Nội dung

cN-crystallin and the evolution of the bc-crystallin superfamily in vertebrates Graeme Wistow1, Keith Wyatt1, Larry David2, Chun Gao1, Orval Bateman3, Steven Bernstein4, Stanislav Tomarev1, Lorenzo Segovia5, Christine Slingsby3 and Thomas Vihtelic6 1 2 3 4 5 6 National Eye Institute, National Institutes of Health, Bethesda, MD, USA Oregon Health Sciences University, Portland, OR, USA Department of Crystallography, Birkbeck College, London, UK Department of Ophthalmology, University of Maryland School of Medicine, Baltimore, MD, USA IBT ⁄ UNAM, Col. Chamilpa, Cuernavaca, Morelos, Mexico Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA Keywords crystallin; eye; gene structure; intron loss; lens Correspondence G. Wistow, Section on Molecular Structure and Functional Genomics, National Eye Institute, Bg 7, Rm 201, National Institutes of Health, Bethesda, MD 20892-0703, USA Tel: +1 301 402 3452 Fax: +1 301 496 0078 E-mail: graeme@helix.nih.gov (Received 21 January 2005, revised 23 February 2005, accepted 8 March 2005) doi:10.1111/j.1742-4658.2005.04655.x The b and c crystallins are evolutionarily related families of proteins that make up a large part of the refractive structure of the vertebrate eye lens. Each family has a distinctive gene structure that reflects a history of successive gene duplications. A survey of c-crystallins expressed in mammal, reptile, bird and fish species (particularly in the zebrafish, Danio rerio) has led to the discovery of cN-crystallin, an evolutionary bridge between the b and c families. In all species examined, cN-crystallins have a hybrid gene structure, half b and half c, and thus appear to be the ‘missing link’ between the b and c crystallin lineages. Overall, there are four major classes of c-crystallin: the terrestrial group (including mammalian cA–F); the aquatic group (the fish cM-crystallins); the cS group; and the novel cN group. Like the evolutionarily ancient b-crystallins (but unlike the terrestrial cA–F and aquatic cM groups), both the cS and cN crystallins form distinct clades with members in fish, reptiles, birds and mammals. In rodents, cN is expressed in nuclear fibers of the lens and, perhaps hinting at an ancestral role for the c-crystallins, also in the retina. Although well conserved throughout vertebrate evolution, cN in primates has apparently undergone major changes and possible loss of functional expression. Much of the complexity and diversity of life arises from the multiplication and evolution of gene families, increasing the functional repertoire of the genome. By gene duplication, a single protein function (or set of functions) can be expanded into a broader set of more specialized functions. The c-crystallins are a gene family with a complex history in vertebrate evolution. They encode proteins that are highly abundant components of the eye lens but are also expressed at lower levels in other parts of the eye, perhaps with a stress-like role [1–6]. Together with the related b-crystallins, the c-crystallins belong to an ancient superfamily (known as the bc-crystallin superfamily) with members ranging from the prokaryotic sporulation protein, Protein S of Myxococcus xanthus [7], to AIM1, a protein implicated in the control of malignancy in melanoma in man [8,9]. In the vertebrate lens, the b and c crystallins together account for the majority of the soluble proteins (the other major family being the a-crystallins, members of the small heat-shock protein superfamily [10,11]). Although it has been suggested that proteins of this superfamily may have roles in maintenance of cellular architecture [8], little is known about their function. Like the c-crystallins, b-crystallins have also been detected in the retina [12], and both have been identified in Drusen bodies, which form with age in retinal pigment epithelium (RPE) [13]. Abbreviations EST, expressed sequence tag; RPE, retinal pigment epithelium. 