Nasal Vaccination Drives Modifications of Nasal and Systemic Antibody Repertoires in Rainbow Trout

In vertebrates, mucosal surfaces constitute dynamic environments continuously exposed to foreign components. These barriers are armed with MALT, which protects them against pathogen attack while establishing symbiotic relationships with commensal microorganisms (1–5).

In mammals, the induction of Ag-specific mucosal immune responses occurs in special microenvironments, where immunocompetent cells (T, B, and APCs) form well-defined structures known as secondary lymphoid organs (SLOs) (6, 7). Peyer patches and nasopharynx-associated lymphoid tissue (NALT) are the best known immune mucosal inductive sites in mammals (8). Mammalian NALT includes well-organized lymphoid structures (organized NALT [O-NALT]) as well as scattered lymphoid cells (diffuse NALT [D-NALT]) (9). In humans, O-NALT consists of several tonsils arranged in a ring known as the Waldeyer ring, whereas in mice, a paired lymphoid tissue structure located along both sides of the nasopharyngeal duct is considered the functional equivalent of the human Waldeyer ring (9–11).

A hallmark of nasal Ab responses in mammals is the production of local specific IgA responses. In mice, O-NALT is considered an inductive site and D-NALT is considered the effector site, and O-NALT was reported to be essential for the μ to α class switching in the upper respiratory tract (12). Nasal vaccination induces IgA specific responses in both O-NALT and D-NALT in mammals (13, 14). Further comparisons between live and inactivated vaccines has shown that intranasal live vaccines such as FluMist or Sendai virus vaccines induce specific IgA-secreting cells in D-NALT and lung but not in systemic tissues or serum in cotton rats. In contrast, inactivated vaccines generate a systemic IgG-mediated response with negligible production of IgA (15, 16). Because adaptive IgA responses were constitutive and long-lasting in the upper respiratory tract, it appears that nasal IgA-secreting cells were NALT residents and not circulating from systemic sites.

The specific contributions of MALT with well-organized SLOs (O-MALT) and with diffuse lymphoid cells (D-MALT) to the overall local IgA production have been further investigated using surgical ablation of O-MALT in mice. Although these studies have shown a nonessential role for O-MALT in the generation of influenza-specific IgA (17), O-NALT appears to be the main contributor of staphylococcal enterotoxin B–specific IgA (18). Thus, O-NALT and D-MALT appear to play more or less critical roles in local IgA production depending on the type of Ag as well as the animal model of choice.

Bony fish represent the most basal vertebrate branch with a known dedicated mucosal Ig, IgT, that plays a similar role to mammalian IgA (19–22) but where no Ab class switching takes place. Teleost MALT include NALT, which has been described in detail in rainbow trout (Oncorhynchus mykiss) as a rich network of innate and adaptive immune cells in the olfactory epithelium (10, 23–25). Because teleost NALT lacks O-NALT structures, rainbow trout can be used as a model to elucidate how D-NALT can mediate local mucosal Ab responses without the need of surgical procedures (23).

Trout D-NALT contains abundant B cells that are mostly located intraepithelially, 48.5% being IgM⁺ B cells and 51.5% IgT⁺ B cells (26). Interestingly, specific IgT but not IgM responses against parasites are detected in the trout nasal mucosa following a chronic infection, a common theme with other trout mucosal barriers such as skin, gut, and gills (19, 22, 27). These findings may suggest that mucosal adaptive responses trigger activation and proliferation of specialized B cells expressing the mucosal isotype IgT and elicit the strong protection observed after nasal vaccination with live or killed vaccines (26, 28). However, Ag-specific IgM⁺ B cells may be induced at systemic sites or migrate to systemic sites once activated in the mucosal periphery and also play a role in protection. The mechanisms underlying the specific protection observed in fish following nasal vaccination are not well understood.

As in mammals, the basis of teleost humoral adaptive immune response is the clonal expression by B cells of somatically diversified Igs, which can recognize a broad range of Ags (29, 30). Theoretically, the process of somatic recombination of variable (V), diversity (D), and joining (J) segments encoding for Ig and the combination of H and L chains has the potential to create >10¹⁴ clonotypes in humans (31, 32). However, the repertoire expressed in a given tissue at a given time is much less diverse (∼10⁷ different clonotypes). High-throughput sequencing of the adaptive immune receptor rearrangements expressed in a lymphocyte population therefore overcomes the previous technical limitations that only allowed the analysis of small fractions of B-lineage lymphocytes (33–35) and may allow a quasicomplete description of expressed Ab repertoires. Despite these recent advances in repertoire sequencing (29, 30, 36–40), teleost mucosal IgM and IgT repertoires remain poorly characterized.

The goal of the current study is to characterize the Ig repertoire of trout NALT and spleen under control conditions as well as following i.p. or intranasal prime immunization with a bacterial vaccine. We generated isotype-specific Ig repertoire libraries using a cDNA 5′-RACE–based protocol with unique molecular identifiers (UIDs). This approach has shown to be robust and useful, even when the goal is to analyze minor lymphocyte subsets present in tissues, as we can expect in NALT (41, 42). Our findings reveal conserved principles of nasal mucosal Ab responses in vertebrates and provide insights into the spatial dynamics of mucosal and systemic adaptive responses in the absence of organized SLOs such as O-NALT.

