than to ALDH/-crystallin of the elephant shrew or even to ⍀-crystallin of the octopus despite the fact that both the scallop and octopus proteins are mollusc lens crystallins.
Structural Comparison of Scallop ⍀-Crystallin and Human ALDH2-We constructed a homology model of scallop ⍀-crystallin based on the x-ray structure of human mitochondrial ALDH2 (pdblcw3.ent) (41). The model is shown in Fig. 7. For 331 (67.3%) of the 492 residues in the scallop ALDH model, the nearest residue in the template structure is an identical amino acid. The root mean square displacement between the corresponding C-␣ atoms in these identical residues is 0.54 Å. The homology model of scallop ALDH permits an analysis of the sequence identity between the scallop and human proteins in three dimensions. The fASA of residues was used to categorize each as belonging to either the core or the surface of the protein.  enhanced sequence identity in the protein core relative to the identity at the surface suggest that the model is reasonable and that the structure of scallop ⍀-crystallin is indeed very similar to that of human ALDH2.
Tests for ALDH Enzymatic Activity-Tests for ALDH enzyme activity were conducted with different substrates. Both the 10,000 ϫ g supernatant fractions and whole homogenates of the sea scallop lens were examined. Assays were performed at 25°C in the presence of 0.006% bovine serum albumin. The following substrates were tested (data not shown): acetaldehyde, glyceraldehyde, benzaldehyde, p-nitrobenzaldehyde, onitrobenzaldehyde, trans-S-cinnamaldehyde, ␤-phenylcinnamaldehyde, and retinal. Since ALDH/-crystallin has been reported to have retinaldehyde dehydrogenase activity (24), retinal was tested extensively as a possible substrate for sea scallop ⍀-crystallin. The enzyme activity with retinal was tested at 37°C as well as at 25°C. In some instances, retinal was mixed either with the retinal-binding protein, interphotoreceptor-binding protein, or with ␥-crystallin. In addition, enzyme activities were tested with NADP as well as with NAD. No significant activity was observed in any case. Parallel controls using Baker yeast ALDH always gave positive results (data not shown). Finally, Baker yeast ALDH was added to the scallop eye extracts in order to test for the possible presence of an inhibitor to enzymic activity using acetaldehye as a substrate under our reaction conditions. In all cases, the yeast ALDH functioned normally when added to the scallop eye extract (data not shown).
Tests for NAD and NADP Binding to Scallop ⍀-Crystallin-We next tested the binding of NAD or NADP to scallop ⍀-crystallin by measuring the quenching of tryptophan fluorescence as described elsewhere (61). The experiments were done with ⍀-crystallin purified from sea scallop eyes by two cycles of HR-12 chromatography. The tests were conducted at 25°C. Fluorescence excitation was at 295 nm and the emission was monitored at 340 nm. Unexpectedly, no binding was found for either NAD or NADP (data not shown). Moreover, no binding was detected to either a Cibacron blue-agarose or NADPagarose affinity column (data not shown); both typically bind dehydrogenases that bind NAD or NADP. These data indicate that scallop ⍀-crystallin either does not bind or interacts very weakly with NAD or NADP, consistent with its enzymatic inactivity in our tests.
We attempted to account for the absence of NAD(P) binding to scallop ⍀-crystallin by comparing residues at the positions known to interact with the co-factor in other ALDHs (60). These NAD interacting residues, indicated by vertical arrows in Fig. 5, are highly conserved between scallop ⍀-crystallin and the other ALDHs. However, an interesting difference is seen at position 282 of scallop ⍀-crystallin. The enzymatically inactive ⍀-crystallins from scallop (this study), octopus, and squid (21,22) have a valine at position 282, but all other sequences listed in Fig. 5 have an isoleucine at this position. Thus, isoleucine 246 appears to correlate with enzymatic activity.
Our homology model shows that valine 282 of scallop ⍀-crystallin is one of only a few residues in the NAD binding pocket that is not conserved relative to human ALDH2 (Fig. 8). The packing of the adenine moiety against the larger isoleucine side chain, when present at position 282, may significantly stabilize NAD binding to the enzyme. An exception to this pattern is rat ALDH3 (60), which possesses the smaller valine residue at position 282, but this enzyme also has a valine in place of glycine at position 254 in Fig. 5. Thus, a valine in each of these positions on opposite sides of the adenine may produce a binding pocket of comparable volume to the pocket formed by a glycine on one side and an isoleucine on the other (see Fig. 8). Future characterization of NAD binding to the V282I recombinant mutant of scallop ⍀-crystallin might be helpful in assessing the significance of these packing interactions between the adenine and neighboring side chains.