2276 FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS G. Wistow et al. b and c crystallins are related in their highly symmetrical structures built from four characteristic bc-motifs (modified Greek keys) arranged as two similar domains [14]. The two families differ in that c-crystallins are monomers that lack lengthy N-terminal or C-terminal extensions, whereas b-crystallins have long N-terminal arms and form dimers and higher oligomers. The two families also differ in gene structure [3,10,15]. In b-crystallins, each protein structural motif is encoded in a separate exon, with other exons encoding the N-terminal arms. The same organization is seen in the AIM1 gene, and as such presumably represents the ancestral condition [8].The gene structures of c-crystallins are clearly related to those of b-crystallins and AIM1, with precisely conserved intron positions delineating protein domains; however, in the c-crystallins the introns that divide the sequences encoding the two motifs of each domain are missing. Thus a c-crystallin gene has ‘fused’ exons corresponding to each two-motif domain. The c-crystallins present a particularly interesting example of the dynamic evolution of a gene family. They play a key role in determining the optical properties of the lens, a tissue that is subjected to strong environmental selective pressures as species move from water to land, from dark to light, from the ground to the air, and the requirements for vision change accordingly. As it has adapted in different evolutionary lineages, the lens has changed its protein composition [3,16]. This has led to considerable variability in the content and sequence of c-crystallins in different vertebrates, which are abundant in species with hard lenses (such as fish and rodents) but at much lower levels or missing in other terrestrial species. This is in contrast with the b-crystallins which are well conserved and have clear orthologs in all vertebrate orders (see for examples [17–20]). In fish and amphibians, there are multiple, divergent c-crystallin genes that may exhibit only about 50% identity at the protein level [17,19,21]. This is similar to the level of divergence among the b-crystallins [18,19,22,23] and suggests a similar antiquity of these gene families in the vertebrate lens. In contrast, birds, with soft accommodating lenses, lack the embryonically expressed c-crystallins that in other vertebrates are major components of the developing lens, and have replaced them with the taxon-specific ‘enzyme crystallins’, d and e crystallin [16,24–26]. In placental mammals there is a closely linked cluster of six c-crystallin genes (cA–F) which are generally expressed in the embryo, and these show 77–97% identity at the protein level, implying a relatively recent FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS cN-crystallin origin for this family in this lineage. It has been suggested that, as in birds, c-crystallins may have been on their way to extinction in the ancestors of mammals but were perhaps ‘reinvented’ by successive duplication of a surviving gene as mammals adapted to principally nocturnal, burrowing habits before the extinction of the dinosaurs [3,16]. Indeed, as some mammals have now become diurnal, these genes may again be in a process of change or loss; in humans two of these genes (cE and cF) are pseudo and others (particularly cA) seem to be expressed at lower levels than in other mammals [15,27]. Mammals also express cS-crystallin, a divergent outlier of the family with a short N-terminal arm, which is the major c-crystallin expressed in the secondary fiber cells of the mature mammalian lens [27–29]. Little is known about the c-crystallins of marsupials or reptiles, but it is clear that this family has undergone considerable changes, particularly during mammalian evolution. These changes illustrate the way in which gene families may expand, contract and adapt. They may also help us understand the functions of the c-crystallin family in vision and elsewhere as most changes are presumably driven by specific adaptive requirements in the eyes of different vertebrates. Here we describe a survey of the evolutionary history of c-crystallins in vertebrates, including a large analysis of crystallin gene expression in zebrafish lens, and the discovery of a new member of the family, cNcrystallin, which seems to be an evolutionary bridge between the b and c families. Results c-Crystallin and b-crystallin sequences from several vertebrate species were cloned and sequenced in NEIBank genomics projects or predicted from bioinformatics analysis of genome sequences. The novel sequences are shown in Fig. 1, aligned using the clustalw algorithm, and their relatedness is illustrated in the phylogenetic tree in Fig. 2, drawn using the neighbor