Noncoordinate expression of J-chain and Blimp-1 define nurse shark plasma cell populations during ontogeny



B-lymphocyte-induced maturation protein 1 (Blimp-1) is the master regulator of plasma cell development, controlling genes such as those encoding J-chain and secretory Ig heavy chain. However, some mammalian plasma cells do not express J-chain, and mammalian B1 cells secrete “natural” IgM antibodies without upregulating Blimp-1. While these results have been controversial in mammalian systems, here we describe subsets of normally occurring Blimp-1 antibody-secreting cells in nurse sharks, found in lymphoid tissues at all ontogenic stages. Sharks naturally produce large amounts of both pentameric (classically “19S”) and monomeric (classically “7S”) IgM, the latter an indicator of adaptive immunity. Consistent with the mammalian paradigm, shark Blimp-1 is expressed in splenic 7S IgM-secreting cells, though rarely detected in the J-chain+ cells producing 19S IgM. Although IgM transcript levels are lower in J-chain+ cells, these cells nevertheless secrete 19S IgM in the absence of Blimp-1, as demonstrated by ELISPOT and metabolic labeling. Additionally, cells in the shark BM equivalent (epigonal) are Blimp-1. Our data suggest that, in sharks, 19S-secreting cells and other secreting memory B cells in the epigonal are maintained for long periods without Blimp-1, but like in mammals, Blimp-1 is required for terminating the B-cell program following an adaptive immune response in the spleen.


In jawed vertebrates, or gnathostomes, transmembrane (TM) IgM is the antigen receptor on naïve B cells, and plasma cells secrete multimeric (tetramer, pentamer, or hexamer) IgM. Secretory IgM is produced predominantly by B1 cells in mammals, comprising the majority of “natural,” polyreactive antibody that is present in both normal mice and animals raised in germfree conditions [1-7]. Conversely, “immune” secretory IgM is secreted by the relatively short-lived plasma cells of the B2 lineage [8]. In mice, approximately half of all secreted IgM is “natural” IgM, which is produced predominantly by BM B1 cells, while the rest is B2-cell-derived IgM [3, 9]. For B2 cells that form germinal centers, activation-induced cytidine deaminase expression induces somatic hypermutation and class-switch recombination, resulting in class-switched cells that secrete high-affinity, non-IgM isotypes (IgG/A/E) possessing different effector functions [10, 11].

Cartilaginous fish, including sharks, skates, rays, and ratfish are the oldest living animals with an adaptive immune system featuring an Ig-based BCR, along with TCR and MHC. Despite 450 million years of divergence from mammals, cartilaginous fish have both TM and secretory forms of IgM H chains [12, 13]. Besides the pentameric form of secreted IgM classically called “19S,” shark B cells also secrete a monomeric form, classically “7S” [14-18]. 7S IgM is thought to be a mammalian IgG equivalent, both with respect to its ability to penetrate extravascular spaces and its enhanced binding strength late in adaptive immune responses, whereas previous work also suggested that 19S IgM does not participate in antigen-specific responses [19, 20].

In addition to IgM, there are three other Ig H chain isotypes expressed in nurse sharks (Ginglymostoma cirratum): IgM1gj, an H chain encoded by an invariant “germline-joined” variable (V) gene and lacking the second constant domain (C2) found in conventional IgM [21]; IgW, related to mammalian IgD but having an unknown function in sharks [22-24]; and IgNAR, an H-chain-only isotype unique to cartilaginous fish [25]. In newborn nurse sharks, almost all of the serum Ig is composed of IgM1gj, and some 19S IgM. Low levels of serum 7S IgM appear around 8 days after birth and rise throughout development [26]. By adulthood there is very little serum IgM1gj, and 19S and 7S IgM are present in roughly equal proportions.

Whether or not shark plasma cells secrete 19S or 7S IgM seems to be dependent upon the expression of the joining chain (J-chain) [27]. J-chain is involved in Ig mulitimerization in mammals, but unlike sharks, mammalian IgM is secreted as disordered oligomers rather than monomers in J-chain KO mice [28-32]. Mammalian IgA, however, is secreted either as a dimer with J-chain (in mucosae) or as a monomer is the absence of J-chain (in serum) [33, 34]. In addition to multimerization, J-chain is required for Ig transport across the mucosal epithelium in tetrapods: the C-terminal domain of mammalian J-chain is necessary for association with the secretory component (SC), a portion of the poly-Ig receptor that remains associated with IgA and IgM after transcytosis across the mucosal epithelium [35]. The homologous region of shark J-chain, however, is divergent in sequence and lacks amino acid (aa) residues that play a role in binding to SC [27]. When combined with the absence of reported shark SC and the lack of J-chain expression in plasma cells in the shark intestinal lamina propria (Castro and Flajnik, unpublished data), the major role of shark J-chain is likely for IgM mulitimerization rather than Ig transport.

In mammals J-chain is upregulated in all Ig-secreting cells as part of the plasma cell transcriptional program, but in sharks it should only be expressed by 19S-secreting cells. To gain insight into the regulation of J-chain expression in these cells, we have examined expression of canonical B cell/plasma cell transcription factors in nurse sharks, especially B-lymphocyte-induced maturation protein 1 (Blimp-1), the master regulator of plasma cell development in mammals [36]. Blimp-1, encoded by the gene prdm1, is one of the two founding members of the multigene prdm family (along with prdm2/RIZ1) [37]. These two homologs are the only members present in invertebrates. The family was greatly expanded in vertebrates, and there is a primate-specific prdm member. All prdm family members contain similar domain structures including a PR (PRDI-BF1 and RIZ) domain and zinc (Zn) finger domains. The N-terminal PR domain is 20–30% similar to an SET domain, and also like an SET domain, can function as a methyltransferase in some members [37-40]. The multiple C-terminal Zn-finger domains aid in protein–protein interactions, and are present in all but one of the prdm genes [37].

The working model of Blimp-1 function in plasma cells involves a combination of cessation of the cell cycle and repression of the “master regulator” of B-cell development, Pax5, a transcription factor that suppresses expression of non-B-cell genes [41-44]. Downregulation of Pax5 results in de-repression of genes required for Ig secretion and the plasma cell phenotype, including XBP1, J-chain, and the upregulation of IgH/L transcription [44-52]. Blimp-1, while most frequently associated with plasma cells, also functions in the differentiation of many other cell types including primordial germ cells [53, 54], DCs [55], osteocytes [56, 57], myeloid cells [58], T cells [59-64], and NK cells [65]. For example, Blimp-1 is expressed in CD8+ effector T cells, and involved in survival and terminal differentiation [64]. Pax5 and Blimp-1 have also been studied in another nonmammalian model, the teleost fish, where they have been shown to play similar roles to their mammalian counterparts [66]. Unfortunately, teleost fish have lost J-chain and thus its regulation cannot be examined in this taxon.