Tissue Distribution of Scallop ⍀-Crystallin-Northern blots were performed using total RNAs from different sea scallop tissues using the 5Ј half of the ⍀-crystallin cDNA as a probe FIG. 7. Homology model of scallop ⍀-crystallin, colored by sequence identity relative to the human ALDH2 template structure. The 67.3% of residues having identical nearest neighbors in the template structure are colored red; others are colored blue. The side chain of scallop Cys 331 in the center of the figure is represented as ball-and-stick. The enhanced conservation in the protein core is evident. This figure was prepared using the programs MolScript (80) and Raster (81). Blue, surface residues; red, interior residues; green, scallop Cys 331 (active site in homologous proteins). (Fig. 9). Strong hybridization was obtained to RNA approximately 2.5 kilobases in length from the eye. This RNA size is consistent with the 2.4-kilobase pair cDNA (Fig. 4). Much weaker hybridization was observed with 2.5-kilobase RNAs from the gill and mantle (Fig. 9, upper panel). Approximately similar amounts of hybridization were evident with RNAs from the stomach and digestive gland in other tests (data not shown). Considerable RNA degradation was evident with the stomach and digestive gland samples (data not shown). Hybridization to an actin cDNA probe was used for normalization in order to estimate the relative amount of ⍀-crystallin RNA present in the different tissues ( Fig. 9, lower panel). Scans of these gels (data not shown) indicated that the relative expression of ⍀-crystallin RNA was at least 10 times greater in the eye than in the gill or mantle. Similar calculations were not attempted for the digestive gland or stomach due to the considerable amount of RNA degradation that was consistently obtained for these tissues.
Spatial Distribution of ⍀-Crystallin within the Scallop Eye-The anti-⍀/L-crystallin antibody of the squid light organ lens was used in immunofluorescence tests to determine the spatial distribution of ⍀-crystallin within the scallop eye. This antibody reacted with similar specificity to the ϳ50-kDa scallop ⍀-crystallin in Western immunoblots of the water-soluble proteins of the bay scallop eye (data not shown) as did the antihuman ALDH2 antibody in Fig. 2. As expected, the lenses of both the bay (Fig. 10A) and sea (Fig. 10B) scallop reacted strongly with the antibody. The corneal epithelial cells of both species also reacted to a lesser extent with the antibody, as did the acellular layer between the cornea and lens. Since ⍀-crystallin is not a secreted protein, we assume that the extracellular immunofluorescence represents nonspecific trapping of the antibody. However, the control sea scallop section treated with preimmune serum did not immunofluoresce in the eye section except for a slightly positive signal in the retina (Fig. 10C). It is likely that the low reaction in the retina is mostly background in Fig. 10, A and B, although we cannot rule out the presence of some of this protein or a cross-reacting ALDH in the retina. Together, these tests show that ⍀-crystallin exists at high concentration in the scallop lens and suggest its presence at a lower concentration in the scallop cornea.

DISCUSSION
The present investigation confirms that scallops have complex eyes with cellular lenses and a single-layered corneal epithelium containing surface microvilli (34 -36). Corneal microvilli (62) associated with the glycocalyx (63) and immunoglobulin A (64), have also been demonstrated by scanning electron microscopy in mammals and may be one mechanism of protecting against corneal infection (65,66). We propose that the microvilli in scallops may contribute to corneal transparency. Regular surface projections that are smaller than half the wavelength of incident light have a uniform refractive index that is the average of the bumps (cytoplasm) and the medium (water). Moreover, if the microvilli are more separated at their tips than at their base, a refractive index gradient reducing reflection would be generated between the water and the corneal cellular cytoplasm (67,68). In the present study the width of the corneal microvilli in the electron micrographs is approximately 100 nm, well below the wavelength of blue light (400 nm).
The optical properties of transparent cellular lenses depend upon the accumulation of diverse, water-soluble crystallins. Vertebrates crystallin may differ among species and are often related or identical to metabolic enzymes (1)(2)(3)(7)(8)(9). The discovery that the major cephalopod lens crystallins are related to glutathione S-transferase established that enzyme-crystallins are also present in invertebrates (17)(18)(19)(20). Here we show that the only crystallin in the cellular lens of the scallop eye is ⍀-crystallin, an ALDH family member. ⍀-Crystallin is also present in cephalopods, where it is a minor crystallin in eye lenses (21,69). However, ⍀-crystallin (called L-crystallin) is FIG. 8. Homology model of scallop ⍀-crystallin, colored by sequence identity to the human ALDH2 template structure as in Fig. 7, except for glycine 257 (yellow) and valine 282 (light green). NAD, drawn as ball-and-stick, was docked by copying its coordinates from the superimposed ALDH2 complex. The NAD pocket is highly conserved relative to human ALDH2. Sequence differences (colored blue or light green) near the adenine moiety, i.e. valine 282, valine 278, and their neighbors, may disrupt local packing and NAD binding (see text). the major crystallin in the muscle-derived lens of the light organ of the squid, E. scolopes (22). Thus, ALDH/⍀-crystallin is present throughout the molluscs, while the glutathione S-transferase/S-crystallins appear to be confined to cephalopods. Indeed, since -crystallin is an ALDH1 in the elephant shrew lens (23,24), ALDH is the only protein family known to have been recruited as a lens crystallin in both vertebrates and invertebrates.