joining option in the program mega [30]. Some previously described sequences are also included to illustrate the overall distribution of the superfamily members expressed in vertebrate lenses. cDNA Libraries Approximately 1500 clones were sequenced from the un-normalized mouse whole eye ioip libraries and a further 1000 and 1300, respectively, from the two equalized libraries jajbjc and lglh. A total of 1000 clones (code designation mw) were sequenced from a cDNA library made from western grey kangaroo 2277 cN-crystallin 2278 G. Wistow et al. FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS G. Wistow et al. cN-crystallin Fig. 1. Protein sequences for representative members of the bc-crystallin superfamily. Sequences are derived from the work described here, with some examples of fish and amphibian sequences taken from GenBank. Sequence names beginning with ‘m’ are from mouse, ‘ig’ from iguana, ‘kan’ from kangaroo and ‘zf’ from zebrafish, while carp and chick sequences are so labeled. Sequences were aligned by CLUSTAL W. The positions of N-terminal and C-terminal arms, the four structural motifs (I–IV) and the connecting peptide between N-terminal and C-terminal domains are indicated below the alignment. Also shown are the approximate positions of the four b-strands of each motif (a–d), by analogy with known structures. Yellow highlights show the principal conserved positions of each motif essential for correct folding. Note that the long C-terminal arm of zfbB3 has been truncated to fit the page. kangB mgB 45 Terrestrial γ mgC 53 50 mgD 99 mgE 94 29 100 mgF γ−type gene mgA kangD 46 frgM1-1 100 frgM1-2 zfgM1 100 74 carpgM1 zfgM2a 92 91 zfgM2b 99 Aquatic γM zfgM2c 100 carpgM2 56 40 66 zfgM3 98 carpgM3 100 γ−type gene zfgM4 γ zfgM5 99 zfgM6 zfgM7 64 100 zfgMX 100 zfgSa zfgSaL 89 zfgSb 100 γs mgS 32 chickgS iggS 71 57 γ−type gene kangS 34 zfgSc zfgSd 72 zfgN1 90 γN zfgN2 mgN 100 βγ−hybrid iggN 69 chickgN 61 zfbgx Fig. 2. Phylogenetic tree of b and c crystallins in vertebrates. Calculated from the alignment in Fig. 1 and drawn using MEGA, by neighbor-joining with Poisson correction. Bootstrap values are indicted for each node. The major clades are identified. Cartoons of the exon structure of the motif-encoding regions of genes in each clade are shown. Red boxes show the typical c-crystallin exon encoding two motifs, and blue boxes show the single-motif exons of b-crystallin genes. FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS 99 zfbA1-1 mbA1 94 βA zfbA4 98 66 mbA4 64 zfbA2 β−type gene β mbA2 76 100 100 zfbB2 βB mbB2 zfbB3 100 100 mbB3 β−type gene 0.1 2279 cN-crystallin (Macropus fuliginosus) lens [31]. In a small pilot survey of expressed genes in a reptile lens,  1000 clones were sequenced from a PCR-derived library (code designation hm) made from the lenses of a single individual iguana. Almost 4000 clones from an adult zebrafish lens library (code designation nab) were sequenced. Further details of this and other zebrafish eye libraries will be described elsewhere. Vertebrate c-crystallins Among the collections of sequences obtained for vertebrate lens and eye proteins were a large set of clones for b and c crystallins. Even the small set of clones from iguana lens yielded complete coding sequences for two members of the c-crystallin family. One of these (GenBank accession number AY788911) was the iguana ortholog of cS-crystallin, with 178 amino-acid residues, including initiator methionine. Indeed, our analyses confirm that cS is well conserved and highly expressed throughout the vertebrates. The other iguana sequence (GenBank accession number AF445457) was a novel protein similar in size to cS (183 residues) and with an N-terminal arm of the same length. However, the new protein is clearly distinct from cS; its overall sequence identity to iguana cS is only 44% and, relative to cS, the protein has insertions (mainly of glycines), in the c–d loops of motifs I and III and in the connection between motifs III and IV (Fig. 1). This protein was given the name cN, for c-new. From the combined mouse whole eye data, complete sequences were obtained for cA–F and cS from C57Bl ⁄ 6 mice. In addition, an almost full-length clone was obtained for a protein very similar to the iguana lens cN. The complete coding sequence of mouse cN (GenBank accession number AF445456) was deduced from this clone, mouse