In support of the model that Blimp-1 is obligatory for plasma cell function, mice lacking Blimp-1 in the B-cell lineage have low (but detectable) levels of serum Ig [36, 49]. The canonical view is that Blimp-1-negative cells have increased levels of Pax5 and decreased Ig secretion, and that for B1 as well as B2 cells, Blimp-1 is required for the plasma cell program [67-69]. However, the role of Blimp-1 in antibody secretion from B1 cells is controversial, and as mentioned, there are still detectible levels of secreted IgM in Blimp-1 KO mice [36, 49]. Additionally, others have shown that there is no augmentation of Blimp-1 expression in antibody-secreting B1 cells, and that even in the absence of Blimp-1, Pax5 expression is downregulated [70]. Finally, it has been shown that Pax5 is also downregulated in early B2 plasmablasts, and that even in the absence of Blimp-1, antibody secretion is initiated in these cells [68].

In this paper, we first embark on determining whether the forms of shark serum IgM would correlate with the types of plasma cells found in shark lymphoid tissues, primarily the spleen (the only shark secondary lymphoid tissue) and epigonal (the shark BM equivalent). Based on the ontogenic appearance of 19S and 7S IgM in plasma, we predicted that all secretory cells in neonates would express J-chain, followed by distinct subsets of J-chain+ and J-chain cells in adulthood. We also initiated studies of classical B-cell transcription factors in sharks, predicting that Blimp-1 would be expressed in all secretory cells, whether or not they expressed J-chain. Our results conformed well to the first prediction of the types of neonatal and adult plasma cells, but not in the predicted expression of Blimp-1. We discuss our data in sharks within the greater context of Blimp-1 expression in B cells subsets in all vertebrates.


Changes in IgM and J-chain expression in nurse shark spleen throughout development

Neonatal shark serum contains predominantly 19S IgM, along with IgM1gj, an IgM isotype that has a prerearranged (germline-joined) V domain [21]. 7S IgM is absent from sera during early development, but 7S IgM levels increase steadily as the animals age, so that by adulthood serum IgM is comprised of roughly equal levels of 19S and 7S IgM and negligible IgM1gj [26]. To examine how IgM expression at the cellular level matches the IgM composition in serum, we performed in situ hybridization using spleen tissue at different developmental points. The spleen is the only fish secondary lymphoid tissue, and in nurse sharks it contains both red pulp and white pulp (illustrated in Fig. 1A), the latter composed of T- and B-cell zones [71]. In contrast to the low IgM signal observed in white pulp B-cell zones, we have classified IgM+ cells localized near red pulp blood vessels or along the splenic capsule as antibody-secreting cells based on their intense IgM signal. This difference in staining intensity is clear when riboprobes specific for either the TM or constant domains (C2 or C3/C4) of IgM are used. TM probes only stain B-cell zones with low intensity, whereas C-specific probes intensely label cells near blood vessels and stain the white pulps weakly (Fig. 1A). Using riboprobes specific for IgM1gj, “conventional” 7S and 19S IgM (exon encoding the C2 domain, lacking in IgM1gj: Fig. 1B), or J-chain, which presumably discriminates between the 19S- and 7S-secreting cells, we have examined these secretory B-cell populations throughout development (Fig. 1).

Figure 1.

Expression of immunoglobulin genes in nurse shark spleen tissue throughout development. (A) In situ hybridization of 8 μm (nonconsecutive) sections of spleen tissue from an adult shark (shark W), stained with riboprobes specific for IgM-TM (left) or the C2 domain of IgM (right). Regions of white pulp B-cell zones (WP), red pulp (RP), and blood vessels (BV) are marked. Each photo was taken at 4× magnification, and shown with 150 μm scale bar. (B) Schematic mockup of domain structure of IgM and IgM1gj with approximate probe position marked. Positions of probes specific for conventional IgM (C2-domain specific), both IgM and IgM1gj (C3-C4-domain specific), IgMigj, or TM are noted. (C–E) In situ hybridization of 8 μm (nonconsecutive) sections of spleen tissue from three sharks ages: (C) 5 days (shark TH), (D) 1–2 months (shark LA), and (E) adult (shark LJ). Probes are specific for: IgM1gj (left), C2-domain of conventional IgM (middle), and J-chain (right). All photos were taken at 10× magnification and one image from each tissue shown with 100 μm scale bar. Each shark is representative of at least one other shark of that age (data not shown). Positive staining cells are dark. Negative controls with sense probes showed no staining (not shown).

In young animals (5 days postpartum and 1–3 months old), the number and location of cells expressing J-chain correlate well with the number of IgM- plus IgM1gj-secreting cells (Fig. 1C and D); thus, all IgM+ cells in these young animals are J-chain+ and IgM1gj was the predominant species expressed in a 5-day-old animal. At this stage, the few cells expressing canonical IgM H chains were not associated with the capsule or blood vessels where the IgM1gj+ cells were found. By 1–3 months of age, more canonical IgM-expressing cells were detected, localized near blood vessels and the capsule with the existing IgM1gj-expressing cells (Fig. 1D). By the time sharks reached 1 year of age, very few splenic IgM1gj-producing cells were observed. Since the total number of IgM+ cells greatly exceeded the number of J-chain+ cells at this stage, only a portion of all adult IgM+ cells must coexpress J-chain (Fig. 1E).

Our in situ hybridization results concur with previously reported serum levels of IgM1gj, 19S IgM, and 7S IgM: neonatal/young shark plasma cells express either IgM1gj or 19S IgM (both J-chain+); by 1 year of age, splenic IgM1gj expression is extinguished and only approximately half of IgM-secreting cells are J-chain+ (19S-producers) and the rest are J-chain (presumed 7S-producers). To study the control of splenic J-chain expression, we began by examining expression of the “master regulator” of plasma cell development, Blimp-1, by in situ hybridization.

