Graphical abstract

   graphic file with name fx1.jpg
   [41]Open in a new tab

Highlights

     * •
       RAG1^−/− chicken spleen shows dysregulated inflammation without
       adaptive lymphocytes
     * •
       Single cell RNA sequencing of RAG1^−/− chicken spleen revealed two
       NK cell subsets
     * •
       Cross-species comparison of chicken NK cells showed divergent gene
       expression
     * •
       Chicken NK cells exhibited conserved functional features with
       humans and mice
     __________________________________________________________________

   Immunology; Evolutionary biology; Cell biology

Introduction

   Comparative immunology has revealed that immune systems exhibit
   conserved features across species, offering valuable insights into the
   evolution of immune mechanisms and bridging the translational gap
   between animal models and humans.[42]^1 Recent studies have
   demonstrated the conservation of various aspects of immunity across
   diverse species. For instance, the neutrophil maturation program in
   zebrafish shows substantial similarities to that in mammals,[43]^2
   cell-intrinsic immune responses are conserved between human and
   nonhuman primates,[44]^3 and innate immunity is conserved among
   crustacean food crop species.[45]^4 These findings underscore the
   preservation of immune functions across evolutionary distances.

   However, studies comparing avian immunology with mammalian immunology
   are limited by gaps in our understanding of the avian immune systems.
   Compared to mammals, birds have evolved unique characteristics in both
   adaptive and innate immunity.[46]^5 Interestingly, although gene
   expression in T cells is highly conserved across various species,
   including chickens, natural killer (NK) cells exhibit greater
   divergence among vertebrates.[47]^6 This disparity highlights the need
   for comparative studies on avian NK cells and their mammalian
   counterparts.

   NK cells play a crucial role in enhancing antiviral responses against
   zoonotic pathogens such as Zika virus and influenza A virus (IAV).
   Recent studies have revealed that Zika virus infection induces
   memory-like NK cells expressing CD27, termed "NK memory stem cells."
   When these cells were adoptively transferred to Zika virus-infected
   mice, they demonstrated superior antiviral efficacy compared to naive
   NK cells.[48]^7 Similarly, research on IAV has shown that NK cells from
   mice exposed to low-dose IAV infection can confer protection when
   transferred to recipients challenged with high-dose IAV.[49]^8 These
   findings underscore the significant potential of NK cells to combat
   emerging infectious diseases and suggest promising avenues for NK
   cell-based immunotherapies.

   NK cells in mammals exhibit heterogeneity, with distinct subsets
   identified in humans and mice based on surface markers and functional
   characteristics. Human NK cells are categorized into CD3^−CD56^dim
   (cytotoxic) and CD3^−CD56^bright (cytokine-producing)
   subsets,[50]^9^,[51]^10 while mouse NK cells are distinguished by CD27
   and CD11b expression.[52]^11 However, comparable NK cell subsets in
   chickens remain unidentified due to two main challenges: the lack of
   definitive NK cell markers and significant transcriptomic overlap
   between chicken NK and T cells.[53]^12^,[54]^13 These factors have
   substantially hindered the comparative studies of NK cells among
   chicken, human, and mouse, thereby limiting our understanding of the
   evolution and function of NK cells across vertebrates.

   In this study, we identified two distinct NK cell subsets in the spleen
   of recombination activating gene 1-deficient (RAG1^−/−) chickens[55]^14
   using the single cell RNA sequencing analysis without T cell
   interference. We then compared these data with those obtained from
   humans and mice. We observed substantially biased genes between the two
   chicken NK cell subsets when compared with their human and mouse
   counterparts, but found that these subsets retained conserved functions
   across the species. These results expand our understanding of conserved
   NK cell biology by incorporating chickens and filling translational
   gaps in the chicken immune system compared to humans and mice, thereby
   laying the groundwork for the use of chicken models to study immunology
   relevant to humans.

Results

Lack of adaptive lymphocytes and enhanced natural killer cell-mediated
immunity in RAG1^−/− chicken spleens

   While human and murine NK cells have well-defined subsets, similar
   subpopulations in chickens remain unidentified due to the incomplete
   differentiation from T cells. Chicken NK cells lack CD3 but express CD8
   or IL2R-α, which are also expressed by T cells, complicating their
   isolation. These challenges have hindered single-cell transcriptomic
   analysis of chicken NK cells. To overcome this, we hypothesized that
   RAG1^-/- chickens would allow the isolation of pure NK cell populations
   without T cell contamination. This approach aimed to identify chicken
   NK cell subsets and compare them with human and murine counterparts,
   potentially advancing our understanding of NK cell evolution across
   species. The RAG1^−/− chicken model, used in this study, was
   established through the precise integration of a tdTomato gene cassette
   into RAG1 gene. RAG1-disrupted progeny were identified by red
   fluorescence, and their genotypes were confirmed by PCR analysis using
   RAG1 donor-specific primers, as previously described[56]^14 ([57]Figure
   1A; [58]Table S1).

Figure 1.

   [59]Figure 1
   [60]Open in a new tab

   Single cell RNA sequencing reveals innate immune cell clusters in
   RAG1^−/− chicken spleen

   (A) The RAG1 genome-edited chicken was generated by inserting the
   tdTomato gene to disrupt the RAG1 gene. RAG1^−/− progeny was produced
   by crossbreeding previously established RAG1^+/− chickens[61]^14 and
   identified by the emission of red fluorescence. Comparison of (B)
   helper T cells (CD3^+CD4^+), cytotoxic T cells (CD3^+CD8^+), (C) αβ T
   cells (CD3^+TCRαβ^+), γδ T cells (CD3^+TCRγδ^+), and (D) B cells
   (CD45^+Bu-1^+) percentages in 1-week-old wild-type and RAG1^−/− chicken
   spleens.

