Teleost IgD⁺IgM⁻ B Cells Mount Clonally Expanded and Mildly Mutated Intestinal IgD Responses in the Absence of Lymphoid Follicles

Summary

Immunoglobulin D (IgD) is an ancient antibody with dual membrane-bound and fluid-phase antigen receptor functions. The biology of secreted IgD remains elusive. Here, we demonstrate that teleost IgD⁺IgM⁻ plasmablasts constitute a major lymphocyte population in some mucosal surfaces, including the gut mucosa. Remarkably, secreted IgD binds to gut commensal bacteria, which in turn stimulate IgD gene transcription in gut B cells. Accordingly, secreted IgD from gut as well as gill mucosae, but not the spleen, show a V(D)J gene configuration consistent with microbiota-driven clonal expansion and diversification, including mild somatic hypermutation. By showing that secreted IgD establishes a mutualistic relationship with commensals, our findings suggest that secreted IgD may play an evolutionary conserved role in mucosal homeostasis.

Graphical Abstract

Highlights

• •IgD⁺IgM⁻ B cells constitute the main non-IgT B cell subset in rainbow trout guts
• •Gut IgD responses establish a two-way interaction with the local microbiota
• •Mucosal but not splenic IgD undergoes clonal expansion and diversification
• •Despite the lack of germinal centers, mucosal IgD is mildly mutated in rainbow trout

Perdiguero et al. show that IgD⁺IgM⁻ plasmablasts constitute a major lymphocyte population in the teleost intestine, as in gills. In these two tissues, IgD molecular signatures reflect a clonal expansion not detected in the spleen. Finally, secreted IgD in the intestine establishes a two-way interaction with the local microbiota.

Introduction

Although initially described as a recently evolved immunoglobulin (Ig), the identification of IgD in teleost (Wilson et al., 1997) and cartilaginous fish (Ohta and Flajnik, 2006) revealed that IgD is an ancient antibody that has been preserved throughout evolution. This implies that IgD may play some important but still elusive role in adaptive immunity (Gutzeit et al., 2018).

In mammals, immature B cells from bone marrow first express surface IgM receptors and then enter a transitional maturation phase involving a progressive upregulation of surface IgD receptors. Because of its co-synthesis with IgM through alternative splicing of a long precursor mRNA, IgD shares an antigen-binding variable region identical to that of IgM in peripheral mature B cells (Enders et al., 2014). Upon antigenic stimulation during cognate T-cell interaction, mammalian follicular B cells initially enter an extrafollicular Bcl-6⁻ differentiation program that transcriptionally downregulates IgD expression to generate short-lived IgM-secreting plasmablasts (Bohannon et al., 2016). Subsequently, antigen-stimulated follicular B cells enter a follicular Bcl-6⁺ differentiation program that replaces the constant region of IgM and IgD with that of IgG, IgA, or IgE through class-switch recombination (CSR) (Nera et al., 2015). The ensuing germinal center (GC) reaction further involves antibody affinity maturation through somatic hypermutation (SHM), a DNA-editing process that, like CSR, requires the enzyme activation-induced cytidine deaminase (AID) (Muramatsu et al., 2000). After completing CSR and SHM, B cells exit the GC as long-lived plasma cells or memory B cells, which express IgG, IgA, or IgE with a high antigen affinity (Nera et al., 2015).

Remarkably, a small subset of mammalian B cells undergo an unconventional form of CSR from IgM to IgD and thereafter differentiate into IgD⁺IgM⁻ plasmablasts specifically secreting IgD (Arpin et al., 1998, Chen et al., 2009, Choi et al., 2017, Koelsch et al., 2007, Rouaud et al., 2014, Zheng et al., 2004). Although also detected in the general circulation, these IgD class-switched cells mostly inhabit the organized lymphoid tissue from nasopharyngeal cavities, including human tonsils (Arpin et al., 1998, Chen et al., 2009, Koelsch et al., 2007). While mouse IgD⁺IgM⁻ plasmablasts are virtually unmutated (Rouaud et al., 2014), human IgD⁺IgM⁻ plasmablasts are extremely mutated, at least in adults (Arpin et al., 1998, Koelsch et al., 2007).

Despite the apparent lack of IgD⁺IgM⁻ plasmablasts in the human intestinal mucosa (Chen et al., 2009), unconventional IgM-to-IgD CSR has been recently identified in mesenteric lymph nodes (MLNs) from mice made unable to initiate canonical CSR due to a deletion of the 3′ regulatory region (3′RR) of the Ig heavy chain (IgH) locus (Rouaud et al., 2014). Similarly, enhanced IgM-to-IgD CSR has been detected in MLNs and gut mucosa from mice lacking p53-binding protein 1 (53BP1), a DNA damage-response protein required for conventional CSR (Choi et al., 2017). The same study revealed a mutualistic relationship of IgD with the gut microbiota, the latter being targeted by and required for IgD secretion (Choi et al., 2017). Finally, IgD-secreting plasma cells reactive to food antigens have been recently described in humans (Shan et al., 2018), which raises the possibility that IgD class-switched B cells inhabit not only nasopharygeal cavities, but also yet-unexplored areas of the human digestive tract.

Only three antibody isotypes have been described in teleosts: IgM, IgD, and IgT. This last is a paradigmatic mucosal antibody equivalent to mammalian IgA (Zhang et al., 2010). In the IgH locus, IgT generates antigen recognition diversity through a private V(D)J gene cassette; therefore, IgT production is completely independent of that of IgM and IgD (Hansen et al., 2005). Accordingly, CSR has never been reported in teleosts (Fillatreau et al., 2013). Of note, teleost B cells mount mucosal IgT and IgM responses to gut bacteria despite having no organized lymphoid follicles (Zapata et al., 2006), while these structures are key to generating a large fraction of bacteria-coating antibodies in the mammalian intestine.

Notwithstanding these peculiarities, most B cells from teleosts co-express surface IgM and IgD and downregulate IgD after encountering antigen as mammalian B cells do (Tafalla et al., 2017). In addition to IgM⁺IgD⁺ B cells, IgD⁺IgM⁻ B cells have been detected in catfish peripheral blood, where they may amount to up to 70% of peripheral B cells (Edholm et al., 2010), and in rainbow trout gills (Castro et al., 2014). Yet, a detailed characterization of this B-cell subset has never been attempted.

Here, we demonstrate that IgD⁺IgM⁻ B cells account for the main non-IgT B-cell subset in the gut mucosa from rainbow trout. IgD from IgD⁺IgM⁻ B cells bound to the gut microbiota. Conversely, gut bacteria enhanced IgD transcription in gut B cells. Remarkably, gut and gill IgDs showed a tissue-specific V(D)J gene signature that reflected clonal expansion and mild SHM, which were absent in splenic IgD. Thus, in rainbow trout, IgD⁺IgM⁻ plamablasts mount microbiota-regulated IgD responses that likely integrate those mediated by IgT and IgM. These IgD responses may promote gut homeostasis though a two-way interaction with the local microbiota.