Cloning of shark Blimp-1

The full-length sequence of nurse shark Blimp-1 gene, prdm1, was amplified from spleen RNA using primers to the Zn-finger-rich region from Xenopus laevis prdm1 (green highlighted region, Supporting Information Fig. 1), and used to probe a nurse shark cDNA library [72, 73] to obtain the full-length sequence (Supporting Information Fig. 1). The nurse shark sequence contains all major domains found in mammalian Blimp-1, including the PR domain and four Zn-finger domains (Supporting Information Fig. 1). The conserved domain structure within nurse shark Blimp-1 is also apparent when aligned to Blimp-1 from several other vertebrates (Supporting Information Fig. 2). A Blastp search matched nurse shark Blimp-1 to that of other vertebrate species with more than 60% aa similarity (E-values 0 in full-length), with highest conservation (nearly identical) in the Zn-finger domains (Supporting Information Fig. 2). To confirm its orthology, we performed phylogenetic tree analysis. We selected three prdm genes (PRDM4, 10, and 15) that are closely related to Blimp-1/prdm1 [37] and used PRDM2 as the outgroup. The nurse shark sequence clusters with those of Blimp-1 from other species rather than with other members of the prdm gene family (Supporting Information Fig. 3) and the nurse shark gene is the Blimp-1/prdm1 ortholog.

Blimp-1 is not expressed in a population of neonatal and adult IgM-secreting cells

In mammals, Blimp-1 is thought to be absolutely required for plasma cell development and function; it indirectly induces J-chain expression by repressing Pax5, a negative regulator of the J-chain promoter [46]. Thus, we were surprised to find that the majority of J-chain+/IgM+ and J-chain+/IgM1gj+cells in neonatal spleen were Blimp-1 by in situ hybridization (Fig. 2A, Table 1); note specifically the absence of Blimp-1 staining in the neonatal splenic capsule (dashed box inset) where many IgM1gj and 19S-secreting cells were found (Fig. 1 and 2).

Table 1. Summary of J-chain and Blimp-1 expression in Ig-secreting cellsa
  1. a

    In situ hybridization from Figures 1-5 and 7 are summarized, encompassing expression in the neonatal spleen along with adult spleen and epigonal tissue.

  2. (+) indicates presence of transcript by in situ hybridization, (-) indicates no expression was detected by in situ, and (N/A) indicates that the cell type in question is not detected in said tissue.

Neonatal spleen19S IgM+
Adult spleen19S IgM+
 7S IgM+
Epigonal19S IgMN/AN/A
 7S IgM
Figure 2.

Expression of IgM, J-chain, and Blimp-1 in adult nurse shark spleen and epigonal tissue. In situ hybridization of 8 μm (nonconsecutive) sections from (A, B) spleen (C) or epigonal tissue from (A) neonatal (shark LA) or (B, C) adult (shark EL) nurse sharks. Probes are specific for: IgM C3-C4 (left), which recognizes both IgM1gj and rearranging IgM, J-chain (middle), or Blimp-1 (right). Images taken at 10× magnification with 100 μm scale bar indicated in C3-C4 stain. Dotted boxes represent insets of 3× digital zoom from each 10× image. Positive staining cells are dark. Negative controls with sense probes showed no staining (not shown). Data are representative of at least two experiments for each stain set and age, each performed with two sharks.

Unlike neonatal sharks, Blimp-1 was expressed in adult spleen, but the number of Blimp-1+ cells was lower than the IgM-secreting cells (Fig. 3B). As mentioned, J-chain transcripts were also expressed in only about half as many adult cells as IgM (Fig. 1E and 2B). By inference from the young sharks, we predicted that there would be a J-chain+/Blimp-1 19S-secreting cell population in adults (Table 1). In this scenario, Blimp-1+ cells would therefore be the J-chain, 7S IgM-secretors, like conventional mammalian plasma cells. However, since splenic IgM-secreting cells formed large clusters under the splenic capsule and near blood vessels, it was difficult to determine whether, as predicted, Blimp-1 cells were truly J-chain+ and vice versa.

Figure 3.

Coexpression of IgM with either J-chain or IgLκ in epigonal tissue. In situ hybridization of 8 μm sections of epigonal tissue from an adult shark (shark W), double stained with either (A) IgM-DIG and J-chain-FLU or (B) IgM-DIG and Kappa light chain-FLU probes. IgM-positive cells (top) are green, and positive staining for (A) J-chain or (B) IgLκ staining (middle) is purple. Images are merged at bottom. Images on the left were taken at 10× magnification with 100 μm scale bar shown in IgM stain, and insets on the right are at 40× magnification with 50 μm scale bars. Black boxes denote the regions of magnified insets. Negative controls with sense probes showed no staining (not shown). Data are representative of at least two experiments for each stain set and age, each performed with one shark.

Due to this dense clustering of splenic plasma cells, we decided to study Blimp-1 and J-chain expression in another lymphoid organ. The epigonal organ is a primary lymphoid tissue, the shark BM equivalent. Since plasma cells are evenly dispersed throughout this tissue, this tissue seemed ideal for examining Blimp-1 expression in individual secreting cells. However, none of the epigonal IgM+ or J-chain+ cells expressed Blimp-1 (Fig. 2C, Table 1). Approximately equal numbers of epigonal IgM+ and J-chain+ cells were present, suggesting that all plasma cells in this tissue were IgM+/J-chain+/Blimp-1. However, upon closer examination of consecutive sections and with in situ hybridization double staining, most of the epigonal IgM+ cells were J-chain (Fig. 3A). Thus, surprisingly, almost all of the IgM+ cells were J-chain in this tissue, and none of them expressed Blimp-1 (Fig. 2C, Table 1). In addition to IgM and IgM1gj, nurse sharks also express two other Ig isotypes, IgW and IgNAR. Since there were about twice as many IgLκ+ plasma cells in the epigonal tissue as there were IgM+ secreting cells (Fig. 3B), it was likely that the IgM/J-chain+ cells expressed another H chain isotype. As IgNAR does not associate with L chains, it was predicted that the J-chain+ cells expressed IgW. Indeed, the majority of epigonal J-chain+ cells did coexpress IgW (Fig. 4A). Some J-chain+/IgM1gj+ cells were also found in the adult epigonal, but the IgNAR cells were all J-chain (Fig. 4B and C). Therefore, although the composition of plasma cells in the epigonal and spleen is quite different, Blimp-1 secreting cells are found in both lymphoid tissues. Specifically, J-chain+ plasma cells in both organs lack Blimp-1, but these cells coexpress predominantly IgW in the epigonal and IgM in the spleen. Also, the epigonal contains Blimp-1/IgM+/J-chain cells whereas these cells are predicted to be Blimp-1+ in the spleen (see Discussion and Table 1). Regardless of this major difference in Blimp-1 expression between IgM+/J-chain cells in adult spleen and epigonal, we emphasize that similar to young animals, adult secretory cells can be Blimp-1 and J-chain expression is not regulated by Blimp-1.