Rainbow trout homozygous Swanson clone were obtained from Dr. Thorgaard’s laboratory at the Washington State University. Fishes with a mean weight of 5 g were placed in individual aquaria that received single-pass spring water at a constant temperature of 14.5°C and a dissolved oxygen content of 9.2 ppm. Fish were fed twice daily a commercial rainbow trout diet (Clear Springs Foods).

Vaccination experiments were conducted using a vaccine model, the enteric red mouth (ERM) vaccine, a Yersinia ruckeri bacterin (43). A mock-immunized group (control) received 25 μl of saline (0.85%) intranasally (i.n.) in each nare and 50 μl of saline i.p. A second group (ERM_i.n.) received 25 μl of i.n. delivery into both nares of ERM vaccine diluted 1:10 in saline. Same vaccine and dilution were used to immunize i.p. a third group of fish (ERM_i.p.). After 1 mo, trout were sampled. They were sacrificed by overexposure to 2-phenoxyethanol diluted 1/1000. Blood was extracted and let to clot at 4°C overnight for serum extraction. The olfactory organ and spleen were removed, fixed in RNAlater (Sigma-Aldrich), and kept at −20°C. Serum extraction was performed by centrifugation at 200 × g for 10 min; supernatants were collected and centrifuged at 1000 × g for 20 min. Serum was frozen at −20°C to determine the presence of specific trout IgM against Y. ruckeri by ELISA. All procedures were approved by the Office of Animal Care Compliance at the University of New Mexico protocol number 16-200384-MC.

Specific IgM and IgT Abs against Y. ruckeri were detected by ELISA. A volume of 100 μl of ERM vaccine (5 × 10⁷ cfu/ml) in bicarbonate coating buffer (pH 9.6) was used for coating microtitre plates overnight at 4°C. Plates were washed three times with PBS, and nonspecific binding sites were blocked by incubation with 5% nonfat milk in PBS pH 7.4 with 0.05% Tween 20 (PBT) for 2 h at room temperature. Plates were washed in washing buffer (PBT + 10 mM EDTA, pH = 7.4) and serial dilution of each serum sample in PBS + 10 mM EDTA (100 μl/well) were added to each well and incubated overnight at 4°C. For IgT detection, serum samples were diluted 1:20 in PBS + 10 mM EDTA containing mouse monoclonal anti-trout IgM Ab. Samples were incubated overnight at 4°C under rotation, and then 10 μl of protein G–agarose beads (Pierce) were added per 100 μl reactions. Samples were incubated at room temperature for 1 h and then centrifuged for 5 min at 10,000 × g to separate IgM–anti-IgM–agarose beads complex from the IgM-free supernatants. Supernatants were then used in the ELISA plates. Serum dilutions >1:20 did not yield any detectable specific IgT titers. Serum samples from all groups were analyzed in duplicate wells. After washing three times, each well was incubated with mouse mAb (1:14) anti-trout IgM diluted or mouse monoclonal anti-trout IgT (2 μg/ml dilution in PBS EDTA) for 1.5 h at room temperature on an orbital shaker (50 rpm). After washing three times in PBT + EDTA, HRP-conjugated goat anti-mouse IgG diluted 1:500 (Jackson ImmunoResearch) was added to each well for 1 h at room temperature. Next, 75 μl of tetramethyl benzidine substrate reagent (BD OptEIA) were added to each well, and the reaction was stopped after 15 min by adding 50 μl of 2 M H₂SO₄. Absorbances were recorded at 450 nm using a plate reader (BioRad). Internal positive and negative control samples were included in all assays. Values were normalized to the highest OD value of each assay, which was considered 1.

Total RNA was individually prepared from spleen and nares by disruption of the tissues in TRIzol reagent (Life Technologies). Total RNA was purified and DNase treated using RNeasy Mini or Micro Kit (Qiagen). cDNA was obtained from 2 μg of total RNA by reverse transcription with template switching. We use the 5-RACE SMARTer kit (Clontech, Mountain View, CA) and specific primer encoding the Cmu2 (5′-AGAGACGGCTGCTGCAGATATTCC-3′) and Ctau1 (5′-GATGTCGTTAGAAGGGGTTCCA-3′) domains. Second-strand synthesis was done using HiFi HitStart ReadyMix (Kapa Biosystems, Wilmington, MA) and primer containing 15 random nucleotides (UID), partial Illumina adapter, and the SMARTer II A oligo from the 5RACE kit sequences (98°C for 3 min, 58°C for 3 min, and 72°C for 10 min). ds cDNA was purified using AMpure XP beads at a ratio of 1:1 and eluted in 30 μl of H₂0.

