All birds were obtained from the National Avian Research Facility (NARF) at The Roslin Institute, University of Edinburgh. All birds were hatched and housed in premises licensed under a UK Home Office Establishment License in full compliance with the Animals (Scientific Procedures) Act 1986 and the Code of Practice for Housing and Care of Animals Bred, Supplied or Used for Scientific Purposes. CSF1R-eGFP transgenic chickens (28), from newly hatched chicks to 20 weeks of age, were used for initial analysis of XCR1⁺ cDCs. Production of founder birds, the XCR1-reporter line and the XCR1 knockout line was carried out under UK Home Office Licenses (70/8528; 70/8940 and PP9565661). Inducible ablation of XCR1-iCaspase9-RFP⁺ cDCs by B/B homodimerizer drug was carried out under UK Home Office Licenses (PCD70CB48 and PP3522089). Experimental protocols and studies were approved by the Roslin Institute Animal Welfare and Ethical Review Board.

Chicken splenocytes were isolated from birds as described previously (20). To isolate the peripheral blood mononuclear cells (PBMCs), 2-5 ml blood was collected into Eppendorf tubes containing 50 μl of 0.5M EDTA (Sigma-Aldrich). Blood was diluted in phosphate-buffered saline (PBS) at a 1:1 ratio, layered on Histopaque (1077-1, Sigma-Aldrich) and spun at 400 × g for 30 min without braking. Mononuclear cells were collected from the gradient interface and washed twice with 1.0% bovine serum albumin (BSA, Sigma-Aldrich) in PBS (BSA/PBS).

To isolate bone marrow cells, femurs and tibias were flushed with PBS and the cells disaggregated by forcing through a 100 μm nylon cell strainer (Corning Inc.). Cells were spun at 500 × g for 10 min. To remove red blood cells, the cell pellet was re-suspended in an appropriate volume of PBS and carefully layered on Histopaque (1077-1, Sigma-Aldrich) and spun at 400 × g for 30 min without braking. Cells were collected from the gradient interface and washed 2 times with PBS/BSA.

To isolate cells from the skin, after removing fat and feathers, the skin was cut into small pieces (2.5 cm²) and digested with 2.5 mg/ml Dispase (Roche) while incubating in a 37°C water bath for 1 h with brief vortexing every 10 minutes. Remaining adipose and subcutaneous tissues were removed under a dissecting microscope and the epidermis/dermis layer was digested with 1 mg/ml Collagenase/Dispase/0.1 mg/ml DNase (Roche) in 5 ml of Hanks’ Balanced Salt Solution (HBSS, Thermo Fisher Scientific) with pulse vortexing and incubation in 37°C water bath for another 1-1.5 h. The skin samples were then minced in a petri dish using two scalpels and the resulting cell suspension passed through 100 μm strainer. Cells were then washed with 10 ml PBS/BSA.

To isolate immune cells from Ileum and Peyer’s Patches, the small intestine from CSF1R-eGFP transgenic birds was initially flushed with PBS to remove remaining intestinal content. Tissues were then dissected under a fluorescence microscope to isolate ileal Peyer’s patch from the non-lymphoid ileum (28). Tissue samples were cut into 2.5 cm² pieces and washed three times in 50 ml Falcon conical tubes with 20 ml of complete media [CM, HBSS with 2% heat-inactivated fetal bovine serum (FBS, Sigma-Aldrich)] by vortexing for 30 s and then replacing the CM. To remove the mucus, 25 ml of CM containing 1 mM dithiothreitol (DTT, Thermo Fisher Scientific) was added to the tissue samples in 50 ml Falcon tubes which were incubated in a shaking incubator at 37°C and 220 rpm for 20 min, followed by vortexing for 30 s. To remove epithelial cells, tissue samples were incubated in 50 ml Falcon tubes with 25 ml of CM/EDTA (CM with 1.3 mM EDTA) at 37°C in a shaking incubator at 220 rpm for 40 min, followed by vortexing for 30 s. Tissue samples were then rinsed with CM and then digested and homogenised using a Potter-Elvehjem Polytetrafluoroethylene (PTFE) pestle and glass mortar as described for isolation of cDC from spleen (20).