Figure 4.

Coexpression of J-chain and non-IgM isotypes in epigonal tissue. In situ hybridization of 8 μm sections of epigonal tissue from an adult shark (shark W), double stained with (A) J-chain-DIG and IgW-FLU, (B) IgM1gj-DIG and J-chain-FLU, or (C) IgNAR-DIG and J-chain-FLU probes. Cells positive for DIG-labeled probes (left) stain green. Cells positive for FLU-labeled probes (middle) stain purple. Images are merged at right. All photographs were taken at 40× magnification with 50 μm scale bar marked. Negative controls with sense probes showed no staining (not shown). Data are representative of two experiments performed with one shark.

Secretion of IgM protein

Although shark plasma cells express high levels of IgM RNA and lack TM IgM mRNA (Fig. 1A), studies so far have focused solely on mRNA expression. Therefore, we wanted to determine whether Blimp-1 cells truly secrete 19S IgM and IgM1gj. In order to examine protein secretion together with Blimp-1 and/or J-chain expression in the adult, we would need to isolate live 19S- and 7S-secreting cells. Unfortunately, mAbs that would unambiguously identify these populations or the secreted product are not available, and thus we focused on IgM1gj-secreting cells from young animals, which are also Blimp-1/J chain+ (Fig. 2A).

ELISPOTs were performed using plasma cell-enriched percoll fractions from spleen or epigonal tissue of a shark that was approximately 1-year-old. In this assay, secreted Ig was captured with the IgM1gj-specific mAb, TG4, and detected with biotinylated mAb GA16, which also binds to IgM1gj [21]. Significantly more IgM1gj-secreting cells were detected from epigonal than from spleen when at least 2.5 × 105 cells/well were plated. Average positive spots ± SEM at this dilution of spleen cells was 9.3 ± 3.9, whereas epigonal cells yielded 28 ± 3 spots per well (p-value = 0.01) (Fig. 5A). Presence of IgM1gj-secreting cells in the epigonal, but not spleen, of adult sharks is consistent with a previous study of Ig expression in the epigonal organ [21] and in situ hybridization data from another shark of this age (Fig. 1E). We were also able to detect conventional IgM secreted from epigonal cells via ELISPOT (data not shown); since all epigonal secretory B cells are Blimp-1 (Fig. 2C), and previous work demonstrated that Ig is secreted from epigonal cells [21], these cells clearly secrete Ig in the absence of Blimp-1 expression.

Figure 5.

IgM1gj protein secretion. ELISPOTs were performed to measure Ig-secretion in multiple tissues from sharks of different ages. Cells were enriched for plasma cells with a percoll gradient before plating. One shark Ig-specific mAb was used to capture secreted Ig from plasma cells and another biotinylated mAb was used for detection. (A) Number of positive ELISPOTs using cells from spleen (gray bars) or epigonal (white bars) tissue from a 1-year-old shark (shark EN), where secreted Ig was captured with anti-IgM1gj mAb, TG4, and detected with biotinylated mAb, GA16. (B) IgM1gj secretion from PBLs was also measured by ELISPOT using blood from an adult (shark B, gray bars) or young (shark R, white bars) shark, LK14 capture, and biotinylated TG4 detection. Data are shown as mean ± SEM of triplicate cell platings for that condition and *p < 0.05, T-test. (C) Representative ELISPOT wells of secreted IgM1gj from (B) or from CB5-coated, LK14-biotin-detected Ig are shown. (D) Immunoprecipitations were performed on metabolically labeled supernatants from the same cells used in (B) and (C). Samples from protein G preclears or immunoprecipitations with CB5 (anti-IgM), TG4 (anti-IgM1gj), or GA10 (anti-NAR) were run nonreduced on 5% gels and film was exposed for 1 week [21]. Data shown are from one experiment representative of two performed.

IgM1gj-secreting cells were also detected in the blood of both a young (∼6-month-old) and adult (9-year-old) shark, although significantly more IgM1gj-secreting cells were found in the blood of the young animal than the adult (p-values = 0.01, 0.02, and 0.03 for 1.25 × 105 cells/well, 2.5 × 105 cells/well, 5 × 105 cells/well, respectively) (Fig. 5B). There was less of this secreted IgM1gj (captured with anti-L chain mAb, LK14, and detected with IgM1gj-specific, biotinylated TG4) than total conventional IgM (CB5 capture, LK14-biotin detection) in both of these sharks, and the size of the IgM1gj spots was smaller (Fig. 5C), which could indicate that these Blimp-1, IgM1gj-secreting cells are producing less protein than conventional IgM-secretors. Since there are no IgM1gj-secreting cells in the spleen of adult sharks (Fig. 1E) [21], it was surprising to find a substantial amount in the blood of an adult shark. However, immunoprecipitation with TG4 of radiolabeled PBL cell supernatants from the same 6-month and 9-year-old sharks confirmed this IgM1gj secretion (Fig. 5D). In addition to IgM1gj, 19S and 7S IgM were seen in both sharks after immunoprecipitation with mAb CB5, and a large amount of 19S IgM was also depleted during preclear with protein G beads, which is expected due to the “stickiness” of high avidity 19S [20]. There was little IgNAR secreted from cells in the young animal (GA10 IP), which is consistent with its age. Since there is negligible IgM1gj in the spleen of adult sharks, it is possible that these secreting cells may be trafficking through the blood from the epigonal organ.

Pax5 is not expressed in any secretory cells

As mentioned, Blimp-1 represses Pax5 in mammalian plasma cells resulting in a plethora of effects [74], including upregulation of J-chain expression; however, since many shark Ig-secreting cells do not express Blimp-1, we wondered whether Pax5 might not be extinguished in these cells, similar to what has been reported for some mammalian secretory cells [70]. Pax5 was expressed in the splenic B-cell zones of both neonatal and adult nurse sharks, but no staining was detected in areas where plasma cells congregate (i.e. around blood vessels, black arrows, Fig. 6B), whether or not they expressed Blimp-1 or J-chain, as expected based on the mammalian model (Fig. 6). Therefore, Pax5 expression is extinguished in all shark Ig-secreting cells, and some other factor besides Blimp-1 must repress Pax5 expression in a subset of nurse shark Ig-secreting cells.