The enzymatic inactivity of scallop ⍀-crystallin in the present investigation is presumably due to its lack of or very weak binding to NAD(P). The absence of binding to NAD(P) is surprising due to the high sequence similarity of scallop ⍀-crystallin to mammalian ALDH1/2, including conservation of the residues known to interact with this moiety (60), and its similar structure to human ALDH2 by homology modeling. Further studies are necessary to test the possibility that valine at position 282 of scallop ⍀-crystallin, rather than the bulkier isoleucine at the equivalent position in mammalian ALDH1 and -2, disrupts the tight binding of NAD(P) to scallop ⍀-crystallin. It is of interest that valine substitutes for isoleucine at this position in the enzymatically inactive ⍀-crystallins in the octopus eye lens (21) and the squid light organ lens (22).
The only ALDH/lens crystallin that has been shown to have enzymatic activity is -crystallin, which is a retinaldehyde dehydrogenase (24). It is noteworthy that -crystallin has isoleucine at the position homologous to valine 282 of scallop ⍀-crystallin. In view of the similarity between scallop ⍀-crystallin and mammalian ALDHs and the conservation of residues associated with the active site, it remains possible that scallop ⍀-crystallin has an enzymatic activity requiring a substrate, a co-factor, or a reaction condition that we have not tested. It is also possible that ⍀-crystallin plays a functional role in the scallop lens by binding thyroxine analogs (71)(72)(73) or androgen (74). It is noteworthy in this connection that homologies have been noted between peptides of a human kidney thyroid hormone-binding protein (71) and -crystallin, an enzymatically inactive kangaroo lens enzyme-crystallin homologous to bacterial ornithine cyclodeaminase (75). The binding of 3,4-didehydroretinol by -crystallin, a crystallin related to cellular retinolbinding protein type 1, to create an ultraviolet filter in lenses of diurnal geckos is a striking case of the importance of a crystallin binding a ligand for a biological function (76).
Scallop ⍀-crystallin, as other enzyme-crystallins, is preferentially expressed in the lens but appears to be also present in other tissues at lower levels, as judged by Western and Northern blots. Another possibility, however, is that there are two closely related genes, one encoding the inactive lens ⍀-crystallin and the other an enzymatically active cross-reactive ALDH that is expressed outside of the lens. Gene duplication has resulted in two enzymatically active ALDHs in the case for -crystallin in elephant shrews, where one of the genes is specialized for eye expression and the other is expressed in non-ocular tissues (24). A similar situation exists for the small heat shock protein, ␣B-crystallin, and ␣A-crystallin, which is specialized for lens expression (4,5). Current Southern blot analyses of the scallop ⍀-crystallin gene are consistent with (but do not prove) the presence of a single gene that hybridizes with the ⍀-crystallin cDNA. 2 If the identical scallop ⍀-crystallin protein is indeed expressed outside of the lens, the question of its function remains as an intriguing puzzle for future study.
Finally, while it is commonly accepted that the optical properties of the transparent lens are dependent upon crystallins, the concept of corneal crystallins is still under investigation (32,33). The unexpected relative abundance of a few watersoluble proteins, often enzymes, in corneal cells supports the possibility that their optical properties depend, at least partially, on the accumulation of crystallins, as in the lens. Moreover, these abundant corneal proteins differ among species as do the lens crystallins. For example, ALDH3 (25,26,28,30,31), transketolase (77), and isocitrate dehydrogenase (78)  epithelial cells (29), gelsolin accounts for approximately half of the water-soluble protein in zebrafish corneal epithelial cells (79), and glutathione S-transferase/S-crystallins are the major proteins in both the lens and cornea of squids (29). In the present immunofluorescence tests ⍀-crystallin is preferentially expressed in the lens and to a lesser extent in the cornea of the scallop eye. Since immunofluorescence data are not quantitative and lack the reliability of direct examination, which was hampered by the limited amount of corneal tissue, further tests are necessary to establish the extent and authenticity of ⍀-crystallin expression in the scallop cornea. Nonetheless, our results raise the possibility that ⍀-crystallin is preferentially expressed in the cornea as well as the lens in scallops.