Single-cell suspensions from tissues or blood were prepared and resuspended in cold FACS buffer (PBS, 1.0% BSA (w/v) and 0.05% sodium azide (w/v); Sigma-Aldrich) and placed on ice for 10 min. Cells were then incubated with reagents listed in Table 1 in FACS buffer for 30 min on ice in the dark. If required, cells were washed and incubated with secondary antibodies for 20 min on ice in the dark. Cells were then washed three times, resuspended in cold FACS buffer and stained with SYTOX^® Blue Dead Cell Stain (Invitrogen; 1.0mM stock, 1/4000 dilution) prior to analysis with a LRSFortessa flow cytometer (BD Biosciences). Data were analysed using FlowJo V10 software. Dead cells were excluded by SYTOX^® Blue staining and doublets were then discriminated based on signal processing (FSC-A/W). Fluorescence minus one controls (FMO) were used to confirm gating strategies.

For DNA content staining, splenocytes were counted, washed with PBS and stained with Zombie Violet™ Fixable dye (Biolegend) at 1/500 in PBS at room temperature (RT) for 15 min in the dark. Cells were then washed twice with FACS buffer (with 1.0% (w/v) BSA) and stained with synthetic chicken XCL1 peptide conjugated to Alexa Fluor 647 (XCL1^AF647) for 30 min on ice. Cells were then washed twice with PBS and fixed with 2.0% paraformaldehyde (PFA, Sigma-Aldrich) at RT for 20 min. Cells were further washed twice with PBS and fixed with ice-cold 70% ethanol for 24 h. Cells were then washed twice with PBS and stained with 0.5 ml of FxCycle™ PI/RNase Staining Solution (Invitrogen™ F10797) per sample. Samples were incubated in the dark for 15–30 min prior to analysis using a BD LRSFortessa.

Tissue samples were trimmed into 1.0 cm² blocks and fixed overnight at 4°C in 4% PFA/PBS. Samples were removed from PFA/PBS and placed in 10% sucrose (w/v; Sigma-Aldrich)/PBS at 4°C overnight. Samples were then placed in 15/20/25/30% sucrose/PBS (w/v) for 24 h at 4°C for each sucrose concentration. Tissue samples were embedded in Cellpath™ OCT embedding matrix (Fisher Scientific UK Ltd, Loughborough, UK) and snap-frozen at −80˚C for two hours. 10µm sections were cut onto Superfrost Plus slides (Menzel-Gläser, Braunschweig, Germany) and air-dried for 1 h at RT. All primary antibodies used in this study are shown in Table 1 . All slides were blocked for one hour in 10% normal horse serum (Sigma-Aldrich), 0.1% Triton X-100 (Sigma-Aldrich) in PBS (HST-PBS). All primary antibodies were diluted in blocking reagent (above) and incubated at 4°C overnight, washed for 20 min in PBS, followed by incubation with secondary antibodies for two hours (donkey anti-mouse IgG Alexa Fluor 594, donkey anti-mouse IgG1 Alexa Fluor 594, donkey anti-mouse IgG2a Alexa Fluor 647; Thermo Fisher Scientific) used at 1/300 dilution and mounted in ProLong^® Gold Antifade Mountant (Thermo Fisher Scientific). Where appropriate, sections were counterstained with 1 μg/ml 4′, 6′-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) in the final incubation step. Samples were imaged using an inverted confocal microscope (Zeiss LSM710) and images were analysed using Zeiss ZEN 3·1 software.

Splenocytes were separately prepared from two 20-week-old female CSF1R-eGFP transgenic birds. CSF1R-eGFP⁺ cells were sorted using a BD FACS Aria IIIu sorter with the target of 5000 cells per sample. Libraries were prepared using a Chromium™ Single Cell 3’ Library & Gel Bead Kit v3 using the 10X Chromium Single Cell RNA Sequencing Platform at the University of Edinburgh. The single-cell libraries were sequenced at a depth of 50,000 reads per cells on an Illumina NovaSeq machine at Edinburgh Genomics, University of Edinburgh.

Cell Ranger (version 3.1.0) was used to generate transcriptomic reference index from Chicken reference genome (Gallus gallus GRCg6a, Ensembl version 101) through cellranger mkref, and gene expression matrix through cellranger count. Downstream analyses were performed on R Seurat (version 3.2.2) (29). Cells with low number of features (gene; <200) and outlier counts (unique molecule identifier; <300 or >12,000), or high mitochondrial count (>40% counts from mitochondrial genes) were removed from further analysis. Principal component analysis were performed on the cells using 50 components, and the top 30 components were used for generating a network graph using Graphia (version 2.0) (30) retaining nodes and edges fulfilling the following parameters: r ≥ 0.75, knn = 10, node degree > 5, and component size > 5. The resultant cells remained in the Graphia network graph were retained in the Seurat analysis, which was subjected to further quality control using DoubletDecon to remove doublets (31).