   (E) Single-cell RNA sequencing (scRNA-seq) was performed on spleen
   cells isolated from 1-week-old wild-type and RAG1^−/− chickens to
   analyze individual leukocytes. A total of 7,367 wild-type and 8,853
   RAG1^−/− chicken cells were aggregated using uniform manifold
   approximation and projection (UMAP).

   (F) 25 distinct clusters (c0-c24) were identified and annotated based
   on gene markers. Wild-type (orange) and RAG1^-/- (blue) spleen cells
   were visualized, respectively.

   (G) Cluster-biased genes among the 25 clusters were displayed through
   heatmap.

   (H) Gene markers used to annotate each cluster.

   (I) Comparison of B and T lymphocyte clusters between wild-type and
   RAG1^−/− chickens, showing percentage of each cell type.

   To investigate the proportion of adaptive lymphocyte in RAG1^−/−
   chickens, flow cytometry was performed. The results revealed a
   significant reduction in the percentage of adaptive lymphocytes,
   including CD4^+ helper T cells, CD8^+ cytotoxic T lymphocytes (CTLs),
   αβ T cells, γδ T cells, and CD45^+Bu-1^+ B cells in 1- or 3-week-old
   RAG1^−/− spleens compared to wild-type spleens ([62]Figures 1B–1D;
   [63]S1A–S1C). Conversely, inflammation was observed in 3-week-old
   RAG1^−/− chicken spleens, as evidenced by increased spleen weight and
   expression of immune-related genes ([64]Figures S2A and S2B). Given the
   known high expression of IFNG and TNFA by NK cells,[65]^15 we
   hypothesized that NK cells induce innate immune responses in RAG1^−/−
   chicken spleens. Indeed, we found that NK cell-related genes were
   upregulated significantly in 3-week-old RAG1^−/− spleens ([66]Figure
   S2C). Flow cytometry further revealed a significant increase in the
   percentage of CD3^−PRF1^+ and CD3^−TNF-α^+ NK cells in RAG1^−/− chicken
   spleens ([67]Figures S2D and S2E). Collectively, these results confirm
   the absence of adaptive lymphocytes (B and T lymphocytes) alongside an
   increase in innate immune responses mediated by CD3^− NK cells in the
   spleen of RAG1^−/− chickens aged 1 or 3 weeks.

Single-cell RNA sequencing of RAG1^−/− chicken spleen reveals innate immune
cell atlas

   Because the absence of adaptive lymphocytes, particularly T
   lymphocytes, was observed to be coupled with an elevated innate immune
   response likely mediated by NK cells in the RAG1^−/− chicken model, we
   inferred that innate immune cell populations, especially NK cells,
   could be clearly identified without the confounding presence of
   adaptive lymphocytes. For this purpose, we subjected spleen cells from
   1-week-old wild-type and RAG1^−/− chickens to single-cell RNA
   sequencing. The viability of the single-cell suspensions obtained from
   both spleens was >85%. Next, 7,367 cells from wild-type and 8,853 cells
   from RAG1^−/− spleens were sequenced using a 10x Chromium system
   ([68]Figures 1E; [69]S3). Clustering analysis of 16,220 cells
   identified 25 distinct clusters (referred to as c0–c24) that were
   visualized using uniform manifold approximation and projection (UMAP).
   Both wild-type and RAG1^−/− cells were present in all clusters, with
   some clusters (c3, c10, c22, and c23) observed predominantly in
   wild-type cells, and others (c0, c12, and c18) in RAG1^−/− spleen
   ([70]Figure 1F). Gene expression analysis of each cluster revealed an
   average of 955 biased genes at the 5% significance level after
   adjusting for multiple comparisons. In particular, c7 and c19
   significantly expressed T-cell- and NK-cell-related genes (BCL11B,
   TCF7, and CD247 in c7, GNLY, RORA, and IL2RB in c19), indicating that T
   cells and NK cells are mixed in these two clusters ([71]Figure 1G;
   [72]Table S2).

   Furthermore, an in-depth characterization of the 25 clusters based on
   cell-biased gene expression patterns and previously established chicken
   cell markers was conducted, which led to the identification of six
   distinct cell types (B lymphocytes, macrophage/monocyte and
   granulocyte, dendritic cells, thrombocytes, T lymphocytes, and NK
   cells). Once again, T and NK cells were confirmed to be mixed in
   clusters c7 and c19, based on the expression of both T cell marker
   genes (CD3D, CD3E, CD4, and CD8A) and NK cell marker genes (IL2RA, BNK,
   SLAMF4, GZMA, and PRF1). Therefore, clusters c7 and c19 were
   categorized as comprising both T and NK cells. Clusters c3, c8, c10,
   c22, and c23 were annotated as B cells, another adaptive lymphocyte
   population based on B cell markers (Bu-1 and CD79B) ([73]Figures 1F and
   1H). As expected, the proportion of adaptive lymphocytes (B and T
   cells) in RAG1^−/− chicken spleens was lower than that in wild-type
   spleens, while the percentage of innate immune cells (including
   macrophages, monocytes, and granulocytes) was higher ([74]Figure 1I).