Results

IgD⁺IgM⁻ B Cells Are a Major Lymphocyte Subset in Rainbow Trout Intestines and Gills

To establish if mucosal IgD integrates canonical mucosal IgT and IgM responses in teleosts (Salinas et al., 2011), we analyzed IgD⁺IgM⁻ B cells in rainbow trout spleen, gills, gut, skin, head kidneys, and peripheral blood. A flow cytometry approach determined that in the spleen, IgD⁺IgM⁻ and IgM⁺IgD⁻ B cells represented 4% and 9% of IgT⁻ B cells, respectively, whereas conventional IgM⁺IgD⁺ B cells accounted for 87% of IgT⁻ B cells (Figures 1A–1C). In gills, IgD⁺IgM⁻ B cells represented 49% of IgT⁻ B cells, while IgM⁺IgD⁻ B cells generally amounted to 8% (Figures 1A–1C). Finally, IgM⁺IgD⁺ B cells generally corresponded to 43% of IgT⁻ B cells (Figures 1A–1C). In the gut, IgD⁺IgM⁻ B cells accounted for 83% of IgT⁻ B cells, whereas IgM⁺IgD⁻ and IgM⁺IgD⁺ B cells amounted to 3% and 14%, respectively (Figures 1A–1C). As expected, IgM⁺IgD⁺ B cells constituted the dominant B-cell subset in head kidneys (86%) and peripheral blood (96%) (Figure S1). In the skin, although IgD⁺IgM⁻ cells accounted for 32% of IgT⁻ B cells, IgM⁺IgD⁺ B cells constituted the main B-cell population among IgT⁻ B cells (Figure S1). Thus, IgD may play some important role in both intestines and gills from teleosts.

Figure 1.

Rainbow Trout IgD⁺IgM⁻ Cells Constitute the Most Abundant Non-IgT B-Cell Population in Gills and Gut

(A) Representative dot plots. The percentages of live IgM⁺IgD⁺, IgM⁺IgD⁻, and IgD⁺IgM⁻ B cells among cells in the lymphocyte gate are shown.

(B) Graphs show the mean percentages of B-cell subsets among total lymphoid cells (mean + SD, n = 13). ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001.

(C) Relative percentages of the different B-cell subsets among the total IgM⁺ and IgD⁺ cells were calculated and plotted as pie charts.

(D) Visualization of spleen, gills, and gut IgM⁺IgD⁺, IgM⁺IgD⁻, and IgD⁺IgM⁻ B cells. Examples of each subset are shown. Scale bars in left-hand images: 20 μm; scale bars in right-hand images: 5 μm.

See also Figures S1 and S2.

The specific mucosal tropism of IgD⁺IgM⁻ B cells was further evaluated by immunofluorescence analysis (IFA), confirming that IgD⁺IgM⁻ B cells were more frequent in gut and gills compared to in the spleen (Figures 1D and S2). Interestingly, IFA also evidenced that IgD⁺IgM⁻ B cells had a larger cytoplasm-to-nucleus ratio than did IgM⁺IgD⁺ or IgM⁺IgD⁻ B cells (Figures 1D and S2). As suggested previously (Castro et al., 2014, Shapiro-Shelef and Calame, 2005), this observation points to the initial commitment of IgD⁺IgM⁻ B cells to a plasmablast/plasma cell differentiation program. Accordingly, gut IgD⁺IgM⁻ B cells showed less surface major histocompatibility complex class II (MHC class II) than did gut IgM⁺IgD⁺ or IgM⁺IgD⁻ B cells (Figure S3A). Of note, attenuated MHC class II expression is viewed as a teleost-specific hallmark of plasmablast/plasma cell differentiation (Abós et al., 2016, Granja and Tafalla, 2019). Similarly, IgD⁺IgM⁻ B cells from peripheral blood transcribed significantly more plasma-cell-inducing transcription factor Blimp-1 than did blood IgD⁺IgM⁺ B cells (Figure S3B). Thus, teleost IgD⁺IgM⁻ B cells display a plasmablast-like phenotype and account for most IgT⁻ B cells in some mucosal tissues, including intestines and gills.

IgD Targets Bacteria from the Rainbow Trout Gut Microbiota

Earlier studies show that both IgM and IgT from teleosts coat the intestinal microbiota (Zhang et al., 2010). To test whether IgD also does so, we adapted a previously published bacterial flow cytometry strategy (Zhang et al., 2010). We found that IgD coated a substantial fraction of the intestinal rainbow trout microbiota (Figures 2A–2C and 2E). Remarkably, the percentage of bacteria coated by IgD was similar to the percentage of bacteria coated by IgM (Figure 2C). However, the mean fluorescence intensity of IgD, which corresponds to the number of bound IgD per bacterial cell, was higher compared to that of IgM (Figures 2A and 2B).

Figure 2.

Rainbow Trout Secreted IgD Co-targets the Intestinal Microbiota, Which in Turn Enhances Secreted IgD Transcription

(A) Representative flow cytometry histograms showing staining of intestinal bacteria with anti-IgM and anti-IgD.

(B) Mean fluorescence intensity (MFI) values of IgM- or IgD-coated bacteria (+SD, n = 9).

(C) Percentage of intestinal bacteria coated with IgM or IgD (mean ± SD indicated by black lines, n = 9).

(D) Levels of secreted and membrane IgM and IgD transcription in the intestines of fish orally treated with antibiotics compared to levels in control fish. Data are shown as the gene expression relative to the expression of EF-1α (mean + SD, n = 9).

(E) Confocal images of trout gut luminal bacteria stained with SYTO BC (green) and anti-IgM-APC (top images), anti-IgD-APC (middle images), or an isotype control (lower images). White arrows point to Ig-coated bacteria. Scale bars represent 5 μm.

Having shown that secreted IgD binds to gut bacteria, we next determined whether the gut microbiota had any impact on IgD responses, as shown in mice (Choi et al., 2017). To this purpose, we orally administered broad-spectrum antibiotics to rainbow trout for 15 days and then evaluated the transcription rate of both secreted and membrane IgD and IgM. Having confirmed the gut-sterilizing properties of our treatment protocol (not shown), we observed that antibiotic-treated fish had a lesser transcription of membrane IgD, secreted IgD, and membrane IgM, but not secreted IgM, compared to control animals (Figure 2D). These data suggest that IgD establishes a bi-directional and possibly mutualistic relationship with the gut microbiota.

Mucosal IgD⁺IgM⁻ B Cells Express Unique Junction Sequence Types

IgD⁺IgM⁻ B cells are prevalent in gut and gills but not in the spleen, raising the possibility that the IgD gene repertoire exhibits distinct molecular signatures in microbiota-colonized mucosal organs, compared to the sterile environment of the spleen. To test this hypothesis, we analyzed the gene repertoires of IgD and, for comparison, IgM and IgT by next-generation sequencing (NGS) (Abos et al., 2018). The sequences obtained (Table S1) were analyzed to identify tissue- and isotype-specific molecular signatures.

First, we defined unique Ig V(D)J gene sequences as gene rearrangements associated with a specific complementarity-determining region 3 (CDR3) amino acid sequence and grouped these sequences into isotype-specific junction sequence types (JSTs) (Fillatreau et al., 2013). These JSTs were then quantified in each tissue under study. We found that the number of JSTs for IgM was higher in spleen than in gills or gut (Figure 3A). This finding reflects the greater diversity of splenic IgM compared to mucosal IgM. The latter may exhibit a relatively constrained diversity due to the expansion of mucosal plasmablasts and plasma cells expressing clonally affiliated IgM. Of note, IgD and IgT showed a tissue-specific JST profile largely similar to that of IgM (Figure 3A). Thus, mucosal IgD differs from splenic IgD and has a JST signature similar to that of canonical mucosal antibodies such as IgM and IgT.

