SINGLE CELL RESOLUTION OF THE ADULT ZEBRAFISH INTESTINE UNDER CONVENTIONAL CONDITIONS, AND IN RESPONSE TO AN ACUTE VIBRIO CHOLERAE INFECTION.

Vibrio cholerae is an aquatic bacterium that primarily infects the gastrointestinal tract, causing the severe and potentially deadly diarrheal disease, cholera. Despite the impact of Vibrio on global health, our understanding of host mucosal responses to the pathogen at the site of infection remains limited, highlighting a critical knowledge gap that must be addressed to develop more effective prevention and treatment strategies. Using a natural infection model, we combined physiological and single-cell transcriptomic studies to characterize adult zebrafish guts raised under conventional conditions and after a challenge with Vibrio . We discovered that Vibrio causes a mild mucosal immune response characterized by T cell activation and enhanced antigen capture in the epithelium. Additionally, we discovered that Vibrio suppresses host interferon signaling, and that ectopic activation of interferon significantly alters the course of infection. Notably, we also found that the adult zebrafish gut shares many similarities with mammalian counterparts, including the presence of previously undescribed Best4+ cells, tuft cells, and a population of basal cycling cells. These discoveries provide important insights into host-pathogen interactions and emphasize the utility of zebrafish as a natural model of Vibrio infection.


INTRODUCTION
The aquatic bacterium Vibrio cholerae (Vc) is a non-invasive, deadly human pathogen that infects millions, claiming roughly 100,000 lives each year through the diarrheal disease cholera, with disproportionately high mortality rates among children 1 . Although rehydration can prevent death, access to clean drinking water is frequently limited during cholera outbreaks, especially in regions impacted by civil conflict or natural disaster 2,3 . Antibiotics and vaccines are effective against Vc, but antibiotic-resistant strains have emerged 4 , and the efficacy of existing vaccines wanes with time 5,6 . Despite the importance of developing therapies that protect from disease, we have limited understanding of the immediate impacts of Vc on immune responses within the host gut. A paucity of information on intestinal anti-Vc responses significantly impedes our ability to develop rational, effective therapies or markers for disease severity. Gene expression data and histological examination of patient biopsies indicate a mild mucosal immune response that likely includes granulocyte engagement and T cell mobilization [7][8][9] . However, assigning host responses directly to Vc is complicated. Many patients were co-infected with intestinal parasites, treated with antibiotics, and may have had prior Vc exposures. As a result, the intestinal response to a natural, primary encounter remains almost entirely unknown.
Several experimental models have been deployed to characterize the impacts of Vc on a host gut 10 .
Drosophila melanogaster is excellent for studying Vc pathogenesis and pathogen-commensal interactions during disease [11][12][13][14] . However, flies lack immune-regulatory Intestinal Epithelial Cell (IEC) types found in humans, and the fly gut does not have associated populations of lymphoid and myeloid cells. Rabbits and mice are valuable mammalian models [15][16][17] , but require work with immunologically immature juveniles, exposure to antibiotic cocktails, or surgical manipulation to replicate the diarrheal disease observed in humans 18 .
In recent years, the zebrafish, Danio rerio, has emerged as an excellent vertebrate model of Vc pathogenesis [19][20][21] . Fish offer several advantages that other infection models lack. Wild and lab-raised fish naturally co-associate with Vibrio strains, and fish are a candidate vector for the transmission of pathogenic Vc in the wild 22,23 . Of equal significance, the simple addition of live Vc to fish water results in a host disease hallmarked by rapid diarrheal shedding of live pathogen 24 . Thus, the fish is ideal for studies that characterize host responses to Vc in a natural infection. Despite general progress with the fish-Vibrio model, we have a rudimentary understanding of the gut immune response to infection. Physiologically, the fish intestine appears quite similar to mammalian counterparts 25 . However, critical gaps in our appreciation of adult intestinal physiology impair our ability to understand how Vc impacts gut defenses.