   Next, we asked whether highly recruited innate immune cells contributed
   to increased inflammation as observed in [75]Figure S2. We investigated
   the expression of inflammation-related genes in the innate immune cells
   recruited to the RAG1^−/− chicken spleen. Our analysis revealed a
   significant upregulation of genes involved in regulating inflammation
   (IRF1, IRF7), leukocyte trafficking (CCL19, CCL21), pro-inflammatory
   cytokines (IL6, IL18), and signaling pathway components (STAT1, STAT3,
   JAK2) in the RAG1^−/− chicken spleen ([76]Figure S4A; [77]Table S3).
   These genes were mostly expressed by macrophage/monocyte and
   granulocyte clusters (inflammation: IL6, NOS2; immune regulation: IRF1,
   STAT1, STAT3, JAK2; chemokine: CCL19, CCL21) as well as T and NK cell
   clusters (inflammation: IL18, GNLY, GZMA; immune regulation: IRF1,
   STAT1, IL2RA, LYN; chemokine: CCL4) ([78]Figure S4B; [79]Table S3).
   Given that RAG1^−/− spleens lack T cells, the clusters identified as T
   and NK cells in RAG1^−/− spleens likely represent NK cell populations
   exclusively. These results suggest that highly recruited innate immune
   cells, including macrophage/monocyte and granulocyte and NK cells,
   contributed to dysregulated inflammation in the RAG1^−/− spleen.

Single-cell transcriptomics reveals two chicken natural killer cell subsets
in RAG1^−/− spleen

   Previous scRNA-seq studies on chicken peripheral lymphoid organs could
   not definitively identify NK cell populations because of the
   co-clustering of NK cells with T lymphocytes.[80]^16^,[81]^17
   Therefore, to identify chicken NK cells, we refined our analysis of
   clusters c7 and c19 by subdividing them into six subpopulations
   (subclusters 1–6; [82]Figure 2A). Subclusters 1, 3, 4, and 5 expressed
   T cell marker genes and were annotated as different T cell subsets
   based on the expression of CD4 and CD8A (subcluster 1: double-negative
   (DN) T cells; subcluster 3: double-positive (DP) T cells; and
   subclusters 4 and 5: cytotoxic T lymphocytes (CTL-1 and CTL-2)
   ([83]Figures 2B; [84]S5). GO term enrichment analysis revealed
   significant expression of genes associated with T cell activation and
   development in subclusters 1, 3, 4, and 5 ([85]Table S4), thereby
   delineating subclusters 1, 3, 4, and 5 as distinct T cell populations.

Figure 2.

   [86]Figure 2
   [87]Open in a new tab

   Distinct chicken NK cell subsets are identified in RAG1^−/− spleen

   (A) Clusters c7 and c19 were divided into six subclusters (1–6):
   subcluster 1 (brown), subcluster 2 (red), subcluster 3 (gray),
   subcluster 4 (black), subcluster 5 (yellow), and subcluster 6 (purple).
   Based on T or NK cell marker gene expression, subclusters 1–6 were
   annotated as follows: subcluster 1 = double-negative (DN) T cells;
   subclusters 2 and 6 = NK-1 and NK-2, respectively; subcluster 3 =
   double-positive (DP) T cells; and subclusters 4 and 5 = cytotoxic T
   cells (CTL)-1 and CTL-2, respectively.

   (B) Expression of representative chicken NK cell-related genes (IL2RA,
   BNK, SLAMF4, IL2RB, CHIR-AB1, PRF1, and CD8A) and genes required for
   the commitment to NK cell lineages in mice (ID2 and ZBTB16) are shown
   in a violin plot.

   (C) The percentage of each T cell subpopulation, NK-1 and NK-2, in the
   wild-type and RAG1^−/− chicken models was compared.

   (D) Heatmap showing the biased genes in NK-1 or NK-2 versus T cell
   populations.

   (E) The significance level (-log[10] (p-value)) of the selected GO
   terms and KEGG pathways for NK-1 and NK-2 compared with those of the T
   cell populations. The dash line indicates the cutoff for significance
   (-log[10](0.05)).

   (F) Flow cytometric analysis of PRF1-expressing NK-1 and NK-2 cells.

   Although subcluster 2 expressed CD8A but not CD4 (similar to CTLs), it
   showed marked expression of NK cell marker genes (BNK, SLAMF4, and
   IL2RB) and PRF1, but lacked IL2RA, suggesting cytotoxic characteristics
   distinct from those of CTLs. Subcluster 6 lacked T cell-related genes
   (TRAT1, CD4, and CD8A) but expressed IL2RA and the potential chicken NK
   cell receptor CHIR-AB1. Additionally, subclusters 2 and 6 showed marked
   downregulation of T cell development-related genes (GRAP2 and LEF1),
   whereas subcluster 6 significantly upregulated NK cell
   development-related genes (ID2 and ZBTB16)[88]^18^,[89]^19 ([90]Figures
   2B; [91]S5; [92]Table S5). Notably, T cell subsets (subcluster 1, 3, 4,
   and 5) were dominant in wild-type spleens but were almost absent in
   RAG1^−/− spleens. Conversely, subclusters 2 and 6 were dominant in
   RAG1^−/− spleens, and were annotated as NK-1 and NK-2, respectively
   ([93]Figures 2A and 2C).

