A Novel Organized Nasopharynx-Associated Lymphoid Tissue in Teleosts That Expresses Molecular Markers Characteristic of Mammalian Germinal Centers

This article is featured in Top Reads, p.

Organized lymphoid structures support the maturation of the adaptive immune response in endotherms. In mammals, secondary lymphoid organs (SLOs) such as the MALTs include organized structures such as Peyer’s patches (PPs) and tonsils. Mammalian SLOs have well-defined B and T cells zones and, upon Ag exposure, germinal center (GC) reactions form within the B cell follicles (1). SLO architecture is, however, not universal to all vertebrate groups. The best studied SLO across different vertebrate taxa is the spleen; however, comparative studies of other SLOs such as MALT across vertebrates are scant. In the spleen of Xenopus, B cell follicles are surrounded by a loose T cell zone (2). Following immunization, however, the presence of scattered, activation-induced cytidine deaminase (AID)⁺ B cells can be observed in Xenopus spleen without development of GC-like structures (2). In reptiles, distinct B and T cell zones are present in SLOs but no GC reaction has been reported (3). In birds, the spleen contains B and T cell zones, and GCs with light and dark zones form in response to Ag exposures (4). Yet, the organization of these two zones is markedly different from that of mammals, with the light zone being in the center of the avian GC (4). Thus, the architecture of B and T cell zones in SLOs and whether and where GC reactions develop within these structures appear to be highly variable among vertebrate groups. As such, further studies across different branches of the vertebrate lineage are needed to understand SLO architecture and function beyond the spleen, particularly MALT.

Ectotherms, similar to teleost fish, were long thought to lack organized mucosa-associated lymphoid structures. However, a lymphoid structure associated with the gills of salmonids and coined the interbranchial lymphoid tissue (ILT) was discovered in 2008 (5). Later, three-dimensional imaging of teleost gills revealed the presence of another organized lymphoid structure, the amphibranchial lymphoid tissue (ALT) (6). Furthermore, a lymphoid structure resembling the avian bursa was recently described near the salmon cloaca (7). These discoveries underscore that other lymphocytic aggregates in teleost mucosal barriers likely exist and that careful histomorphological examination of intact tissues in multiple orientations combined with specific Ab labeling are critical for their discovery. Despite multiple anatomical and histological descriptions of teleost lymphoid aggregates at mucosal sites, their functions remain essentially unknown. Combined, the current body of work should motivate the search for mucosal lymphoid structures and their immunological function in teleosts.

Mammalian MALTs are the inductive sites for mucosal immune responses, playing a vital role in populating effector sites with Ag-experienced B cells. For instance, Ag-activated B cells exit the PP and seed the gut lamina propria after traveling through the mesenteric lymph nodes and blood (8), whereas nasopharynx-associated lymphoid tissue (NALT) B cells seed the nasal passages (9). The distinction between inductive and effector sites in the teleost mucosal immune system has been hard to delineate, as bona fide MALT structures have not been functionally characterized (10, 11). However, teleost mucosal lymphoid aggregates may have a similar function to mammalian MALT. For instance, following immunization T cell numbers in the salmonid ILT and ALT decrease, whereas circulating ZAP70⁺ cell numbers increase, perhaps suggesting lymphocytes exit from these structures and seeding of other tissues (6). Yet, the fate of these cells following immunization is yet to be investigated in teleosts, and functional characterizations of B cells from the teleost ILT, ALT, or bursa have not been conducted.

Upon Ag exposure, unique microenvironments known as GCs emerge in the center of B cell follicles within the SLOs of endotherms. In the case of MALT, GCs are universally present due to continuous exposure to mucosal-derived Ags (i.e., microbiota) (12, 13). GCs support the diversification and maturation of the adaptive immune response, a dynamic, complex process involving multiple cell types and transcriptional programs that drive B cell differentiation, maturation, and survival (14). First, resting B cells will encounter Ag and will move to the T/B border where CD4 T helper cells will send costimulatory signals (15). Next, stimulated B cells will proliferate intensely, and a fraction of these B cells will contact a critical stromal cell type, the follicular dendritic cell (FDC), forming the first vestige of the GC (16). FDCs are important players in the GC because they retain Ag for long periods of time and are responsible for B cell survival (17–20). As GCs mature, light and dark zones can be distinguished based on the density of FDCs and B cell distribution, at least in mammals (21). The dark zone contains tight clusters of proliferative B cells and the highest expression of AID in B cells, which mutates DNA at the Ig locus, causing changes in Ag binding affinity (22, 23). Thus, AID activity in the dark zone of the GC suggests that this is the site for generation of clonal variants and somatic hypermutation (SHM) (24, 25). Segregated from these mutational processes, the light zone is the site where diverse B cell clones are selected, with high-affinity clones ultimately outcompeting low-affinity clones (21). Current models, however, suggest that this process is more complex than originally thought, because both low- and high-affinity clones can persist in mammalian GCs and may be subject to similar rates of apoptosis (26). High proliferative bursts of clones receiving T cell help may instead lead to the expansion of B cells with high-affinity receptors (1).