Developmental, transcriptional and functional studies have outlined broad similarities between fish and mammalian intestines [26][27][28][29][30][31][32][33] . However, we have an incomplete picture of adult fish IEC composition and arrangement within the gut, and we lack a detailed characterization of the identity and state of gutassociated leukocytes. We believe a comprehensive characterization of cell functions encoded within the adult gut is essential for developing the zebrafish as a Vibrio infection model.
To advance our understanding of Vc pathogenesis in a natural infection, we completed structural and transcriptional characterizations of the adult fish gut in the presence or absence of pathogenic Vc. We discovered previously unknown IEC types, including lineages with human counterparts such as tuft cells, and Best4+ absorptive cells. We simultaneously generated an unbiased atlas of the transcriptional states of gut-associated leukocytes that included granulocytes, macrophages, dendritic cells, B cells, and specialist T cell subpopulations. We resolved the transcriptional response of each cell type to Vc at the level of individual cells and discovered that Vc prompts a moderate immune response in mature epithelial cells, activates protective anti-bacterial responses in gut-associated leukocytes, and suppresses expression of key interferon pathway elements throughout the gut. We consider possible relationships between infection and interferon signaling noteworthy, as interferon is the subject of natural selection in a human population from a cholera-endemic region 34 . In follow-up work, we showed that pre-treatment of infected adults with an interferon agonist reprogrammed the transcriptional response to infection and significantly modified host colonization, suppressed epithelial death, and increased diarrheal shedding of Vc by the host, suggesting a link between interferon and intestinal disease. Our work expands the value of zebrafish for studying enteric pathogens and resolves the gut response to Vc in a natural infection model. We believe a detailed resolution of the intestinal immune response to infection will be particularly valuable for the development of effective vaccines and biomarkers for disease.

The Adult Intestine Associates with Sub-and Intraepithelial Leukocytes
To determine how Vc affects intestinal function and morphology in adult fish, we first used tissue histology, immunofluorescence, and immunohistochemistry to characterize gut architecture in wildtype adults. H&E stains confirmed earlier reports that the adult intestine includes an epithelial monolayer that protrudes into the lumen 25 ( Figure 1A). Each fold contained physically distinct cell types that included mucus-filled goblet cells, columnar epithelial cells, and inter-epithelial leukocytes ( Figure 1B-C). Apical intestinal epithelial cells (IECs) were marked by membrane-enriched beta-catenin staining, while basal goblet cells displayed a nuclear enrichment of beta-catenin signal ( Figure 1D-E), possibly reflecting a role for beta-catenin in secretory cell specification as described previously for mammals [35][36][37] . We also detected basal clusters of PCNA-positive proliferating cells that parallel a recent description of cycling cells in the adult gut and are highly reminiscent of the intestinal stem cell/transit amplifying cell compartment of mammalian small intestinal crypts ( Figure 1F). Further emphasizing similarities between zebrafish and mammals, we also detected an apical accumulation of pSMAD-positive cells ( Figure 1G), indicating a basal to apical gradient of Bone Morphogenetic Protein activity in IECs.
Immunofluorescence staining of the pan-leukocyte marker L-plastin revealed a regular distribution of intraepithelial leukocytes ( Figure 1H-J), showing that, like mammals, fish IECs interdigit with specialist immune cells. We also observed occasional subepithelial aggregates of intestinal leukocytes ( Figure 1A, K-M) that included round, lymphoid-like cells, large, eosin-rich cells ( Figure 1L), and cells that expressed the antigen-presenting cell marker mpeg1.1 ( Figure 1N). On average, we detected one leukocyte aggregate per four folds throughout the adult intestine ( Figure 1O). In combination, our data highlight the adult fish gut as an elaborate structure with overt similarities to mammalian counterparts, including subepithelial aggregates of gut-associated leukocytes.