   Analysis of gene expression revealed that NK-2 cells express several
   innate lymphoid cell (ILC)-associated genes, including IL22, IL17A, and
   IL7R.[94]^20^,[95]^21 To explore the potential overlap between ILCs and
   chicken NK cell subsets, we examined the expression of both ILC and NK
   cell marker genes in NK-1 and NK-2, based on an extensive literature
   review ([96]Figure S6). Both subsets showed significant upregulation of
   GATA3, a transcription factor crucial for ILC2 differentiation,[97]^22
   and NK-2 uniquely expressed RORA and RORC, which are important for
   ILC1, ILC3 and lymphoid tissue inducer cell (LTi)
   development.[98]^22^,[99]^23^,[100]^24 However, the expression of core
   functional genes for each ILC subtype was largely absent in both
   subsets ([101]Figure S6). ILC1 markers such as CD49A, CD200R1A, and
   ITGB3[102]^25^,[103]^26^,[104]^27 were not significantly expressed, nor
   were key ILC2 genes such as BCL11B, IL4, IL9, or
   AREG.[105]^28^,[106]^29^,[107]^30^,[108]^31 Although NK-2 expressed
   RORA and RORC, both subsets lacked AHR1A (another essential
   transcription factor for ILC3)[109]^32 and other critical ILC3 and LTi
   genes.[110]^33^,[111]^34^,[112]^35 Conversely, both NK-1 and NK-2
   robustly expressed NK cell core genes, with the exception that NK-2
   lacked EOMES, and cytotoxic genes (PRF1, GZMA). Additionally, NK-1
   significantly upregulated BNK, a C-type lectin gene most homologous to
   human NKRP-1 ([113]Figure
   S6).[114]^36^,[115]^37^,[116]^38^,[117]^39^,[118]^40^,[119]^41^,[120]^4
   2^,[121]^43^,[122]^44 These findings indicate that NK-1 likely
   represents mature NK cells, whereas NK-2 may be an immature NK cell
   population with partial ILC3-like characteristics due to RORC
   expression, corresponding to previous report.[123]^45

   We also evaluated classical mammalian NK cell markers in chicken NK-1
   and NK-2 subsets. While orthologs of canonical markers such as NCR1,
   FCGR3A, KLRF1, NKG7, and KLRB1 remain unclear in the chicken
   genome,[124]^46^,[125]^47 we assessed NCAM1/CD56 (human NK cell
   marker)[126]^48 and ITGA2/CD49b (murine NK cell
   marker).[127]^49^,[128]^50 Neither NK-1 nor NK-2 significantly
   expressed these genes, highlighting evolutionary divergence in chicken
   NK cell markers ([129]Figure S7). Instead, chicken NK-1 uniquely
   expressed CD8A and BNK but lacked IL2RA, whereas NK-2 specifically
   upregulated IL2RA ([130]Table S5). These findings identify novel
   markers that distinguish chicken NK subsets, underscoring
   species-specific distinctions from human/murine NK cells.

Functional characterization of chicken natural killer cell subsets: cytotoxic
natural killer -1 cells and immunoregulatory natural killer -2 cells

   To characterize the newly identified NK-1 and NK-2 subsets, we compared
   their gene expression profiles with those of T cells. The NK-1 biased
   genes include those associated with cytotoxicity (PRF1, GZMA, and
   FCER1G), whereas those of NK-2, included genes related to
   cytokines/receptors and immune regulation (IL22, IL17A, and SOCS1)
   ([131]Figures 2D; [132]Table S5). NK-1 is enriched in active immune
   responses, such as “Defense response,” and “Inflammatory response,”
   while NK-2 is involved in the regulation of immune responses through
   cytokine production, including “Regulation of immune system process,”
   and “Cytokine-cytokine receptor interaction” ([133]Figures 2E;
   [134]Table S4). We further compared the proportion of cells expressing
   PRF1, a representative protein involved in NK cell-mediated
   cytotoxicity,[135]^51 between NK-1 and NK-2 subsets. The results showed
   that NK-1 cells expressed significantly more PRF1 than NK-2 cells
   ([136]Figure 2F), indicating that NK-1 is more actively involved in
   cytotoxicity than NK-2. These results suggest that NK-1 cells exert
   greater cytotoxic effects than NK-2 cells and that NK-2 is the major
   cytokine producers.

   When comparing NK-1 and NK-2 cells directly, we detected 389
   significantly upregulated genes in NK-1 cells and 554 in NK-2 cells
   ([137]Figure 3A). The NK-1 biased genes included cytolytic genes (GZMA,
   PRF1, and GNLY), whereas NK-2 included genes involved in immune
   regulation through cytokines (IL2RA, IL22, and IL1R1) ([138]Figure 3B).
   The identified genes were categorized based on their core functions in
   each NK cell subset. NK-1 showed significant upregulation of
   cytotoxicity-associated genes (GZMA, PRF1, and GNLY), actin remodeling
   genes (ITGAD, PXN, and VCL), and phosphatidylinositol signaling pathway
   genes (SYK, PIK3CD, and PIK3C2G) ([139]Figures 3C–3E). This contrasted
   with the notable upregulation of genes related to
   cytokine/chemokine-mediated immune regulation (IL7R, IL2RA, and IL18R1)
   and TGF-β/SMAD signaling pathway genes (SMAD1, SMAD7, and TGFBR3) in
   NK-2 ([140]Figures 3F and 3G). The function of NK-1 was enriched in
   actin filament-based processes, immune activation, and the
   phosphatidylinositol signaling pathway, supporting its role in NK cell
   activation and immunological synapse formation. Conversely, NK-2 was
   enriched in cytokine production and regulation of immune responses such
   as the TGF-β receptor signaling pathway, suggesting a role in immune
   modulation ([141]Figures 3H; [142]Table S4). These findings indicate
   that chicken NK cells can be divided into two subsets, mirroring the
   dichotomy observed between human and mouse NK cells. One subset
   exhibited high cytotoxicity, whereas the other functioned primarily as
   a cytokine producer, which is consistent with previous observations in
   humans and mice.[143]^9^,[144]^10^,[145]^11

Figure 3.