Figure 6.

Pax5 expression in young and adult spleen tissue. In situ hybridization staining of 8 μm (nonconsecutive) sections of spleen tissue from (A) a 1-month-old (shark H) or (B) an adult shark (shark W). From left to right probes are specific for: IgM spanning C3-C4, J-chain, Blimp-1, or Pax5. Black arrows indicate regions near blood vessels where plasma cells congregate. All photographs taken at 10× magnification with 100 μm scale bar shown. Positive staining cells are dark. Negative controls with sense probes showed no staining (not shown). Data shown are representative of two experiments performed with one shark.

J-chain and Blimp-1 are noncoordinately expressed in adult spleen

While it was clear from the results in neonatal spleen and adult epigonal that J-chain+ Ig-secreting cells were Blimp-1 (Fig. 2A and 5 and Table 1), staining of adult spleen was problematic because of the great number of clustered plasma cells. We attempted double-stain, in situ hybridization experiments with Blimp-1 and J-chain probes in spleen, but this also proved difficult due to the vastly different expression levels between the transcription factor and Ig/J-chain. We therefore performed single stains on consecutive, 6 μm spleen sections and “field-stitched” images from consecutive slices together to examine coexpression of IgM with J-chain, IgM with Blimp-1, and, most importantly, Blimp-1 with J-chain.

Since most plasma cells are approximately 10 μm in diameter, they are large enough, in many cases, that one cell will be visible in consecutive 6 μm sections, however some cells will obviously not carry over to the next slide. In order to determine how often one cell could be detected in consecutive sections, we performed control stains using the same probe on successive slices. For these controls, we used probes specific for the previously mentioned IgNAR isotype that is expressed in approximately 90% fewer splenic plasma cells than IgM, allowing detection of nonclustered IgNAR+ cells.

Because both consecutive sections were photographed in bright field, they were pseudocolored; one image in orange and the other in blue, which allowed us to distinguish the images once they were overlaid. Because the pseudocolor affected only the white space in the photos, any cell that was positive in image 1 (orange) but negative in image 2 (blue) appeared blue in the overlay. Conversely, any cell that was negative in image 1 and positive in image 2 appeared orange. Tissue sections with no positive cells were neutral-colored in the merge and, most importantly, cells that were positive in both remained black. When the IgNAR-specific probes were used on consecutive sections, five cells were found in a representative field of view that were IgNAR+ in both sections and appeared black in the merged image (Fig. 7A, black arrows), whereas only two cells were found in the first but not second image, and appear blue in the merge (Fig. 7A, black circles). Therefore, the majority (five out of seven) of IgNAR+ cells detected in one tissue section were also found in the following section, giving us confidence in the ability to use this procedure with other probes.

Figure 7.

In situ hybridization for coexpression using consecutive spleen tissue sections. In situ hybridization of consecutive 6-μm sections of adult spleen tissue (shark W). Consecutive sections were stained with the probes indicated and two images were pseudocolored and field stitched together (right column). Probe pairs were (A) IgNAR:IgNAR, (B) J-chain:IgNAR, (C) IgM:J-chain, (D) Blimp-1:J-chain, and (E) Blimp-1:IgM. Vertical white arrows indicate same probe used as photo above. Presence of black stain in the merge (far right) column indicates coexpression. (A) Arrows indicate cells expressing IgNAR in both fields. Circles indicate IgNAR+ cells only seen in one of the two fields. (B) Arrows indicate IgNAR+/J-chain cells (orange in merge). All images were taken at 40× magnification and 100 μm scale bar is indicated in left-hand column. Each region is representative of multiple stains. Data shown are representative of at least three experiments performed with two sharks.

Since IgNAR is a monomer in serum, and since no J-chain+/IgNAR+ double staining cells were found in the epigonal tissue (Fig. 4), we did not expect coexpression of J-chain in IgNAR plasma cells in the spleen. Furthermore, like 7S IgM, IgNAR makes up an enhanced-affinity, antigen-specific response [20], so we predicted that the J-chain expression state of IgNAR+ plasma cells would be similar to the Blimp-17S secretors (also see Discussion). Therefore, as an additional control, we examined IgNAR and J-chain expression in these consecutive sections and indeed, the majority of IgNAR-secreting cells in the spleen were J-chain. Black arrows highlight IgNAR single-positive cells, which appear orange in the merge; whereas all blue cells in the merged image are J-chain+/IgNAR (Fig. 7B).

As expected based on raw numbers of positive cells (Fig. 1 and 2), approximately half of the IgM-secreting cells in the spleen were J-chain+ (Fig. 7C) and, similarly, half of the IgM+ cells coexpressed Blimp-1 (Fig. 7E). However, very few cells coexpressed J-chain and Blimp-1 together, as evidenced by the lower number of black cells in the Blimp-1/J-chain merge when compared with that in the IgM/J-chain (Fig. 7C) or IgM/Blimp-1 (Fig. 7E). Therefore, the J-chain and Blimp-1 expression patterns in the adult spleen are consistent with our observations in the neonatal shark: J-chain+ cells are predominantly Blimp-1 and vice versa (Table 1).

J-chain+, Blimp-1 cells express lower levels of IgM

Since Blimp-1 is traditionally believed to regulate high levels of IgM secretion, and since the J-chain+ cells were found to be Blimp-1, it is possible that, without Blimp-1, these cells may secrete lower amounts of IgM than the Blimp-1+ cells; similar to what has been suggested for mammalian B1 cells [69, 75]. Additionally, the ELISPOT spot size of Blimp-1 IgM1gj secreting cells was noticeably smaller, than other cell types, suggesting that they may be secreting low levels of protein. If true, the J-chain+/Blimp-1 19S cells may also have fewer IgM transcripts than the Blimp-1+ 7S cells.

When the IgM mRNA signal in plasma cells was visualized with a fluorescently conjugated secondary antibody instead of our standard chromogenic in situ hybridization method (Fig. 126, and 7), we found variability, almost a dichotomy, in the IgM signal intensity in secreting cells (Fig. 8A). To determine whether the J-chain+ cells were, in fact, staining less intensely for IgM, we performed in situ hybridization double-stains using a digoxigenin (DIG) labeled IgM probe, detected fluorescently, and followed by chromogenic detection of a fluorescein (FLU) labeled J-chain probe. We found that the J-chain+ cells did indeed have a lower IgM signal than many J-chain cells (Fig. 8A). However, it should be noted that the level of IgM transcript in these IgMlow secreting cells is still much higher than that seen in nonsecreting B cells found in the white pulp (Fig. 1).