FIGURE 1: Arrangement of Epithelial Cells and Associated-Leukocytes in the Adult Zebrafish Gut.
(A-C) H&E stains of sagittal posterior intestinal sections from adult zebrafish. Scale bars are indicated in each panels. In A, a prominent subepithelial leukocyte aggregate is marked with an arrowhead. In C, tissue is also stained with alcian blue to visualize goblet cells. Representative inter-epithelial leukocytes and alcian blue-positive goblet cells are marked with arrowheads and asterisks, respectively. (D-G) Immunohistochemical images of a sagittal posterior intestine section stained without a primary antibody (D, control), for beta-catenin (E), PCNA (F), and pSMAD (G). (H-J) Immunofluorescence imaging of two posterior intestinal folds stained for DNA (H) and the pan-leukocyte marker, L-plastin (I). In the merged, false-colored image (J), DNA is labeled blue, leukocytes are labeled red, and the tight junction marker ZO-1 is labeled green. (K) H&E stain of a subepithelial leukocyte aggregate from the posterior intestine. (L) Digital magnification of the boxed region from K. Round, lymphoid-like cells are marked with asterisks and larger, eosin-rich cells are marked with arrowheads. (M) A merged, false-colored image of a subepithelial aggregate with DNA labeled blue, leukocytes labeled red, and the tight junction marker ZO-1 labeled green. (N) Immunofluorescence imaging of a leukocyte aggregate from mpeg1.1:GFP fish with DNA labeled blue, and filamentous actin labeled red. GFP-positive APCs within a lymphoid aggregate are labeled with a white arrowhead. (O) Quantification of sub-epithelial lymphoid aggregates per ten folds in serial sections prepared from five fish.

Cells.
To uncover the range of cellular functions encoded within such a complex epithelial structure, we resolved the transcriptomes of single cells purified from healthy adult fish guts. We successfully generated high-quality expression data for 18,358 cells that partitioned into 16 transcriptional states through graph-based clustering (Figure 2A). We assigned a cellular identity to each state based on expression of established cell type markers from fish and mammals ( Figure 2B, Supplementary table 1).
Among the transcriptional clusters, we identified leukocytes, differentiated stromal cells (marked by col1a1b, vim, kdrl), and nine distinct IEC types. Two IEC types that we have provisionally labeled progenitor 1 and 2 ( Figure 2A-B), were characterized by prominent expression of genes associated with proliferation (e.g. mki67, pcna, ccnd1, Figure 2C), alongside markers of Notch pathway activity (e.g. dld, ascl1a), including her15.1 ( Figure 2D), a pan-stem cell marker in zebrafish [38][39][40] . The progenitor two population also expressed the recently identified ISC marker, prmt1 ( Figure 2E), adding support to the hypothesis that prmt1 marks adult stem cells in zebrafish. Fluorescence in situ hybridization confirmed that her15.1+ cells are proliferative residents of the interfold base ( Figure 2F), with 55% of all her15.1positive cells incorporating the S-phase label EdU ( Figure 2F). We believe our transcriptional identification of cycling cells with enriched Notch pathway activity that express stem cell markers ( Figure   2A-E), alongside our immunohistochemical characterization of multiple proliferative cells at each fold base ( Figure 1E) suggests that, like mammals, the adult fish intestine contains two populations of basal, cycling, progenitor cells. However, we emphasize that additional work is needed to unambiguously determine if the cycling populations directly correspond to ISC/transit amplifying cells observed in mammalian intestinal crypts.