genome sequence, from an apparently full-length expressed sequence tag (EST) in dbEST (CK795274), and from interspecies comparisons. Several other ESTs for both mouse and rat cN eye are also present in dbEST. The mouse gene is on chromosome 5 at about position 23.2 Mbp, cA–F are on chromosome 1, and cS is on chromosome 16 [2]. Three full-length c-crystallins were obtained from kangaroo lens. One was the ortholog of cS, identical in length and 72% identical in sequence with that of mouse (GenBank accession number AY898646). The other two were more similar to the cA–F crystallins, with no N-terminal arm. Based on their closest matches in blast searches, these were designated cB, with the longer connecting peptide and a length of 175 codons (accession number AY898644), and cD, with 174 codons (accession number AY898645), although a 2280 G. Wistow et al. more systematic nomenclature will probably be needed when all kangaroo c-crystallin sequences are known. In this small sample, no clones for an ortholog of cN were detected. From zebrafish, a total of 16 distinct c-crystallins were identified along with a b-like sequence (with a long N-terminal arm) that had some sequence similarity to c-crystallins and several b-crystallins (GenBank accession numbers AY738742–AY738756). Nine of the c-crystallins were generally similar to the cM-crystallins previously cloned from carp lenses [32] and were named accordingly (Figs 1 and 2). Two were named zfcM1 and zfcM3, and three sequences related to carp cM2, including one closely related pair, were named zfcM2a, zfcM2b and zfcM2c. The other cM-like sequences were named zfcM4–7. All of these sequences lack the N-terminal arm seen in cS, cN and b crystallins and are similar in size to the cA–F group of mammals. As shown in the phylogenetic tree (Fig. 2), the cM-crystallins form a distinct clade of ‘aquatic’ c-crystallins. An additional zebrafish c-crystallin was found to be generally similar to the cM class in size but is more divergent in sequence. As shown in Fig. 2, this protein does not group with either the terrestrial vertebrate c-crystallins or the aquatic cM class. Provisionally this has been named zfcMX. Mammalian species possess just one gene for cS-crystallin. However, the zebrafish lens has four proteins of the cS class. Although these proteins, named zfcSa–d, show clear sequence similarity to known cS-crystallins, they have considerable variability at the N-terminus. Most of the clones for zfcSa, the single most abundant species in the zebrafish lens cDNA library, lack an N-terminal arm altogether. However, 13% of the sequences for zfcSa (zfcSaL) revealed an alternative splice at the end of the first exon that added four codons to the coding sequence, enough to make an N-arm of the same length as in mammalian cS-crystallins. Like zfcSa, the closely related zfcSb also lacks an N-arm and so far there is no evidence of alternative splicing in this gene. A third member of the cS family in zebrafish, zfcSc has an N-arm that is longer than in other species (although it contains three methionines near the N-terminus that could potentially give rise to alternative translation products), and a fourth member, zfcSd, has a short arm of just a single residue. In addition to the cM and cS crystallins, two of the zebrafish c-crystallins, zfcN1 and zfcN2, are members of the cN family. These two proteins have N-terminal arms identical in length with those of mouse and iguana cN and they also exhibit the characteristic FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS G. Wistow et al. cN-crystallin insertions in the c–d loops of motifs I and III and in the link between motifs III and IV. As shown in the phylogenetic tree, these sequences group with other members of the cN class in a distinct clade that is essentially separate from both b and c crystallins. In the tree shown, the cN branch is weakly linked to the b-crystallin family, but overall the cN family is an intermediate in the wider bc superfamily. The b-crystallins are represented in the phylogenetic tree by five members cloned from the zebrafish lens library (zfbA1-1, zfbA2, zfbA4, zfbB2 and zfbB4) and their orthologs from mouse. As expected, each fish sequence is closely related to its mammalian ortholog, in marked contrast with the relationships among cA–F and cM crystallins. The designation of zfbA1-1 