The resulting dscDNA was amplified by two consecutive PCR amplifications in a 50 μl volume with 5 PRIME HotMaster Taq DNA Polymerase (Quantabio). The first reaction was designed with primers containing both a region complementary to the C region of the Cmu1 (5′-CACATTGCGCAAGAGGGAACAA-3′) or Ctau1 (5′-GTTCCACAGTTCATAAGAGTG-3′) domains and a primer complementary to the Illumina adaptor sequence (5′-GACTGGAGTTCAGACGTGTGCTCTTCCGATCT-3′). The amplification program was as follows: 94°C for 3 min, then 94°C for 30 s, 59°C for 30 s, 72°C for 30 s (25 cycles), with a final elongation at 72°C for 10 min. PCR products were loaded onto a 1% agarose gel. After electrophoresis, the IGHV-Cmu or IGHV_Ctau product at 450–600 bp were extracted using Nucleospin Gel and PCR Clean-Up Kit (Clontech). A second PCR reaction was then used to add the remaining Illumina adaptor sequence and unique sample indexes (30). The amplification program was as follows: 94°C for 3 min, then 94°C for 30 s, 64°C for 30 s, 72°C for 30 s (10–15 cycles), with a final elongation at 72°C for 10 min. The second PCR products from the same first PCR were pooled and purified using AMpure XP beads at a ratio of 1:1 and eluted in 20 μl of H₂0. The PCR product concentration in each library was determined using a Qubit fluorimeter (Invitrogen). Sequencing was performed at Dr. D. Dinwiddie’s laboratory at the University of New Mexico Health Sciences Center on the Illumina MiSeq instrument using paired-end 2 × 300-bp runs and the MiSeq Reagent Kit v3 (600 cycles) (Illumina).

Sequences were filtered and merged using pRESTO (44) and ImMunoGeneTics (IMGT)/High V-QUEST tool was used to determine V(D)J assignments (45). In the current state of annotation of the rainbow trout Ig gene repertoire and polymorphism, our V and C annotation defines V family subsets and isotypes, respectively, rather than unique genes. Because we used the Swanson clone of rainbow trout in which every locus is in homozygous configuration, the complexity of annotation was limited to a single haplotype.

Nonfunctional sequences were removed from the data, and the PCR biases were offset by developing a unique molecular identifier (MID) based on the UID and the CDR3 nucleotide sequence determined by IMGT/HighV-QUEST, as previously described (30). The total number of MID barcoded consensus sequences is provided in Supplemental Table I. The expression level for each IGHV family was computed by counting the corresponding MID barcodes on consensus sequences.

Each clonotype was defined by IGHV family, CH isotype, the JH annotation, and a CDR3 sequence in amino acid. Their expression level was computed by counting the corresponding MID barcodes, as defined above. Species accumulation or rarefaction curves were conducted using the Vegan R package (46), with clonotypes representing species and sampling done without replacement.

Previous to analyzing the clonotype diversity and sharing, we normalized datasets by random subsampling. For each IGHV family/CH isotype combination and tissue, 10 subsamples of minimum MID were performed, and the average results were computed and analyzed. The number of MID subsampled was imposed by the smallest individual MID count per each IGHV family (considering IgM and IgT isotypes and both tissues: NALT and spleen), see Table I. The conclusions of the analyses are supported by the highly consistent results produced by distinct subsamples.

To get insight about the diversity of clonal repertoire we used indices such as clonotype richness and Shannon entropy_H (47). These parameters were calculated on normalized (subsampled) datasets. The Shannon entropy_H takes into account each clonotype (i), the proportion of MIDs for a given clonotype over all of the different clonotypes being measured (p), and the number of different clonotypes (R), as shown in equation: H = ∑^R_{i = 1} p_i Ln_i. If nearly all of the sequences in a sample are found in one clone, the Shannon entropy approaches 0. Conversely, if all of the clones are equally abundant, the Shannon entropy approaches the natural logarithm of R. It was calculated using the Vegan R package (46).

All sequencing data have been submitted to the National Center for Biotechnology Information https://www.ncbi.nlm.nih.gov/sra (SRA accession: PRJNA551127 or https://dataview.ncbi.nlm.nih.gov/object/PRJNA551127?reviewer=90oeuvdps2dlj9ibjf0c4dkvro).

Statistical analysis was performed by one-way ANOVA with post hoc Tukey honestly significant difference (HSD) test and Bonferroni using different R packages. In some comparisons Student t test was also used.

A summary of the quality-filtered reads corresponding to each tissue and individual is provided in Supplemental Table I. Sequence analysis by IMGT High/V-QUEST showed a lower level of productive IgT transcripts (with an average of 50.5% in spleen and 58% in NALT) compared with IgM transcripts (83% in spleen and 87% in NALT). The analysis of these IgT nonproductive sequences revealed that a large proportion of them lacked the IGHV region and contained the JH segment spliced directly to the constant exon with genomic sequence at the 5′JH location (data not shown). Similar sterile transcripts have also been identified from Ig L and H chain loci in other teleosts (48, 49).

