Identification of the Fc-alpha/mu Receptor in Xenopus provides insight into the emergence of the poly-Ig receptor (pIgR) and mucosal Ig transport

Abstract

The poly-Immunoglobulin Receptor (pIgR) transcytoses J chain-containing antibodies through mucosal epithelia. In mammals, two cis-duplicates of PIGR, FCMR and FCAMR, flank the PIGR gene. A PIGR duplication is first found in amphibians, previously annotated as PIGR2 (herein xlFCAMR), and is expressed by antigen-presenting cells. We demonstrate that xlFcamR is the equivalent of mammalian FcamR. It has been assumed that pIgR is the oldest member of this family, yet our data could not distinguish whether PIGR or FCAMR emerged first; however, FCMR was the last family member to emerge. Interestingly, bony fish ‘pIgR’ is not an orthologue of tetrapod pIgR, and possibly acquired its function via convergent evolution. PIGR/FCAMR/FCMR are members of a larger superfamily including TREM, CD300, and NKp44, which we name the ‘double-disulfide Ig superfamily’ (ddIgSF). Domains related to each ddIgSF family were identified in cartilaginous fish (sharks, chimeras) and encoded in a single gene cluster syntenic to the human pIgR locus. Thus, the ddIgSF families date back to earliest antibody-based adaptive immunity, but apparently not before. Finally, our data strongly suggest that the J chain arose in evolution only for Ig multimerization. This study provides a framework for further studies of pIgR and the ddIgSF in vertebrates.

Graphical Abstract

Phylogenetic, functional, and syntenic analysis reshape our understanding of antibody transcytosis and related receptors. The canonical poly-Immunoglobulin Receptor (pIgR) emerged first in amphibians, concurrent with FcamR, while we suggest J chain emerged initially for antibody multimerization.

Introduction

The polymeric immunoglobulin receptor (pIgR) is an antibody-transport receptor. pIgR is displayed on mucosal epithelial cells facing the lamina propria where it binds multimeric antibodies, dimeric IgA or pentameric IgM, in a J-chain dependent fashion. Upon binding, the receptor/Ig complex is internalized by epithelial cells and transcytosed to the lumen(1–4). During transcytosis, pIgR is cleaved, resulting in the lumenal secretion of Ig associated with the retained fragment of pIgR, called the secretory component(1). Transcytosed IgA (and to a lesser extent IgM) is vital for preserving homeostasis as it protects epithelia from toxins and microbes, and contributes to regulating commensal microbiota(5). While pIgR performs a unique and vital role in immune protection and homeostasis, it is one of a family of three related receptors in mammals. This family, referred to here as the PIGR family, is comprised of PIGR, FCAMR (CD351), and FCMR (FAIM3, TOSO). In the genome the PIGR family genes are in a single syntenic region. PIGR is located on human chromosome 1q32, and flanked by the two paralogues, FCAMR, and FCMR (Fig. 1A). FCMR and FCAMR encode the IgM-restricted receptor FcmR(6, 7) and the IgA/IgM-restricted receptor FcamR(8), respectively. These three genes and their genomic organization are conserved in all placental mammals examined. Like pIgR, FcamR binds IgM and IgA, but unlike pIgR, the binding is J-chain independent(9). FcamR also has cellular expression distinct from pIgR. In mammals, pIgR is expressed on mucosal epithelia, while FcamR is expressed by follicular dendritic cells (FDC), subsets of macrophages and B cells, and kidney tubules(8, 10, 11). FcmR aids the development of B-2 cells, and like FcamR limits production of autoantibodies(12–14).

Figure 1. The polymeric immunoglobulin (PIGR) family is comprised of three receptors in mammals and two receptors in amphibians.

Annotations based on prior literature. 1A. Mammalian FCAMR and FCMR are single-domain receptors homologous to D1, the N-terminal domain, of PIGR. Two PIGR family members are identifiable in amphibians. One is canonical PIGR, and the other has been annotated as PIGR2, but is established in this work as the equivalent of mammalian FCAMR. 1B. Mammalian FCAMR and equivalents in ectothermic vertebrates have sequence similarity. The alignment shows FCAMR and “PIGR2” or “PIGRL” N-terminal or “D1” domains. Background shading represents conservation of the residue at that site. Canonical cysteines in IgSF domains indicated with a filled star and those specific to ddIgSF with and open star. This second intrachain disulfide bond is the defining characteristic of this family, hence the double-disulfide Ig superfamily.

The PIGR family genes are members of an under-studied superfamily defined by a specialized Variable (V)-type immunoglobulin superfamily (IgSF) domain, which has the canonical IgSF disulfide bond between the two sheets in the B and F strands, but also an additional disulfide bond linking the C and C’ strands(15) (Fig. 1B). These receptors have been given various names including: modular domain immune-type receptors (MDIRs)(16), soluble immune-type receptor (SITR)(17), PIGRL (PIGR-like) genes in bony fish(16–18), novel immunoglobulin-like transcripts (NILTs) in zebrafish(19), and the CD300/CMRF35 receptors. To create a unified naming convention based on shared structure, herein we propose to rename them the double-disulfide Immunoglobulin SuperFamily (ddIgSF). However, respecting the previous studies of this gene family we continue to use prior names that apply to families within the ddIgSF. Many of the ddIgSF functions have only recently been investigated(20–23).

