J Chain in the Nurse Shark: Implications for Function in a Lower Vertebrate¹

J chain is a small polypeptide present only in polymeric Ig. Despite its discovery >30 years ago, the function of J chain has remained enigmatic. In humans and mice, it appears to facilitate the transport of polymeric Ig into secretions by promoting the noncovalent association of the Ig with the polymeric Ig receptor (1, 2, 3). J chain also appears to have a role in modulating the extent of Ig polymerization, but the precise nature of this role is not known and the regulation of Ig polymerization remains unclear (3, 4, 5, 6). A polypeptide identified as J chain has also been identified in a number of lower vertebrate species (7, 8, 9), and was reported to be present even in some invertebrates (10), an unexpected finding because only jawed vertebrates are believed to produce Ig.

J chain does not resemble any other known protein. Its amino acid sequence, determined in several mammalian species, is rich in acidic residues and is highly conserved, including eight residues of half-cystine (Cys).⁷ In human IgA and IgM, two of these Cys residues form disulfide bridges to the H chain, the remainder forming intrachain disulfide bridges (11, 12, 13). The conservation of Cys residues is retained in all species examined except in Xenopus laevis, in which one of the previously invariant Cys residues is replaced by serine (8). A carbohydrate acceptor tripeptide sequence is present at the same position in each sequence.

Cartilaginous fish, including sharks, skates, and rays, are the most ancient vertebrate group in which Ig has been identified. IgM from these species, which resembles mammalian IgM in many respects, exists in two forms, monomeric or 7S IgM (a single unit of two H and two L chains) and pentameric or 19S IgM (five such units) (14, 15). In the nurse shark, Ginglymostoma cirratum, as in mammals, J chain is present only in pentameric IgM (16). Toward our goal of determining which features of J chain are retained in all vertebrates that produce polymeric Ig, we have cloned cDNA encoding J chain in the nurse shark, compared the translated amino acid sequence with J chains of other species, and examined its tissue-specific expression.

A cDNA library, prepared from the spleen of a single nurse shark, has been described (17). Approximately 1.2 × 10⁵ PFU were screened with X. laevis J chain probe JCH3A. Hybridizations were performed overnight at 42°C in 25% formamide, 5× standard saline citrate phosphate/EDTA (SSPE) (1× SSPE: 150 mM NaCl, 10 mM sodium phosphate, 1.0 mM EDTA, pH 7.4), 2.5× Denhardt’s solution (1× Denhardt’s solution: 0.02% Ficoll, 0.02% polyvinylpyrrolidone, 0.02% BSA), 1% SDS, and 1 mg/ml calf thymus DNA. Filters were washed twice for 15 min with 3× SSC (1× SSC: 150 mM NaCl, 15 mM sodium citrate, pH 7.0) and 0.1% SDS at room temperature, and twice for 20 min with 2× SSC, 0.1% SDS at 50°C. After plaque purification, pBluescript was rescued from the λZAPII vector, according to the manufacturer’s instructions.

Nurse shark splenic RNA was reverse transcribed using a polydT-adapter primer (5′-GGCCACGCGTCGACTAGTACT₁₇-3′) and Superscript II reverse transcriptase (Life Technologies, Rockville, MD). The resulting first-strand cDNA was tailed with dGTP using TdT. To obtain the 3′ end of the J chain sequence, the cDNA was amplified by 3′ RACE using 0.2 μM each of an adapter primer (5′-GGCCACGCGTCGACTAGTACT-3′) and sense primer NShJ1 (5′-CCCATTCGGACCAAGTTTGTC-3′; Fig. 1, positions 308–328). To obtain the 5′ end, cDNA was amplified by 5′ RACE using 0.4 μM each of an adapter primer (5′-CGGCGAATTC₁₈-3′) and antisense primer NShJ2 (5′-TCTGCATGATTGGTCGTC-3′; Fig. 1, positions 448–431). Cycling conditions were 94°C for 30 s, 50°C for 30 s, and 72°C for 1 min for 30 cycles, followed by a final extension at 72°C for 10 min. Amplified products (3′ RACE, ∼450 bp; 5′ RACE, ∼420 bp) were ligated into a cloning vector (pCR2.1; Invitrogen, Carlsbad, CA). Three clones derived from 3′ RACE and four clones derived from 5′ RACE were isolated, and the inserts were sequenced.