Figure 3.

Rainbow Trout IgD from Gut and Gills Have a Unique VHDJH Gene Configuration as Well as Distinct CDR3 Spectratype Compared to IgD from the Spleen

(A) Total number of junction sequence types (JSTs) in different organs. Bar charts show the total number of JSTs identified for each isotype (mean + SD, n = 4). ^∗p < 0.05; ^∗∗p < 0.01.

(B) Heatmap representation of differential germline Ig VH, DH, and JH genes. Heatmaps represent the mean (n = 4) relative frequency observed for VH genes reaching at least 2.5% of sequences. Statistical differences are highlighted in bold (p < 0.05).

(C–E) CDR3 spectratyping of IgD transcripts was performed using sequences corresponding to the most expanded IgD VH families, namely, IGHV1S3 (C), IGHV2S1 (D), and IGHV8S7 (E). Observed p values are indicated in the spectratyping representation.

See also Figure S4.

Mucosal IgD⁺IgM⁻ B Cells Show a Unique V(D)J Family Usage

To gain additional insight into the molecular configuration of mucosal IgD, we dissected the heavy chain variable region (VH) family usage of IgD clones, as well as that of IgM and IgT clones, from gut and gills and used a systemic lymphoid organ such as the spleen as comparison. We found that IgM exhibited a highly diverse VH gene repertoire both in the spleen and mucosal tissues (Figure 3B). In all these tissues, IgM showed a predominance of IGHV1S3, IGHV2S1, and IGHV4S1 (Figure 3B). Compared to IgM, IgD displayed a more restricted VH gene family usage, with IGHV1S3, IGHV2S1, and IGHV8S7 gene segments present in approximately 32.76% of all splenic IgD sequences. Compared to splenic IgD, mucosal IgD showed an even higher IGHV1S3, IGHV2S1, and IGHV8S7 gene usage, as these genes were detected in 49.41% and 49.6% of IgD sequences from gills and gut, respectively (Figure 3B). At the same time, mucosal IgD from both gills and gut used less IGHV5S2, IGHV5S3, IGHV5S4, IGHV11S1, and IGHV13S1 compared to splenic IgD (Figure 3B). As for IgT, a preferential IGHV1S1 and IGHV4S1 gene usage was observed in all tissues under study (Figure 3B). Thus, mucosal IgD shows a VH gene repertoire strikingly different from that of splenic IgD.

When D segments were analyzed, we found similar IgM profiles in all tissues, with some 25% of IgM clones using IGHD4 (Figure 3B). This gene segment was also present in an elevated percentage of IgD clones (Figure 3B). However, the number of IgD clones lacking IGHD was increased in both gills and gut mucosae compared to the spleen and accounted for about 20% of mucosal IgD clones (Figure 3B). Of note, IGHD1 and IGHD3 marked approximately 75% of IgT clones in all tissues. Thus, mucosal IgD shows a D gene family usage distinct from that of splenic IgD.

Finally, we determined the J gene usage by IgM and found no differences in spleen, gills, and gut (Figure 3B). Similar to IgM clones, IgD clones frequently used IGHJ4 and IGHJ5. However, compared to IgD clones from the spleen, IgD clones from gills used more IGHJ1 and IGHJ2, and IgD clones from gut used less IGHJ7 (Figure 3B). As for IgT, although IGHJ1 accounted for about 70% of IgT clones in all tissues, IGHJ4, IGHJ5, and IGHJ6 were increased in gills, whereas IGHJ3, IGHJ5, and IGHJ6 were increased in the gut (Figure 3B). Thus, similar to mucosal IgT, mucosal IgD shows a J gene repertoire distinct from that of splenic IgD.

Collectively, these results indicate that mucosal IgD from rainbow trout features a molecular signature largely distinct from that of systemic IgD. This contrasts with the rather coherent molecular composition of splenic and mucosal IgM.

Mucosal IgD⁺IgM⁻ B Cells Undergo Antigen-Driven Clonal Expansion

Given the extreme sequence variability of the CDR3 segment from Ig VH regions and its different length in antigen-selected B-cell clones (Xu and Davis, 2000), the determination of CDR3 length by spectratyping is a powerful tool to estimate the overall diversity of the Ig gene repertoire (Castro et al., 2013). In general, any deviation of the CDR3 length from a bell-shaped Gaussian distribution indicates B-cell clonal expansion. To determine whether mucosal IgD⁺IgM⁻ B cells from rainbow trout underwent antigen-driven clonal expansion, we analyzed the CDR3 length distribution of unique IgD sequences by focusing on the most expanded VH families from spleen, gills, and gut, including IGHV1S3, IGHV2S1, and IGHV8S7. The CDR3 length from IGHV1S3 presented a Gaussian distribution in the spleen (Figure 3C). In contrast, in most fish, an oligoclonal expansion was observed in the gills and gut, with the exception of some fish that exhibited a monoclonal expansion (Figure 3C). Similar results were observed for IGHV2S1 (Figure 3D) and IGHV8S7 (Figure 3E). These results indicate that unlike IgD from splenic IgD⁺IgM⁺ B cells, IgD from mucosal IgD⁺IgM⁻ B cells exhibits molecular hallmarks of antigen-driven clonal expansion. This conclusion was further corroborated by the analysis of CDR3 from IgM and IgT, which showed length patterns consistent with antigen-driven clonal expansion of mucosal IgM⁺IgD⁻ or IgT⁺ (IgD⁻IgM⁻) cells, respectively (Figures S4A and S4B).

To further confirm the clonal expansion of mucosal IgD, we dissected the clonal size distribution of JSTs from gills, gut, and, for comparison, spleen. Studies published previously show that JSTs detected fewer than 3–5 times in a given tissue correspond to non-expanded antigen-naive B cells. In contrast, JSTs detected more than 100 times in a given tissue reflect expanded antigen-experienced B-cell clones, including antibody-secreting cells (Castro et al., 2013). When IgM was analyzed, the number of JSTs detected only once was lower in mucosal tissues than in the spleen (Figures 4A and 4B). Conversely, the number of JSTs detected more than 6 times was increased in mucosal tissues, compared to the spleen, including JSTs detected more than 500 times (Figures 4A and 4B). These results confirm the presence of IgM-secreting plasmablasts and/or plasma cells in mucosal rainbow trout tissues (Salinas et al., 2011). As for IgD, the number of IgD JSTs detected more than 100 times was significantly increased in both gills and gut compared to the spleen (Figures 4A and 4B). These data confirm that IgD⁺IgM⁻ B cells from both gut and gills correspond to plasmablasts undergoing antigen-driven mucosal clonal expansion. Finally, the number of IgT JSTs detected 101–500 times was also increased in gills and gut, compared to the spleen (Figures 4A and 4B). Thus, the rainbow trout shows that mucosal IgD has tissue-specific molecular signatures similar to those of mucosal IgM or IgT but different from those of splenic IgD. These signatures may reflect antigen-driven clonal expansion of mucosal IgD⁺IgM⁻ B cells.

Figure 4.

Rainbow Trout IgD from Gut and Gills Shows Molecular Hallmarks of Clonal Expansion Compared to IgD from the Spleen

(A) Grouped clonal size distribution for IgM, IgD, and IgT. Bar charts show the relative frequency of JSTs observed n times in the sequence datasets for each isotype (mean + SD, n = 4). ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

(B) Total clonal size distribution. Bar charts show the global distribution of JSTs observed n times in the sequence datasets.