   [146]Figure 3
   [147]Open in a new tab

   NK cell functional heterogeneity: cytolytic NK-1 cells and
   immunoregulatory NK-2 cells

   (A) Volcano plot showing representative genes (highlighted in red)
   associated with each function of NK-1 and NK-2 cells.

   (B) The heatmap showing NK-1 and NK-2 biased genes. The identified
   genes were categorized according to their signature functions, and are
   shown in the heatmap, violin plot, and UMAP plots: genes related to (C)
   NK cytolytic function, (D) actin cytoskeleton remodeling for
   immunological synapse formation, and (E) the phosphatidylinositol
   signaling pathway for NK cell activation are upregulated in NK-1, while
   genes related to (F) immune regulation and (G) the TGF-β/SMAD signaling
   pathway are upregulated in NK-2.

   (H) Level of significance (-log[10] (p-value)) of selected GO terms and
   KEGG pathways for NK-1 and NK-2. The dash line indicates the cutoff for
   significance (-log[10](0.05)).

   Recent study have identified two distinct NK cell developmental
   pathways: one originating from early NK progenitors (ENKPs) and another
   from ILC progenitors (ILCPs).[148]^52 To determine the origin of
   chicken NK-1 and NK-2 subsets, we analyzed ENKP- and ILCP-biased
   signature genes. NK-1 exhibited significant enrichment of
   ENKP-associated genes, while NK-2 showed the preferential expression of
   ILCP-associated genes, despite opposing expression patterns in CCL5 and
   XCL1 ([149]Figure S8). Notably, ZBTB32 (encoding PLZF, a canonical ILCP
   lineage marker) and TCF7 (a transcription factor specific to
   ILCP-derived NK cells) were significantly upregulated in NK-2
   ([150]Figure S8). These findings suggest that NK-1 likely arises from
   the ENKP pathway, whereas NK-2 originates via the ILCP trajectory,
   highlighting evolutionary conservation of dual NK developmental
   pathways in chickens.

The majority of genes exhibit highly specific expression patterns in chicken,
human, and mouse natural killer cell subsets

   Most of our knowledge of NK cell heterogeneity is limited to mammalian
   species, such as humans and mice.[151]^53 However, we found that two
   subpopulations of NK cells exist in chickens, similar to what has been
   observed in humans and mice. Under these circumstances, we hypothesized
   that integrating and comparing our new findings with previously
   published scRNA-seq data from humans and mice, in which NK cells are
   well characterized, could further expand our understanding of NK cells.
   Based on this hypothesis, we performed a cross-species integrated
   analysis of scRNA-seq data from all three species to identify genes
   that are differently expressed between the two NK cell subgroups (NK-1
   and NK-2 cells). We found that 5,921 and 3,025 genes were
   differentially expressed in human and chicken NK-1 and NK-2 cells,
   respectively ([152]Table S6). The total number of orthologous genes
   assessed in the integrated interspecies analysis across the three
   species was 7,110, suggesting that most genes showed different
   expression patterns between the two species, with these differences
   being more pronounced in NK-1 cells ([153]Figure 4A; [154]Table S6).
   Compared to humans, the genes that were significantly upregulated in
   chicken NK-1 cells included EOMES, HIC2, and HIP1. Conversely, genes
   such as CD28, IL2RA, and RORA, which were exclusively upregulated in
   chicken NK-2 cells, were newly identified.

Figure 4.

   [155]Figure 4
   [156]Open in a new tab

   Cross-species comparison of the transcriptome of NK cell subsets in
   chickens, humans, and mice reveals species-specific gene expression

   The gray dotted line represents where the log[2] fold-change value is
   0. The color of the dots is red when the difference in log2
   fold-changes between the two species is large, and purple when the
   difference is small. The top 5% genes with the largest log[2]
   fold-change value between the two species were labeled with the gene
   symbols. Comparison of log[2] fold-changes (NK-1 – NK-2 cells) between
   (A) human and chicken species.

   (B) mouse and chicken species.

   Similar results were observed after a cross-species integrated analysis
   between chicken and mouse cells, which again examined the differences
   in expression patterns between the two types of NK cell subsets
   ([157]Figures 4B; [158]Table S6). We found that 6,118 and 4,290 genes
   were differentially expressed in the NK-1 and NK-2 cells of chickens
   and mice, respectively. This suggests that, as with the differences
   observed between chicken and human NK cells, most genes have
   significantly different expression patterns depending on the species.
   Compared to mice, genes specifically upregulated in chicken NK-1 cells
   include SH2D1A and INPP4B. Similarly, genes specifically upregulated in
   chicken NK-2 cells included PLCG1 and IL2RA when compared with mice.
   Although these genes exhibit distinct expression patterns in chicken NK
   cell subsets compared with their human and murine counterparts, they
   play crucial roles in various aspects of NK cell function.
   Collectively, these results indicated that chicken NK cells have
   developed unique molecular mechanisms to perform functions analogous to
   those of their human and murine counterparts.

   Because the three species are very different and the technical pipeline
   for sequencing NK cells is not uniform across all three species, it is
   expected that the majority of genes are upregulated exclusively in NK-1
   and/or NK-2 of a specific species. Notably, while the majority of genes
   were expressed specifically in certain species of NK-1 and/or NK-2
   cells, we found several genes, such as GZMA and PRF1, that showed a
   common directional expression pattern across the three species.