reflects the fact that the lens library also contains cDNAs for a second, and possibly a third, bA1-like protein which has not yet been completely characterized. One remaining zebrafish sequence from the lens library is an outlier of the b-crystallin family. Indeed, in simple blast comparisons, this sequence is slightly more closely related to cN-crystallins than to other crystallins, although it has a long N-terminal arm like a b-crystallin. In the phylogenetic tree it is placed as an early offshoot of the main b-crystallin lineages. For its currently ambiguous status, this has been named zfbcX. In current versions of the zebrafish genome, not all assembled regions have been assigned to specific chromosomes. However, there is evidence of some clustering of crystallin genes. zfcM1, zfcM2b, zfcM2c and zfcM6 are all located between positions 12.79 and 12.86 Mbp on chromosome 2, and zfcM3 and zfcM5 are both close to position 29.24 Mbp on chromosome 8. It has been known for a long time that bird lenses lack most c-crystallins, although there has been evidence for the presence of cS in its former guise as bS-crystallin [33–35]. Although no cDNAs are yet available, inspection of the chicken genome using blat reveals the presence of well-conserved genes for both cS and cN crystallins. The predicted sequences are presented in Fig. 1, and in the phylogenetic tree (Fig. 2) they group in their respective subfamilies. Chicken cN is located at about position 5.7 Mbp on chromosome 2, and cS is at about 9.3 Mbp on chromosome 9. The expression of these genes in the chicken remains to be examined in detail. gene in all these genomes, conserved across over 400 Myr of evolution, is its exon ⁄ intron structure (Fig. 3). The first half of the gene has the typical structure of a c-crystallin gene [3], with a short first exon encoding the start codon and the short N-terminal ‘arm’ similar to that of cS. A phase 0 intron separates that exon from a larger exon, exon 2, which encodes the first two structural motifs, and hence the N-terminal domain of the protein, just as in the genes for cS and cA–F. However, the second half of the gene has the structure of a b-crystallin gene, with two exons encoding the two motifs of the C-terminal domain. The crygn gene is thus a hybrid of b and c crystallin gene structures, apparently an intermediate in the evolution of the c-crystallins from the b-crystallins. This observation is concordant with the position of the cN family in the protein sequence phylogenetic tree, where it is an intermediate between the b and c crystallins. Novel gene structure of cN-crystallins Fig. 3. Correspondence of exon structure and protein structure for cN-crystallin. Red boxes show coding sequence, green boxes show untranslated regions. Protein structure is indicated by the stylized Greek key folding pattern of the bc motif. The four motifs are labeled I–IV, as are the corresponding exon sequences that encode those motifs. Genes for cN orthologs are present in mouse, rat, chicken, and zebrafish, and, as described below, orthologous genes are also present in the human and chimp genomes. The most striking feature of the crygn FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS cN in primates: a nonfunctional gene? A search of the human genome for an ortholog of cN located a highly conserved gene sequence on chromosome 7q36.1, and a very similar gene is present in the chimp genome, located in an unassembled portion of chromosome 6. This gene sequence contains a wellconserved coding sequence for the cN protein, with one notable exception. The stop codon in rodents is TAG, but in the human and chimp genes, this codon has a single base change to CAG (glutamine) (Fig. 4). In principle, this could allow the translation of a larger version of cN with an 11-kDa C-terminal extension rich in glycine and proline (not shown). However, no cDNA clones for a human cN transcript have emerged from any of the NEIBank analyses. In an attempt to 2281 G. Wistow et al. cN-crystallin in humans. At the very least, the human gene has clearly changed its expression and may indeed be heading for extinction, joining cE and cF [15]. Fig. 4. Variant splice forms of the human CRYGN gene. Exons are shown by boxes and labeled as in Fig. 3. For the full-length cDNA from human testis, red boxes show the coding sequence corresponding to the cN