The affinity maturation of the adaptive immune response in teleosts has been a matter of debate for decades. Teleosts express aicda, and teleost AID is able to induce class switch recombination when expressed in mammalian B cells (27). However, the genomic organization of IgH loci, as well as the absence of true switch regions, prevents class-switch recombination in teleost fish, and AID activity in teleosts therefore is thought to only support SHM (28, 29). Expression of aicda at mucosal sites in teleosts has not been well investigated. Thus, whether mucosal lymphoid structures such as the ILT, ALT, or bursa express aicda, and whether this correlates with SHM and affinity maturation of the adaptive immune response in teleosts at these sites remains unknown, raising questions about the immunological function of these structures.

We first described teleost nasopharynx-associated lymphoid tissue (NALT) in rainbow trout in 2014 (30). We reported that teleost NALT consists of myeloid and lymphoid cells scattered throughout the olfactory epithelium that respond quickly to nasal vaccination (30). Subsequent studies by our group revealed microenvironments within the olfactory epithelium; notably, the mucosal epithelium at the tips of the olfactory lamellae in rainbow trout contains clusters of CD8α⁺ T cells, and these clusters are not present in the lateral neuroepithelium (31). Of note, all of our previous work lacked full anatomical context, as we isolated the olfactory rosette from the nasal cavity.

In this study, to our knowledge, we broaden the understanding of the teleost NALT, as we report a novel lymphoepithelium that lines the entire nasal cavity, including the presence of lymphocyte aggregates that we refer to as organized NALT (O-NALT) for the remainder of this study. We further investigate in detail the structure and function of trout O-NALT and provide experimental evidence that trout O-NALT is a site of active B cell proliferation and apoptosis following intranasal vaccination. Furthermore, we show that the trout O-NALT expresses molecular markers characteristic of mammalian GCs.

Juvenile rainbow trout (Oncorhynchus mykiss) (body weight 1–3 g) were obtained from Troutlodge (Bonney Lake, WA) and housed in the University of New Mexico aquatic animal housing facility until they reached 30 g. Animals were maintained at a 12-h dark/12-h light photoperiod and fed a commercial pellet diet (Skretting). Water temperature was maintained at 16°C, and fish were not fed the day prior to any vaccination procedures. No mortalities were recorded during any experiment. All experiments were reviewed and approved by the Institutional Animal Care and Use Committee at the University of New Mexico (protocol 19-200863-MC).

Wild-type AB adult zebrafish (Danio rerio) (1 y old) used in the current study were maintained at the Centre for Molecular Medicine Norway zebrafish facility (Oslo, Norway) in accordance with the European Union, as well as Norwegian animal care guidelines and ethical standards. Handling and euthanasia of the specimens were performed by trained facility staff in accordance with the same regulations, and in contact with the local Ethics Committee.

Rainbow trout (mean weight 30 g) were anesthetized in buffered MS-222 solution in tank water (100 mg/l; Syndel, Ferndale, WA) and administered 25 μl of live attenuated infectious hematopoietic necrosis virus (IHNV) vaccine (2 × 10⁸ PFU/ml) or vehicle (DMEM containing 10% FBS used to grow the IHNV) in each nare as previously described (32). Four weeks later, fish that had received intranasal (i.n.) IHNV vaccine either received vehicle (control as well as prime groups) or a second dose of i.n. IHNV vaccine (prime and boost group). All fish were sampled 2 wk later. For each experiment, n = 10 animals per treatment group were used. Experiments were repeated three independent times. On the day of the sampling, fish were euthanized with an overdose of buffered MS-222 (250 g/l, Syndel, catalog no. Tricaine1G), bled from the caudal vein, and each half of the snout was removed and either fixed in 4% paraformaldehyde (PFA) or flash-frozen in OCT (Sakura Finetek USA, catalog no. 4583) and stored at −80°C until use.

Serum was left to clot for 6 h on ice and then spun at 7000 × g for 15 min at 4°C and stored at −80°C until used for Ab titration.

Serum samples were serially diluted 2-fold from 1:20 to 1:640 on 24-well plates. Neutralizing Ab titers were quantified using a complement-dependent plaque neutralization assay for IHNV as previously described (33). Titers of 1:20 were recorded as 0.

Whole snouts from control rainbow trout were dissected after bleeding animals and fixed in 4% PFA for 48 h. Following decalcification in 10% EDTA for 5 d, samples were processed in a tissue processor and embedded in paraffin in two different orientations: parasagittally and transversely. Paraffin sections (5 μm) were stained with routine H&E staining and imaged in a Nikon Ti inverted microscope. Large-image acquisitions were obtained using Nikon Nis-Elements Advanced Research software image stitching. To estimate the size of the O-NALT at steady state, serial parasagittal sections were taken from an untreated animal, and the size of the O-NALT, as defined by tissues with a high density of lymphocytes opposite the tips of the olfactory lamella, was measured using Nis-Elements Advanced Research software. The depth of the O-NALT was estimated by measurement of the maximum O-NALT length in parasagittal sections.