We assessed the reproducibility of our sequencing approach by analyzing independent libraries prepared in parallel from aliquots of the same RNA and sequenced them in different runs. For both tissues, the rates at which sequences were shared between replicates depend on the frequency class of considered sequences; this rate ranged from 6 to 14% (all sequences being considered) to almost 100% for abundant sequences (Supplemental Fig. 1). Additional tests to compare IGHV family usage and CDR3 length profiles between replicates showed that our protocol provided a robust and highly reproducible description of expressed Ab repertoires (Supplemental Fig. 1).

We characterized the expressed IgM and IgT repertoires in NALT and spleen from six healthy juvenile fish and compared nasal and systemic clonotype populations. We analyzed key components that determine the biological diversity and function of Ab repertoires including: 1) VH usage in expressed V(D)J rearrangements, 2) clonotype diversity that correlates with the range of Ag-binding potential, 3) highly frequent clonotypes, a measure of the likelihood that B cell clones were activated after encountering the Ag, 4) clonotype tissue distribution, related to the dynamics between systemic and mucosal lymphoid “compartments,” and 5) clonotypes shared by several fish, which may reflect convergent response to Ag and/or high probability of generation by VDJ rearrangement (30).

We detected the expression of all 13 IGHV gene families both in spleen and NALT but with different frequency distribution (Fig. 1A). In spleen, the top four IGHV gene families (IGHV1, 5, 6, and 8) expressed by IgM transcripts accounted for over 65% of all MID barcoded consensus sequences (average of 6 fish), with 9, 21, 24, and 11% for IGHV1, 5, 6, and 8, respectively. The observed pattern for IgT was slightly different, with IGHV6 family being the most expressed (29%), followed by IGHV1 (15%), IGHV5 (13%), and IGHV9 (9%). In both isotypes, IGHV7 was the least abundant family, accounting for only 0.05% of all consensus sequences (Fig. 1A). IGHV gene family frequencies were consistent among individuals for μ and τ isotypes in spleen. In contrast, nasal IgM repertoire was very heterogeneous between individuals, often with a strong bias toward one or two IGHV families (Fig. 1A), and this result was reproducible in both sequencing replicates (Supplemental Fig. 1). IGHV usage in IgT NALT repertoire was more similar to that found in spleen, with limited variation between individuals.

To further compare spleen and NALT Ab repertoires, we then analyzed our datasets at the clonotype level as defined by the IGHV family, JH, and CDR3 aa sequence, and we performed rarefaction analysis. Importantly, the size of the B cell compartment in both tissues differs by several orders of magnitude (∼80,000 B cells in NALT of a large trout versus 60 million in the spleen of an adult trout), a factor that may affect the diversity of clonotype datasets. Plots showing the number of unique clonotypes according to the number of subsampled consensus sequences revealed a quick saturation both for NALT IgM and IgT and suggested that the diversity of these repertoires was comparable in control fish (Supplemental Fig. 2). Our data also supported the idea that most clonotypes present in NALT were detected. In contrast, rarefaction curves established from spleen datasets showed that the sequencing did not reach saturation for IgM, but IgT was less diverse than IgM. Overall, these results are in accordance with the size of the IgM⁺ and IgT⁺ B cell populations in trout spleen and NALT, respectively (Supplemental Fig. 2).

We then selected the top seven IGHV families with low interindividual variation from the spleen and performed a comparison of the number and abundance of distinct IgM and IgT clonotypes. To avoid bias related to sample size, we also normalized the datasets for each IGHV family by random subsampling. The number of sequences subsampled was determined considering the smallest individual MID count and the saturation curve. The (normalized) number of distinct clonotypes identified in spleen was higher than in NALT for both IgM and IgT and for all IGHV families analyzed (see Table I). However, there were no significant differences between the normalized number of IgM and IgT clonotypes in both tissues.

We next considered other metrics such as the Shannon entropy index that takes into account the clonotype size distribution in addition to species diversity (50). This analysis indicated that the Ab repertoire is less diverse in NALT than in spleen, which fits with the repertoire size. Additionally, IgM repertoire is significantly less diverse than the IgT repertoire within NALT (Fig. 1B). Overall, the same pattern was observed when each IGHV family was analyzed one by one (Supplemental Fig. 3A).

Constant Ag exposition of mucosa may lead to a high proportion of B cell expansions in these body sites. B cell clonal expansions in NALT of control fish can be evaluated from the proportion of frequent clonotypes. Our data show that 90 to 99% of IgM or IgT clonotypes from subsampled/normalized dataset were represented by <5 consensus sequences (Fig. 1C). Interestingly, this analysis revealed a trend toward a higher proportion of frequent clonotypes in NALT compared with spleen, both for IgM and IgT. Moreover, IgM clonotypes represented at least by 80 consensus sequences were found only in NALT. Cumulative frequency of the 20 most frequent clonotypes confirmed this result (Fig. 1C). The Top20-IgM clonotypes in NALT significantly covered a higher part of the repertoire than in spleen (Fig. 1D). A similar pattern was observed with the Top20-IgT clonotypes, albeit a smaller size compared with IgM (Fig. 1D). Similar findings were observed when each IGHV family was analyzed one by one (Supplemental Fig. 3B, 3C).