Inserts from cDNA clones were sequenced in both directions by the dideoxy chain termination method with the T7 Sequenase 2.0 DNA sequencing kit (Amersham Pharmacia Biotech, Piscataway, NJ) or the ThermoSequenase cycle sequencing kit (Amersham Pharmacia Biotech) and by automated fluorescent DNA sequencing (Tufts Medical School Core Facility, Boston, MA).

Database searches were performed using BLAST programs maintained by the BCM Search Launcher, Human Genome Center, Baylor College of Medicine (Houston, TX): http://searchlauncher.bcm.tmc.edu (18). The charge of proteins at pH 9.5 and 7.0 was estimated by using a program at the EMBL WWW Gateway to Isoelectric Point Service: http://www.embl-heidelberg.de/cgi/pi-wrapper.pl. To determine the most likely site of cleavage by signal peptidase, we applied programs at the SignalP World Wide Web server: http://www.cbs.dtu.dk/services/SignalP (19). An alignment of mature J chain sequences was generated by the Multiple Sequence Alignment version 2.1 at the Biology Workbench 3.2 server maintained by the San Diego Supercomputer Center (San Diego, CA) (http://workbench.sdsc.edu) (20, 21). Percent identities between pairs of sequences in this alignment were determined from the number of identical residues divided by the number of total residues in the shorter sequence. Construction of a phylogenetic tree, based on the alignment, used the PHYLIP programs (22) also found at the Biology Workbench 3.2 server.

Mouse mAb GA15, which preferentially binds the monomeric form of nurse shark IgM, has been described previously (23). To generate J chain-specific Abs, mice were immunized with a pool of three peptides corresponding to the translated ShJCH1 cDNA sequence: VSSKMIVTELPNGEKVEQL, ENISDPTSPIRTKFVY, and GRQESPQESPEPQCKPKPPPTDE. The first and third peptides correspond to sequence that is not conserved between J chain in nurse shark and other species; the second peptide corresponds to a region of sequence that is highly conserved among J chains. The peptides were prepared as multiantigenic peptides in which eight individual peptides are linked together on a branching lysine matrix (24). Peptides were emulsified in CFA, each mouse receiving 50 μg i.p. After 1 mo, the mice were boosted with the same amount of Ag in IFA. Two weeks later, one of the mice was immunized i.v. with 20 μg of soluble peptides, and the fusion was conducted 3 days later (25). Hybridoma supernatants were tested by ELISA against nurse shark serum and the pool of peptides.

Polymeric and monomeric IgM were isolated from serum by SDS-PAGE under nonreducing conditions, followed by excision of these proteins from the gel. The isolated Ig were then applied to SDS-PAGE under reducing conditions. To visualize proteins, gels were stained with Coomassie blue.

Nurse shark serum was precipitated with Ab JC4 or GA15 by combining 10 μl of shark serum with 1 μl mAb ascites fluid and 189 μl of wash buffer (150 mM NaCl, 50 mM Tris, pH 8) and incubating overnight at 4°C; 30 μl of a 50% suspension of protein G agarose beads (Amersham/Pharmacia, Uppsala, Sweden; beads stored in wash buffer) was added, and the tubes were rotated end over end for 1 h at 4°C. The beads were washed three times, and the immunoprecipitates were collected in 30 μl Laemmli sample buffer (2% SDS, 200 mM Tris, pH 8, 5% glycerol, 0.01% bromphenol blue with or without 5% 2-ME) (26) and boiled for 2 min. Nonreduced and reduced precipitates were subjected to SDS-PAGE on 5 and 12% polyacrylamide gels, respectively.