We also identified transcriptional profiles for secretory enteroendocrine (neurod1+, scgn+, nkx2.2a+) goblet (agr2+, CABZ0180550.1+) and tuft cells (pou2f3+, spry2+, alox5a+), as well as absorptive enterocytes (fabp2+, cd36+, chia.1+), lysosome-rich enterocytes (LREs, fabp6+, tmigd1+, slc10a2+), and recently discovered Best4+ cells (best4+, otop2+, cftr+) ( Figure 2B). We found that pou2f3+ IECs did not express the endocrine marker neurod1 and were morphologically distinct from goblet cells ( Figure 2G To fully characterize IEC cell types, we systematically analyzed gene expression profiles of all secretory and absorptive lineage for specialist subsets. In this manner, we resolved the enteroendocrine population ( Figure 2B) into eight subtypes based on their unique peptide-hormone expression profiles ( Figure 2I). We also identified two tuft cell subsets distinguished by expression of si:dkey-61f9.1, a Ctype lectin domain-containing protein that exhibits homology to the IgE receptor fcer2 ( Figure 2J). Upon examination of enterocytes, we identified four unique clusters. Cells within cluster one expressed markers of lipid metabolism, while cluter two cells were enriched for expression of the bile acid binding protein gene fabp6; cluster three cells expressed high levels of endopeptidases including meprin subunits; and cluster four cells expressed interferon-response genes, such as ifit14, isg15 and rsad2 ( Figure 2K). Our identification of interferon-sensitive enterocytes in adults matches our recent discovery of an enterocyte population in conventionally raised larvae that expressed interferon response genes 41 , and suggests enhanced interferon signaling within a dedicated enterocyte population ( Figure 2K). We also discovered transcriptionally distinct LRE and Best4+ populations ( Figure

Gut-Associated Leukocytes Mediate Critical Protective Responses
Given the importance of gut-associated leukocytes for intestinal homeostasis, and for defense against Looking at T cell gene expression profiles, we noted relatively modest enrichment for classical T cell markers such as lck, cd8, or the CD3zeta-like gene cd247l, suggesting heterogeneity within the T cell population. Upon re-clustering the T cell group, we resolved three unique transcriptional states ( Figure   3C). The most prominent cell type was conventional T cells that expressed cd4 or cd8 paralogs ( Figure   3D-G), indicating an abundance of cytolytic and helper T cells in adult fish guts. Intestinal T cells appeared to be fully mature, as we detected co-expression of cd4 or cd8 paralogs within individual cells ( Figure   3H), but never co-expression of cd4 and cd8 ( Figure 3H). The remaining two subsets were marked by expression of numerous nitr genes ( Figure 3I-K) or gata3 and il17a/f3 ( Figure 3L-M), respectively, matching the recently characterized type three and two populations of gut-associated innate lymphoid cells 42 (ILCs, Figure 3P). Functionally, ILC2s and ILC3s displayed clear distinctions, with ILC3s primarily expressing genes linked with T cell activation and type 2 antibacterial responses, whereas ILC2s expressed genes associated with adaptive immune responses (Supplementary Figure 2G).
To chart the intestinal distribution of gut-associated leukocytes, we used cell type-specific in situ hybridization probes to visualize each cell type in the adult gut ( Figure 3 O-T). Although more extensive mapping is needed, our data suggest a degree of spatial coordination among gut-associated leukocytes.   Figure 5F). However, we observed signatures of T cell activation that included enhanced expression of the il2 receptor beta ortholog il2rb, and elevated expression of ccr9a, a gene linked with maturation of mucosal T cells ( Figure 5H).
Looking at phagocytes, we detected lineage-specific responses to infection. For example, dendritic cells modified expression of genes required for peptidoglycan metabolism and cellular chemotaxis ( Figure 5F), enhanced expression of the pro-inflammatory granulins grn1 and grn2 ( Figure 5I), and increased expression of key components of the antigen processing and presentation machinery (e.g. ctsh, ctsz, mhc2a, Figure 5I). In agreement with a previous report, unchallenged macrophages expressed pro-and anti-inflammatory M1 and M2 markers at the population level 46  receptor, il10ra, and increased expression of ccl19a, a chemokine-encoding gene associated with T cell recruitment ( Figure 5J).