Combined, our analysis of trout NALT Ab repertoires in control fish revealed a particular structure of these repertoires characterized by two main features: 1) a fish-specific VH usage pattern for IgM and 2) a limited repertoire diversity due to both the small size of the B cell population and to the higher proportion of clonotypic expansions in NALT compared with spleen.

To further characterize the B cell responses induced by mucosal and systemic vaccination, we sequenced IgH repertoires from spleen and NALT samples collected 1 mo following ERM i.p. or i.n. immunization of juvenile rainbow trout. Both vaccination routes have previously been shown to be very effective at protecting trout against bacterial challenge (51). Accordingly, most immunized fish showed increased levels of serum ERM-specific IgM compared with controls (Fig. 2A), the highest titers detected after i.p. administration compared with i.n. vaccination. ERM-specific IgT titers, albeit much lower than those detected for IgM, were also detected after i.p but not i.n. ERM administration.

In spleen, the relative usage of the four dominant IgM IGHV families was different in controls and i.p. immunized fish. Specifically, vaccination induced a relative increase of IGHV4 family usage (Fig. 2B), which became part of the top four IGHV gene families along with IGHV5, 6, and 8. In contrast, the relative expression of IGHV6 and IGHV13 decreased in the spleen IgM repertoire. No differences in IgT IGHV usage were observed in the spleen. In NALT, i.p. immunization apparently modified the IGHV family usage for both IgM and IgT, reducing the observed interindividual variation of naive controls. Because of the high heterogeneity observed in control animals, statistically significant trends in IGHV usage were difficult to define (Fig. 2B).

We next investigated whether i.p. immunization modifies the clonotypic diversity expressed in spleen and NALT. Rarefaction analyses suggested a substantial decrease in diversity of the spleen IgM repertoire after i.p. ERM immunization compared with controls (Supplemental Fig. 2), whereas no clear trend was observed for IgT. Strikingly, NALT repertoire responses seemed to follow an opposite trend to those observed in the spleen with higher IgM clonotypic diversity but modest changes in IgT repertoire. Clonotype counts and Shannon entropy values from normalized subsampled datasets confirmed these conclusions for all the seven previously selected IGHV families.

In spleen, the total number of detected IgM clonotypes was significantly lower after i.p. (829.48 ± 558.67) ERM administration compared with controls (1098 ± 552.04) (Table II). When each IGHV family was considered, the number of IgM clonotypes decreased in all families, but the decrease after ERM i.p. immunization was only significant for IGHV4. A similar trend was observed for IgT clonotypes, but no difference was statistically significant (Table II). Shannon entropy values further showed the overall reduction (Fig. 3A) and the individual IGHV family reduction (Fig. 3B) of spleen IgM diversity induced by the immunization. No significant differences were detected for IgT. i.p. vaccination induced a dramatic decrease of IGHV4 IgM clonotype diversity in spleen (Fig. 3B), which was further illustrated in the clonotype size distribution analysis (Fig. 3C). For example, the frequency of splenic IGHV4 IgM clonotypes represented by more than five consensus sequences increased after immunization compared with controls (Fig. 3C).

An opposite trend (i.e., an increase in clonotypic diversity) was confirmed in NALT both by count and Shannon entropy analysis (Fig. 3A, 3B). The total number of IgM clonotypes after normalization increased after i.p. ERM vaccination (433.62 ± 298.92) compared with controls (298.50 ± 329.67) (Table II), and the increase was statistically significant for IGHV5 family. Shannon entropy values revealed a similar result. Accordingly, the fraction of the repertoire made up by the most expanded IGHV5 IgM clonotypes (those with >20 consensus sequences) decreased after i.p. ERM vaccination (Fig. 3C). No significant differences were detected for IgT.

Taken together, these results reveal that B cell responses to i.p. immunization lead to contrasted changes in spleen versus NALT IgM repertoires. Whereas the diverse IgH repertoire expressed in spleen is significantly narrowed by the expansion of large clones in response to vaccination, the small NALT repertoire becomes enriched by additional responding clonotypes that, remarkably, do not displace the existing diversity. The modifications of the IgT repertoire are, however, much less evident, possibly due to the high interindividual heterogeneity observed in controls.

In the spleen, the IGHV family usage analysis revealed a perturbation of IgT repertoire with a significant increase of IGHV2 usage after ERM i.n. vaccination compared with controls (Fig. 2). No significant differences were found for IgM, including the IGHV4 relative frequency, which was similar in controls and immunized fish. In NALT, IGHV6 and IGHV11 usage tended to increase within the IgM repertoire, whereas IGHV8 usage increased for IgT.

Rarefaction analyses suggested a similar effect of i.n. and i.p. vaccination on IgM and IgT clonotypic diversity (Supplemental Fig. 2). We observed decreased diversity in spleen, especially for IgM clonotypes, and increased diversity in NALT upon vaccination. These results were confirmed when the number and abundance of IgM and IgT clonotypes were analyzed in normalized subsampled datasets of seven IGHV families (Table II).