Euthanasia of wild-type AB adult zebrafish was conducted with an overdose of buffered MS-222. The labeling of the zebrafish T/NK cells was performed as previously described (6). Briefly, euthanized specimens were fixed in a solution of formaldehyde (4%) in HEPES buffer (60 mM, pH 7.4) and cryoprotected with two incubations in sucrose (32% w/v) in distilled water. Heads were severed, embedded in Tissue-Tek OCT compound (Sakura Finetek USA, catalog no. 4583) and flash-frozen using isopentane (Sigma-Aldrich, catalog no. M32631). Samples were then processed into 30-µm cryosections with a CM1950 cryostat (Leica, Wetzlar, Germany) and Superfrost Plus slides (Thermo Fisher Scientific, catalog no. 12-550-15). Cryosections displaying the nasal cavity were blocked for 1 h with BlockAid blocking solution (Thermo Fisher Scientific, catalog no. B10710) and incubated for 1 h 40 min with a rabbit anti-ZAP70 mAb (1:300, Cell Signaling Technology, catalog no. 99F2). Sections were then rinsed and incubated 30 min in a mix of Alexa Fluor 647–conjugated goat anti-rabbit secondary Ab (1:250, Jackson ImmunoResearch, catalog no. 111-605-003), fluorescent phalloidin–tetramethylrhodamine isothiocyanate (3 U/ml, Sigma-Aldrich, catalog no. P1951), and DAPI (5 µg/ml, Thermo Fisher Scientific, catalog no. D1306). Following final rinses, slides were mounted with ProLong glass mounting medium (Thermo Fisher Scientific, catalog no. P36980) and cured at room temperature for 24 h. Three-dimensional multifields of view images were acquired using the 40-µm pinhole of a Dragonfly 500 spinning-disk confocal microscope (Andor, Belfast, U.K.) equipped with a Zyla 4.2 sCMOS camera and a ×60/1.4 oil immersion objective. Acquisitions, stitches, iterative deconvolutions, and image analysis were performed with Fusion, IMARIS and ImageJ software at the Norwegian Molecular Imaging Consortium imaging platform (University of Oslo, Oslo, Norway).

Ten-micrometer-thick, parasagittally oriented cryosections of trout olfactory organs from either control, prime, or prime and boost groups (n = 3–4) were first processed by successive immersion in 70% ethanol, ddH₂O, 75% ethanol, 95% ethanol, and 100% ethanol for 30 s each, followed by immersion in xylene for 5 min.

Using the Arcturus XT laser capture microdissection (LCM) system (Applied Biosystems), we captured either the tip regions of the olfactory lamellae or the O-NALT as previously described from the same section (31). The tips of the trout olfactory lamellae are enriched in CD8α⁺ T cells, but lymphoid aggregates consisting of B cells and CD4⁺ T cells have not been described at this location. Thus, we hypothesized that the O-NALT may be acting as a mucosal inductive site and the olfactory lamellae as an effector site and compared these two regions in the rest of this study. A total of 9–10 cryosections per animal were used to capture tips and O-NALT samples onto Arcturus CapSure macro LCM caps (Applied Biosystems, catalog no. LCM0211) and immediately processed to extract total RNA using the Arcturus PicoPure RNA isolation kit (Applied Biosystems, catalog no. KIT0204) per the manufacturer’s instructions.

RNA quality and quantity was determined by NanoDrop (Thermo Fisher Scientific), and cDNAs were synthesized using the SuperScript III first-strand synthesis system (Invitrogen, catalog no. 18080051). Two hundred fifty nanograms of total RNA was added to 1 µl of oligo(dT)₂₀, 1 µl of 10 mM 2′-deoxynucleoside 5′-triphosphate mix with a final volume of 10 µl followed by denaturation at 65°C for 5 min. To complete cDNA synthesis, the following reagents were added to the denatured RNA: 1 µl of SuperScript III reverse transcriptase (RT) enzyme, 2 µl of 0.1 M DTT, 2 µl of 10× RT buffer, 4 µl of 25 mM MgCl₂, and 1 µl of RNase OUT enzyme (40 U/µl) in a final volume of 20 µl. Samples were then incubated at 50°C for 50 min followed by 85°C for 5 min. After a cooldown on ice for 5 min, 1 μl of RNase H enzyme was added to each tube and incubated at 37°C for 20 min.

cDNA was used to measure relative expression in tips and O-NALT via RT–quantitative PCR (qPCR). At the steady state, we measured mRNA levels of aicda, cd4-2b, cd8a, ck12a, ighm, ight, tcra, and tcrb in tips and O-NALT. For vaccination trials (control, primed, and boosted animals), we measured expression levels of aicda, bcl2, bcl6, cxcr4, cxcr5.x1, and tnfa in tips and O-NALT. For all RT-qPCRs, ef1a was used as the reference gene. Gene-specific primers are reported in Table I. The RT-qPCR reaction was prepared in a 96-well plate with 3 µl of each cDNA, 2 µl of a specific forward primer, 2 µl of a reverse primer, 5 µl of nuclease-free water, and 10 µl of SsoAdvanced Universal SYBR Green supermix (Bio-Rad, catalog no. 1725270). Reactions were performed using the detection system Bio-Rad CFX96 C1000 Touch real-time PCR (Bio-Rad).

For scanning electron microscopy, the entire snout of an untreated rainbow trout (n = 1) was fixed overnight at room temperature in 2.5% (v/v) glutaraldehyde in PBS. After washing in PBS (three times, 10 min), the tissues covering one of the two olfactory cavities were removed with a scalpel whereas the other side was left intact. Samples were dehydrated in a graded series of ethanol (10–100%), then acetone. Samples were coated with gold/palladium by vaporization in a vacuum chamber, then scanned with a TESCAN VEGA3 scanning electron microscope at the University of New Mexico Earth and Planetary Sciences Department.