Evolution of the various families within the ddIgSF, particularly the polymeric Ig receptor (PIGR) family, has been discussed in recent reviews and primary papers(24, 25), but the analyses have neither comprehensively included all ddIgSF families nor given special attention to transitional species important in evolution, such as amphibians, lobe-finned fish, and cartilaginous fish. In our previous work on multifunctional antigen presenting cells (APC) in Xenopus, we suggest that the PIGR duplicate found in amphibians, previously annotated as PIGR2 (herein xlFCAMR), is likely the FCAMR equivalent(26). To further evaluate this hypothesis, we assessed the identity and evolutionary history of the Xenopus xlFCAMR via synteny, position of each ddIgSF domain in phylogenetic trees, and binding capacity to the three Xenopus laevis Ig isotypes. It has been proposed that PIGR is the oldest member of this family, arising in bony fish, with FCMR emerging in reptiles and birds, and FCAMR subsequently arising in mammals(24). Herein we also reevaluate and revise this hypothesis, using new data from all jawed vertebrates (gnathostomes), including the oldest living member of this taxon, the cartilaginous fish (sharks, rays, chimeras).

Although pIgR-mediated transport requires the J-chain, the appearance of IgM and the J chain in cartilaginous fish predates the earliest described IgM receptor (the presumed PIGR equivalent in ray-finned fish) by 50 million years. It was shown previously that the ddIgSF family clearly emerged earlier than PIGR in gnathostomes, but apparently not in agnathans or lower deuterostomes(16). We confirm the results of this earlier study and evaluate cartilaginous fish genomes for domains orthologous to PIGR. In cartilaginous fish we identify a cluster of ddIgSF genes that are syntenic to the typical PIGR cluster found in tetrapods, and we perform phylogenetic analyses in an attempt to determine the relationship of these genes to all other vertebrate ddIgSF genes. From our examination of these genes, we propose new paradigms for the emergence of Ig transcytosis via pIgR in vertebrates.

Results

xlFCAMR Binding Resembles Mammalian FCAMR

As described above, the pIgR family in humans and other mammals is composed of three members, pIgR, FcamR, and FcmR (Fig. 1A). pIgR has five ddIgSF domains while FcmR and FcamR each have only one domain, previously shown to be homologous to pIgR domain 1 (D1, alignment in Figure 1B, which also shows the general structure of the ddIgSF domain). Xenopus laevis pIgR was described previously as a four-domain molecule, lacking mammalian pIgR domain 3(27). The Xenopus PIGR duplicate, xlFCAMR, was previously identified as a three-domain molecule, which was thought to have lost pIgR domains 4 and 5 and was proposed to be FcmR(24). However, we previously demonstrated by in situ hybridization that xlFCAMR is expressed at high levels by activated Xenopus myeloid-derived APC (XL cells) that are similar to mammalian FDC, and bind frog IgM and the IgA-equivalent IgX(26). In addition, xlFCAMR is not expressed at mucosal sites, but is expressed at low levels by B cells(26, 28). Tissue expression by qPCR showed highest levels of xlPIGR expression in the intestine and pancreas, with lower levels in thymus and lung (Fig. 2). While mucosal expression in intestine and lung was expected, high expression in the pancreas was a surprise and requires further study, as Igs are found at varying levels in the pancreas of bony fish and especially cartilaginous fish(29, 30) (Flajnik unpublished observation). xlFCAMR was expressed only in the spleen.

Figure 2. Xenopus xlFCAMR is expressed in the spleen, similar to mammalian FCAMR.

qPCR expression is shown for Xenopus laevis PIGR1 and FCAMR (xlPIGR2). PIGR1 has canonical expression at mucosal surfaces, as previously described. FCAMR is expressed solely in the spleen. Data are pooled from three separate experiments (n=1 per experiment), each with three technical replicates. Mean expression with 95% confidence intervals is shown. Expression is relative to LMPY.

In mammals and birds FcamR is a single-domain receptor that binds to IgM and IgA independent of the J-chain(23). Consequently, binding studies were performed to evaluate whether xlFcamR isotype binding is similar to that of mammalian FcamR. xlPIGR was used as a control for binding to IgM because mammalian and Xenopus PIGR binding to polymeric immunoglobulins is J-chain dependent(31–33). In Xenopus, IgM associates with the J chain, but the Xenopus IgA homologue IgX does not(34–36). Therefore xlPIGR is a positive control for IgM binding, but is not expected to bind IgX (IgA homologue), or IgY (IgG homologue).

FLAG-tagged constructs of both xlPIGR and xlFcamR, were transfected into HEK-293 cells. Intracellular flow cytometry confirmed the expression constructs (Fig. 3A, Supplemental Fig. 1). Receptor-expressing cells were incubated with purified IgM, IgX, or IgY from an immunized adult X. laevis and then were stained with the corresponding isotype-specific mAb(36, 37). Consistent with previous data, Xenopus pIgR bound IgM(27, 31) , but not IgX which forms a multimer without J-chain(34) (Fig. 3). In contrast, xlFcamR bound both IgM and IgX, demonstrating J chain-independent, multimeric Ig-binding, like mammalian FcamR.

Figure 3. Xenopus Immunoglobulin Isotype binding by PIGR and xlFCAMR.