To determine its amino-terminal sequence, J chain was isolated from reduced polymeric IgM that had been immunoprecipitated with JC4 and analyzed by SDS-PAGE on a 13% polyacrylamide gel. Proteins were transferred to a polyvinylidene difluoride membrane (Immobilon-Psq; Millipore, Bedford, MA) by electroblotting for 2 h at 250 mA constant current in 10 μM CAPS (3-cyclohexylamino-1-propanesulfonic acid), pH 11, with 10% methanol. Proteins were visualized by staining the membrane with 0.1% (w/v) Amido Black 10B (Bio-Rad Laboratories, Hercules, CA) in methanol/acetic acid/water (40:10:50) and destained with methanol/acetic acid/water (50:10:40). Each of two closely spaced stained bands migrating just ahead of L chains was excised from four lanes on the polyvinylidene difluoride membrane and applied directly to an Applied Biosystems 494 Protein Sequencer (Foster City, CA).

To generate a nurse shark J chain probe, the insert of cDNA clone ShJCH1 was excised with XhoI and EcoRI. A Xenopus J chain probe JCH3A (8), nurse shark conventional IgM V_H probe (23), and nucleoside diphosphate kinase (NDPK) probe S1-3 (27) have been described. All probes were labeled with [α-³²P]CTP by random priming.

Peripheral blood leukocytes were separated from whole blood with Lympholyte-H (Cedarlane Laboratories, Hornby, Ontario, Canada; specific density, 1.077; diluted 2/1 in elasmobranch saline (350 mM urea, 200 mM NaCl in PBS)). Total RNA was extracted using TRIzol reagent, as directed by the manufacturer (Life Technologies); 20 μg was subjected to electrophoresis through a 1% agarose gel and transferred to a reinforced nitrocellulose membrane. The membrane was incubated sequentially with the J chain, V_H, and NDPK probes overnight at 42°C in 50% formamide, 6× SSC, 5× Denhardt’s solution, 10 mM EDTA, 5% SDS, and 100 μg/ml sheared salmon sperm DNA; washed with 2× SSC, 1% SDS for 20 min at room temperature, followed by 0.2× SSC, 0.1% SDS for 15 min at 65°C; and exposed to film for 22 h. Hybridization with the J chain and NDPK probes was performed when the radioactivity on the membrane from the previous hybridization reached background levels.

Genomic DNA (10 μg), extracted from erythrocytes from a single nurse shark, was digested with BamHI, EcoRI, EcoRV, HindIII, PstI, or SacI, and subjected to electrophoresis through a 0.8% agarose gel. The DNA was denatured, neutralized, and transferred onto a reinforced nitrocellulose membrane. The membrane was incubated overnight with the J chain probe at 42°C in 30% formamide, 6× SSC, 5× Denhardt’s solution, 0.5% SDS, and 100 μg/ml sheared salmon sperm DNA; washed with 2× SSC, 1% SDS for 20 min at room temperature, followed by 2× SSC, 0.1% SDS for 15 min at 55°C; and exposed to film for 15 h.