Five-micrometer-thick cryosections were used for all stainings. We used whole nasal cavity cryoblocks oriented in two different ways: for O-NALT stainings, blocks were cut parasagittally, whereas for the nasal cavity opening, blocks were sectioned transversely.

For immunostaining with rat anti-trout CD8α (34), mouse anti-trout IgM (35), and mouse anti-trout IgT (36), cryosections were postfixed in ice-cold acetone, followed by 4% PFA, and labeled with rat anti-trout IgG CD8α (1:80), mouse anti-trout IgM (1.4 μg/ml), and mouse anti-trout IgT (5 μg/ml) labeled with CF555 using a Mix-n-Stain labeling kit, followed by FITC donkey anti-rat IgG (1:300, Jackson ImmunoResearch, catalog no. 712-095-153) and Cy5 goat anti-mouse IgG1 (1:300, Jackson ImmunoResearch, catalog no. 115-175-166).

For immunofluorescence (IF) staining to detect CD4-2b and IgM cells, we incubated cryosections with polyclonal guinea pig anti-trout CD4-2b (1:250) (37) and mouse anti-trout IgM (1.4 μg/ml) (35) as primary Abs, followed by secondary Ab labeling with Alexa Fluor 488 goat anti-guinea pig IgG (1:300, Jackson ImmunoResearch, catalog no. 106-545-003) and Cy5 goat anti-mouse IgG1 (1:300, Jackson ImmunoResearch, catalog no. 115-175-166). Isotype controls consisted of slides incubated with rat IgG, guinea pig IgG, mouse IgG, or rabbit IgG as primary Abs diluted in PBST at the same concentration as the corresponding primary Abs. Cell nuclei were labeled with DAPI (1 μg/ml, 10 min, Thermo Fisher Scientific, catalog no. D1306).

To determine the numbers of apoptotic IgM⁺ B cells in tips and O-NALT, we performed double immunostaining with polyclonal rabbit anti–caspase-3 and mouse anti-trout IgM (35). Briefly, cryosections were postfixed in 4% PFA, blocked in T20 starting block (Thermo Fisher Scientific, catalog no. EN37543) for 13 min at room temperature, and incubated overnight at 4°C with rabbit anti–caspase-3 (1:300, Abcam, catalog no. ab4051) and mouse anti-trout IgM (1.4 μg/ml). After washing, slides were incubated with Cy3 goat anti-rabbit IgG (Jackson ImmunoResearch, 1:300, catalog no. 111-165-003) and Cy5 goat anti-mouse IgG1 (Jackson ImmunoResearch, 1:300, catalog no. 115-175-166). Cell nuclei were labeled with DAPI (1 μg/ml, Thermo Fisher Scientific, catalog no. D1306) for 10 min. After mounting, samples were observed under a Nikon Ti microscope.

For proliferation assays, 30-g rainbow trout were divided into the same three experimental groups described above (vehicle, prime, and prime and boost, n=10 each). Twenty-four hours prior to sampling, trout received 500 mg of 5-ethynyl-2’-deoxyuridine (EdU, Thermo Fisher Scientific, catalog no. C10340) by i.p. injection. Each trout nasal cavity was snap-frozen in OCT and cryosectioned to 5 μm. Prior to EdU detection, IgM⁺ B cells were labeled with mouse anti-trout IgM IgG1 (1.4 μg/ml) (35), followed by detection with Cy3 donkey anti-mouse IgG (Jackson ImmunoResearch, catalog no. 715-165-150). Following immunostaining, proliferating cells were detected using the Click-iT EdU Alexa Fluor 647 imaging kit (Thermo Fisher Scientific, catalog no. C10340) per the manufacturer’s instructions. As a negative control, cryosections from animals injected with EdU were stained as described above but omitting the Click-iT EdU Alexa Fluor 647 step.

For quantification of IF staining, positive cells were counted per ×60 field (area, 145 × 110 μm). Fields were chosen based on the anatomical and structural characteristics of the O-NALT and tips of the olfactory lamellae. The O-NALT was defined as an area of the nasal cavity lining with high density of lymphocytes located opposite the olfactory lamellae in between the basal membrane and the epithelium. Tips of lamellae were imaged at the same depth as the O-NALT to capture images from the same cryosection. Images were analyzed using Fiji with the BioFormats plugin, and positive cells were manually counted. Cells positive for IgM, IgT, CD4-2b, and CD8α were defined as a blue DAPI nucleus surrounded by a fluorescent ring membrane signal. Any cells that were only partly surrounded by fluorescent signal, or any fluorescent signal that was not surrounding a clear DAPI nucleus, were not counted. Cells positive for EdU were only considered positive when the signal was present in the cell nucleus. The percentage of proliferating IgM⁺ cells in the O-NALT and tips was calculated by counting the number of Alexa Fluor 647⁺, Cy3⁺, and double-positive cells per field. Caspase-3⁺ cells were counted as described in Kraus et al. (38). The percentage of apoptotic IgM⁺ and apoptotic IgM⁻ cells in the O-NALT and tips was calculated by counting the number of Cy5⁺Cy3⁺, Cy5⁻Cy3⁺, and Cy5⁺Cy3⁻ cells per field. The background of the captured images corresponding to the lumen of the nasal cavity was cleaned in Adobe Photoshop Elements version 14 to remove any cell debris generated from tissue sectioning. Ten images for each animal were captured, and counts for each animal in the O-NALT and tips were averaged. All counts were performed in a blinded fashion, and counts were validated by a second investigator.