(C) Heatmap representation of differential germline Ig VH family usage for each isotype identified in naive (JST 1–5 times) and expanded (Exp.; JST > 100 times) B cells. Heatmaps represent the mean (n = 4) relative frequency observed for VH genes reaching at least 2.5% in a sample. Statistical differences (p < 0.05) in naive versus expanded sequences are bolded in data from expanded columns.

(D) Physicochemical properties of amino acid sequences from CDR3 in naive and expanded sequences. Bar charts show the mean of averages (+SD, n = 4) observed in naive (JST 1–5 times) and expanded (JST > 100 times) sequences. ^∗p < 0.05; ^∗∗p < 0.005; ^∗∗∗p < 0.001.

Clonally Expanded IgD⁺IgM⁻ B Cells Express a Unique V(D)J Repertoire

Having shown that mucosal, but not splenic, IgD experiences clonal expansion as mucosal IgM and IgT do, we went on to comparatively dissect the V(D)J gene repertoire in non-expanded (i.e., naive) and expanded (i.e., antigen-activated) B-cell clones from the spleen, gut, and gills. Non-expanded B-cell clones were defined by JSTs detected 1–5 times, whereas expanded B-cell clones were defined by JSTs detected more than 100 times.

We first found that non-expanded and expanded IgM clones used largely overlapping VH gene families, with a few exceptions that included a higher IGHV5S3 usage in expanded IgM clones from gills and gut, compared to non-expanded IgM clones (Figure 4C). Remarkably, the use of some VH families by expanded IgD clones greatly differed in mucosal tissues compared to the spleen. For instance, non-expanded IgD clones from all tissues frequently use IGHV2S1 (Figure 4C). However, IGHV2S1 is never used by expanded splenic IgD clones, while its use is increased in expanded IgD clones from gills and gut (Figure 4C). Of note, expanded IgD clones from the spleen preferentially used IGHV3S3 and IGHV6S4, which were not frequently used by expanded IgD clones from gills or gut (Figure 4C). These latter two showed a bias for IGHV1S3, which was less evident in gills (Figure 4C). Finally, VH family use by non-expanded and expanded IgT clones was quite diverse in all tissues under study (Figure 4C).

Next, we analyzed the expression patterns of D and JH gene segments and found that a wide range of them were paired with IGHV2S1 (Figure S5), which is frequently used by expanded IgD clones from gills and gut, but not spleen. This finding likely reflects an antigen-driven mucosal expansion of multiple IGHV2S1-using B-cell clones. In contrast, IGHV3S3 and IGHV6S4 from expanded splenic, but not mucosal, clones used a variety of D segments but a more restricted range of J segments (Figure S5). Of note, IGHV1S3 used no D segment in expanded IgD clones from gills or gut (Figure S5). Thus, clonally expanded mucosal IgD clones show unique V(D)J gene repertoires, which may reflect the selection of a fraction of IgD⁺IgM⁻ B cells by intraluminal antigens.

Mucosal IgD⁺IgM⁻ B Cells Show Antigen-Induced CDR3 Changes

To further understand how naive IgD sequences differ from antigen-induced IgD sequences in rainbow trout, we compared the physicochemical properties of IgM, IgD, and IgT CDR3s from non-expanded and expanded B-cell clones inhabiting the spleen, gills, and gut. Antigen-experienced Igs from mammals undergo physicochemical changes relative to the antigen-naive Ig repertoire (Wu et al., 2010). One of these changes involves the length of the CDR3 segment, which is smaller in canonical antigen-experienced B cells than in naive B cells (DeKosky et al., 2016, Wu et al., 2010). We found that CDR3s were significantly smaller in expanded IgM clones than in naive IgM clones in all tissues under study (Figure 4D). Splenic IgD also showed smaller CDR3s in expanded clones than in naive clones (Figure 4D). However, in mucosal tissues, CDR3s from expanded IgD clones showed no significant differences in length compared to CDR3s from naive IgD clones (Figure 4D). As a result of this, CDR3s from expanded mucosal IgD clones were longer compared to CDR3s from expanded splenic IgD clones (Figure 4D). This finding echoes published human studies showing that a large fraction of mucosal IgD clones have longer-than-average CDR3s (Duty et al., 2009). Finally, the length of CDR3s from naive and expanded IgT clones showed no significant differences in any tissue studied (Figure 4D).

An additional antigen-induced physicochemical alteration of CDR3 consists of its lower aliphatic index in antigen-experienced memory B cells than in antigen-naive B cells (Wu et al., 2010). Accordingly, we found a lower aliphatic index in the CDR3s of expanded IgM clones from all tissues under study, compared to the CDR3s from naive IgM clones (Figure 4D). Similarly, the aliphatic index of CDR3s from expanded IgD clones in the spleen or gills, but not the gut, was lower than the aliphatic index of naive IgD clones from identical tissues (Figure 4D). In contrast, no significant differences were observed in the aliphatic index of CDR3s from splenic and mucosal IgT clones (Figure 4D).

Another antigen-induced physicochemical alteration of CDR3 relates to its lower hydrophobicity in antigen-experienced B cells (Wu et al., 2010). We found that the hydrophobicity of CDR3s from expanded IgM clones was reduced compared to that of naive IgM clones, but only in mucosal tissues (Figure 4D). As for IgD, a decreased hydrophobicity was detected in expanded IgD clones compared to naive IgD clones in the spleen and gills, but not in the gut (Figure 4D). Finally, naive and expanded IgT clones showed no differences in their hydrophobicity in any of the tissues under study (Figure 4D). Thus, in rainbow trout, clonally expanded IgD⁺IgM⁻ B-cell clones from some, but not all, mucosal tissues show physicochemical changes compatible with IgD selection by antigen.

Mucosal IgD⁺IgM⁻ B Cells Are Somatically Hypermutated

In mammals, the induction of Ig V(D)J gene SHM in antigen-activated B cells from the specialized microenvironment of the GC provides a structural correlate for the selection of antibody mutants with a higher affinity for antigen. The mutational machinery expressed by antigen-activated B cells includes the DNA-editing enzyme AID, which targets RGYW and WRCY hotspots within Ig V(D)J genes (Faili et al., 2002). In addition, the error-prone DNA polymerase eta (Pol-η) targets WA and TW hotspots within Ig V(D)J genes (Delbos et al., 2007). Despite lacking lymphoid follicles and GCs, teleosts express AID, which triggers some degree of SHM in antigen-activated B cells (Abos et al., 2018). Remarkably, antigen-activated B cells from mammals can also undergo clonal expansion and some SHM outside the follicular environment from GCs, at least under specific conditions (Di Niro et al., 2015, William et al., 2002). This prompted us to compare the mutational rate of naive and clonally expanded IgD clones from systemic (i.e., spleen) and mucosal (i.e., gills, gut) compartments.

The total number of V(D)J gene mutations in naive and expanded IgD clones from all tissues under study did not evidence significant differences (not shown). In splenic IgD clones, mutations within AID-targeted RGYW and WRCY motifs were readily identified (Figure 5A). Remarkably, expanded IgD clones showed fewer WRCY mutations than did naive IgD clones (Figure 5A). Mutations of Pol-η-targeted WA and TW motifs from splenic IgD clones were also detected, but no differences between naive and expanded clones were observed (Figure 5A). In contrast, expanded IgD clones from both gills and gut showed increased mutations at both AID-targeted and Pol-η-targeted hotspots compared to naive IgD clones (Figure 5A).