ORF; the blue box shows the coding sequence of the cryptic exon; green boxes show untranslated regions; red lines show the testis splice pattern. The change of the stop codon seen in other species to CAG causes an increase in the potential ORF of the exons encoding motif IV. Potential coding sequence from intron read-through seen in clones from human RPE is shown in orange, and the alternative splice of this variant is shown by the orange lines. identify human transcripts, PCR was performed on template made from human lens, retina, and RPE ⁄ choroid libraries. No products were obtained from lens or retina (a negative result in this procedure is not proof of absence). A product was obtained from RPE ⁄ choroid template using primers located in exons 2 and 3, equivalent to the N-terminal and C-terminal domains (Fig. 4). However, sequencing showed that the amplified product contained sequences for the N-terminal domain of human cN with splice-site skipping and use of a cryptic splice junction in intron 2. Such an alternative splice could produce a truncated, one-domain protein (Fig. 4). However, the crystallin coding sequence does not appear to be part of a translatable ORF in this transcript. Interestingly, an EST apparently corresponding to a similar transcript from human hippocampus (BM548090) is present in dbEST, suggesting that the PCR product from RPE may represent a transcript found at low levels in neural tissue, but not one that could produce a viable protein. Screening of available Invitrogen full-length GeneTrapper-ready cDNA libraries showed that cN transcripts were detectable only in human testis. Two positive clones were obtained from this tissue (GenBank accession number AF445455). The testis transcript included the first exon (the N-terminal arm region) and exon 2 (the N-terminal domain). For the C-terminal domain, however, only exon 3 (the third motif) was included. The exon corresponding to the fourth motif was skipped and instead an unrelated, cryptic downstream exon was included (Fig. 4). This exon has no similarity to the bc motif sequence and could not produce a polypeptide capable of completing the C-terminal domain of the protein. Currently there is no evidence for expression of canonical cN in primates. This leaves open the question of whether the gene for cN retains any function 2282 Recombinant mouse cN Recombinant mouse cN was synthesized in a bacterial host (Fig. 5A), purified and verified by MS. Initial A B Fig. 5. Recombinant mouse cN-crystallin is less stable than cD-crystallin. Expression of recombinant mouse cN. Lane 1, pET-cN E. coli whole cell lysate (uninduced); lane 2, pET-cN E. coli whole cell lysate (induced); lane 3, purified cN. Denaturation profiles for recombinant mouse cN and human cD in increasing urea concentrations. FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS G. Wistow et al. attempts at obtaining a mass measurement of the covalent structure of cN by MS resulted in a spectrum with a large number of peaks (including the value calculated from the sequence), probably due to binding of multiple sodium ions, making deconvolution difficult. However, prewarming the sample at 37
C resulted in an almost single peak spectrum of 21 270 Da, corresponding to cN lacking the initiator methionine. In solution studies, this protein behaved as a monomer, similar to c-crystallins and in contrast with the multimeric b-crystallins. Indeed, c-crystallins in solution tend to behave as if they were even smaller than expected for 20-kDa monomers and cN exhibits the most extreme version of this behavior seen so far. An estimate of the protein oligomeric size was gained from gel filtration using two different chemical supports. On both columns, cN was eluted with a higher elution volume than human cD-crystallin. On preparative gel filtration on Sephacryl S300, cN was eluted at 103.7 mL. Under the same conditions, human cD-crystallin was eluted at 98.5 mL. On analytical gel filtration on Superose 12, cN was eluted at 16.05 mL whereas human cD-crystallin was eluted at 14.96 mL. These results indicate that cN is eluted at a smaller apparent size than another monomeric c-crystallin. As the two polypeptide chains have a similar molecular mass, these data suggest that cN behaves even more anomalously on gel filtration than other c-crystallins, possibly through interactions with the