1A. Intracellular staining for FLAG tag validates the ectopic expression of the Xenopus receptor. Permeabilized cells were stained with mouse anti-FLAG, followed by goat-anti-mouse-FITC (GaM-FITC) for detection. 1B. Binding of Xenopus laevis isotypes (IgM, IgX, IgY) to PIGR and xlFCAMR, quantified in panel 3C. Colors indicate increasing concentrations of isotype: red indicates no Xenopus antibodies; blue indicates 3uL of purified isotype; orange 10uL of purified isotype; green 30uL of isotype used. IgM and IgY were used at a stock concentration of 0.1 mg/mL; IgX was used at a stock concentration of 0.3 mg/mL. Binding was detected using monoclonal antibodies against the X. laevis isotypes: 10A9 (anti-IgM), 4110B3 (anti-IgX), and 11D5 (anti-IgY). 1C. Quantification of binding shown in panel B. Percent of cells bound at the highest Ig concentration used for each isotype. Data are pooled from three experiments, one technical replicate per experiment. Fisher’s exact test was used to evaluate difference in binding proportions, bar graphs show mean and 95% CI, and **** indicates P<.0001. There was no appreciable binding to monomeric IgY.

xlFcamR binding of IgM was saturable, and binding strength was qualitatively higher for IgM than IgX (Fig. 3c). Neither xlPIGR nor xlFcamR bound the Xenopus IgG-equivalent, monomeric IgY. These data show that xlFcamR binding is functionally similar to that of mammalian FcamR, despite the additional domains found in Xenopus xlFcamR (Fig. 1A). Binding data, combined with the tissue expression, synteny, and phylogenetic trees (shown below) prove that xlFcamR is the equivalent of human FcamR.

PIGR Family Synteny is Conserved in Tetrapods and Coelocanth

The functional similarities of mammalian FcamR and xlFcamR prompted us to evaluate whether FCAMR is conserved in vertebrate Orders that emerged between amphibians and mammals using synteny and phylogenetic analyses (below). In addition, we evaluated whether FCAMR or domains giving rise to the PIGR family could be identified in cartilaginous fish and a lobe-finned fish.

The synteny of PIGR family is conserved in tetrapods (Fig. 4), from amphibians through eutheria, with the exception of FcmR which appears first in reptiles but was lost in various tetrapod species (Figs. 4, S2). Neighboring genes seen in the human PIGR region include members of the IL-10 family and complement receptors and regulators in the RCA complex. These neighboring genes are also conserved, even in the coelacanth and elephant shark “PIGR regions” (see below). This region on human chromosome 1 has been identified as an “MHC paralogous region”(38, 39), which emerged early in vertebrate evolution as a consequence of two genome-wide duplications(40–42). Previous studies in bony fish and birds suggested an association between the PIGR region and the MHC(25, 43), which is further confirmed here (see Conclusions).

Figure 4. Conserved synteny of PIGR family genes.

PIGR family receptor genes are flanked by scaffold genes including C1orf116, YOD1, complement receptors in the regulators of complement activation (RCA) locus, and the IL-10 family. The PIGR family locus is conserved in genera from amphibians though placental mammals. PIGR and FCAMR/PIGR2 are both present in amphibians, however FCMR appears in reptiles (present in turtle species and platypus, see Fig. S2). No reptile species were identified with all three PIGR family receptors, consequently anole and tortoise syntenies are depicted to represent the presence of both FCMR and FCAMR in reptiles. Genes that may have given rise to PIGR family domains were identified in L. chalumnae and C. milli and are indicated in gray boxes connected by dashed lines. Additional C. milli ddIgSF genes are also shaded in gray but did not cluster with PIGR family domains in phylogenetic analyses. Data were compiled from the NCBI Genomes page (https://www.ncbi.nlm.nih.gov/genome).

Assessment of ddIgSF Domains in Cartilaginous Fish, Bony Fish, and Tetrapods

To determine the ancestry of ddIgSF and formally address the domain assignation of xlFCAMR and xlPIGR in relation to other members of the ddIgSF, we performed phylogenetic analyses. Protein sequences for vertebrate ddIgSF members were downloaded from NCBI for inclusion in the analyses. In addition, we included an assessment of cartilaginous fish and lobe-finned fish genomes to evaluate the origins of ddIgSF receptors and the PIGR family.

As described, PIGR family receptors have been conserved in tetrapods(24, 44). In addition, a family of ddIgSF receptors designated as “modular domain immune receptors” (MDIRs) has previously been described in an elasmobranch, the clearnose skate (Raja eglanteria)(16). Sequences were also downloaded for known ddIgSF genes including the mammalian and amphibian PIGR and its duplicates, the CD300 family, CMRF35-like receptors, TREM receptors, NILT and MDIRs. All sequences were selected based on similarity to known ddIgSF receptors and the four cysteine residues forming the two disulfide bonds characteristic of this superfamily (Fig. 1B)(15). The complete list of domains and receptors used is in Supplemental Table 1.

Based on the presence of ddIgSF domains in an elasmobranch, we reasoned that domains orthologous to FCAMR, FCMR, or PIGR domains, might be identifiable in other cartilaginous fish species and transitional species that were not previously evaluated. BLAST searches using mammalian and Xenopus PIGR were used to identify possible IgM receptors in cartilaginous fish, coelacanth, and lungfish. Where possible, synteny was also assessed (Fig. 5). BLAST searches identified nine ddIgSF receptors in the holocephalan elephant shark(45) (Figs. 5 and S3) and preliminary data from the elasmobranch bamboo shark (Fig. S4). ddIgSF domains could not be identified in genera that emerged prior to gnathostomes, agnathans (e.g. lamprey) and lower deuterostomes (e.g. sea urchin) or in any protostomes. All elephant shark ddIgSF (and bamboo shark) genes that we could identify clustered on a single scaffold. As mentioned above, neighboring genes correspond to human chromosome 1p13 and 1q32 (1q32 contains the human PIGR loci) (Fig. 5b). The genes at this single locus are related to the various ddIgSF families, indicating this locus may have given rise to all tetrapod ddIgSF genes (discussed below in phylogenetic analysis).

Figure 5. ddIgSF loci and synteny in Coelocanth and Elephant Shark.