Figure 5.

Increased SHM in Antigen-Expanded IgD Clones from Gills and Gut but Not the Spleen Compared to Naive IgD Controls

(A) Bar charts show the average (+SD, n = 4) percentages of mutations identified in AID and DNA polymerase eta (pol eta) motifs in naive and expanded IgM, IgD, and IgT sequences. ^∗p < 0.05; ^∗∗∗p < 0.001.

(B) Bar charts show the percentage of mutations in IgD sequences producing replacement in amino acids and silent mutations identified in CDR2 and FR3 for each donor fish.

Finally, we evaluated the distribution of somatic mutations within distinct IgD V(D)J segments. In splenic naive IgD clones, both silent and replacement mutations were detected within complementarity-determining region 2 (CDR2) and framework 3 (FR3) segments (Figure 5B). However, silent CDR2 mutations decreased in expanded splenic IgD clones compared to naive splenic IgD clones (Figure 5B). In gills and gut, silent CDR2 mutations from naive IgD clones were greatly reduced compared to the spleen (Figure 5B). Indeed, virtually no CDR2 mutations were silent in IgD clones from gills or gut (Figure 5B). In addition, replacement FR3 mutations were more abundant in gills or gut compared to in the spleen (Figure 5B). Thus, the mutational profile of IgD V(D)J regions from rainbow trout raises the possibility that mucosal IgD⁺IgM⁻ B cells are exposed to discrete antigenic forces capable of promoting the selection of higher-affinity IgD mutants.

Discussion

IgD remains the most enigmatic antibody in vertebrates. Yet, IgD is involved in severe diseases such as hyper-IgD syndrome (van der Meer and Simon, 2016) or IgD multiple myeloma (Sinclair, 2002). Additionally, it could have an unexpected impact on spontaneous or vaccine-induced immunity to infections in both humans and animals, including fish. Thus, a better understanding of IgD is necessary because harnessing its functions may improve human health in addition to reducing costs in the food industry.

We found that both gut and gill mucosae from rainbow trout contained IgD⁺IgM⁻ cells with canonical plasmablast-like properties (Granja and Tafalla, 2019), including a larger cytoplasm-to-nucleus ratio, decreased surface MHC II expression, and higher Blimp-1 transcription. In addition to suggesting an evolutionary conserved role of IgD in mucosal homeostasis, our findings show that teleost B cells can mount clonally expanded and mildly mutated IgD responses. This is remarkable, considering that the putative protective role of IgD has so far been confined to the mammalian aerodigestive tract. This mucosal compartment may be involved in the induction of tolerance to airborne and oral antigens during the first years of life. Gut and gill IgD could implement similar tolerogenic functions in teleosts. Similar to mouse mucosal IgD (Choi et al., 2017), teleost mucosal IgD was partly dependent on inductive signals from the local microbiota. Indeed, treatment of rainbow trout with broad-spectrum antibiotics attenuated gut IgD transcription. Thus, IgD may establish a mutualistic relationship with commensals to enhance mucosal homeostasis.

The demonstration of IgD in the teleost intestine represents a “quantum leap” in our overall understanding of the humoral control of microbiota-colonized mucosal surfaces and may stimulate a thorough reassessment of IgD responses in the mammalian gastrointestinal tract. Indeed, identifying and functionally characterizing IgD in gut-associated organs could have far-reaching implications in both health and disease, including allergic and inflammatory disorders, as the presence of some IgD secretion in gut-associated lymphoid compartments could also occur in humans. This hypothesis may explain why some patients with hyper-IgD syndrome develop abdominal symptoms as well as enlarged MLNs during attacks of periodic fever, which is the hallmark of this autoinflammatory disorder (Oretti et al., 2006). Finally, in healthy individuals, gut IgD might contribute to the clearance of some food antigens and to their tolerance, possibly in cooperation with nasopharyngeal IgD (Shan et al., 2018).

In addition to indicating the potentially complementary role of IgD to canonical mucosal IgT and IgM responses, our data reveal that clonally organized and mildly mutated IgD responses can take place in the diffuse lymphoid tissue from the rainbow trout intestine, which is devoid of GC-rich lymphoid structures. In this regard, IgD is the most mutated human antibody, possibly due to the continuous re-entry of pre-existing human IgD⁺IgM⁻ memory B cells into mucosal GCs (Liu et al., 1996). This iterative process would allow IgD⁺IgM⁻ B cells to adjust their V(D)J gene repertoire to transient changes in food and commensal antigens (Shan et al., 2018). Our data indicate that in teleosts, the dispersed lymphoid tissue from gut and gill mucosae might be sufficient to support some degree of AID-mediated SHM, possibly aimed at adjusting IgD responses to small changes in commensal and waterborne antigens (Salinas et al., 2011). Consistent with this possibility, we detected molecular traces of antigen-mediated IgD selection.

Aside from showing IgD-secreting IgD⁺IgM⁻ plasmablasts in gills, the rainbow trout displayed IgD-secreting IgD⁺IgM⁻ plasmablasts in the intestine, but not the skin (Xu et al., 2013). This implies that the homing requirements of teleost IgD⁺IgM⁻ plasmablasts are quite different in gills or gut, compared to the skin. An alternative but not mutually exclusive possibility is that despite its remarkable diversity (Lowrey et al., 2015), the skin microbiota may not support the homing and/or survival of IgD⁺IgM⁻ plasmablasts.

By showing a coating of gut bacteria by IgD, our study raises questions as to the mechanism whereby IgD translocates across epithelial cells to reach the intestinal lumen. In mammals, monomeric IgD does not associate with the J chain and thereby does not bind to the polymeric Ig receptor (pIgR), the epithelial transporter of polymeric IgM and IgA. In rainbow trout, monomeric IgD coated gut bacteria as much as polymeric IgM did, but the latter has a coating potential inferior to that of polymeric IgT (Zhang et al., 2010). While polymeric IgM and IgT bind to pIgR, it remains unknown how teleost IgD undergoes transcytosis. However, similar to coating IgM and IgT, coating IgD could regulate the attachment of commensals to the gut epithelium and block its invasion by pathogens. In mammals, this process would involve the cooperation of IgA with IgM (Magri et al., 2017). In teleosts, it may entail the cooperation of IgD with IgT and IgM.

As for the molecular properties of IgD, both gut and gill IgD displayed less diversified, more clonally expanded, and more somatically mutated V(D)J family genes than the splenic B cells. These properties suggest that, although not clonally related, IgD⁺IgM⁻ plasmablasts from both gills and gut are exposed to similar antigen-driving forces, which are likely related to activating signals from the commensal microbiota. Remarkably, the mutational profile of rainbow trout mucosal IgD featured mutations mostly concentrated within the FRs and not the CDRs of V(D)J genes. Given the lack of organized lymphoid follicles in fish, these findings indicate that IgD⁺IgM⁻ plasmablasts from gut and gills undergo some degree of antigen-driven SHM and selection in the extrafollicular compartment of the lamina propria, which echoes some previously published findings in mammals (Di Niro et al., 2015, William et al., 2002).