column [36,37]. To provide unambiguous evidence of the oligomer size of cN, light scattering was performed. The molecular mass of the protein at 5 mgÆmL)1 was evaluated by dynamic light scattering. The average over 15 readings gave a diffusion coefficient (DT) of 974.5 · 10)13 m2Æs)1. The data were of high standard, with baseline values within the range 1.000 ± 0.001. Sum-of-squares values were below 5, and the majority were below 2, showing that the quality of the data was statistically valid. This measured diffusion coefficient gives an estimated molecular mass of 23.14 kDa. The results showed clearly that in free solution at  5 mgÆmL)1 the protein was monomeric. The likely explanation for the different molecular sizes in the gel filtration systems is the propensity for the crystallin molecule to interact with the column matrix [36,37]. However, efforts to crystallize the protein were unsuccessful. Recombinant mouse cN is less soluble than other c-crystallins. Although not rigorously tested, it seemed that a concentration of 5 mgÆmL)1 was limiting. Unlike other lens c-crystallins, mouse cN, which has a calculated pI of 6.27, was not soluble when exposed for significant lengths of time to pH 5, precluding the use of cation chromatography for purification. Samples FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS cN-crystallin of cN were turbid after storage at 4
C or when thawed. Cooling-induced precipitation was not fully reversible. Thus, although cN exhibits some of the characteristics of the phase-separation-driven phenomenon known as ‘cold cataract’, its behavior is not consistent with a simple liquid-liquid phase transition seen for some other c-crystallins [38]. Typically, c-crystallins also exhibit very high conformational stability [39]. Recombinant mouse cN was subjected to unfolding in urea under equilibrium conditions and compared with human cD-crystallin (Fig. 5B). The data show that under conditions in which cD is unchanged, as judged by fluorescence, cN completely unfolds, suggesting a much lower conformational stability. In common with other c-crystallins, the tryptophans of cN are more quenched when buried in the folded protein than when exposed to the denaturant [40]. cN expression in mouse eye Eye-specific expression of cN was confirmed by Northern blotting. In mouse multi-tissue Northern blot analysis, cN was detectable only in eye (Fig. 6A). In Northern blot analysis of rat tissues, cN was detected only in retina (Fig. 6B). Lens was not included on these blots. Expression of cN protein in lens was examined by 2D gels and MS. Figure 7A shows a Ponceau S-stained blot of soluble protein from a newborn mouse. The identities of major crystallins were known from earlier work [41]. After destaining, the blot was probed with antibody to cN (Fig. 7B). A single immunoreactive spot was observed just below bA2. The immunoreactive spot was not visible in Ponceau stain, but Coomassie blue staining of a larger 2D electrophoresis gel of soluble protein from newborn mouse lens did detect a protein spot at this position (cN, Fig. 7C). This spot was confirmed to be cN by in-gel digestion and LC ⁄ MS ⁄ MS analysis of tryptic fragments that identified seven distinct cN peptides covering 48% of the protein sequence (data not shown). The antibody to cN was used in immunofluorescence studies of mouse eye sections (Fig. 8). In the anterior segment of the eye (Fig. 8A), cN immunoreactivity was seen specifically in the lens nucleus, the primary site of expression for cA–F crystallins, but not in secondary fibers or lens epithelium, where cS-crystallin is expressed. No expression was evident in other tissues of the anterior chamber. In the retina (Fig. 8B), expression was seen in the outer plexiform layer (containing photoreceptor axons and synapses) and photoreceptor outer segments. 