Figure 8.

Variable strength of IgM signal in J-chain positive and negative spleen cells. In situ hybridization double stain of 6 μm sections spleen tissue from an adult shark (shark W). (A) Slides were double stained with IgM-C3-C4-DIG (left) and J-chain-FLU (middle) riboprobes. Images merged at right. Photographs taken at 40× magnification and marked with a 100 μm scale bar. (B) Data from 50 cells each of J-chain+ and J-chain cells (from multiple fields of view) were pooled to get an average mean fluorescent intensity of the IgM stain (green). Data are shown as mean ± SEM. *p < 0.05, T-test.

In order to quantify the level of IgM signal in the splenic secretory cells, the mean brightness of IgM stain was analyzed in J-chain+ and J-chain cells. All J-chain+ cells and any brightly staining IgM cells that were J-chain were included in this analysis. J-chain/IgMlow cells were excluded from this analysis since we could not be certain whether they were staining dimly because they actually expressed lower amounts of IgM mRNA, or because during sectioning the cell was sliced at its edge, resulting in a deceptively low signal. The average brightness of J-chain+ and J-chain cells was measured using the densitometric mean of the green (IgM) signal for each selected cell. Based on this analysis, the majority of the brightest IgM+ cells were J-chain, while none of the very bright cells coexpressed J-chain; conversely, J-chain+ cells were almost universally IgM-dim (Fig. 8B). J-chain cells had an average densitometric mean of 2173.2 ± 140.1, whereas J-chain+ cells were dimmer, at 1688.7 ± 127.4 (p-value = 0.01). In summary, the J-chain+/Blimp-1 19S-secreting cells express lower levels of secreted IgM transcript than the J-chain/Blimp-1+ 7S-secretors.


We have found a population of Ig-secreting, J-chain+ cells present in both adult and young nurse sharks that do not express Blimp-1. Therefore, unlike conventional mammalian plasma cells, neither the antibody secretion functions nor the expression of J-chain are under the control of this master transcription factor in sharks. Thus, in sharks, Blimp-1 seems to be more important for terminal differentiation and cell cycle arrest than for Ig secretion and J-chain expression. However, Blimp-1/J-chain+ cells do express lower amounts of IgM transcript than Blimp-1+/J-chain cells, and ELISPOTs from IgM1gj-secreting cells are smaller than those of conventional Ig, suggesting that the 19S- and gM1gj-secreting cells produce somewhat lower levels of secreted IgM protein than 7S-secretors, consistent with some recent models in mammals [9, 69, 75].

The secreted high-avidity, 19S IgM is known to be cross-reactive and it may be beneficial for the shark to have a large proportion of cells that can be easily induced to secrete such “innate-like” Ig, even if it is not produced at the same high levels as “adaptive” Ig. This may be especially important for the nurse shark, since the 7S and IgNAR responses, while comprising the typical antigen-specific response, take some time to reach peak levels [20]. Therefore, having cells that can be easily triggered to secrete natural antibodies might be of added importance as a first line of defense in an animal with such slow adaptive kinetics. These 19S-secreting cells could either self-renew like mammalian B1 cells, or be continually derived from precursors after stimulation with commensal bacteria or other common signals.

Based on the differences in gene expression between 19S- and 7S-secreting cells, we hypothesize that the transcriptional expression profile of these cells may be related to how the cells arise during B-cell development and that one of two models may be true: either 19S and 7S producers develop from separate lineages, like B1 and B2 cells in mammals (model 1 below), or plasma cells go through a 19S-secreting plasmablast state before “switching” to become 7S producers by shutting down J-chain expression (model 2 below). Of course, the actual situation may be more complicated than either of these models and could even be a mixture of the two.

Model 1: 19S secreting cells as a B1-like lineage

In adult spleen, J-chain+, 19S IgM-secreting cells do not express Blimp-1, whereas J-chain, 7S-secreting cells express high levels of Blimp-1. The 7S, Blimp-1+ cells may encompass a population of conventional, B2-like cells, which, after undergoing traditional antigen-specific expansion and hypermutation, would require Blimp-1 to extinguish these processes. The 19S-secreting cells, however, may come from a B1-like lineage. Certainly, the presence of 19S IgM-secreting cells early in development is similar to the early appearance of B1 cells in mammals. Additionally, as mentioned above, there are reports that mammalian B1 cells do not require Blimp-1 for antibody secretion [70]. Indeed, if 19S secreting cells continually divide throughout life, it may be a requirement that these cells do not express Blimp-1, since terminally differentiated Blimp-1+ cells would be unable to self-renew. In this model, the 19S-secreting cells, like IgM1gj-secreting cells, may persist long term in the adult epigonal organ (Table 1 and discussed further below).

Model 2: 19S secreting cells as a developmental stage

Alternatively, the 19S-secretors may constitute a plasmablast-like developmental stage, as has also been suggested in mammalian B1 cell models [9, 69]. It has been shown that Blimp-1 is not expressed in mouse preplasmablasts, even though Pax5 is already downregulated at this stage [68]. Nurse shark 19S-secreting cells could therefore represent a similar transitional state between B cells and the end-stage 7S-secreting plasma cells. In this model, during an antigen-driven immune response, J-chain expression would eventually be extinguished, allowing 19S-secreting cells to “functionally class switch” to a 7S-IgM secreting cell type, coinciding with high levels of Blimp-1 and IgM expression, characteristic of true plasma cells. If model 2 is correct, this transitional stage appears to be more pronounced and longer lasting in the nurse shark than in mammals, since 19S-secreting J-chain+ cells make up a major proportion of adult spleen cells regardless of immunization status.

Regardless of whether J-chain+/Blimp-1 19S cells represent a separate lineage or developmental state, additional transcription factors must be required to play a role in controlling Pax5, Blimp-1, and J-chain expression. Although it is currently unclear what these factor(s) could be, identification of such elements will be important for clarifying whether these cells truly come from distinct lineages or if 19S cells somehow “mature” into 7S secretors (as in model 2). If model 1 is true, we hypothesize that shark B cells are clonally marked from their earliest stages for J-chain expression, and perhaps committed to separate lineages early, as suggested for mouse B cells by Leanderson [30].