In summary, we have demonstrated that both gut and gills include a clonally expanded and mildly mutated population of mucosal IgD⁺IgM⁻ plasmablasts that secrete IgDs reactive to the commensal microbiota. These IgD-secreting cells receive IgD-inducing signals from commensal bacteria, suggesting a mutualistic relationship between secreted IgD and the local microbiota that is possibly relevant to mucosal homeostasis. The increased presence of intestinal IgD in teleosts compared to mammals could reflect a compensatory response to the lack of mucosal IgA. More importantly, our results provide evidence that IgD is an evolutionarily conserved mucosal antibody that, in principle, could be harnessed to enhance mucosal homeostasis and/or clear mucosal pathogens.

STAR★Methods

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rainbow trout anti-IgM specific monoclonal Ab	DeLuca et al., 1983	N/A
Rainbow trout anti-IgD specific monoclonal Ab	Ramirez-Gomez et al., 2012	N/A
APC-labeled mouse IgG1	BioLegend	Cat#400119; RRID: AB_10679040
Biological Samples
Healthy rainbow trout (Oncorhynchus mykiss), adults of approximately 100 g	Piscifactoria Cifuentes (Cifuentes, Guadalajara, Spain)	N/A
Chemicals, Peptides, and Recombinant Proteins
Benzocaine	Sigma-Aldrich	Cat#E1501
Heparin	Sigma-Aldrich	Cat#H3149
Leibovitz’s medium (L-15)	GIBCO	Cat#11415049
Penicillin Streptomycin solution	GIBCO	Cat#11548876
Fetal calf serum	GIBCO	Cat#10270106
Percoll	GE Healthcare	Cat#17089101
Trypan blue	Sigma-Aldrich	Cat#T6146
R-PE Lightning-Link labeling kits	Innova Biosciences	Cat#703-0015
FITC Lightning-Link fluorescein	Innova Biosciences	Cat#707-0010
APC Lightning-Link labeling kits	Innova Biosciences	Cat#765-0005
4’,6-diamine-2′-phenylindole dihydrochlorid (DAPI)	Sigma-Aldrich	Cat#D9542
poly-L-lysine	Sigma-Aldrich	Cat# P4707
Fluoromount	Sigma-Aldrich	Cat#F4680
Protease inhibitors	Roche	Cat#1169749800
SYTO BC	Thermo Fisher	Cat#S34855
Amoxicillin	Sigma-Aldrich	Cat#A8523
Oxytetracycline	Sigma-Aldrich	Cat#BP782
Oxolinic acid	Sigma-Aldrich	Cat#O0877
TSA agar	Sigma-Aldrich	Cat#22091
Critical Commercial Assays
TRI-Reagent	Thermo Scientific	Cat#AM9738
DNase I RapidOut DNA Removal Kit	Thermo Scientific	Cat#K2981
RevertAid Reverse Transcriptase	Thermo Scientific	Cat#EP0441
FastStart Essential DNA Green Master reagents	Roche	Cat#06402712001
RNeasy Mini Kit	QIAGEN	Cat#74104
DNA polymerase	Biotools	Cat#10013-4104
GelRed® Nucleic Acid Gel Stain	Biotium	Cat#41003
TruSeq DNA PCR-Free Library Prep Kit	Illumina	Cat#20015962
QuBit DNA quantification system	Invitrogen	Cat#Q33238
Agilent 2100 Bioanalyzer	Agilent Technologies	Cat#G2939BA
Power SYBR Green Cells-to-Ct Kit	Ambion	Cat#4402953
Deposited Data
Illumina MiSeq Raw data	This paper	GenBank: PRJNA548246
Oligonucleotides
oligo(dT)23VN	Sigma-Aldrich	Cat#O4387
Primers for rainbow trout IgM membrane, IgM secreted, IgD membrane and IgD secreted (Experimental genes in RT-PCR)	Castro et al., 2014, Granja et al., 2017	N/A
Primers for rainbow trout EF-1α gene (Reference gene in RT-PCR)	Castro et al., 2014, Granja et al., 2017	N/A
Primers for rainbow trout β-actin amplification (Reference gene in RT-PCR)	Dong et al., 2017	N/A
Primers for immune repertoire amplification located in rainbow trout IGHV, IGHM, IGHD and IGHT genes	Abos et al., 2018, Castro et al., 2013	N/A
Primer Blimp1F GenBank: LOC110486999 F: CATTCGGCCCTATGTGTGG	This paper	N/A
Primer Blimp1R GenBank: LOC110486999 R: CCCCTCGGTAGTCAACATGG	This paper	N/A
Software and Algorithms
BD FACSDiva software	BD Biosciences	https://www.bd.com/en-us
FlowJo® v.10	FlowJo LLC, Tree Star	https://www.flowjo.com/
Zeiss Zen software	Carl Zeiss	https://www.zeiss.com/corporate/int/home.html
Adobe Photoshop CS6 software	Adobe Systems	https://www.adobe.com/
MiSeq Analysis pipeline	Illumina	https://www.illumina.com/
FASTX toolkit (Barcode splitter, Trimmer and Quality filter tools)		http://hannonlab.cshl.edu/fastx_toolkit/
FASTQ interlacer tool	Blankenberg et al., 2010	https://usegalaxy.org/
PEAR software	Zhang et al., 2014	https://cme.h-its.org/exelixis/web/software/pear/
IMGT/HighV-QUEST tool	Alamyar et al., 2012	http://www.imgt.org/HighV-QUEST/home/action
Antigen Receptor Galaxy (ARGalaxy)	IJspeert et al., 2017	https://bioinf-galaxian.erasmusmc.nl/argalaxy
BRepertoire software	Margreitter et al., 2018	http://mabra.biomed.kcl.ac.uk/BRepertoire
Office Excel 2010	Microsoft	https://www.microsoft.com/en-us
GraphPad Prism version 7.00	GraphPad	https://www.graphpad.com/scientific-software/prism/

Lead Contact and Materials Availability

This study did not generate new unique reagents. Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Carolina Tafalla (tafalla@inia.es).

Experimental Model and Subject Details

Healthy rainbow trout (Oncorhynchus mykiss) adults of approximately 100 g were obtained from Piscifactoria Cifuentes (Cifuentes, Guadalajara, Spain) and maintained at the animal facilities of the Animal Health Research Center (CISA-INIA, Spain) in an aerated recirculating water system at 16°C, with a 12:12 h light:dark photoperiod. All animals used were females and the influence of sex was not considered in the analysis of the data. Fish were fed twice a day with a commercial diet (Skretting) and were acclimatized to laboratory conditions for at least 2 weeks prior to any experimental procedure. During this period no clinical signs were ever observed. All the experiments described comply with the Guidelines of the European Union Council (2010/63/EU) for use of laboratory animals and were approved by the Ethics Committee from INIA (Code PROEX 002/17). All efforts were made to minimize suffering.

Method Details

Fish sampling procedures

Fish were anaesthetized with benzocaine (Sigma) and prior to sampling, a transcardial perfusion was conducted to remove all circulating blood from tissues. For this, the heart was cannulated through the ventricle into the bulbus arteriosus with approximately 30 mL of 0.9% NaCl, using a peristaltic pump (Selecta, Spain), while the atrium was cut to drain the blood out of the circulatory system. After perfusion, tissues were sampled for RNA extraction to analyze the Ig repertoire (gills, spleen and gut) and for leukocyte isolation to characterize the different non-IgT B cell populations by flow cytometry (gills, spleen, gut, kidney and skin) and immunofluorescence (gills, spleen and gut). To obtain blood leukocytes for flow cytometry, peripheral blood was extracted from the caudal vein of freshly killed rainbow trout using a heparinized syringe (Sigma-Aldrich).