In sum, our data uncover a coordinated engagement of specialist immune cells by Vc that promotes inflammatory defenses within the gut, increases the capacity to capture microbial antigens, and enhances recruitment and activation of gut-associated T cells. Notably, we discovered that, like the IEC response to infection, Vc attenuated expression of multiple interferon-response genes, including cxcl11 paralogs in B cells and macrophages ( Figure 5G, J), ifi30 in T cells ( Figure 5H) and isg15 in macrophages ( Figure   5J), indicating that suppressed interferon signaling is a unifying feature of the host intestinal response to Vc.

Vibrio cholerae Suppresses Interferon Signaling in Infected Intestines.
Our single cell gene expression data indicated an intestinal immune response to Vc characterized by pro-inflammatory gene expression, and attenuated interferon signals in IECs ( Figure 4) and gutassociated leukocytes ( Figure 5). Consistent with links between infection and diminished interferon activity, we detected significantly suppressed expression of the interferon-sensitive viperin ortholog rsad2 in mature IECs, including endocrine, goblet, and enterocyte cells ( Figure 6A). We also discovered that challenges with Vc greatly diminished the extent of STAT phosphorylation in IECs ( Figure 6D-E) relative to unchallenged IECs ( Figure 6B-C), further supporting an inhibitory effect of Vc on intestinal IFN signaling.
We consider possible interactions between Vc and host interferon intriguing, as interferon is the subject of natural selection in a human population where Vc is endemic 34 . Therefore, we examined the relationship between interferon activation, gut immunity, and disease severity in infected fish. In initial experiments, we used quantitative gene expression analysis to compare the intestinal immune response of infected adults (Vc) to an experimental group that we pre-treated with the interferon pathway agonist poly I:C prior to infection (I:C/Vc, Figure 6F). We confirmed that pre-treatment with poly I:C alone was sufficient to induce an interferon response in the intestines of wildtype adult fish ( Figure 6G). In agreement with our single cell data, we found that infection with Vc alone drove an immune response characterized by induction of pro-inflammatory cytokines (e.g. ccl25b, ccl19a.1, ccl19b and ccl20a.2, Figure 6H), enhanced expression of genes associated with macrophage function (e.g. grn2, fcer1gl, mpeg1.1 and traf6, Figure 6I), and supported B cell activation marked by production of the IgM heavy chain, ighm, and the IgA-related IgZ heavy chain, igzh ( Figure 6J). Notably, we confirmed that infection with Vc greatly attenuated expression of numerous interferon response genes ( Figure 6J). Thus, the data in Figure 6B-J confirm that Vc infection establishes a host intestinal environment that includes broad suppression of interferon-response genes.
From there, we quantified the impacts of pretreatment with poly I:C on the gut response to Vc. We found that poly I:C blocked Vc-dependent induction of many pro-inflammatory cytokines (e.g. ccl20b, ccl25b and il1fma, Figure 6H), and prevented Vc-mediated expression of critical signatures of B cell activation (e.g. ighm, b2m, relb, ighz Figure 6I). Instead, exposure to poly I:C enhanced expression of multiple IFN pathway components (e.g. jak2a, jak2b, irf3 Figure 6J), and increased expression of a number of interferon-response genes (e.g. isg20, ifitm1, rsad2, socs2, socs3b, Figure 6J), indicating that poly I:C treatment tilted the gut Vc response towards an interferon signature.
To test if pre-treatment with poly I:C impacts Vc-mediated pathology, we then looked at cell death in intestines of adult fish that we pre-treated with poly I:C for five hours, followed by a sixteen-hour infection with Vc. In the absence of infection, we detected a limited number of TUNEL+ apoptotic cells in the gut ( Figure 6K). Consistent with data from Figure 4, infection disrupted epithelial integrity, and increased the amount of dying cells, including material shed into the lumen ( Figure 6L). In contrast, pre-treatment of fish with poly I:C restored epithelial integrity, and greatly diminished the appearance of TUNEL+ cells ( Figure 6M), indicating that a poly I:C treatment protects the host epithelium for Vc-dependent damage.