The transcriptomic signatures of natural killer cell subsets were conserved
across species

   Although transcriptomic differences between NK-1 and NK-2 subsets in
   chickens show extensive divergence when compared with those in humans
   and mice, we found that the NK cell subset-biased genes across species
   reflect conserved functional features. NK-1-biased genes shared among
   all three species were associated with cytotoxicity (PRF1, GZMA, IFNG,
   CARD11) ([159]Figure 5A), actin cytoskeleton remodeling for
   immunological synapse formation (PXN, PLEK, ACTR2, ADD3) ([160]Figure
   5B), and phosphatidylinositol signaling for NK cell activation (PRKCB,
   PIP4K2A, LYN) ([161]Figure 5C). Transcription factors upregulated in
   NK-1 cells across species (TBX21, RUNX1) are involved in mature NK cell
   development ([162]Figure 5D). Some NK-1 biased genes were conserved
   only between chickens and humans (cytotoxicity: SYTL1; actin
   remodeling: VAV3, VCL) or chickens and mice (cytotoxicity: FCER1G;
   actin remodeling: ITGA4, ITGB1; phosphatidylinositol signaling pathway:
   ITPR3, SYK; transcription factor: EOMES). These genes exhibited
   signature NK-1 functions such as "Regulation of actin cytoskeleton" and
   "Natural killer cell mediated cytotoxicity" in both chicken-human and
   chicken-mouse comparisons, reflecting conserved functional features
   despite extensive transcriptomic divergence between NK subsets across
   species ([163]Figures S9A, S9B; [164]Table S7).

Figure 5.

   [165]Figure 5
   [166]Open in a new tab

   NK cell function is conserved within each NK cell subset across
   chickens, humans, and mice

   (A) Expression patterns of a total of six cytotoxicity-related genes,
   including GZMA, in NK-1 and NK-2 cells, respectively. All genes are
   up-regulated in NK-1 cells across three species except for the SYTL1 in
   mice.

   (B) A total of nine actin cytoskeleton remodeling genes exhibit an NK-1
   biased pattern in three species commonly, except for the ITGB1 and
   ITGA4 genes in humans and for the VAV3 gene in mice.

   (C) A total of six genes involved in the phosphatidylinositol signaling
   pathway show an NK-1 biased pattern, except for the SYK in humans.

   (D) Transcription factors specifically upregulated in NK-1 cells across
   three species, except for the EOMES in human.

   (E) A total of 13 regulation of immune response genes, including CD28,
   are commonly upregulated in NK-2 cells across the three species, except
   for RORA, ZAP70, and MAP3K5 genes in humans.

   (F) Transcription factors such as TCF7 and TOX are commonly upregulated
   in NK-2 cells of three species.

   (G) Oxidative phosphorylation-related genes are upregulated in NK-2
   cells of three species except for ATP6, COX1, and COX2 in humans.

   NK-2 biased genes shared across chickens, humans, and mice showed the
   upregulation of immune regulation genes (IL7R, CXCR4, GPR183)
   ([167]Figure 5E) and transcription factors for early NK cell maturation
   (TOX, TCF7, NFKB1) ([168]Figure 5F). Oxidative phosphorylation genes
   (ND5, COX3) exhibited conserved expression patterns, with higher
   conservation between chickens and mice ([169]Figure 5G). Several NK-2
   biased genes conserved only in chickens and humans (regulation of
   immune response: IL18R1, TNFAIP3, CD81), or chickens and mice
   (regulation of immune response: CD28, RORA, NFKBIZ; oxidative
   phosphorylation: ATP6, ND3, COX1) were also identified. BATF was a
   conserved transcription factor in chicken and mouse NK-2 cells. These
   genes were enriched for functions such as "regulation of signaling,"
   "negative regulation of lymphocyte activation," and "oxidative
   phosphorylation" ([170]Figures S9C, S9D; [171]Table S7). Collectively,
   the patterns of NK cell subset-biased genes that are conserved across
   species suggest that the immunological functions of NK cell subsets are
   highly conserved among chickens, humans, and mice.

Discussion

   Chickens are invaluable models for the investigation of various immune
   functions, vaccine production, and influenza A virus. Chicken NK cells
   play a crucial role in innate immunity against various pathogens and
   tumor cells[172]^13^,[173]^54^,[174]^55; however, the function of
   multiple chicken NK cell subsets remains poorly understood owing to the
   lack of well-defined marker genes. Additionally, chicken NK cell gene
   expression profiles show substantial overlap with T lymphocytes, making
   it difficult to distinguish heterogeneous NK cell subsets at the
   single-cell level. This overlap, combined with the lack of highly
   specific surface markers, limits the ability to resolve chicken NK cell
   diversity.[175]^12^,[176]^13^,[177]^16^,[178]^17^,[179]^56^,[180]^57 To
   address this issue, we utilized a RAG1^−/− chicken model[181]^14 and
   confirmed the absence of B and T lymphocytes in the spleen. In this
   model, we observed extensive recruitment of NK cells to the spleen.
   These NK cells exhibited the upregulation of genes associated with
   cytolysis, inflammation regulation, inflammatory cytokines, and NK cell
   activation signaling. These changes induce hyperinflammation in the
   spleen in the absence of immune regulation by adaptive
   lymphocytes.[182]^58 We hypothesized that the RAG1^−/− chicken model
   would allow for the clear isolation of largely recruited chicken NK
   cell subsets without T cell interference. Indeed, through scRNA-seq
   analysis of RAG1^−/− chicken spleens, we successfully identified two
   distinct chicken NK cell subsets (NK-1 and NK-2).