To isolate cDNA clones encoding nurse shark J chain, we screened a nurse shark spleen cDNA library (17) at low stringency with a X. laevis J chain probe. Of the ∼1.2 × 10⁵ plaques screened, two hybridized to the probe; both inserts were ∼750 bp. The sequence of one of these inserts, ShJCH1, is shown in Fig. 1. Database searches using BLAST programs showed that the best matches to ShJCH1 were J chain sequences of other species. In eukaryotes, translation usually starts at the AUG nearest the 5′ end of the mRNA; however, in this case, the 5′-most ATG (position 71–73 in Fig. 1) does not have neighboring flanking sequences favorable for initiation: most importantly, A (or sometimes G) at −3 and G at +4 (28). Two nearby ATG codons (positions 74–76 and 89–91 in Fig. 1) are slightly more favorable, having A or G at −3. It is possible that under these circumstances, translation initiates at more than one ATG codon. For purposes of this analysis, we have taken the 5′-most ATG as the initiating codon. The resulting polypeptide would contain 167 aa residues. The predicted site of signal peptidase cleavage yields a leader peptide of 23 aa residues and a mature protein of 144 residues. The leader segment is similar in length to leaders in other J chains; however, the mature protein is a few residues longer. The 3′ untranslated region consists of 152 bp with a single polyadenylation signal (AATAAA) 19 bp 5′ to the poly(A) tail.

When polymeric and monomeric IgM isolated from nurse shark serum were separated into their individual peptide chains by SDS-PAGE under reducing conditions, a doublet migrating slightly faster than L chains was observed in the polymeric, but not in the monomeric fraction (Fig. 2,A). The doublet, whose migration in the gel is consistent with that expected for J chain, was isolated as follows. mAbs were generated to a pool of three peptides corresponding to translations of segments of cDNA ShJCH1. One of these Abs, JC4, immunoprecipitated much more polymeric than monomeric IgM from nurse shark serum (Fig. 2,B). When the JC4 precipitate was subjected to SDS-PAGE under reducing conditions, a doublet migrating ahead of L chains was again seen (Fig. 2 C). This doublet was not seen in the precipitate obtained using GA15, a mAb specific for nurse shark IgM that preferentially recognizes the monomeric form of the molecule.

Protein in each of the doublet bands obtained from the JC4-immunoprecipitated IgM was subjected to amino-terminal sequence analysis. The sequence of the slower component was KSERNLLVSSKXKLLEV, and that of the faster component was KSERNLLVSSKXKL. Evidently, cDNA clone ShJCH1 encodes the protein in each band of the doublet. The presence of two bands may be a consequence of glycosylation differences. The correspondence between the translated cDNA sequence and the amino-terminal sequence of these proteins establishes that the insert in clone ShJCH1 encodes nurse shark J chain. The results also confirm the predicted site of signal peptidase cleavage.

An alignment of J chain sequences is shown in Fig. 3. We had previously noted two particularly conserved segments among the J chains then characterized (8). One of these, near position 50, is still conserved when the shark sequence, as well as the more recently determined brushtail possum and chicken sequences, and the completed Rana sequences are added to the alignment; however, in Rana, there are two nonconservative replacements in this region. A site for N-linked glycosylation at positions 50–52 has been found in all J chain sequences to date. In mouse J chain, replacement of Asn⁵⁰ by alanine markedly reduced synthesis of IgA dimers, suggesting that N-linked sugar plays a role in dimerization (37).

The second segment of conserved sequence noted previously, surrounding Cys-6 and Cys-7, is also present in the possum, chicken, and Rana J chains, but this region is substantially different in the nurse shark sequence. From Cys-4 on, only six residues in the shark J chain are also present in all other J chains (i.e., Cys-4, Cys-5, Cys-6, Tyr¹⁰⁵, Arg¹⁰⁷, and Pro¹³⁷). Cys-7 and Cys-8 are both absent from the shark chain. Because of the dissimilarity in the carboxyl-terminal sequence between nurse shark and other J chains, the complete shark cDNA sequence was redetermined by 5′ and 3′ RACE. The sequence was identical with that obtained previously and shown in Fig. 1.

In human J chain, it has been shown that Cys-2 and Cys-3 participate in disulfide bridges with penultimate residues in H chains of polymeric IgM (11) or IgA (12, 13); the remaining six Cys residues pair in identical disulfide bridges within the J chain. Cys residues in chicken and Rana J chains are found at the same positions, but Cys-3 is missing in the Xenopus J chain (Fig. 3). In the shark, Cys-2 is absent (as well as Cys-7 and Cys-8). Thus, only four Cys residues (Cys-1, Cys-4, Cys-5, and Cys-6) are present in all of these J chains. The nurse shark has one additional residue of Cys (positions 84 in Fig. 3) that is not found in any of the other J chains.