We then used the infection paradigm outlined in Figure 6F to test if pre-exposure to poly I:C impacts intestinal colonization by Vc. Consistent with Figure 4A-C, we found that infection alone resulted in approximately equal loads of Vc in each intestinal segment at 24h and 48h ( Figure 6N-O). In contrast, poly I:C-treated fish had a marked shift in the intestinal distribution of Vc. At both 24h and 48h time points, we observed a significant decline in the anterior Vc burden, with a marked increase in posterior accumulation of Vc ( Figure 6N-O). A recent study showed that diminished anterior colonization by commensal Aeromonas leads to increased fecal shedding of the microbe 47 . Therefore, we asked if poly I:C exposure also impacted host shedding of Vc. We were particularly interested in effects on Vc shedding, as fish are candidate vectors for Vc transmission in the wild. For this assay, we challenged poly I:C-treated fish with Vc for 16h, transferred infected fish to a fresh container daily, and measured water contamination by Vc at 24h intervals ( Figure 6F). At both 24 and 48h post infection, we found that pre-treatment with poly I:C significantly increased the shedding of Vc relative to untreated, infected controls ( Figure 6J). Thus, our data indicate that Vc suppresses interferon signaling in an infected fish, Upon examination of gene expression data in conventional adults, we discovered that the gut contains tuft and Best4+ cells that were only recently identified in humans and developing zebrafish larvae 41,53,54 .
The role of Best4+ cells in gut health is unclear. However, excess Best4 activity appears to support progression and metastasis of colorectal cancer 55 , suggesting a role for Best4+ cells in intestinal homeostasis. Mice lack Best4+ cells, preventing us from studying this cell type in a popular vertebrate model. However, with the extensive toolset available for genetic studies, we anticipate that zebrafish will emerge as a critical model to understand the impacts of Best4+ cells on health and disease. Likewise, our identification of IECs that express classical tuft cell markers establishes the adult as a powerful tool to explore the role of tuft cells in intestinal immunity, including in the context of gut-resident innate and adaptive immune cells. We also uncovered two distinct populations of cycling IECs: an undifferentiated cluster that expressed classical stem cell markers, alongside a second group enriched for Notch pathway components. We note similarities between the cycling cell populations of the zebrafish gut and the progenitor compartment of mammalian intestines formed by spatially restricted stem cells and more prevalent transit amplifying cells, and we provide supporting immunohistochemical evidence that the fish gut epithelium contains an abundance of proliferative cells at each fold base. However, we also caution that additional research is needed to clarify the exact identity of the cycling populations in the adult zebrafish gut.
We also discovered striking specializations within mature epithelial cell types, including enteroendocrine cells, enterocytes, and lysosome-rich enterocytes. Each lineage had transcriptionally distinct subsets that allow fine-tuned control of nutrient processing and interorgan communication. Alongside a transcriptional profile of conventional guts, we resolved the host response to Vc to the level of individual intestinal cell types. Vc is a deadly, non-invasive pathogen that impacts human populations globally. Fish are established models of Vc pathogenesis 24,60,61 , but we lack detailed understanding of the impacts of a natural, primary encounter with Vc in patients or biomedically relevant models. Our work supports patient data that Vc causes a mild, inflammatory response that includes T cell engagement, but also clarifies the impact of Vc on host IECs, as well as gut-associated leukocytes.
We discovered a coordinated response of IECs and their associated lymphocytes to infection that resulted in attenuated interferon signaling, and a moderate immune response that included initiation of T cell-based defenses. where optimal PCs were determined using JackStraw scores 70 and elbow plots. After using the "FindMarkers" function in Seurat to identify marker genes for each cluster, clusters were annotated according to known cell type markers in zebrafish, or orthologous markers in mammals.