In spleen, the total number of IgM clonotypes was significantly lower after i.n. (611.64 ± 331.79) ERM administration compared with controls (1098 ± 552.04). When each IGHV family was considered, the number of IgM clonotypes decreased in every family. Specifically, the number of IgM IGHV1, 4, 5, 8, and 9 clonotypes was lower after i.n. administration, suggesting a more drastic effect of i.n. compared with i.p. immunization. Shannon entropy values also showed this reduction in diversity for all the IGHV families studied, especially IGHV5 (Fig. 3A, 3B). Remarkably, the overall diversity of the IgT repertoire significantly decreased in spleen after i.n. vaccination compared with controls. No differences were detected for a particular IGHV family (Fig. 3A, 3B).

In NALT, a slight increase in the number of IgM clonotypes was observed after i.n. ERM immunization (350.83 ± 319.88) compared with controls (298.50 ± 329.67), but we did not find significant differences when data were analyzed per IGHV family (Table II). A similar trend was observed for IgT, although differences were not statistically significant.

Combined, these data unveil several remarkable consequences of i.n. vaccination on trout repertoires. First, this route leads to clear modifications of the spleen repertoire, especially in the case of IgT, revealing that nasal vaccination in trout is effective at triggering systemic adaptive immune responses of a typically mucosal Ig isotype. Second, i.p. and i.n. immunizations had distinct impacts on B cell repertoires. Surprisingly, the perturbation elicited by i.n. immunization in the NALT B cell repertoire was less prominent than that elicited by the i.p. route, at least at the diversity level.

B cell responses to pathogens proceed by expansion of both shared and private clonotypes, which is expected to reduce the clonotypic diversity of the repertoire (29, 30). Large public or shared responses found in multiple fish lead to a “convergence” between individuals. To assess the importance of “highly shared clonotypes” within the response, we first selected for each tissue the clonotypes that were detected in at least four fish after immunization (highly shared); we then selected those that were overexpressed, on average, >10 times (cumulative counts) compared with controls (i.e., leading to a list of “highly shared responding [HSR] clonotypes”). To perform this analysis, we considered the normalized subsampled dataset of seven IGHV families.

First, in all IGHV families, IgM and IgT highly shared clonotypes were detected in the spleen of control individuals, especially in the case of IgM (Fig. 4). After ERM i.p. vaccination the numbers and cumulative counts of IgM clonotypes shared by all individual fish (n = 6) increased in the IGHV1, IGHV4, and IGHV5 families compared with controls. This effect was also observed for IgT clonotypes belonging to the IGHV6 family. In control NALT, highly shared clonotypes were also found, but, unlike the spleen, there were no IgM or IgT clonotypes shared by all individuals, even if a higher fraction of this repertoire was covered in our sequencing compared with the spleen (Fig. 5). After i.p. immunization, a few IgM clonotypes belonging to IGHV1, IGHV4, IGHV5, and IGHV8 families were found in NALT from all six individuals.

Interestingly, i.n. immunization had a different effect on clonotype sharing compared with i.p. vaccination. In spleen, there was no increase in the number of IgM and IgT clonotypes shared by all six fish upon i.n. vaccination, although the number and level of expression of highly shared IgM clonotypes (i.e., found in at least four fish) increased in IGHV1 and IGHV4 families (Fig. 4). In NALT, IGHV5 IgM or IgT clonotypes were shared by six individuals after i.n. ERM vaccination (Fig. 5).

To verify that these highly shared clonotypes (found in ≥4 fish) found after ERM immunization were likely Ag-responding clonotypes, we checked if they were found, on average, >10 times compared to controls. After ERM i.p. immunization, we thus identified 13 and 24 HSR IgM clonotypes in spleen and NALT, respectively (Fig. 6A). These clonotypes expressed IGHV4 and IGHV5 genes, and two of them were found in both tissues (Fig. 6A), further supporting that they were part of the ERM-specific response. We also found eight and four HSR IgT clonotypes in spleen and NALT, respectively, coming from IGHV5 and IGHV6 families (Fig. 6A).

After ERM i.n. immunization, we identified HSR IgM and IgT responding clonotypes in spleen and NALT (Fig. 6B). For IgM, we found 24 in spleen and 25 in NALT belonging to IGHV4 and IGHV5 families and five of them being detected in both tissues (Fig. 6B, IGHM heat maps and Venn diagram). Four HSR IgT clonotypes were found in spleen, whereas 14 were detected in NALT, coming from IGHV5 and IGHV6 families, and one was common to both tissues (Fig. 6B, IGHT heat maps and Venn diagram). The sets of HSR clonotypes found after i.n. and i.p. immunization were largely overlapping, further supporting their role in the Ag-specific Ab responses in trout (Fig. 6A, 6B).

Taken together, our data suggest that ERM vaccine administered either i.p. or i.n. induces IgT as well as IgM responses, identified as HSR clonotypes. Strikingly, the numbers of HSR clonotypes detected support our diversity analysis and underscore that a surprisingly strong B cell response is elicited in the trout spleen after i.n. vaccination despite the fact that circulating specific IgM levels were modest.