Following quality control, cells from two samples were integrated using Seurat SelectIntegrationFeatures, FindIntegrationAnchors and IntegrateData functions. Cell cycle states were labelled using CellCycleScoring using gene list organised by Seurat based on Tirosh et al. (2016) (32). Uniform Manifold Approximation and Projection (UMAP) dimension reduction was performed across all cell types under Seurat, and Potential of Heat-diffusion for Affinity-based Transition Embedding (PHATE) (33) was used to further interrogate the DC populations. The scRNAseq data for the study is available on https://www.ncbi.nlm.nih.gov/(BioProject: PRJNA996296).

PGCs were derived from embryos homozygous for the CSF1R-eGFP reporter transgene (28) at Hamburger–Hamilton (HH) stage 16 and expanded in vitro as previously described (34). Briefly, approximately 1 μl of embryonic blood was aspirated from the dorsal aorta of embryos and placed in FAot (FGF, Activin, ovotransferrin) culture medium to expand PGCs. After culturing for 2-3 weeks, PGCs were co-transfected with 2 μg of CRISPR-Cas9 vector (PX459 V2.0) (35, 36) which included two targeting guides (sgRNA) for the XCR1 locus and a double-stranded donor plasmid ( Supplementary Table 1 ) using Lipofectamine 2000 (Thermo Fisher Scientific) as described previously (37, 38). After 24 h in culture, cells were treated with 0.6 µg/ml puromycin for 48 h for selection of transfected cells. After selection, PGCs were sorted into single wells of 96-well plates using a BD FACS Aria IIIu sorter at one PGC per well. PGCs were clonally expanded for 2 to 3 weeks (36). Genomic DNA (gDNA) was extracted for genotyping as described below. Edited PGCs were cryopreserved in STEM-CELLBANKER (AMSBIO).

Genomic DNA from PGCs was extracted using QiaAmp DNA micro kit (Qiagen) according to the manufacturer’s instruction. Chorioallantoic membrane (CAM) lysate was prepared using a REDExtract-N-Amp Tissue PCR Kit (Sigma-Aldrich) and gDNA from blood was prepared using PUREGENE^® DNA Purification Kit (Flowgen). Primer pairs were designed to amplify the XCR1-transgene or wild type XCR1 ( Supplementary Table 1 ). PCR reactions included 100 ng of gDNA or CAM lysate, primer sets and REDExtract-N-Amp kit. PCR comprised the following cycling parameters: 94°C 3 min; 94°C 30 s, 58°C 30 s and 72°C 2 min for 30 cycles. Genotypes were distinguished by amplicons for the wild-type and edited allele (wild-type, monoallelic edit, and biallelic edit) ( Supplementary Figure 1 ). Cultured PGCs and all birds were sexed using a W-chromosome-specific PCR (39).

Targeted male PGC lines were thawed after storage at −150°C and cultured for 5-10 days before injection into iCaspase9 surrogate host embryos as described previously (38). One male founder was bred to wild type hens to produce birds for following studies and to generate G₁ offspring for breeding. All G₁ offspring were screened by PCR for the presence of iCaspase9-RFP transgene. The expression of RFP in spleen and tissues was analysed by flow cytometry or confocal microscopy. G₁ males and females were bred to produce G₂ offspring. The G₂ progenies were screened by PCR for the presence of iCaspase9-RFP transgene or native XCR1 gene. Homozygous (XCR1 KO), heterozygous (HET) and wild type birds were distinguished by PCR as described above.

B/B homodimerizer (AP20187, Takara) was injected intravenously into XCR1-iCaspase9-RFP reporter birds at a dose range of 0.5, 1 or 2 mg/kg, 6 birds per group. The stock solution of the drug was prepared in absolute ethanol at 62.5 mg/ml. The carrier solution was 10% PEG-400 and 2.0% Tween in sterile water. The control group received carrier solution only. Birds were sacrificed for tissues after 24 hours. Ablation rates were calculated for each dose. The optimised dose of B/B drug was then determined accordingly. For the time course, B/B homodimerizer was injected intravenously into XCR1-iCaspase-RFP birds at a dose of 0.5 mg/kg 6 birds per time-point. Birds were sacrificed for tissues 24 h, 48 h, 4 and 7 days after injection.