All data are presented as the mean ± SEM. Normal distributions of data were checked in Prism version 9. We used paired t tests to determine differences between tips and O-NALT at the steady state, and a two-way ANOVA was used for all analyses between vehicle, prime, and prime and boost animals. For gene expression data, results were analyzed by the Pfaffl method (39) using ef1a as the reference gene. Differences were considered statistically significant when p < 0.05. All data were plotted in Prism version 9.

Our previous studies described lymphoid and myeloid cells in the olfactory rosette of teleost fish (30, 31). These studies always involved the dissection of olfactory rosettes out of the nasal cavity; however, the two trout olfactory rosettes each sit inside a chamber covered by a flap that forms the inlet and outlets for water circulation (Fig. 1A–D). Removal of the covering tissue flap shows the morphology of the olfactory rosette and the cavity, which is lined by a previously undescribed epithelium (Fig. 1B). Histological examination of intact trout nasal cavities revealed that the nasal cavity is lined by cuboidal pavement cells and mucous cells, with abundant lymphocytes embedded deeply between reticulated epithelial cells that extend all the way from the dorsal opening of the nasal cavity to the bottom (ventral) aspect of the cavity (Fig. 1E, 1F). This lymphoreticular epithelium contains a lymphoid aggregate opposite the olfactory lamellae, which we have named the O-NALT. Trout O-NALT was easily identifiable in the nasal cavity because of the bulged appearance of the lining epithelium in the O-NALT region. Based on serial sections of paraffin-embedded tissues, the O-NALT is ∼500 × 550 × 200 μm deep in control 30-g trout (Fig. 1C, 1D, 1G). Assuming an elliptical cylinder shape, this means a volume of 1.7 × 10⁸ μm³. Closer examination of the rainbow trout O-NALT indicates that this structure is rich in lymphocytes, which show a low degree of compaction as evidenced by the marked intercellular spaces (Fig. 1G). For comparison purposes, we show an image of a tip of an olfactory lamella (Fig. 1H), a region we previously described as rich in immune cells compared with the rest of the olfactory neuroepithelium in trout (31). These observations indicate that the olfactory organ of rainbow trout sits in a cavity lined by a previously unknown extensive lymphocytic network. Furthermore, these observations suggest the presence of lymphoid structures in trout mucosal tissues similar to those that support the maturation of adaptive immune responses in endotherms.

To confirm that lymphoid aggregates similar to the O-NALT are present in other teleost species, we stained adult zebrafish nasal cavities for the T cell marker ZAP70, a technique recently used to describe lymphocytic structures in the zebrafish gill cavity (6). We identified a large cluster of ZAP70⁺ cells in the nasal cavity opposite to the olfactory lamellae, in a position anatomically analogous to the trout O-NALT (Supplemental Fig. 1). Although we did not further characterize this structure in zebrafish, these findings indicate that O-NALT is likely not restricted to salmonids.

We next characterized the cellular composition of rainbow trout O-NALT at the steady state. Staining with rainbow trout–specific Abs against IgM (35), IgT (36), CD4-2b (37), and CD8α (34) revealed that rainbow trout O-NALT is mostly composed of CD4-2b⁺ T cells and IgM⁺ B cells (Fig. 2A–F). Double staining with anti–CD4-2b and anti-IgM Abs did not reveal clear B and T cell zones in trout O-NALT. Similar results were observed in O-NALT cryosections stained for IgM, IgT, and CD8α (Fig. 2C, 2D). Quantification of IF stainings from the O-NALT indicates that trout O-NALT lymphocytes are composed of ∼56% CD4-2b⁺ T cells, 24% IgM⁺ B cells, 16% CD8α⁺ T cells, and 4% IgT⁺ B cells (Fig. 2E, 2F). Interestingly, staining of the nasal cavity opening shows that the entire lymphoepithelium is rich in CD4-2b⁺ and IgM⁺ cells, whereas IgT⁺ and CD8α⁺ cells are rare outside the O-NALT. These results indicate that at a steady state the trout O-NALT constitutes a unique immune cell microenvironment.

We next investigated the immune gene expression profile of trout O-NALT at the steady state. As a comparison, we measured immune gene expression in the tips of the olfactory lamella. Both regions were dissected by LCM, and total RNA was extracted to measure gene expression. Expression of the GC marker aicda is ∼15-fold higher in the O-NALT compared with the tips of the lamellae (Fig. 3A). O-NALT also has significantly higher expression levels (2- to 30-fold) of cd4-2.b, cd8a, ighm, tcra, and tcrb compared with the tips (Fig. 3C, 3E–H, Table I). Expression levels of ight (Fig. 3D) and ck12a (Fig. 3B) were not statistically significant in the O-NALT compared with the tips, although there was a trend toward elevated ck12a expression in the O-NALT (Fig. 3B). Taken together, these results indicate that trout O-NALT has a unique immune transcriptional profile and suggest that O-NALT may be a site for SHM in teleosts.