Leukocyte isolation

Total leukocyte populations were isolated from spleen, gills, gut, kidney and skin of blood-depleted (buffer-perfused) naive fish as well as from peripheral blood. Spleen, gill and kidney cell suspensions were obtained by passing the tissues through a 100 μm nylon mesh (BD Biosciences) using Leibovitz’s medium (L-15, GIBCO) containing 100 I.U./ml penicillin, 100 μg/ml streptomycin (P/S, Life Technologies), 10 U/ml heparin (Sigma- Aldrich) and 5% fetal calf serum (FCS, GIBCO). Skin and gut leukocytes were isolated following an enzymatic digestion of the tissues as previously described (Granja et al., 2015, Soleto et al., 2019). For all tissues, cell suspensions were placed onto 30/51% Percoll discontinuous density gradients and centrifuged at 500 x g for 30 min at 4°C. Blood was diluted 10 times with L-15 medium containing antibiotics, 10 U/ml heparin and 5% FCS. Peripheral blood leukocytes (PBLs) were isolated placing blood samples onto 51% Percoll (GE Healthcare) density gradients. In all cases, the interface cells were collected, washed with L-15 supplemented antibiotics and 5% FCS. The viable cell concentration was determined by Trypan blue (Sigma-Aldrich) exclusion and cells were resuspended in L-15 with 5% FCS at a concentration of 1x10⁶ cells/ml.

Flow cytometry analysis

To analyze the distribution of the non-IgT B cell subsets in different tissues, leukocytes isolated from gills, spleen, gut, kidney, skin or peripheral blood were incubated with monoclonal antibodies against IgM and IgD and analyzed by flow cytometry. For this, leukocytes obtained as described above were incubated with the anti-IgM and IgD specific monoclonal antibodies in staining buffer (phenol red-free L-15 medium supplemented with 2% FCS) for 1 h at 4°C. The anti-trout IgM [1.14 mAb mouse IgG₁ coupled to R-phycoerythrin (R-PE), 1 μg/ml] and the anti-trout IgD [mAb mouse IgG₁ coupled to allophycocyanin (APC), 5 μg/ml] used in this study have been previously characterized (DeLuca et al., 1983, Ramirez-Gomez et al., 2012) and were fluorescently labeled using R-PE or APC Lightning-Link labeling kits (Innova Biosciences) following manufacturer’s instructions. After the staining, cells were washed twice with staining buffer and analyzed on a FACS Celesta flow cytometer (BD Biosciences) equipped with BD FACSDiva software. The cell viability was checked by staining the cells with 4’,6-diamine-2′-phenylindole dihydrochlorid (DAPI, 0.2 μg/ml). Flow cytometry analysis was performed with FlowJo® v.10 (FlowJo LLC, Tree Star).

Confocal microscopy

Spleen, gills and gut leukocyte suspensions were collected and seeded on a poly-L-lysine (0.01% solution, Sigma)-coated slide and incubated at room temperature (RT) for 1 h in a humidified chamber. The slides were then fixed in 4% paraformaldehyde solution for 30 min at RT. The fixed samples were incubated for 1 h at RT with blocking solution (TBS, pH 7.5 containing 5% BSA and 0.5% saponin) to minimize non-specific adsorption of the antibodies to the coverslip. The samples were then incubated with mAbs against trout IgM [coupled to fluorescein isothiocyanate (FITC) 17 μg/ml] and trout IgD (coupled to APC, 50 μg/ml) for 1 h at RT in a humidified chamber. Slides were counterstained with 1 μg/ml DAPI (Sigma-Aldrich) for 10 min at RT, rinsed with PBS 1x and mounted with Fluoromount (Sigma-Aldrich) for microscopy.

To visualize the coating of gut luminal bacteria with IgM and IgD, gut bacteria isolated as described before were incubated for 1 h at RT with anti-trout IgM-APC or anti-trout IgD-APC (5 μg/ml each) or the isotype control antibody anti-IgG1-APC (BioLegend). Finally, samples were washed and incubated with SYTO BC (Thermo Fisher; 2 μM) for 30 min at RT before mounting for microscopy.

Laser scanning confocal microscopy images were acquired with an inverted Zeiss Axiovert LSM 880 microscope with Zeiss Zen software. Images were analyzed with Fiji (NIH) software package and processed with Adobe Photoshop CS6 software.

Analysis of Blimp1 transcription

Blood leukocytes, obtained as described above, were incubated with monoclonal antibodies against IgM and IgD. IgM⁺IgD⁺ and IgD⁺IgM^- B cells were isolated by flow cytometry in a BD FACSAria III cell sorter (BD Biosciences). Total cellular RNA was isolated from cell populations using the Power SYBR Green Cells-to-Ct kit (Invitrogen) following the manufacturer’s instructions. RNA was treated with DNase during the process to remove genomic DNA that might interfere with the PCR reactions. Reverse transcription was also performed using the Power SYBR Green Cells-to-Ct kit following the manufacturer’s instructions. To evaluate the levels of transcription of Blimp1 (Uniprot; O75626), real-time PCR was performed with a LightCycler 96 System instrument (Roche) using SYBR Green PCR core Reagents (Applied Biosystems) and specific primers. Each sample was measured in duplicate under the following conditions: 10 min at 95°C, followed by 40 amplification cycles (15 s at 95°C and 1 min at 60°C). A melting curve for each primer set was obtained by reading fluorescence every degree between 60 and 95°C to ensure only a single product had been amplified. The expression of individual genes was normalized to the relative expression of β-actin, and the expression levels were calculated using the 2^−ΔCt method, where ΔCt is determined by subtracting the β-actin value from the target Ct. No template negative controls and minus reverse transcriptase controls were also included.

Bacterial flow cytometry

To determine whether intestinal bacteria were coated by IgD, a flow cytometry method previously described was adapted (Magri et al., 2017, Zhang et al., 2010). Briefly, the intestinal content obtained from rainbow trout was resuspended at 0.1 g/ml in PBS with protease inhibitors (Roche). Samples were homogenized and centrifuged at 400 x g for 5 min to pellet large particles. Supernatants were filtered through a sterile 70 μm cell strainer and centrifuged at 8,000 x g for 5 min to pellet bacteria. Pellets were washed three times with PBS, resuspended in staining buffer with either anti-trout IgM and or anti-trout IgD coupled to APC (5 μg/ml), and incubated at 4°C for 1 h. Bacteria stained with an APC-labeled mouse IgG1 isotype (BioLegend) were used as controls. Finally, samples were washed and incubated with SYTO BC (Thermo Fisher; 2 μM) for 30 min at 4°C before performing flow cytometry analysis.