   The NK-2 subset lacked EOMES (a critical NK cell commitment
   factor[183]^40) but exhibited significant expression of ILC-associated
   genes (IL22, IL17A, IL7R, RORC[184]^20^,[185]^21^,[186]^22^,[187]^23),
   raising questions about potential ILC contamination. However, these
   genes also define immature NK cells in humans and mice, where IL22,
   IL17A, and IL7R regulate survival and
   function.[188]^59^,[189]^60^,[190]^61 Transcriptomically, NK-2 mirrored
   human stage 4a immature NK cells, which express RORγt (encoded by RORC)
   and produce IL-22.[191]^45 Both NK-1 and NK-2 retained NK cell-core
   gene expression while lacking most ILC-specific markers. Notably, NK-2
   upregulated ILCP-derived NK cell signatures (ZBTB32, encoding PLZF, and
   TCF7), whereas NK-1 enriched ENKP-derived genes.[192]^52 These findings
   position NK-1 as ENKP-derived mature NK cells and NK-2 as
   ILCP-originating immature NK cells with transient ILC-like features.

   When comparing chicken NK-1 with NK-2, we found that NK-1 cells
   specifically expressed cytotoxic genes, including PRF1 and GZMA, as
   well as genes associated with cytoskeleton remodeling in cytotoxic NK
   cells (ITGB1, ITGAD, PXN, and VCL).[193]^62^,[194]^63^,[195]^64 The
   phosphatidylinositol signaling pathway is markedly activated in NK-1
   cells (upregulation of SYK, PIK3CD, PIK3C2G, and AKT3), leading to NK
   cell degranulation.[196]^65 In contrast, NK-2 cells do not express
   genes encoding lytic granule components. Rather, they express genes
   encoding various cytokines, chemokines, and their associated receptors,
   all of which are involved in the regulation of immune responses.
   Likewise, IL-22 is expressed by murine NK cells in the absence of T
   cells,[197]^66 and IL-7Rα is expressed specifically by human
   immunoregulatory NK cells.[198]^67 Additionally, the TGF-β/SMAD
   signaling pathway, which plays an essential role in immune homeostasis,
   is also activated in chicken NK-2 cells, implying that NK-2 is an
   immunoregulatory subpopulation.[199]^68^,[200]^69^,[201]^70

   Comparative transcriptomic analysis of human and mouse NK cells has
   previously demonstrated the functional conservation of NK cell subsets
   between these species.[202]^53 Our study expands this comparative
   analysis by including avian species. Comparative transcriptomics of
   human, mouse, and chicken NK cells revealed significant evolutionary
   divergence in gene expression. However, these genes exclusively
   upregulated in chicken NK-1 or NK-2 cells were associated with various
   NK cell functions, and key transcription factors required for NK cell
   maturation showed distinct expression patterns across species. TBX21,
   essential for terminal NK cell maturation,[203]^71 was consistently
   upregulated in NK-1 cells across species, while EOMES, required for
   early NK cell maturation,[204]^72 was upregulated in chicken and mouse
   NK-1 cells but downregulated in human NK-1 cells. Additionally, INPP4B,
   a regulator of NK cell tissue residency in mice with low expression in
   blood and spleen, was specifically upregulated in chicken splenic NK-1
   cells, contrasting its expression pattern in murine NK cells.[205]^73
   Similarly, ITGA1, enriched in tissue-resident human NK cells,[206]^74
   was exclusively upregulated in chicken NK-1. These results indicate
   chicken NK-1 differentially rely on transcription factors for their
   maturation and tissue-resident function, highlighting evolutionary
   divergence in NK cell regulatory pathways and enhanced tissue-residency
   features in chickens.

   IL2RA, which is exclusively expressed in chicken NK-2 cells, has been
   previously proposed as a marker of chicken intestinal NK cells. The
   28-4 antibody, which recognizes chicken IL2RA, is highly reactive with
   CD8α-negative cells, and IL-2Rα^+ NK cells in the spleen are less
   active than 20E5^+ NK cells, aligning with our findings for
   NK-2.[207]^47^,[208]^75 In humans, CD56^bright NK cells (corresponding
   to NK-2) consistently express IL2RA, whereas CD56^dim NK cells
   (corresponding to NK-1) upregulate their expression upon activation;
   however, some CD56^bright NK cells also show low-level expression of
   IL2RA, suggesting potential heterogeneity within this subset.[209]^76
   Our findings corroborate this trend, demonstrating the increased
   expression of IL2RA by CD56^bright NK cells (Log[2]Fold change = 1.32),
   albeit without statistical significance (P[adj] = 1.0). Interestingly,
   chicken NK-2 cells exhibited significantly higher IL2RA expression than
   NK-1 cells (P[adj] = 2.41E-31), suggesting that IL2RA may serve as a
   distinctive marker for chicken NK-2 cells in contrast to the human NK
   cell paradigm. These findings underscore the highly divergent gene
   expression patterns in NK cells across species and the complex and
   species-specific nature of NK cell biology.