To understand how trout O-NALT participates in an immune response, we performed a nasal prime and boost vaccination protocol (Fig. 4A) using live attenuated IHNV, a model intranasal vaccine. Our previous work indicated that i.n. IHNV vaccination confers almost 100% protection 4 wk after primary immunization in rainbow trout (40). We added a second vaccine dose at week 4, then waited a further 2 wk to measure adaptive immune responses in the O-NALT (Fig. 4A). This vaccination regimen showed significantly higher IHNV neutralizing Ab titers in serum of primary vaccinated trout and a trend toward even higher titers following booster vaccination (Fig. 4B). However, at this time point titers of boosted fish were not significantly different from primary vaccinated fish, potentially due to the short time of sampling after boosting. We next measured the numbers of CD4-2b⁺, IgM⁺, CD8α⁺, and IgT⁺ cells in the O-NALT and tips of control, primary vaccinated, and boosted rainbow trout. We found that primary vaccination increases the number of all four cell types in the O-NALT (Fig. 4C). Boosted animals have similar IgM⁺ and CD4-2b⁺ cells in the O-NALT compared with primary vaccinated fish (Fig. 4C), but the numbers of CD8α⁺ and IgT⁺ cells were lower in the O-NALT of boosted trout compared with primary vaccinated trout (Fig. 4C). In the tips, primary vaccination drives an increase in IgM⁺ and CD8α⁺ cells whereas CD4-2b⁺ and IgT⁺ cell numbers do not change compared with unvaccinated controls (Fig. 4D). Boosting did not cause further changes in the tips, which still had significantly elevated IgM⁺ cells compared with the unvaccinated controls, or changes in CD4-2b⁺ and IgT⁺ cell counts. The olfactory lamellae tips of boosted trout, however, had CD8α⁺ cell numbers lower than primary vaccinated fish, but still higher than unvaccinated controls (Fig. 4D). Double staining for CD4-2b and IgM in the O-NALT of all three treatment groups did not reveal clear B and T cell zones (Fig. 4E–G). Taken together, these results indicate that nasal vaccination triggers unique immune cell responses in the O-NALT compared with the tips and suggest that O-NALT may be involved in the maturation of the adaptive immune response within the nasal local environment.

Based on our results, we next asked whether B cell selection may occur within the O-NALT microenvironment. To do so, we quantified the number of proliferating IgM⁺ B cells in the O-NALT and tips, as well as the number of apoptotic IgM⁺ B cells in the O-NALT and tips of control, primary, and boosted trout. In the O-NALT, primary and boost vaccinations cause an increase in the number of proliferating IgM⁺ B cells (Fig. 5A–D). No increase in proliferation of IgM⁺ B cells occurs in the tips of primary or boost vaccinated animals (Fig. 5H). The number of IgM⁺CASP3⁺ cells (apoptotic IgM⁺ cells) increased significantly (∼3-fold) in the O-NALT of primary vaccinated but not boosted trout (Fig. 5E), whereas no significant changes in IgM⁺ B cell apoptosis were observed in the tips (Fig. 5I). The numbers of total CASP3⁺ cells increased only by ∼1.5-fold in the O-NALT but not the tips of both primary and boosted vaccinated trout (Supplemental Fig. 2A, 2C). Importantly, accounting for background apoptosis, the IgM⁺CASP3⁺/CASP3⁺ cell ratio showed a significant increase in the O-NALT (but not tips) of primary vaccinated but not boosted trout compared with controls (Supplemental Fig. 2B, 2D). We did not detect apoptotic CD42-b⁺ cells in any of our samples. In support of the presence of apoptotic B cells in trout O-NALT of vaccinated animals, we detected a significant downregulation of the apoptotic repressor gene bcl2 in the O-NALT of primary but not boosted trout (Fig. 5F), whereas blc2 expression does not change in the tips following intranasal vaccination (Fig. 5J). We also measured expression levels of bcl6, a transcriptional repressor required for mature B cells and a regulator of GC formation in mammals (41, 42). Interestingly, bcl6 expression in the trout O-NALT does not change in primary vaccinated animals but increases significantly in the boosted group (Fig. 5G). This response is also true in the tips, where bcl6 expression is significantly higher in boosted fish compared with control or primary vaccinated fish (Fig. 5K).