2283 cN-crystallin G. Wistow et al. A B Fig. 6. Expression of cN transcripts is eye specific in rodents. (A) Northern blot of multiple mouse tissues with probe for mouse cN. Br, brain; Ey, eye; He, heart; Lu, lung; Li, liver; Sp, spleen; Ki, kidney; Pa, pancreas; Sm, skeletal muscle, Th, thymus. Staining pattern for 28S and 18S rRNA is shown below. (B) Northern blot of multiple rat tissues with probe for mouse cN. Ret, retina (two preparations); Lu, lung; Ki, kidney; Te, testis; Li, liver; Sp, spleen; Br, brain; He, heart. The staining pattern for 28S rRNA is shown below. Discussion In phylogenetic analyses, the b-crystallins form a distinct clade with bA (acidic) and bB (basic) branches. As has previously been observed, and is illustrated in Fig. 2, most b-crystallins have clear orthologs in all vertebrates so that mammalian and zebrafish bA2 sequences, for example, are close together on the same branch of the tree. In contrast, the cA–F-crystallins that are expressed in most mammals have no orthologs in fish and form a distinct branch of their own that includes c-crystallins of similar size from amphibians (two of which, from a frog, Rana catesbeiana [42], are shown), and from marsupials in a ‘terrestrial’ branch of the family. However, even on this branch, different orders do not appear to have truly orthologous crystallins, i.e. the frog sequences are not orthologs of any gene in mammals. Whereas most of the zebrafish c-crystallins are similar in size to the mammalian cA–F group, with no 2284 Fig. 7. Expression of cN protein in newborn mouse lens. (A) Ponceau S-stained blot of 2D electrophoresis gel of soluble protein from a newborn mouse lens showing major crystallin spots. (B) Western blot of destained 2D electrophoresis gel blot shown in (A) using antibody to cN. A single spot is detected at the same relative position occupied by cN in part (C). (C) Coomassie blue-stained large 2D electrophoresis gel of soluble protein from a newborn mouse (from [41]). Protein marked with an arrow was confirmed to be cN by MS. N-terminal arm, they too form a distinct branch of the overall family. This branch includes the cM-crystallins which have previously been identified in carp, so it seems appropriate to name this subfamily the cM-crystallins and to number the new zebrafish sequences FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS G. Wistow et al. cN-crystallin A B Fig. 8. Immunofluorescence localization of cN-crystallin in mouse eye. (A) Expression of cN in the anterior segment. Left panel shows DAPI staining for nuclei; center panel shows immunofluorescence stain (red) for cN; right panel shows a control with no primary antibody. (B) Expression in the posterior segment. Left panel shows combined DAPI (blue) staining of cell nuclei and immunofluorescence (red) signal for cN. GCL, Ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer, OS, outer segments. Right panel shows control with no primary antibody. White arrows show positive stain in OPL and OS layers. accordingly (however, the previous cM nomenclature for Rana may not be appropriate in this overall context). As is seen elsewhere in the family tree, several of the zebrafish sequences (such as cE and cF in mammals) appear to be the result of relatively recent gene duplications, probably the result of large-scale genome duplication events in these and other fish [43]. Mammal, bird and fish cS sequences form a third, more ancient branch. In this subfamily, a pair of zebrafish genes (cSa and cSb) are close in sequence to those of chicken and mouse, and a second pair (cSc and cSd) belong to an earlier offshoot. This may indicate that there were two cS genes in the common ancestor of fish, birds and mammals and that only one of these survived in the terrestrial species while both survived in fish and indeed underwent a subsequent duplication in some species. A predicted cS from chicken and one cloned from iguana belong to this FEBS Journal 272 (2005) 2276–2291 ª 2005 FEBS clade and appear to be orthologs of mammalian cS-crystallin. The cN family is newly discovered. In terms of structure and solution behavior, cN most closely resembles c-crystallins. However, in the phylogenetic tree, the cN sequences do not associate strongly with either the b or c subbranches. Genes from mammals, chicken and zebrafish all show a hybrid gene structure with both c-like and b-like exons. Overall, the cN family appears to be an evolutionary intermediate between the wider b and c crystallin families. It seems likely that the b-crystallin ⁄ AIM1 group of genes represents the original gene organization state of the superfamily, with genes built up by successive duplication from an ancestral gene that encoded an individual motif (although, as an individual bc motif could not be a stable structure alone, the protein product must have been an obligate dimer), resulting in each motif 2285