The mucosal immune system is associated with vast tissue barriers that are located at the interface between the host and the environment. All vertebrate mucosal barriers are colonized by microbiota and exposed to pathogens (4, 52, 53), but each mucosal barrier, depending on the anatomical location, has unique physiological adaptations. The nasal mucosa, an understudied tissue with regards to B cell repertoire and Ab responses, is specialized in detection of chemical cues in the environment. Therefore, nasal immune cells such as B cells are in close proximity to not only epithelial cells and goblet cells but also olfactory sensory neurons (54, 55). Additionally, the nasal mucosa is a highly vascularized tissue that can rapidly change vascular permeability in response to environmental, chemical or physical stimuli (56). Despite the fact that nasal vaccination is a very effective mucosal route to stimulate adaptive immune responses and protect animals against infectious diseases (57, 58), the associated adaptive immune B cell responses are not well understood.

The present study characterizes for the first time, to our knowledge, the nasal B cell repertoire in a nonmammal vertebrate species in which O-MALT structures such as tonsils and adenoids are absent (10, 26). Using NGS, we compared the IgH repertoire of two Ab classes, IgM and IgT, at mucosal and systemic sites of homozygous rainbow trout. Fish IgM is considered the systemic Ab isotype par excellence, whereas IgT is known to play specialized roles in the mucosal immune system of fish (49, 59). However, several lines of evidence support the idea that IgM contributes to some degree to mucosal immune responses in fish (20, 27, 51, 60). Similarly, several studies have highlighted the presence of systemic IgT responses following virus-primed immunization and boost (29, 30).

Our results indicate that NALT Ab repertoires in fish have a particular structure with a fish-specific VH usage pattern for IgM and limited IgM and IgT diversity. This restricted diversity was due to both the small size of the B cell population in NALT and to the higher proportion of clonotypic expansions in this tissue compared with spleen. Interestingly, NALT IgM and IgT repertoires showed characteristics reminiscent of those described for IgM and IgA mucosal repertoires in humans and mice, with a high percentage of clonotypes present at a low frequency and only a few highly expanded clonotypes (3, 61–63). In addition, comparing IgM and IgT repertoires from several individuals, we observed a lower IgT sharing (clonotypes shared by at least two or four fishes) compared with IgM. This suggests that an individual-specific nasal IgT repertoire is established in naive fish, as described for gut IgA repertoire in mice (3). Recent studies have highlighted a cross-talk between microbiota and adaptive humoral immune responses, identifying the important role of microbiota in shaping the mucosal B cell repertoire (3, 5, 64, 65). Our current characterization of the trout NALT repertoire combined with our previous findings showing that trout nasal microbiota is coated by both IgM and IgT (26) suggest that nasal IgM and IgT repertoires are both largely shaped by the unique antigenic environment found at the mucosal barriers of each individual.

Nasal vaccination with ERM bacterin, our model vaccine, is known to confer 100% protection 28 d postimmunization in rainbow trout (26, 28). Additionally, several studies have shown the correlation between serum-specific Abs elicited by immersion or injection vaccination and protection against Y. ruckeri (66). However, the immunological mechanisms that afford this specific protection have thus far not been investigated. Nasal Ab responses have only been characterized in rainbow trout surviving chronic infections to the protozoan parasite Ichthyophthirius multifiliis (27). In this model, specific Ab responses in nasal mucus were largely of IgT isotype, whereas in serum they were largely IgM. Our ELISA results indicate that nasal immunization with ERM triggers weak specific IgM systemic Ab responses compared with i.p. administration. Additionally, nasal vaccination did not trigger specific systemic IgT Ab responses 28 d after immunization. This is in line with previous studies in fish showing that mucosal vaccination routes tend to result in weaker serum Ab responses compared with injection vaccination (67). Similarly, in humans and nonhuman primates, intranasal primed immunization with live attenuated influenza virus vaccine does not elicit a reliable serum Ab response. It also fails to increase the number of abundant clonotype lineages of IgG, IgA, and IgM isotypes in the peripheral blood repertoire (68–70). Taken together, these data also indicate that serum Ab and/or peripheral blood B lymphocyte responses do not thoroughly reflect the immunogenicity of intranasal vaccines.

ERM administration route played a striking influence on both the systemic and mucosal Ab responses in trout. i.p. immunization clearly shifted IGHV family usage by IgM transcripts in spleen, with a drastic increase in the IGHV4 family. The effect was also reflected in the diversity of IgH repertoire expressed in this organ, which was narrowed by the expansion of large clonotypes expressing IGHV4, IGHV5, and IGHV1 in response to ERM i.p. immunization. Importantly, we observed very different changes in the composition of NALT IgM repertoires compared with spleen repertoires following immunization. The i.p. ERM administration led to an enrichment of NALT IgM repertoire, as clearly evidenced by the increased diversity of clonotypes belonging to IGHV5 family. The modifications of the IgT repertoire in spleen and NALT were, however, much less evident compared with IgM. These results may reflect a lower impact on IgT compared with IgM due to the systemic route of immunization.