Related genes are color-coded based on the phylogeny in Figure 5 and Figure S3. Domains shaded in gray lack ddIgSF structure. TREM- and PIGR-related domains were identifiable in one receptor in elephant shark. In Coelacanth these domains were no longer combined in a single receptor, and instead domains related to PIGR D4 and D5 were found on contig NW_005819616, and domains potentially related to TREM receptors were identified on contig NW_005819114 and NW_005819048. Data were compiled from NCBI (http://www.ncbi.nlm.nih.gov), peptide sequences are available in Supplemental Tables 1 & 3.

Phylogenetic Analysis of ddIgSF Genes

Multiple sequence alignments were generated in the GUIDANCE2 server with MAFFT, PRANK, and ClustalW algorithms. MAFFT produced the best Guidance score (Table 1), however all alignments were used to generate maximum likelihood phylogenies in IQ-TREE. The full alignment is shown in Supplemental Figure 2, and sequences are provided in Supplemental Table 1. Consensus trees were used for interpretation. Bootstrap values and agreement between multiple phylogenies guided interpretations. A simplified version of the MAFFT phylogeny is shown in Figure 6, with full phylogenies for all alignments in Supplemental Figure 3.

Table 1.

GUIDANCE2 alignment scores for amino acid sequences of ddIgSF genes. A full record of sequences used can be found in Supplementary Table 1.

ClustalW	MAFFT	PRANK
0.735143	0.779683	0.635451

Figure 6. Phylogeny of ddIgSF domains emphasizing PIGR2 domains related to FCAMR and FCMR.

A simplified version of the MAFFT phylogeny is shown with node labels and grayscale denoting bootstrap support. The ddIgSF genes broadly fall into two clades, one containing CD300/CMRF35 genes, and the other containing PIGR family genes. TREM 1 and 2 most often cluster at the base of the PIGR family clade. Cartilaginous fish sequences are seen throughout the families and highlighted in yellow. Data were compiled from NCBI (http://www.ncbi.nlm.nih.gov), complete phylogenies are shown in Supplemental Figure 3 and peptide sequences are listed in Supplemental Table 1.

ddIgSF Genes form Four Major families; Elasmobranch Genes Cluster with All Major ddIgSF Families

The ddIgSF genes clustered into four families: 1) PIGR and PIGR-related receptors (FCMR and FCAMR); 2) TREM receptors; 3) CD300 receptors; and 4) NKp44 (Fig. 6). The relationships among these families were poorly resolved, but clades within the families reliably grouped together in the different phylogenies. NKp44 genes did not reliably cluster with any subfamily (Fig. S2).

The clade described here as the “CD300 Family” includes CD300, CMRF35L, MDIR, NILT, and the ray-finned fish PIGR (rfPIGR). This analysis is consistent with published data indicating that the MDIRs are most related to CMRF35-L receptors(16), and our data expand that family to include CD300s and rfPIGR. The PIGR Family included PIGR, FCAMR, and FCMR receptors from all genera besides ray-finned fish (Fig. S2). TREM and TREM-like genes reliably clustered forming TREM1 and TREM2 clades as previously described by Cannon et al. (2006)(16), but the relationship between TREMs and other ddIgSF families was not consistent across the phylogenies. Similarly, the relationship between the PIGR family and CD300 family was not well resolved. It is unclear which of the four families represents the most ancient family of receptors. Interestingly, this failure to predict an ancestral domain may reflect the fact that all four families arose from ddIgSF genes already present in the Chondrichthyans, the oldest extant vertebrates with ddIgSF genes. In agreement with that hypothesis, domains from cartilaginous fish cluster with the PIGR family, CD300 family, and TREMs, perhaps indicating that the “primordial raw material” for these families was present from the earliest Ig-based adaptive immunity (Fig. 5, and see highlights in Fig. 6).

The Earliest PIGR Family Orthologue is Clearly Identifiable in Amphibians

A two-domain pIgR has been identified and characterized in ray-finned fish and is generally regarded as the earliest PIGR(15, 18, 25, 46–48). rfPIGR is a functional mucosal and skin receptor for IgM and IgT, while other teleost PIGRL genes have been identified, some of which bind other ligands such as phospholipids(18, 25, 49–52). The rfPIGR cluster is flanked by ABHD4, DAD1-like (DAD1L) and LRRC24 or LRRC24-like (LRRC24L) genes. These are not the scaffold genes associated with canonical PIGR in all other species, including sharks(25) (Fig. 4). Previous phylogenetic analysis compared the rfPIGR to mammalian PIGR and concluded the teleost PIGR domains are related to mammalian D1 and D5(15, 46, 48, 52). However, when we compared rfPIGR domains to the broader ddIgSF family members, both domains clustered within the CD300 family and thus are not orthologous to any mammalian PIGR domains (Fig. 6, Fig. S2). The relationship of teleost PIGR to other clades in the family varies between phylogenies, but in all iterations of the phylogenetic analyses the rfPIGR domains cluster within the CD300 genes. rfPIGR often clusters with reptile CMRF35L genes in our phylogenies. It is likely, therefore, that the rfPIGR Ig-binding and -transport function (for IgM and IgT) was derived by convergent evolution with canonical pIgR.

FcamR Emerged in Amphibians

As described, two genes homologous to mammalian PIGR are found in Xenopus, canonical xlPIGR with four ddIgSF domains (D1, D2, D4, and D5), and xlFCAMR with three ddIgSF domains (D1, D2, and D4) (Fig. 1A). The phylogeny of ddIgSF domains indicated the xlFCAMR D1 is more closely related to mammalian FCAMR than PIGR (Fig. 6, Fig. S2). The xlFCAMR domains not present in mammalian FCAMR form separate clades within the PIGR family. In addition, the D2 and D4 domains of xlFCAMR cluster with “PIGR2” or “PIGRL” receptor domains in reptiles, not with canonical PIGR domains (Fig. S2). This indicates that along with xlFCAMR, the reptile PIGR2 genes are FCAMR domain orthologues and the D2 and D3 domains of these receptors evolved separately from PIGR D2 and D3 before being lost in mammals.