Antibiotic treatment

Rainbow trout of approximately 5 g were fed twice a day with feed pellets supplemented with antibiotics to achieve daily doses of 15 mg/kg fish amoxicillin, 55 mg/kg fish oxytetracycline and 15 mg/kg fish oxolinic acid. All antibiotics were obtained from Sigma-Aldrich. After 15 days, antibiotic-treated fish were sacrificed along with control fish maintained in the same conditions but not treated with antibiotics. The intestinal content was removed and the hindgut sampled and placed in TRI-Reagent (Thermo Fisher Scientific) to isolate total RNA following manufacturer's instructions. Microbiota reduction was confirmed by culture of intestinal contents on TSA agar (Sigma-Aldrich). One μg of hindgut RNA was treated with DNase I to remove any genomic DNA traces using a RapidOut DNA Removal Kit (Thermo Scientific) and then used to synthesize cDNA using the RevertAid Reverse Transcriptase (Thermo Scientific), primed with oligo(dT)₂₃VN (1.6 μM), following the manufacturer's instructions. cDNA was diluted in nuclease-free water and stored at −20°C until use.

To evaluate the levels of transcription of the membrane and secreted forms of IgM and IgD, a real-time PCR was performed in a LightCycler 96 System instrument (Roche) using FastStart Essential DNA Green Master reagents (Roche) and specific primers previously described (Castro et al., 2014, Granja et al., 2017). Each sample was subjected, in duplicate, to an initial cycle of denaturation (95°C for 10 min), followed by 40 amplification cycles (95°C for 10 s, 60°C for 10 s and 72°C for 10 s). A dissociation curve was obtained by reading fluorescence every degree between 60°C and 95°C to ensure only a single product was amplified. Negative controls with no template and minus reverse transcription controls (-RT) were included in all experiments. Gene expression was normalized to the relative expression of the rainbow trout elongation factor (EF-1α) gene amplified using specific primers (Castro et al., 2014, Granja et al., 2017). Expression levels were calculated using the 2^-ΔCt method, where ΔCt is determined by subtracting the EF-1α value from the target Ct as described previously (Castro et al., 2014, Granja et al., 2017).

Repertoire analysis

Total RNA was extracted from gills, spleen and gut using a combination of TRI-Reagent and RNeasy Mini Kit (QIAGEN). For this, samples were homogenized in 1 mL of TRI-Reagent through mechanical disruption and centrifuged at 12,000 x g for 15 min at 4°C to remove tissue debris. A total of 200 μl of chloroform was then added and the suspension centrifuged again at 12,000 x g, for 15 min at 4°C. The clear upper phase was recovered, mixed with an equal volume of 100% ethanol, and transferred to RNeasy Mini Kit columns. The procedure was then continued following the manufacturer’s instructions and performing an on-column DNase treatment. Finally, RNA pellets were eluted from the columns in RNase-free water and stored at −80°C until use.

Purified RNA was quantified using a Nanodrop spectrophotometer and its integrity verified on a 1.5% denaturating agarose gel. To obtain cDNA from each sample, 2.5 μg of RNA were reverse transcribed using RevertAid Reverse Transcriptase (Thermo Fisher Scientific) and oligo(dT)₂₃VN in a final volume of 20 μl following the manufacturer’s instructions. The resulting cDNA was diluted in a 1:5 proportion with water and stored at −20°C until use.

cDNA samples from spleen, gills and gut obtained from four representative fish were individually amplified using combinations of forward primers specific for each subgroup of IGHV genes and one of the reverse primers specific for IGHM, IGHD or IGHT genes (Cμ, Cδ or Cτ). All primers have been previously designed and validated (Abos et al., 2018, Castro et al., 2013). Mix reactions for the PCR were as follows: 1 μl of cDNA was used as template using 0.2 mM of each dNTP, 0.2 mM of each primer, and 0.03 U/μl DNA polymerase (Biotools) in 1x reaction buffer with 2 mM MgCl_2. The PCR was programmed as follows: an initial step of 95°C for 5 min followed by 40 cycles of 95°C for 45 s, 60°C for 60 s, 74°C for 45 s and a final extension step of 74°C for 10 min. Negative controls without cDNA were included. An 8 μl aliquot of each PCR product was mixed with 2 μl of loading buffer and loaded into a 1.5% agarose gel stained with GelRed® Nucleic Acid Gel Stain, 10,000X (Biotium) for visualization. In parallel, 3 μl aliquots of all PCR products obtained from the same tissue were pooled together for further RNaseq analysis. DNA concentrations were measured using the QuBit DNA quantification system (Invitrogen) and the quality checked using an Agilent 2100 Bioanalyzer (Agilent Technologies). A library was constructed per individual tissues with the TruSeq DNA PCR-Free Library Prep Kit (Illumina) according to the manufacturer’s instructions. Libraries were pooled together and paired-end sequencing was performed on an Illumina MiSeq with a MiSeq Reagent Kit v3 (2 × 300 cycles) cartridge (Illumina).

Sequence analysis

Raw data was demultiplexed and sequencing adapters and barcodes removed from the sequences by the MiSeq Analysis pipeline (Illumina). The first 20 nt from reverse primers used in the PCRs were used as barcodes for the identification of 3′ ends corresponding to the constant gene. Using the FASTQ Barcode splitter tool (http://hannonlab.cshl.edu/fastx_toolkit/) reads from R1 or R2 that matched a primer sequence allowing one mismatch were classified into the corresponding isotypes (IgM, IgD or IgT). The opposite paired end reads, corresponding to the 5′ end of PCR products, were extracted with the FASTQ interlacer tool implemented in Galaxy (Blankenberg et al., 2010). For IgT, given its smaller length, the paired forward and reverse reads were merged using PEAR software (Zhang et al., 2014) with a minimum overlap size of 12 nt. When no overlap was detected, reads corresponding to the 5′ end were retained, adjusted to 250 bp and included together with merged reads. For IgM and IgD, 5′ end reads were adjusted to 250 bp using the FASTQ Trimmer tool and included for successive analysis. A quality filter was applied to retained sequences accepting those with Phred base quality ≥ 20 in at least 90% of the full sequence using the FASTQ Quality filter tool.

The sequences corresponding to each Ig isotype obtained in each tissue for each individual fish were compared with the available information from Oncorhynchus mykiss contained in the International Immunogenetics information system databases (Lefranc et al., 2005) using the IMGT/HighV-QUEST tool (Alamyar et al., 2012). Results were analyzed with the immune repertoire and SHM and CSR pipelines from Antigen receptor Galaxy (ARGalaxy) tool (H et al., 2017) to obtain JST and CDR3 sequences and to analyze V(D)J family usage and SHM. The characteristics and physicochemical properties of CDR3 and an advanced analysis related to VD and VJ usage were also performed with the BRepertoire software (Margreitter et al., 2018).

Quantification and Statistical Analysis

Data obtained from flow cytometry was analyzed using Microsoft Office Excel 2010 and GraphPad Prism version 7.00 for Windows (GraphPad Software). Statistical analyses were performed using an unpaired t test (Welch’s t test). Results from repertoire analysis were compared in different tissues for each Ig isotype using a two-tailed Student's t test. Statistical significance for differences in CDR3 length distribution was calculated using the Kolmogorov-Smirnov test implemented in the BRepertoire software. The differences between the mean values were considered significant on different degrees, where ^∗ means p ≤ 0.05, ^∗∗ means p ≤ 0.01, ^∗∗∗ means p ≤ 0.001. Unless otherwise specified in the figure legends, the data presented represents the mean + the standard deviation (SD). Numbers of individual fish used in each experiment (biological replicates) are detailed in figure legends (n).

Data and Code Availability

The raw data from Immune repertoire sequencing generated in this study is available at NCBI (BioProject: PRJNA548246). All used software is available and listed in the Key Resources Table.

Published: December 24, 2019