Mucosal vaccines are known to elicit systemic Ab responses. Out of all the known mucosal routes of vaccination, nasal vaccines are perhaps the most efficient at mounting systemic-specific Ab responses (58). We found that ERM intranasal administration resulted in a significant perturbation of IgM (IgH μ) repertoire in trout spleen, despite the low level of ERM-specific IgM in serum. We also observed a significant decrease of IgM clonotype diversity in the spleen. The IgT repertoire also showed a lower diversity and higher relative IGHV2 usage compared with controls. These modifications support the notion that systemic IgT responses may be better activated through mucosal rather than systemic vaccination. However, no ERM-specific IgT was detected in serum after intranasal immunization, suggesting a lower capacity to activate plasmatic B cell differentiation or to stimulate their migration to head kidney. Future studies should confirm this hypothesis and address the functions of these systemic IgT clonotypes. Additionally, the question remains as to where such IgT responses occur.

Intranasal administration also resulted in the perturbation of local IgH repertoires, as evidenced by the increase of diversity of the NALT IgM clonotypes. Our data suggest that intranasal ERM vaccination induced a migration and/or redistribution of B cell clones toward the nasal mucosa. Indeed, a local activation and expansion of resident IgM⁺ clones alone would not explain the rise of diversity. Surprisingly, the NALT IgT repertoire suffered a less dramatic perturbation than the NALT IgM repertoire after i.n. immunization with a trend toward increased diversity. These results may suggest that local specific IgM secretion in the olfactory mucosa is more important for protection against Y. ruckeri than nasal IgT. However, future studies are needed to test this hypothesis.

Previous studies using chronic parasitic models of mucosal infection revealed a compartmentalization of Ab responses in trout. This compartmentalization meant that mucosal-specific Ig responses were exclusively IgT-mediated, whereas systemic-specific responses were mostly IgM mediated (22, 27). In this study, we report that both i.p. and i.n. protocols of ERM bacterin immunization induce both IgT and IgM expansions, indicating that both isotypes were responsive to both vaccination routes. Some of these HSR clonotypes were identified in spleen and NALT, further supporting that both immunization routes induce IgT as well as IgM responses. Importantly, because teleost fish do not undergo class-switch recombination (49, 59), further studies are needed to unveil how the participation of both B cell populations are regulated during the course of systemic and mucosal adaptive immune response and how these responses correlate with the Ab production.

A recent study in rainbow trout using Ich infection as a model to study mucosal Ab responses in NALT identified in vivo local proliferation of IgT B cells in the olfactory organ (27). Taking our findings together with those reported by Ye et al. suggest that tissue damage and/or chronic antigenic stimulation may be necessary for the stimulation of strong local IgT B cell responses in fish, which would therefore vary between infections. In this regard, i.n.-delivered Ags such as live attenuated infectious hematopoietic necrosis virus (IHNV) are no longer detectable in the olfactory organ of trout 4 d postadministration (71). Although we currently do not know what is the clearance time for the ERM bacterin in the trout olfactory organ, given that it is a killed vaccine, it likely is 4 d or less. Thus, it appears that short-term antigenic stimulation of the nasal mucosa elicited by nasal vaccination using an ERM bacterin triggers IgM and IgT responses as seen in the spleen but likely is not sufficient to cause large expansions of local clonotypes. Combined, these findings may suggest that the type of pathogen/Ag has an impact on the balance of IgT- or IgM B cell–mediated responses in trout.

To gain further insights into the specificity of the repertoire responses, we looked in detail at HSRs. Many of HSRs that are presumably ERM specific belonged to IGHV4 and IGHV5 families (see Fig. 6). Interestingly, these two IGHV families were also involved in rainbow trout humoral adaptive immune response to viral hemorrhagic septicemia (VHS) virus (29, 30). Because IGHV4 and IGHV5 comprise only one and eight segments, respectively, it does not seem that their implication in the specific Ab responses to various pathogens is due to the size of the VH family (international IMGT information system; (http://www.imgt.org). Interestingly, these observations are reminiscent of the particular convergent properties of the repertoire of VH5DJ rearrangements in trout that are likely to favor their implication in shared/public responses to diverse Ag (30).

In conclusion, the current study provides, to our knowledge, the first characterization of the nasal Ab repertoire of a nonmammalian species that lacks O-NALT. Our findings suggest that the principles that shape mucosal Ab (both IgM and IgT) repertoires in fish are likely similar to those previously reported in mammals and possibly driven by the antigenic contribution of microbiota. Nasal and injection vaccination with ERM revealed unique dynamics of IgM and IgT repertoires at systemic and mucosal sites and the remarkable ability of nasal bacterin vaccines to induce spleen Ig responses (especially for IgT). Our findings provide an important immunological basis for the effectiveness of nasal vaccination in fish and other vertebrate animals and will help the design of future nasal vaccination strategies.

We thank Manuel Mendoza for help with data analysis.

The authors have no financial conflicts of interest.