   Meanwhile, we found that NK cell functions in chicken are remarkably
   conserved with those in humans and mice, despite substantial divergence
   in the overall gene expression profiles during evolution. NK-1
   signature genes related to cytotoxicity (PRF1, GZMA, IFNG, CARD11),
   actin remodeling (PXN, PLEK, ACTR2), and phosphatidylinositol signaling
   (PRKCB, PIP4K2A, LYN) are consistently upregulated across species.
   Transcription factors crucial for NK-1 development, such as RUNX1
   (inducing NK cell cytolytic function[210]^77) and TBX21 (required for
   NK cell maturation[211]^71), showed conserved upregulation. NK-2
   subsets exhibit the upregulation of immune regulation genes (IL7R,
   CXCR4) and transcription factors supporting NK-2 development (TCF7,
   TOX). TCF7 is required for early NK cell development while repressing
   maturation,[212]^71 and TOX plays a vital role in NK progenitor
   differentiation.[213]^78 These findings demonstrate that fundamental
   immunological functions are conserved even in evolutionarily distant
   species, despite overall gene expression divergence.

   In summary, we identified two distinct chicken NK cell subsets for the
   first time, demonstrating core functions conserved across species. This
   discovery is significant as chicken models are valuable for studying
   zoonotic diseases affecting humans. Chickens enable broader validation
   of vaccines and therapeutics against zoonotic pathogens such as Zika
   virus and influenza A subtypes compared to murine
   models.[214]^79^,[215]^80 In particular, it has been reported that the
   highly pathogenic avian influenza (HPAI) H5N1 virus can infect cattle
   and eventually spread to humans through milk.[216]^81 In this regard,
   chickens, the natural host of HPAI H5N1, serve as an ideal avian
   experimental system for developing vaccines and therapeutics against
   emerging viral threats. However, to further apply chicken models in
   human infectious disease research, it is crucial to establish the
   boundaries of evolutionary conservation between avian and mammalian
   immune systems. In this respect, we believe that our findings, which
   identify conserved and species-specific properties of chicken, mouse,
   and human NK cells, pave the way for advancements in combating zoonotic
   infectious diseases.

Limitations of the study

   This study has limitations that current technology cannot fully
   address. First, the NK cell subsets identified in the RAG1^−/− chicken
   model may have lower fitness than those under normal
   conditions,[217]^82^,[218]^83 and the impact of RAG activity on chicken
   NK cell fitness remains unclear. As NK cells were highly abundant in
   RAG1^−/− chicken spleens but scarce in wild-type counterparts, it
   precluded a direct comparison between RAG1-deficient and -sufficient
   conditions. However, we found no significant evidence that NK cell
   subsets in the RAG1^−/− chicken spleen increased cell death-associated
   gene expression. Additionally, prior studies in mice demonstrated that
   RAG deficiency reduced EOMES expression in NK cells, mirroring the lack
   of EOMES in chicken NK-2 cells, which exhibit immature features akin to
   murine RAG-deficient NK populations.[219]^84 Second, the potential
   contamination of ILCs within the NK-2 population cannot be entirely
   ruled out due to the absence of chicken ILC-specific markers. While
   NK-2 cells lacked significant expression of most core ILC genes, it
   remains uncertain whether they express unidentified chicken ILC markers
   or represent a minor ILC-NK hybrid population. Additionally, we could
   not evaluate classical NK cell markers (found in humans/mice, such as
   NCR1, KLRF1, NKG7) in chicken NK cell subsets because orthologs to most
   of these genes are not identified in the chicken
   genome.[220]^46^,[221]^47 Although we demonstrated that chicken NK cell
   subsets significantly upregulated NK cell-biased genes, the definitive
   identification of chicken NK cells through exclusive markers remains
   challenging. This limitation underscores the need for the future
   identification of reliable chicken ILC and NK cell markers to resolve
   this ambiguity. Third, comparative analysis of adult NK subset
   proportions in chickens is challenged by the short survival of
   RAG1-deficient chickens (<1 month). Prior work reported immature
   IL-2Rα^+ NK cells as dominant in 1-week-old chickens, transitioning to
   20E5^+ NK cells by 35 days.[222]^85 However, our data revealed
   comparable proportions of cytotoxic NK-1 and immature NK-2 subsets,
   suggesting that 20E5 alone may not fully capture cytotoxic maturity in
   young-aged chickens. Fourth, the proposed markers CD8A and IL2RA for
   isolating NK-1 and NK-2 cells from splenic single cells require further
   verification, including at the protein level. Fifth, the varying
   numbers of NK-1 and NK-2 cells across species in the scRNA-seq data
   were addressed using harmony correction and hypothesis testing, but
   this approach may not fully resolve all variable factors. Consequently,
   we compared NK-1 and NK-2 cells within species rather than directly
   comparing results between species. We believe that these limitations
   are likely to be resolved as technologies for identifying reliable
   markers and corresponding antibodies for each chicken NK cell subset
   are developed.

Resource availability

Lead contact

   Further information and requests for resources and data should be
   directed to and will be fulfilled by the lead contact, Professor Jae
   Yong Han (jaehan@snu.ac.kr).

Materials availability

   This study did not generate new unique reagents.

Data and code availability

     * •
       All data generated or analyzed during this study are included in
       this published article and its supplementary information files. The
       newly generated datasets for chicken used and/or analyzed during
       the current study are available in the GEO: [223]GSE255238
       . Preprocessed scRNA-seq data from human NK-cells (GEO:
       [224]GSM3377673, [225]GSM3377674, and [226]GSM3377675) and murine
       NK-cells (GEO: [227]GSM3377679, [228]GSM3377680, [229]GSM3377681)
       were collected from the GEO database.
     * •
       This article does not report original code.
     * •
       Any additional information required to reanalyze the data is
       available from the [230]lead contact upon request.

Acknowledgments