Antibody binding and phylogenetic data support the presence of an FcamR equivalent in X. laevis. If this receptor emerged in amphibians, as we propose, then FCAMR should be present across amphibian species, some separated from each other by 300 million years. To test this, tblastn searches identified multiple ddIgSF domains in other amphibian species (Supplemental Table 2). A phylogeny of amphibian domains (Fig. S4) recapitulates the grouping see in the multi-Class phylogeny in Fig. 6. All amphibian genomes had ddIgSF domains with distinct PIGR and FCAMR domains. Most amphibians have multi-domain PIGR and FCAMR. The exception is axolotl, which has orthologues of xlPIGR D2 and xlFCAMR D2 but no other identifiable ddIgSF domains. In other amphibian species, clades for PIGR D1, D2, D4, and D5 are separate but adjacent to clades for FCAMR D1, D2, and D4 (P. ornatum lacks FCAMR D2) (Fig. S4). Evolutionarily distance amphibian species were included and the same clustering of domains is seen in caecilian and ranids. This suggests that FcamR emerged in or before early amphibians. These data show that FCAMR is highly conserved and has evolved separately from the PIGR in amphibians. We conclude that xlFcamR is the mammalian FcamR equivalent, and evaluated the origin and synteny of PIGR-related receptors to better understand the evolution of the entire family.

PIGR Family Domain Evolution

The earliest PIGR-related domain is identifiable in elasmobranchs and is part of the two-domain receptor that likely also gave rise to the TREM family (elephant shark gene 632963074) (Figs. 5b, 6). The second domain of that receptor most often branched at the base of the PIGR family in phylogenies (Fig. 6).

Lacking a PIGR orthologue or conserved syntenic region in ray-finned fish, we evaluated PIGR-related domains and loci in lobe-finned fish (coelacanth, lungfish), which is an intermediate order between ray-finned fish and amphibians. In the coelacanth (Latimeria chalumnae) there are ddIgSF domains related to the PIGR family, but not a 4-domain pIgR (Fig. 5). Coelocanth has one- and two-domain ddIgSF receptors located on four different genome scaffolds. There is a putative coelacanth PIGR annotated in the genome with surrounding genes corresponding to the mammalian PIGR locus. This receptor was not identifiable in the current lungfish transcriptome(53). Phylogenetically, the coelacanth PIGR-related domains are orthologous to mammalian PIGR D4 and D5 domains (Fig. S2). D4 and D5 are directly involved in IgM and IgA binding by PIGR, implying the coelacanth PIGR-related receptor may be capable of binding and transcytosing antibodies, but no functional or expression data exist for this receptor.

A model of PIGR domain evolution is shown in Figure 7. A closer evaluation of PIGR family domain evolution did not reveal the origins of the domains D1 or D2. The PIGR and FCAMR domains form well supported monophyletic clades, indicating that D4 of xlPIGR and xlFCAMR are paralogues originating from the single PIGR-related receptor perhaps represented by the coelocanth molecule (Fig. S2). xlFCAMR has a D1-D2-D4 organization that is retained in reptiles. D2 and D4 of FCAMR were lost in early mammals, giving rise to the single domain FCAMR in eutheria (Fig. 7). xlPIGR has a D1, D2, D4, D5 organization. A duplication of PIGR D2 gave rise to D3, which first appears in monotremes (Fig. 7).

Figure 7. Model of ddIgSF family evolution, with an emphasis on PIGR Family domains.

Related domains are color-coded to match Figure 1A and Figure 5. PIGR and FCAMR appeared first in Amphibians. However, ddIgSF domains related to the PIGR Family, TREMS, and CD300 Family are identifiable in cartilaginous fish, the oldest vertebrates with adaptive immunity base on Ig/TCR/MHC. This model is based on the phylogenies shown in Figure 6 and Supplemental Figure 3.

CD300 Family Clades are Related to Batoid Ray MDIRs and Elasmobranch Sequences

CD300 family genes, including CMRF35-like (CMRF35L) genes, form a set of clades (Fig. 6). The relationships between these clades were not consistent or well supported between the different phylogenies. However, some distinct clusters emerged, often reflecting species- or Order-specific duplications, not distinct subfamilies of receptors. The MDIRs from the skate Raja eglanteria comprise one such set of taxon-specific duplications closely related to a cluster of four nurse shark genes. The exception is MDIR1 which clusters with a set of elasmobranch genes (Fig. S3). This cluster is related to the large clade of CD300 genes. This indicates that the CD300 clade may originate with a set of receptors arising within cartilaginous fish and potentially giving rise to CD300 genes (discussed in greater detail below). Within mammals, the CD300 genes do form specific subfamilies. These include CD300D, CD300LF and CD300LB in one cluster; CD300A and CD300C with a closely related CD300E cluster; and a single clade for CD300LG. The CD300 family receptors appears to have multiple recent “birth-and-death” events giving rise to the diversity of CD300 genes in mammals and some reptiles(54, 55). Notably the teleost pIgR domains cluster with CD300 Family genes, as discussed above (Fig. 6, Fig. S2).