We next evaluated transcriptional changes in the O-NALT and tips of control, primary, and boost vaccinated trout, assessing expression of aicda, cxcr4, cxcr5, and tnfa. Because class switching recombination does not occur in teleosts, expression of aicda is thought to be a marker only for SHM in fish (28, 29). Expression of aicda in primary vaccinated fish is 12-fold higher than in control groups, whereas aicda expression in boosted animals is at the same level as in control animals (Fig. 6A). Interestingly, a similar expression profile is observed in the tips, albeit with a 3-fold increase in aicda expression (Fig. 6B). CXCR4 and CXCR5 regulate the organization of dark and light zones in mammalian GCs (43). We therefore measured expression of cxcr4 and cxcr5 in the O-NALT and tips of our three experimental treatments. Expression of cxcr4 is elevated in the O-NALT of both primary and boost vaccinated trout compared with controls (Fig. 6C). In the tips, cxcr4 expression is only higher in boosted animals (Fig. 6D). Cxcr5 expression patterns are complex. Cxcr5 is downregulated in the O-NALT of primary vaccinated trout, while the O-NALT of boosted animals has cxcr5 expression levels similar to those of controls (Fig. 6E). The combinations of upregulation of cxcr4 and downregulation of cxcr5 have been noted as a marker of B cell differentiation to plasmablasts in mammals (44). In the tips, cxcr5 expression does not change in response to i.n. vaccination (Fig. 6F). We finally measured tnfa expression because teleosts do not have an ortholog for TNFSF1 (lymphotoxin-α <) (45). Both TNF and LT are generally thought to be critical for SLO and PP formation, as well as GC development in mammals (46), but LT does not appear to be necessary for NALT formation (47). In this study, we did not observe significant changes in tnfa expression in the O-NALT or tips of primary vaccinated or boost vaccinated trout, at least in our intranasal prime and boost vaccine model (Fig. 6G, 6H). Combined, our results show that the transcriptional profile of the trout O-NALT is consistent with that of sites where B cell selection and maturation to plasmablasts occur in mammals.

Mucosal immunity is a universal arm of the immune system of all vertebrates, from agnathans to mammals. During vertebrate evolution, the complexity of the mucosal immune system appears to have increased, resulting in complex and organized lymphoid structures where selection of high-affinity B cell clones occurs within the GC reaction (48–51). However, MALTs have been far less studied in nonmammalian vertebrates, and their contributions to the adaptive immune response are still unclear. For instance, the fish mucosal immune system has long been thought to be devoid of any organized mucosal lymphoid structures and lack GC reactions. However, several studies including the identification of nasal lymphoid aggregates in lungfish (52), the discovery of salmonid ILT (5), and the recent discoveries of the ALT and a bursa-like organ in teleosts (6, 7) indicate that lymphocytes do aggregate at mucosal sites in ectotherms. Although histomorphological descriptions of lymphoid aggregates at teleost mucosal sites exist, the function of these structures is still poorly understood. The present work therefore fills important knowledge gaps in our basic understanding of mucosal immune systems and their evolution in the vertebrate lineage.

To our knowledge, this study reveals the presence of a novel nasal lymphoid aggregate referred to as O-NALT that was previously overlooked in rainbow trout. Although we cannot definitively conclude whether O-NALT is a SLO or a tertiary lymphoid organ, the fact that O-NALT was observed in all animals examined and that these animals were all healthy, uninfected individuals suggests that these are in fact SLOs. Additionally, trout O-NALT is located at an intraepithelial level, directly below a covering epithelium and above the basal membrane of the epithelium, an indication of an SLO. Of note, trout O-NALT is not encapsulated, as is the case in many MALT structures in mammals (53–55). Future studies will investigate the development of teleost O-NALT and how it changes with age, as the present description was limited to juvenile trout.

It is interesting that trout O-NALT shares some similarities with the salmonid ILT, in that immune cells extend forming a trailing edge along the primary filament to its very distal end (56). Similarly, we report that trout O-NALT is a cluster of immune cells that continues to cover the entire nasal cavity, forming a lining of lymphoepithelial tissue. The trout nasal lymphoepithelium is mostly composed of CD4-2b⁺ T cells and IgM⁺ B cells whereas the O-NALT also contains CD8⁺ T cells and IgT⁺ B cells. The ILT including the trailing edge, in contrast, contains mostly T cells with very few B cells (56), which may suggest that the ILT and O-NALT have separate functions in the adaptive immune response. An important caveat in our study is that we did not perform three-dimensional imagining of all immune cells in intact nasal cavities. Tridimensionally, the olfactory rosette is inserted within the nasal cavity, but the tips of the olfactory lamellae are also extensively attached to the nasal cavity depending on the depth of the sections. This means that the lymphoid epithelium found at the tips of the lamellae could well be a continuous extension of the O-NALT described in the present study. This continuation of the lymphoid tissue is again reminiscent of the salmonid ILT found in the gill, where a diffuse lymphoid tissue is observed not only at the base of the gills but also all along the trailing edge of the filaments (56). Our future work will attempt to resolve this “continuum” question using tissue clearing and high-resolution microscopy of the entire nasal cavity in three dimensions.

Compared to the tips of the olfactory rosette, trout O-NALT constitutively expresses high levels of B and T cell markers as well as aicda. This observation contrasts with previous studies in the salmonid bursa, where no aicda transcripts could be detected by in situ hybridization (7). However, RT-qPCR may be a much more sensitive detection method. In support of our findings, aicda expression is also highly expressed in mammalian MALT structures (57–59). Although AID was originally proposed to be exclusively expressed in MALT B cells, this idea was later challenged by the detection of AID expression in gut lamina propria B cells (60). In ectotherms, aicda expression has also been detected in non–B cells such as shark and teleost T cells, where SHM was reported (61–63). In this study, we did not attempt to identify what cell types are aicda expressors in the trout O-NALT, a limitation of our study. Mammalian PPs and tonsils and their GCs are unique because of the continuous exposure to mucosal microbial Ags, which establishes immune tolerance and responses to pathogens (48, 64). Thus, constitutive aicda expression in trout O-NALT suggests microbial Ag sampling and perhaps induction of tolerance in this microenvironment, a question worth investigating in future studies.