The arrangement of disulfide bridges in human J chain and the number and the location of the Cys residues in other J chains are diagrammed in Fig. 4. If one assumes that the conserved Cys residues that are present in nurse shark J chain form the same disulfide bridge linkages as those in human J chain, then the intrachain disulfide bridge joining Cys-7 and Cys-8 is absent in the shark. Furthermore, the retention, in both shark and Xenopus J chain, of only one of the two Cys residues that, in humans, join J chain to either μ- or α-chain suggests that a single disulfide bridge links J and μ-chains in polymeric IgM in both species. The extra Cys in nurse shark J chain may also participate in an interchain disulfide bridge or its sulfhydryl group may be blocked by a low m.w. mercaptan.

The importance of certain structural features of J chain in mouse and human IgA has been tested by site-directed mutagenesis (37, 38). Johansen et al. (38) showed that elimination of segments near the carboxyl-terminal end of human J chain did not eliminate incorporation into IgA and had variable effects on polymerization of coexpressed IgA. However, these changes had profound effects on the in vitro binding of IgA to secretory component, the fragment of the polymeric Ig receptor that remains bound to secretory IgA. Replacement, by serine, of both Cys residues (Cys-2 and Cys-3), which form disulfide bridges to H chain, nearly abolished incorporation of J chain into human IgA, but the presence of Cys-3 alone (as in nurse shark) was sufficient to allow substantial incorporation. Although the resulting IgA was still able to form polymers that bind secretory component, transcytosis across epithelial cells was greatly reduced. When either Cys-7 or Cys-8 (both missing in shark J chain) was replaced by serine, formation of IgA polymers was only slightly reduced, but binding to secretory component was abolished. Thus, these changes introduced by site-directed mutagenesis diminished the secretory functions attributed to J chain more severely than the incorporation of J chain into polymeric IgA.

A characteristic feature of J chains is rapid migration upon electrophoresis in alkaline-urea gels, the result of a relatively high content of acidic residues. This property was critical in the initial identification of J chain in polymeric Ig and has subsequently been useful in establishing the presence of J chain in Ig from a variety of species. In an early study, Weinheimer et al. (7) searched for J chain in IgM of several lower vertebrates. From the lack of a fast-migrating component in alkaline-urea electrophoresis, they concluded that J chain was absent from nurse shark Ig. However, McCumber and Clem (16) isolated a component that appeared to be J chain from nurse shark polymeric IgM by gel filtration; the mobility of this protein on alkaline-urea gels was less than that of human J chain. They suggested that the previous failure to identify J chain in this species was a consequence of its lower negative charge at pH 9.4. This supposition was supported by comparing the amino acid composition of the putative shark J chain with that of human J chain. The composition reported for shark J chain by McCumber and Clem corresponds reasonably well to that of the mature protein encoded by cDNA ShJCH1, although there are substantial differences in a number of residues, especially Cys, Lys, Arg, His, and Gly. From the translated sequences, the mole percent of Asp plus Glu in shark J chain is 14.6, and the mole percent of Lys plus Arg is 16.0. Comparable values for human J chain are 18.2 and 11.7.