Data were analysed using GraphPad Prism 7.00 (GraphPad, US). Statistical analysis was conducted using unpaired non-parametric Mann-Whitney test or Multiple t-test. Statistical significance was defined as follows: no significant (ns), p > 0.05, ∗, p < 0.05; ∗∗, p < 0.01; and ∗∗∗, p < 0.001.

Previously we demonstrated that the chemokine receptor XCR1 is selectively expressed on chicken cDCs and staining for XCR1 on CSF1R-eGFP transgene-expressing cells could be used to identify and characterise these cells (20). In this study we extend these observations by examining the dynamics of XCR1⁺ cDC development in the spleen of chickens aged between day 1 post-hatch and sexual maturity ( Figure 1 ). As a percentage of the CD45⁺ cell population the XCR1⁺ cDC population was relatively stable between hatch and sexual maturity (week 20), ranging between 0.7-2.0% of CD45⁺ cells ( Figure 1A ). While the proportion of CSF1R-eGFP transgene-expressing cells in the CD45⁺ cell population decreased over time, likely reflecting the maturation and emergence of B and T-cell populations, the proportion of XCR1⁺ cells within the CSF1R-eGFP⁺ population increased from ~3% on the day of hatch to ~40% in sexually mature males ( Figures 1B, C ). Compared with sexually mature females, males consistently showed a higher percentage of XCR1⁺ splenic cDCs, expressed as either a percentage of the CD45⁺ or CSF1R-eGFP⁺ cell populations ( Figures 1A-C ). We found that within the splenic CSF1R-eGFP⁺ population the ratio of XCR1⁺ cDCs to MRC1L-B⁺ macrophages increased from ~0.025 at hatch to ~0.5 in sexually mature birds ( Figure 1C ). MRC1L-B (also known as “KUL01”) is a classical marker of chicken splenic macrophages (40, 41) and is expressed on multiple splenic macrophage subsets (20, 42). Our data therefore imply that the XCR1⁺ cDC population becomes the most abundant CSF1R-eGFP⁺ subset in the chicken spleen by sexual maturity. Previously (20), we showed that ~80% of XCR1⁺ cDCs expressed the CSF1R-eGFP transgene in three-week-old chicks. We confirmed this result and showed that the proportion of CSF1R-eGFP⁺ XCR1⁺ cDCs is similar in all age groups ( Figure 1D ). We also examined the relative abundance of XCR1⁺ cDCs in the ileum, Peyer’s patches (PP), skin, bone marrow (BM) and blood ( Supplementary Figures 2 , 3 ). We have found that the prolonged enzymatic digestion required for extraction of cells from intestinal tissues and skin results in the loss of FLT3 staining. As such, FLT3 staining is not shown for cells derived from these tissues. XCR1⁺ cDCs are a relatively rare cell population in the BM and blood ( Supplementary Figure 2 ), a minor (~1%) component of the CSF1R-eGFP⁺ cell population in the skin, but a major (30-40%) CSF1R-eGFP⁺ cell population in the ileum and PP ( Supplementary Figure 3 ).