Long-term survival in the primary lymphoid organ

We have also found a population of Ig-secreting cells in the epigonal organ that is J-chain and Blimp-1 (Fig. 3), thus differing in Blimp-1 expression status from the J-chain cells in the spleen (Table 1). This apparent contradiction may be explained by the proposed nature of cells in the epigonal; like mammalian BM, the epigonal organ likely contains a subset of long-lived secreting cells that persist throughout life. For example, the IgM1gj-secreting cells that develop early in life do not express Blimp-1, and persist only in the epigonal tissue of adult sharks ([21] and Table 1). Perhaps, not just IgM1gj but all epigonal-resident plasma cells continue to slowly divide throughout the life of the animal. In this scenario, IgW+/J-chain+ cells would make up one lineage (with unknown function), while IgM+/J-chain cells may be a second long-lived, antigen-specific lineage that continue to persist in the epigonal after B-cell activation in the spleen. In either case, the epigonal appears to be an organ in which Blimp-1 plasma cells of various types survive and are capable of long-term self-renewal.

Future studies in shark

The ability to separate 19S- from 7S-secreting cells would greatly aid future characterization of these cell types, and identification of the factors controlling J-chain expression. Additionally, by separating 19S and 7S cells, we would be able to analyze the IgM repertoires, including CDR3 length and N-region additions. If the two forms do originate from different cell lineages, we would predict observation of signature differences between the two, with the 19S IgM repertoire perhaps being less diverse, as has been described for natural antibodies in mammals [76]. Shark 7S, but not 19S IgM, shows augmented affinity to antigen after immunization [20], so it would also be interesting to follow mutations in the IgM-secreting cells throughout the course of an immune response. We would predict that only 7S-secretors would contain mutations, but by following an antigen-specific response it may be possible to identify antigen-specific 19S cells found early in a response, and to compare these to 7S secreting cells arising later. If IgM secreting cells switch as per model 2, we would likely find unmutated and mutated sequences from the same antigen-specific BCR within the 19S and 7S populations, respectively.

Relevance of shark Blimp-1 and J-chain expression to mammalian and other vertebrate systems

We predict that this dichotomy of J-chain and Blimp-1 expression is not a shark-specific phenomenon. Unlike monomeric Ig H chain classes in mammals, such as IgG, where the expression of J-chain has no effect on the structure of the secreted isotypes, in sharks, regulation of J-chain expression is important because 7S IgM is secreted only in the absence of J-chain. Even in mammalian systems, it is known that J-chain is only expressed in a subset of plasma cells. For example, some IgA-secreting cells in the mouse are J-chain and secrete IgA monomers. Therefore, transcriptional control of J-chain is likely to be as important in at least a subset of mammalian plasma cells as it is in sharks, and in many mammalian systems, the importance of J-chain may have simply fallen under the radar since it does not associate with IgG.

It has been documented that J-chain+ IgA-secreting cells of the mammalian lamina propria are derived from B1 cells, whereas B2 cells secrete J-chain IgA [77]. We therefore predict, based on our results in nurse shark, that the J-chain+ (B1-derived, IgA+) cells may be Blimp-1 negative, whereas the J-chain (B2-derived, IgA+) plasma cells would express Blimp-1. From a broader evolutionary perspective, similar predictions can also be made for other nonmammalian vertebrate models. Both teleosts and Xenopus have dedicated mucosal Ig classes (IgT and IgX, respectively) [78-80], and these mucosal Ig-secreting cells may contain B1-derrived secreting cells, similar to mammalian J-chain+, IgA secretors. Although it is important to note that while the teleosts genome has lost J-chain, mucosal Ig secreting cells may lack Blimp-1, similar to shark 19S secretors, and in Xenopus, mucosal IgX+ cells may be J-chain+/Blimp-1.


While results with Blimp-1 expression in plasma cells have been controversial in mammalian systems, here we describe naturally occurring antibody secreting cells, found in multiple lymphoid tissues in both young and adult nurse sharks, which do not express Blimp-1. Additionally, we found that J-chain+, 19S IgM-secreting cells are Blimp-1, while Blimp-1 is expressed normally in 7S secretors. Although IgM transcripts are expressed at lower levels in the J-chain+ cells, these cells nevertheless secrete detectable 19S IgM in the absence of Blimp-1. Our observation, that Blimp-1 is not expressed in all antibody-secreting shark cells, supports the noncanonical model of Blimp-1 expression in the Ig secretion program of mammalian B1 cells. Thus, our results not only make inroads to the age-old problem of the production of 19S and 7S IgM in secretory B cells in sharks, but also have implications for the role of Blimp-1 in Ig secretion in B-cell lineages in all vertebrates.

Materials and methods

Animals, tissue collection, and cell preparations

Nurse sharks (G. cirratum) were captured off the coast of the Florida Keys and maintained in artificial seawater at approximately 28°C in large indoor tanks at the Institute of Marine and Environmental Technology, in Baltimore, MD. As per animal protocol (IACUC 1012003), animals were euthanized and bled out after an overdose of MS222. Spleen and epigonal tissues were harvested and fixed in 4% paraformaldehyde in SBP. Tissue was then washed in sucrose gradient solutions for more than 24 h before mounting in OCT and freezing in a methyl butane bath cooled with liquid nitrogen. Once frozen, tissue was cut into either 6- or 8-μm sections using a Leica Cryostat and mounted on slides. Cut tissue was kept at –80°C until ready to use for in situ hybridization. Frozen, fixed tissue used was taken from six sharks of various ages. Shark TH and H were approximately 1 week old. Shark LA was 1–3 months old, and LJ, EL, and W were all adults over 1 year of age.

For immunoprecipitation and ELISPOT studies, shark cells were collected from dissociated spleen or epigonal tissue or from blood and washed in “shark PBS plus” (PBS supplemented with 0.35 M urea, 0.14 M extra NaCl, 10% FCS, and DNase). Plasma cells were enriched by fractionating over a 1.054 density percoll gradient, and washed again in shark PBS plus before either radiolabeling or resuspending in RPMI with 10% FCS, 0.35 M urea, and 0.14 M extra NaCl, for overnight culture at 27°C. Cells were taken from three sharks of various ages: shark R is 6 months in age, EN is approximately 1-year-old, and B is an adult shark of at least 9 years.