ddIgSF Gene Identification Revealed a Single Genomic Region in Chondrichthyans

Most Chondrichthyan ddIgSF genes are the product of expanding gene families in a given species, or perhaps within that Class or Order (Supplemental Fig. 4). This is seen in a cluster of bamboo shark genes, and the majority of skate MDIRs (excluding MDIR1). The exception is a cluster of genes conserved in multiple species. This cluster includes holocephalan and elasmobranch sequences: R. eglantaria MDIR1; C. plagiosum gene 11350436; C. milii genes 632963066, 632963068, and gene fragment 632963302_C-terminus; and G. cirratum spleen transcript 320504, and gill transcript c519607. In the comprehensive phylogeny (Supplemental Fig. 3), MDIR1 is related to CD300 family members. C. milii gene 632963058 is related to mammalian CD300 genes, similarly to MDIR1. The other Chondrichthyan genes in the conserved elasmobranch clade were not clearly related to tetrapod genes. However, several notable Chondrichthyan genes branch at the base of this clade: C. milii gene 632963074 domains 1 and 2 are related to the PIGR family and TREM domains in tetrapods. C. milii gene 632963058 also branches below this clade with an intermediate cluster of C. plagiosum genes. In summary, there is a conserved clade in cartilaginous fish that is most closely related to CD300 genes, based primarily on the position of MDIR1 in the full phylogeny and the conserved clade. This clade may be the ancestral ddIgSF genes giving rise to multiple domains with multiple functions.

No Ig Transport Receptor in Cartilaginous Fish?

While domains related to all ddIgSF families were detected in Chondrichthyes, only one domain closely related to PIGR (domain 2) was identified. As mentioned, PIGR D1, D4, and D5 are involved in IgM binding in mammals(56), so while the coelocanth PIGR could transport IgM, this is highly unlikely for the cartilaginous fish ddIgSF family members. Cartilaginous fish are the oldest living taxon with antibodies, including IgM. A large percentage of shark IgM is associated with the J chain(57–59), and thus lumenal transport could be considered in the classical fashion. However, alignment of J chains from a large number of cartilaginous fish species shows strong conservation of the N-terminal and central regions that associate with IgM, but not of the C-terminal region, important for pIgR binding(60, 61) (Fig. 8A). Disulfide bridges made between the J chain and IgM tail, as well as within the J chain itself are clearly modified in cartilaginous fish (Figs. 8A and 8B). These data strongly suggest that J chain is used for multimerization in sharks, but not lumenal transport, at least not in the same manner as mammalian PIGR transcytosis.

Figure 8. High conservation of the N-terminal but not C-terminal region of the J chain in sharks in comparison to tetrapods.

8A. Alignment of tetrapod J chain with many cartilaginous fish sequences. Strands in the three structural (hairpin) domains of the J chain are indicated with black bars at the top alignment and numbered according to their position in beta-hairpin 1 or beta-hairpin 2. Cysteines are numbered, based on the human sequences. Dots indicate identity and gaps as dashes. 8B. Crystal structure of human J chain showing the three structural (hairpin) domains of which the last domain is missing or entirely different in cartilaginous fish (circled). The ribbon diagram of J chain structure is used from Kumar N, Arthur CP, Ciferri C, Matsumoto ML. Structure of the secretory immunoglobulin A core. Science. 2020;367(6481):1008–14. Reprinted with permission from AAAS.

Cartilaginous fish express two forms of IgM, monomeric (7S) and pentameric (19S), the latter form containing the J chain. Thus, expression of the J chain is crucial in shaping the quaternary structure of IgM. Consistent with J chain’s crucial role in biosynthesis, the 7S IgM form is involved in that antigen-specific response and memory while the multimeric form has been shown to be polyreactive(62). We speculate that if there is a ddIgSF family member that transports Chondrichthyan IgM, that this would likely have emerged by convergent evolution and be J chain-independent. Note that this is also the likely scenario for bony fish IgM/T transport, because bony fish have lost the J chain gene(45). These data argue against direct lines of antibody transport evolution from both the cartilaginous fish to bony fish, and from the bony fish to tetrapods.

Discussion

pIgR is a conserved antibody-transport receptor in tetrapods vital for intestinal homeostasis and mucosal barrier maintenance. It was previously proposed that PIGR appeared first in teleosts, and that the related receptors FCAMR and FCMR emerged in reptiles/mammals. Here, we establish by function, synteny, and phylogenetic analysis that FCAMR is also an ancient and conserved receptor, and that the equivalents of PIGR and FCAMR are first identifiable in amphibians.

The amphibian PIGR and FCAMR share essential features with their mammalian counterparts. Amphibian PIGR was previously described and is a J-chain dependent, mucosal transport receptor(27). Amphibian FcamrR binds IgM and IgX, does not require the J-chain, and is expressed by amphibian splenic APCs (called XL cells) but not by mucosal epithelia(26, 63). FcamrR is the only identified IgM or IgA receptor on mammalian FDC, and modulates antibody responses by mechanisms that are not fully understood(64, 65). Although the exact function of FcamrR on FDC remains unclear, it may restrict availability of antigen to B cells to facilitate affinity maturation, while still contributing to IC retention. A similar function is described for CD32B (FcgRIIB) on FDC, which restricts antigen availability, thereby creating competition favoring affinity matured B cells(66–69). Xenopus XL cells are myeloid cells that display intact antigen, likely in the form of immune complexes, in B cell follicles after immunization(26, 70). We have proposed that XL cells serve the function of mammalian FDC prior to the emergence of FDC in the vertebrate line. Notably, mammalian FcamrR is expressed in multiple mucosal organs, as well as in kidney tubules(8, 10, 11), while Xenopus FCAMR is only expressed at high levels in the spleen. This suggests that the splenic role for FcamR is most ancient, and that the role of FcamrR on splenic APCs may be conserved over the transition from myeloid XL cells to stromal FDC.

In vivo, XL cells bind all three Xenopus Ig isotypes, and based on our data (Fig. 3) it is likely that xlFcamR binds IgM and IgX in vivo. As IgY did not bind to Xenopus FcamrR in our transfection study, another unidentified FcR is probably responsible for that binding. We predict that the display of immune complexes, especially IgM-based immune complexes early in an adaptive immune response, is important for B cell activation in cold-blooded vertebrates.