Alkaline-urea gel electrophoresis is usually conducted at about pH 9 (39). At this pH, Asp and Glu are negatively charged, Arg is expected to be mainly positively charged, and Lys to carry a partial positive charge. From the amino acid composition of the J chains shown in Fig. 3 and the pK values of the charged side chains of amino acid residues, it is possible to calculate the approximate net charge of J chains at any pH. At pH 9.5, the predicted charges for the mammalian and chicken J chains range from −7.6 to −13.3; for X. laevis and Rana catesbeiana, they are −7.6 and −4.5, respectively; but for the nurse shark, only −0.9. We have omitted the Cys residues from these calculations because, during the isolation of J chain, these residues are ordinarily reduced and alkylated. Weinheimer et al. (7) blocked the reduced Cys residues with iodoacetamide, which is uncharged. Accordingly, shark J chain would not be expected to have the rapid anodal mobility on alkaline-urea gels that is characteristic of human and other J chains. At pH 7, the nurse shark J chain is predicted to have a net charge of +2.3, and the acidic mouse and human chains to have net charges of −10.8 and −8.9, respectively. Thus, the nurse shark polypeptide is the only J chain identified to date that is not acidic.

A commonly used method to estimate sequence conservation is to determine, for all positions, the fraction of identical residues. These values, based on the alignment shown in Fig. 3, are shown in Table I. The values for pairwise comparisons between the mammalian J sequences range from 63 to 79%. Comparing chicken and mammalian J chains, the values range from 56 to 64%; Xenopus vs mammalian, 52 to 57%; and Rana vs mammalian, 48 to 51%. However, the values for shark vs mammalian J chain are much lower, ranging from 33 to 36%.

The values shown in Table I for the sequence comparison between earthworm J chain and the other J chains are puzzling. We would expect that a sequence in an invertebrate would be approximately equally divergent from the orthologous sequence in any mammal, or for that matter, in any vertebrate, reflecting the time since the separation of the vertebrate and invertebrate lineages. The range in identities between the earthworm and any of the mammalian J sequences is 52–72%, much greater than the ranges noted in the previous section for comparison of any of the other nonmammalian J chains with the mammalian sequences. Comparison of the earthworm J vs any other (i.e., vertebrate) J sequence ranges from 32 to 72%. Another puzzle is that the earthworm J sequence is more similar to mouse and human J than are the chicken, Xenopus, Rana, or nurse shark J sequences.

A more revealing method of exploring relationships among sequences is to construct a cladogram or phylogenetic tree. Fig. 5 is such a cladogram for the sequence alignment shown in Fig. 3. As expected from the discussion in the preceding paragraph, but unexpected from our knowledge of the relatedness among these species, the earthworm J sequence clusters with the mammalian J sequences and indeed lies on a common branch with mouse J. The earthworm J sequence is based on cDNA sequence obtained by RT-PCR from total RNA extracted from Eisenia foetida (10). A possible explanation for this result is contamination of material derived from the earthworm with that from another rodent. However, if this were the case, other aspects of this report would be difficult to explain, e.g., detection of putative J chain in earthworm extract by reaction with polyclonal Abs to human J chain, as well as the localization, by immunohistochemistry, of J chain in cells of the earthworm. Northern blot analysis, with either an earthworm or human J chain probe, revealed J chain transcripts of ∼1.4 kb in a variety of invertebrates, the same size as a human J chain control. However, the J chain transcript in the nurse shark is ∼0.8 kb (see below), similar in size to the transcript encoding Xenopus J chain (8). The size of the Rana J chain transcript is slightly smaller (∼0.7 kb; V. Hohman and L. Steiner, unpublished results); the J chain transcript in the chicken is 1.8 kb (29). These results, taken together, indicate that the existence of J chain in the earthworm needs to be re-evaluated.

With the exception of the position of the earthworm, the remaining branching pattern in the cladogram is consistent with the relationships among these species. The length of the shark branch is indicative of the considerable difference in sequence of this J chain from the others. The two amphibian sequences branch from a short common stem. This is consistent with the relatively distant relationship of ranid and pipid frogs (40).

To determine the multiplicity of genes encoding nurse shark J chain, liver DNA was digested with a panel of restriction enzymes and analyzed by Southern blotting. Following hybridization with the labeled insert of cDNA clone ShJCH1, one to three bands were detected in each lane (Fig. 6). The data suggest that the shark J chain is encoded by no more than one or two genes.