In situ hybridization

DIG- or FLU-labeled riboprobes were made with the following primers: IgM1gj, IgM C2, IgM C3-C4, J-chain, IgL Kappa, Blimp-1, or Pax5. For regular in situ hybridization, DIG-labeled probes were detected with anti-DIG-alkaline phosphatase (AP) and nitroblue tetrazolium and 5-bromo-4-chloro-3′-indolyphosphate (NBT/BCIP) substrate. Experiments including transcription factors were developed using Tyramide Signal Amplification (TSA) Plus Biotin or DNP System (Perkin Elmer) with either streptavidin-AP or anti-DNP-AP and NBT/BCIP substrate. For in situ hybridization staining of consecutive 6-μm sections, probes were detected with TSA Plus Biotin System (Perkin Elmer) with streptavidin-AP and NBT/BCIP substrate. The slides were photographed and the images were field-stitched together using Zeiss's AxioVision software, version For double stains, DIG- and FLU-labeled probes were hybridized together, and the fluorescent signal (DIG-labeled probe) was detected first using anti-DIG rabbit olioglonal (Invitrogen) primary antibody and AlexaFluor 488 goat anti-rabbit secondary antibody. The slides were photographed before completing chromogenic detection of the FLU-labeled probe using PerkinElmer's TSA Plus Biotin System with streptavidin-AP and NBT/BCIP substrate. Slides were then rephotographed and images were aligned using the previously mentioned software.

Blimp-1 cloning

Primers were designed to amplify the Zn-finger-rich region of X. laevis Blimp-1 (Genbank accession AF182280, xBlimpF1: 5′-CTGGTTATAAGACTCTCC-3′, xBlimpR1: 5′-GCACAGGATGTAGCTC-3′) from X. laevis spleen RNA. The 428 bp product was labeled and used to probe a nurse shark cDNA library, as previously described [72, 73, 81]. Two identical Blimp-1 clones were sequenced that were both truncated at the 5′ and 3′ ends. The full-length sequence was obtained using Clontech's SMART RACE cDNA amplification kit (Clontech) using gene-specific primer (GSP):5′ RACE used GcBlimpRR1 (5′-CTTCGAGGCAGGTCAATGCACTTTAGC-3′) for first round PCR with the kit-supplied universal primer, followed by the second PCR with the nested GSP GcBlimpRR2 (5′-CTGCAAACTCGCTGCTGTACCAGAC-3′) with the kit-supplied universal primer. 3′ RACE was performed by using GcBlimpRF1 (5′-CAACCTAAACCTACCTCAGCTGGCTTG-3′) for first round PCRs with the kit-supplied universal primer and GcBlimpRF2 (5′-CAAGGTCCATTTGAGAGTCCATAGTGGG-3′) for nested second round PCR with the kit-supplied universal primer. In order to complete the sequence, 3′ RACE was repeated using the primers GcBlimpRF5 (5′-CCTCAAAGGAAACTGCCCCGTCGCC-3′) for first round PCR and GcBlimpRF6 (5′-GACCTGAACCGCCTGAACGAGG-3′) for nested second round PCR (with kit-supplied primers as above).The full-length sequence was confirmed using the primers GcBlimpFullF1 (5′- GTGGATGAAAATGGACATGACCC-3′) and GcBlimpFullR1 (5′-AGGTCAGAGACACGTTGCATCTC-3′).

Blimp-1 tree

The V domains of the deduced aa sequences were aligned using ClustalX, and neighbor-joining bootstrapping trees (500 trials) [82] were made and viewed in the TreeView 1.6.6. program [83].


Ninety-six-well plates (Millipore MultiScreen HA plates #MAHAS4510) were coated with 100 μL/well mAbs (CB5, TG4 or LK14) at 10 μg/mL, and incubated overnight at 4°C to coat. The following day the plates were washed five times with 100 μL/well sterile PBS, and then blocked with 200 μL/well DMEM with 10% serum for 2–3 h at 37°C. Plates were washed again five times with 100 μL/well sterile PBS before plating cells in shark RPMI and incubating overnight at 27°C in humidified incubator with 5% CO2. The following day, cells were washed away with 3× PBS washes followed by 3× PBST (PBS with 0.05% Tween) of 200 μL/well, with 1-min incubations between washes. A total of 100 μL/well of biotinylated detection antibodies (LK14-biotin, GA16-biotin, or TG4-biotin) were added at 5 μg/mL in PBST with 1% FCS and plates were incubated for 2 h at room temperature. 3× PBS and 3× PBST washes were repeated and plates were incubated with 100 μL/well streptavidin-AP secondary at 1:2000 in PBST with 1% FCS for 1 h at room temperature. Plates were washed again five times with PBS, and AP signal was detected with 100 μL/well NBT/BCIP diluted in 0.1 M Tris pH 9.5, 0.1 M NaCl buffer. Plates were washed in water to stop development reaction and left to dry before reading and analyzing spots using CTL Immunospot capture software version 6.3.5 and analysis software version 5.0. Significance was determined using an independent two-tailed Student's t-test.


Nurse shark cells were incubated in RPMI minus MET for 30 min at 27°C followed by addition of 1 μCi S35 to cells and incubation at 27°C overnight. Supernatant was collected the following day. Supernatants were precleared with protein G sepharose beads (Amersham #17–0618–01) (100 μL beads/500 μL supernatant) for 1 h at room temperature, before incubating with mAbs (2 μL purified CB5, TG4, or GA10) overnight at 4°C, while rotating [21]. The following day supernatants were incubated with 30 μL of protein G beads each for 1 h at 4°C with rotation. Beads were washed two times with NET-NON and one time with NET-N, before boiling in 30 μL Laemmli sample buffer and loading (nonreduced) on a 5% sodium dodecyl sulfate (SDS) acylamide gel. After running, gels were fixed for 1 h and incubated two times with dimethyl sulfoxide (DMSO) at room temperature for 30 min while shaking. Gels were then washed with DMSO-PPO (2,5-dipheynl oxazole) overnight followed by washing with water and dehydration before 1-week film exposure and development.


We would like to thank Rezwanul Wahid and Harold Neely for their generous help with ELISPOTs and antibody purifications. This work was funded by NIH RR006603. C.C. was a trainee under the Institutional Training Grant T32AI007540 from the National Institute of Allergy and Infectious Diseases and the Institutional Training Grant in Cardiac and Vascular Cell Biology #5 T32 HL072751 from the NIH/NHLBI while this work was completed.

Conflict of interest

The authors declare no financial or commercial conflict of interest.


alkaline phosphatase




B-lymphocyte-induced maturation protein 1


constant domain 2






secretory component




Tyramide Signal Amplification