Emergence of FCAMR concurrent with PIGR raises the question of why PIGR would maintain domains and become a five-domain mammalian receptor, while mammalian FCAMR has lost domains to produce a single-domain receptor. Both pIgR and FcamR use D1 for antibody binding, however pIgR binding is J chain-dependent, which may account for domain conservation in pIgR. In addition, pIgR retains the ability to directly bind bacteria with different domains than those directly binding antibodies or J-chain(71, 72). Recent structural analysis has shown that pIgR directly contacts IgM and IgA with D1, D3, D4, and D5 and an alternately spliced form of pIgR with D1, D4, and D5 is naturally present in mammals(44, 73). This suggests D2 and D3 are dispensable for aspects of pIgR function, particularly Ig binding. When pIgR is transcytosed without antibody cargo, it is still cleaved resulting in release of the pIgR extracellular domains, called secretory component (SC)(74, 75). Free SC and membrane-attached pIgR are both able to directly bind bacteria. N-linked glycosylation on D1, D2, and D4 of SC, as well as glycosylation on polymeric immunoglobulins, have been implicated in non-specific bacterial binding(71, 72). It is possible that PIGR D2 and D3 are maintained because they allow non-specific binding of SC to bacteria in mucosal lumens, leading to exclusion of those bacteria from the mucosal epithelia.

Ray-finned (bony) fish clearly have an IgM-transport receptor that is a ddIgSF family member, annotated as PIGR (herein rfPIGR). However, this receptor, neither by phylogenetic analysis nor synteny is the PIGR equivalent. In addition, the J chain was lost in the bony fish lineage and their IgM (and IgT) have unique mechanisms for multimerization(45, 76). We suggest that the conserved transport function of bony fish “pIgR” with tetrapods arose by convergent evolution. rtPIGR is phylogenetically related to CD300 receptors and is not orthologous is any PIGR family gene identified.

In contrast to ray-finned fish, the cartilaginous fish do have the J chain and are the oldest taxon of vertebrates in which it has been recognized. We identified ddIgSF domains in cartilaginous fish to examine whether a regulatory or transport receptor could be detected. While there was no clear Ig-binding receptor identified in sharks, evaluation of ddIgSF domains did reveal an ancient origin for all of the ddIgSF receptor families in a cartilaginous fish gene complex that matches the PIGR gene complex in tetrapods (see next section). The lack of a bona fide pIgR in sharks is in keeping with the fact that the C-terminal region of shark J-chain, required for pIgR binding in mammals (and likely all tetrapods), is highly divergent or entirely lacking in cartilaginous fish (Fig. 8). Cartilaginous fish are the only vertebrate group having high levels of both monomeric and multimeric IgM, that latter associated with the J chain(57–59). The monomeric Ig is induced in the course of an adaptive immune response(62), while the multimeric form has been proposed to be utilized as a first-line of defense, perhaps predominantly in T-independent responses(57, 59, 62). We propose that the J chain first emerged to regulate IgM polymerization and was later co-opted for binding to pIgR in tetrapods or sarcopterigians. Consistent with this proposal, we have found IgM-secreting cells in the shark olfactory epithelium that appear to secrete Ig directly into the nasal mucosa (Salinas, Neely, & Flajnik, unpublished observation).

These scenarios in cartilaginous and bony fish do not exclude the possibility that primordial ddIgSF members bind Ig. Clearly bony fish have an IgM-binding and -transport ddIgSF family member. Some shark ddIgSF molecules may also have an Ig-binding function.

This study provides more comprehensive data on the evolution of PIGR/FCAMR and MHC-PIGR association. These data support specific theories on the origins of PIGR and Ig transport, the emergence of the ddIgSF family, and the primordial MHC, and may inspire further studies of PIGR family evolution and function. We posit that ddIgSF receptors have been involved in immune regulation from the earliest immunoglobulin-based adaptive immunity. The earliest identified ddIgSF domains appear to be encoded in a single genetic region. This single region was identified in the elephant shark and bamboo shark. The elephant shark complex has six ddIgSF genes and the bamboo shark has several more. Tblastn searches have not identified any ddIgSF domains prior to cartilaginous fish, and the data suggest that the superfamily emerged in jawed vertebrates (gnathostomes). The shark gene complex (in contrast to the ray-finned fish) is clearly in the same genetic region as in tetrapods (human chr 1q32). This region has previously been identified as an MHC paralogous region, arising as a consequence of two genome-wide duplications early in vertebrate evolution. In mammals and other vertebrates, two other ddIgSF members, NKp44 and TREM, are encoded in the bona fide MHC (human chr 6p21)(77, 78). Even in ray-finned fish the “PIGR” gene is closely linked to a class I gene (and chicken PIGR is in an MHC paralogous region)(25, 43). These associations demonstrate that the ddIgSF family was part of the primordial MHC, whichever function was important for adaptive immunity when gnathostomes emerged. Note that the region flanking the PIGR family genes includes the regulator of complement activation (RCA) complex. it is likely that the RCA complex, encoding complement regulators and receptors, was also part of the primordial MHC, in keeping with the MHC association of other complement genes such as C3/4/5, factor B/C2, and others.

Some subfamilies within the ddIgSF are conserved (PIGR family, TREMs), and in other cases ddIgSF subfamilies exhibit multiple “birth and death” events(54, 55) (CD300 clades). Of the conserved receptors, pIgR has been extensively studied for its role in antibody transport and mucosal homeostasis. The conserved TREM family has more recently been described as important regulators of homeostasis versus inflammation(79–81). In contrast, the birth and death events of genes in the CD300 family may reflect a mechanism of “fine tuning” of myeloid and granulocyte function(20). Overall, the ddIg superfamily has likely modulated the adaptive immune responses from their earliest stages of vertebrate evolution, when gnathostomes emerged ~500 million years ago.