To evaluate the tissue-specific expression of J chain, total RNA from various organs was analyzed by Northern blotting with a nurse shark J chain probe. By far, the highest level of expression was in the spleen, but there was some expression in most of the tissues, especially epigonal tissue, gills, and peripheral blood (Fig. 7). The length of the J chain transcript was estimated to be 0.8 kb. To correlate the expression of J chain with that of IgM, the blot was also hybridized with a V_H probe that is expected to detect most μ-chain transcripts (23, 41, 42). In most tissues, μ expression correlated with that of J chain. An exception was muscle, in which there was expression of J chain, but not μ-chain. Muscle from another nurse shark showed the same expression pattern (data not shown). Further investigation is needed to explore the reason for J chain, but not μ-chain, expression in muscle. For all tissues, results obtained by hybridizing with a Cμ probe were similar to those obtained with V_H (M. Flajnik, unpublished observations).

The high level of J and μ expression in the spleen and epigonal organ is consistent with the functions and cellular compositions of these organs. The spleen is the main peripheral lymphoid organ and a major site of Ab synthesis in cartilaginous fish; the white pulp of the nurse shark spleen contains large populations of lymphocytes at various stages of differentiation, and the red pulp contains most of the splenic plasma cells (43, 44, 45). The epigonal organ (the tissue surrounding the testes and ovaries) and the Leydig organ (located in the esophagus) are sites of lymphohemopoiesis in cartilaginous fish (43). Although some species have both organs, others, such as the nurse shark, have only an epigonal organ (43). The nurse shark epigonal organ is a site for recombination-activating gene 1 expression (23) and contains differentiating lymphocytes as well as plasma cells (44, 45).

In mice, the level of J chain expression in the small intestine is second only to that in the spleen (V. Hohman and L. Steiner, unpublished results). In X. laevis and R. catesbeiana, the highest level of J chain expression is in the intestine (8) (V. Hohman and L. Steiner, unpublished results). These observations are consistent with a role for J chain in these species in binding polymeric Ig and mediating its transport into secretions. It was therefore surprising to detect neither J nor μ-chain transcripts in the nurse shark spiral valve (intestines). Spiral valves were obtained from two additional nurse sharks with the same result. Gut-associated lymphoid tissue has been described in the intestine of cartilaginous fish; there are plasma cells in the lamina propria, and both B cells and plasma cells in the epithelium (46). In our immunohistological examination of nurse shark intestine, we have also observed lymphoid cells in the epithelium (L. Rumfelt and M. Flajnik, unpublished observations). Perhaps the proportion of lymphoid cells is too low for detection of J or μ transcripts by Northern blotting.

The function of J chain in the nurse shark may differ from its function in mammals. To date, there are no reports of a secretory component homologue in the nurse shark nor in any lower vertebrate. Although such lack of identification is not conclusive, it would be consistent with the possibility that J chain is not necessary for promoting transport of IgM into shark secretions. The supposition that nurse shark J chain may not participate in secretory-mediated transport is in general agreement with the results of site-directed mutagenesis conducted with human J chain that has been incorporated into polymeric IgA (38). The most striking sequence differences between shark and mammalian J chains are the lack of conservation near the carboxy terminus and the absence of three Cys residues. When similar changes are made in human J chain, the product retains some capacity for entering into IgA polymers, but the binding of the IgA to secretory component is severely reduced. These observations, taken together with the overwhelming preponderance of J and μ-chain transcripts in the spleen, compared with the intestine, suggest that J chain in the nurse shark may not be involved in IgM secretion. Further studies are required to explore the function of J chain in this species.

Since submission of this paper, a report describing the cloning and expression of J chain in the turtle, Trachemys scripta, has been published (47). In addition, a sequence that appears to encode part of the J chain in the skate, Raja eglanteria, has been deposited in GenBank, accession AF520475.