A hallmark of the mammalian GC reaction is the dynamic presence of proliferating and apoptotic B cells (65). Our findings indicate that proliferation and apoptosis of IgM B cells take place in the trout O-NALT in response to i.n. vaccination, which suggests that O-NALT is a microenvironment that may support affinity maturation. This observation requires careful validation via repertoire sequencing analysis in future studies. Additionally, recent studies in mammals have illuminated the importance of fine timescales when quantifying rates of apoptosis in GC B cells (66). Our study only sampled animals at one time point and tested a single prime and boost protocol. Thus, proliferation and apoptosis rates reported in the present study may have differed if different intervals and sampling times after boost had been evaluated.

GC B cells undergo complex transcriptional programs that dictate their differentiation and selection (67). For instance, downregulation of the apoptotic repressor bcl2 occurs in B cells that are eliminated in GCs due to low affinity (68, 69). CD4 T cells also downregulate BCL2 in mammalian GCs (70). Thus, our results may indicate that decreased bcl2 expression is the result of T and/or B cell bcl2 downregulation in the O-NALT. Furthermore, the transcriptional repressor BCL6 is also a marker of mammalian GCs (71). BCL6 prevents premature activation and differentiation of GC B cells and facilitates DNA in the process of high-affinity Ab selection (72). In mammals, BCL6 is downregulated in B cells that differentiate into plasmablasts (73, 74), and elevated BCL6 expression promotes persistence of activated B cells and GC commitment (75). We found that bcl6 expression increases in both the tips and O-NALT of boosted but not primary vaccinated trout, suggesting an accrual of Ag-experienced B cells in both the tips and O-NALT upon boosting.

A major caveat of our work is that we did not analyze B cell repertoire changes and SHM in control and immunized animals. Performing B cell repertoire sequencing analyses is necessary to ascertain whether GC-like reactions occur within teleost mucosal structures such as O-NALT. Of note, we previously conducted repertoire studies in response to nasal vaccination but only collected olfactory rosettes to evaluate BCR repertoire perturbation (76). Overall, we believe that comparisons between BCR repertoire perturbations including SHM analysis in the O-NALT and the olfactory rosette will provide exciting insights into affinity maturation of B cell responses in teleosts.

As mentioned earlier, upon i.n. vaccination, changes in gene expression in O-NALT were recorded. This study focuses on markers known to define the mammalian GC reaction such as bcl2, bcl6, cxcr5, cxcr4, tnfa, and aicda. Expression studies were done by RT-qPCR of LCM tissues and therefore do not resolve the identities of the cells that were expressing each of these particular gene markers, or whether these changes were due to changes in cell numbers. Future studies using unbiased gene expression approaches at single-cell resolution such as single-nuclei RNA sequencing will help resolve this question. Of note, we observed a striking increase in aicda expression in the O-NALT following primary vaccination, but the response was no longer detected in boosted animals. These results may indicate that, as occurs in mammals (9, 77, 78), boosting induces Ag-activated B and T cells to exit trout O-NALT and seed other body tissues. Alternatively, the differences in prime and boosted animals may be due to the time points at which we sampled animals (6 wk after primary vaccination and 2 wk after boosting). Selecting other intervals between primary and boost vaccination and sampling times after boosting may well have resulted in different aicda and expression patterns of other gene markers as well as all other readouts in this study. Finally, our prime and boost protocol consisted of the same Ag dose in both vaccinations. Simulations have predicted that a lower dose prime could increase the selection stringency in GCs due to reduced Ag availability, resulting in the selection of GC B cells with higher affinities for the target Ag (79). Thus, refining prime-boost dosages in our model as well as dosing intervals is a logical next step in our endeavor to improve fish nasal vaccines. Overall, our findings carry important implications for the design of prime and boost intranasal vaccines for aquacultured fish species.

In conclusion, our work unveils an overlooked lymphoid structure in the nasal cavity of rainbow trout composed by B and T cells. Our findings suggest that teleost O-NALT is a unique microenvironment with molecular markers that resemble mammalian GC reactions. The molecular signatures of O-NALT suggest a role in SMH, but whether this structure is a site for SHM and supports affinity maturation remains to be demonstrated in future studies. Finally, our work informs ways to improve prime and boost mucosal vaccines for farmed fish, as well as where and how mucosal immune responses following mucosal vaccination should be measured.

We thank Dr. J.O. Sunyer for sharing all Abs used in this study. We thank Dr. F. Takizawa for providing the anti-CD8α Ab. We also thank the staff at the T1 laboratory at the University of New Mexico Health Science Center for the LCM instrument. Additionally, we thank Dr. Michael Spilde at University of New Mexico Earth and Planetary Sciences for assistance with scanning electron micrograph imaging. We thank the Centre for Molecular Medicine Norway zebrafish facility (C. Esguerra and A.C. Tavara), the Oslo Norwegian Molecular Imaging Consortium imaging platform (O. Bakke, F. Skjeldal and L. Haugen) and the histology platform of the Section for Physiology and Cell Biology (T. Klungervik).

The authors have no financial conflicts of interest.