Methods

Identification of ddIgSF Sequences

Mammalian and Xenopus pIgR were used for tblastn searches of the cartilaginous fish elephant shark C. milli genome to identify ddIgSF domains in GenBase (http://www.ncbi.nlm.nih.gov), and the bamboo shark (Biorxiv). Identified ddIgSF protein sequences were downloaded from NCBI. Additional sequences from important transitional species (coelacanth, lungfish, marsupials and monotremes) were identified by BLAST using mammalian and Xenopus PIGR on the NCBI genome site, and the African lungfish transcriptome(53). A full list of sequences used is available in Supplementary Tables 1 (sequences corresponding to the tetrapod phylogeny in Fig. 6 and S3), 2 (sequences corresponding the amphibian phylogeny in Fig. S4), and 3 (sequences corresponding to the chondrichthyes phylogeny in Fig. S5).

Multiple Sequence Alignments and Phylogenetic Analysis

The structure of secretory piece (the cleaved extracellular domains of pIgR), as well as ddIgSF domain structure were described previously(15), and now has been further resolved(56, 61). For the alignment ddIgSF peptide sequences were trimmed at the N-terminus to a conserved glycine residue, trimming on C-terminus of the domain was minimal to remove poorly aligning disordered regions predicted by CFSSP server(82). Peptide sequences were aligned with the GUIDANCE2 server, using the MAFFT, ClustalW, and PRANK multiple sequence alignment (MSA) algorithms(83). MAFFT alignments were used for supplemental phylogenies. A multiple sequence alignment demonstrating conservation of FCAMR and PIGR2 domain 1 is shown in Figure 1B.

Maximum likelihood trees for the three MSAs were generated using the IQ-TREE web server(84). Default setting of ultrafast bootstrap analysis with 1000 iterations was used. Consensus trees were visualized with FigTree v1.4.4 and used in subsequent interpretation.

Xenopus Isotype Purification for Binding Studies

Xenopus laevis Ig isotypes were purified by passing fresh adult Xenopus sera over columns with monoclonal antibodies (mAb) specific for IgM, IgX, and IgY covalently coupled to Protein A columns, as described by Schneider et al. (1982). The following mAbs were used: 10A9 (IgM-specific), 4110B3 (IgX-specific), and 11D5 (IgY-specific)(36, 37). IgM and IgX were eluted with 7 mM acetic acid pH=4, and IgY was eluted with 50 mM diethylamine pH=11.5. Purified isotype fractions were immediately neutralized to pH 7.4. Isotypes were concentrated and buffer-exchanged to amphibian osmolarity PBS over size-separation spin columns (MilliporeSigma™ Amicon™ Ultra Centrifugal Filter Units (30,000 MWCO Catalogue: UFC903008). Concentrated isotypes were quantified in a Nanodrop™ 2000 and stored at −20C. IgM and IgY were used at a stock concentration of 0.1 mg/mL; IgX was used at a stock concentration of 0.3 mg/mL. All procedures and animal care were approved by the UMB IACUC committee.

Ectopic receptor expression and binding of Xenopus isotypes

FLAG-tagged constructs of both full-length xlPIGR, xlFCAMR, and empty P3X vector were transfected into HEK-293 cells. Cells were transfected by nucleofection using the Lonza 4D Nucleofector X-Unit. Nucleofection conditions were optimized per manufacturer protocols using the 4D-NucleofectorTM X Kit.

Intracellular staining flow cytometry confirmed expression of each construct. Cells were mechanically dissociated from tissue-culture plates, washed with FACS buffer (mammalian osmolarity PBS, 0.01% sodium azide, 2% FCS), then fixed and permeabilized with the BD™ Cytofix/Cytoperm™ reagent (Catalogue: BD 554714). 1×10⁶ cells per well were stained in 100uL 1x Cytoperm solution with 1:100 dilution of mouse anti-FLAG (Millipore Sigma F1804). Cells were washed three times with 150uL Cytoperm solution. Anti-FLAG binding was detected with a secondary stain of FITC-conjugated Goat-anti-Mouse Ig (Invitrogen Cat: # F-2761). Secondary stain was performed in with a 1:500 dilution of antibody in 100uL 1x Cytoperm solution, incubated 45 minutes on ice and protected from light, then washed as above. Construct expression was evaluated by flow cytometry on a BD™ LSR-II cytometer.

To evaluate isotype binding, receptor expressing HEK-293 cells were mechanically dissociated from culture plates. Vector-transfected cells not expressing a receptor were used as an internal negative control for isotype binding as follows: P3X vector cells were labeled with cell trace violet (CTV) per manufacturer specifications (Invitrogen™ CellTrace™ Violet Cell Proliferation Kit, Catalogue: C34571). For staining, 5×10⁵ CTV labeled P3X cells were mixed with 5×10⁵ receptor expressing cells. Cells were then incubated with varying concentrations of Xenopus isotypes on ice in FACS buffer for 45 minutes. The Ig isotypes were added to FACS buffer at volumes of 0uL (no primary control), 3ul, 10ul, and 30uL, per total volume of 100uL. Cells were then washed three times with 150uL cold FACS buffer. Isotype binding was detecting with 10A9, 4110B3, or 11D5 that had been FITC-conjugated per kit specifications (MilliporeSigma FluoroTag™ FITC Conjugation Kit Cat: FITC1–1KT). Isotype binding was evaluated by flow cytometry on a BD™ LSR-II cytometer.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.