We and others have shown that in chicken, cells that express XCR1 also express high levels of genes associated with cDCs in mice and humans (e.g. FLT3, ZNF366, CIITA, CADM1, ID2, IRF8 and ZBTB46 (16–19);), but also the classic macrophage marker CSF1R (20) which is not detected or weakly detected on mammalian cDC1s (43–46). In our previous analysis of chicken splenic cDCs and macrophages (20), we identified a FLT3^HI cell population that expressed the dendritic cell associated marker XCR1 and lacked expression of the chicken macrophage marker MRC1L-B, suggesting a bona fide cDC identity. Nevertheless, unlike mammalian XCR1⁺ cDCs, this chicken cell population expressed low levels of surface CSF1R and the vast majority of this population also expressed the CSF1R-transgene. In addition, we also described a FLT3^LOW MRC1L-B^LOW population, which expressed high levels of CSF1R and the CSF1R-transgene, but lacked expression of XCR1. Both cell populations expressed high levels of surface MHCII. Previously published data suggests a macrophage/monocytic origin for the MHCII^HIGH MRC1L‐B^LOW (FLT3^LOW) splenic cell population (42). To determine the relationship between these chicken splenic cell populations and other chicken splenic macrophage populations, which lack FLT3 expression, we performed single cell RNA-sequencing (scRNA-seq) of individual CSF1R-eGFP⁺ splenic cells isolated from sexually mature hens (n=2; twenty weeks of age). Across all samples, we generated 16,994 cell transcriptomes with median unique molecular identifier (UMI) and gene counts per cell of 932 and 538, respectively. For our analysis, we used two dimension-reduction techniques, Uniform Manifold and Approximation and Projection (UMAP (47);) and Potential of Heat-diffusion for Affinity-based Transition Embedding (PHATE (33);). UMAP analysis identified 10 clusters ( Figure 2 ), comprising of macrophage and dendritic cells, as expected from our previous analysis of CSF1R-eGFP⁺ cells in the chicken spleen ( Figure 2 ; Supplementary Table 2 ). Proliferating cells (predominately Clusters 7 and 10) were identified on the basis of expression of TOP2A, PCNA, SMC2 and MCM6 ( Figures 2B, C ) (48). Clusters were identified as macrophages on the basis of expression of genes for factors that regulate macrophage development and function (e.g. MAFB, SPIC, CSF3R and MRC1L-B; Figure 2 ; Supplementary Table 2 ). XCR1⁺ cDC clusters were identified on the basis of expression of FLT3, XCR1, IRF8, CADM1, and ID2 ( Figure 2 ; Supplementary Table 2 ). MRC1L-B (also known as MMR1L4) encodes the macrophage mannose receptor recognised by antibody KUL01 (41). We and others have shown that the MRC1L-B is expressed by at least two distinct macrophage populations in the chicken spleen (20, 42). In the present analysis, we found that MRC1L-B is expressed by three distinct cell clusters ( Figures 2B ). Collectively, MRC1L-B expressing clusters comprised 36.8% (6,248 of 16,994 total cells) of total cells analysed. Thus, XCR1⁺ cDCs collectively formed the largest cell subset in this analysis, comprising of 45.5% (7,734 of 16,994 total cells) of the total cells analysed. Due to the complexity of the data, functional analysis of the macrophage clusters is beyond the scope of this paper and will be detailed in a separate publication.

A small cluster of XCR1^- FLT3⁺ IRF8⁺ cells was noted (“Cluster 9”; Figure 2 ; Supplementary Table 2 ). On the basis of high level of expression of JCHAIN (49, 50) and IRF8 (51) and the lack of expression of ID2 and CADM1 (51) ( Figures 2 , 3 ), we putatively identified these cells as chicken plasmacytoid dendritic cells (pDCs). The transcription factors TCF4 (also known as E2-2) and ZEB2 are essential for the development of pDCs (51). We found two TCF4 homologues in chickens ENSGALG00000055022 and ENSGALG00000033770 (referred to here as TCF4-like_A and TCF4-like_B respectively). While these genes are expressed in a subset of Cluster 9 cells ( Figure 3 ), this was not at the high levels seen in mammalian pDCs. Similarly, ZEB2 was not highly expressed in Cluster 9 cells ( Figure 4A ). In mammals, pDCs are potent producers of type I interferon (IFN) (52, 53) in response to viral pathogens via the TLR-MyD88-IRF7 pathway (54, 55). As expected, in unstimulated pDCs, we did not detect expression of type I IFN genes, but we did detect the expression of genes associated with TLR recognition of viral pathogens (TLR3, TLR7 and TLR21) (56, 57) and type I IFN production (IRF7) (54), validating our identification of these cells as chicken pDCs ( Figure 3 ).

Transcription factors: Chicken XCR1⁺ cDC clusters (Clusters 0, 3, 5 and 8) expressed a range of transcription factors associated with mammalian XCR1⁺ cDC development and function, including BATF3 (58), BLC6 (59, 60), SPI1 (61), ZBTB46 (62), ZNF366 (63), ID2 (51), BCL11A (64), and IRF8 (51) ( Figures 2B , 4A ). In contrast to mice (62), ZBTB46 is also expressed in putative chicken pDCs ( Figures 2B , 4A ). The macrophage associated transcription factors SPIC, CEBPA/B/C, and MAFB (46, 65) were not expressed in pDCs or XCR1⁺ cDC clusters ( Figure 2B ).

Cytokines/chemokines: As is expected in unstimulated cells, chicken pDCs and XCR1⁺ cDC clusters expressed few detectable cytokines or chemokines, with the exception of TGFB1, IL18 and CCL1 in XCR1⁺ cDCs, and low levels of TGFB1 and IL16 in pDCs ( Figures 4B, C ). With the exception of XCR1 itself, XCR1⁺ cDC clusters did not globally express any other chemokine receptor ( Figure 4B ), although immature XCR1⁺ cDCs (Cluster 5; see below) expressed low levels of CCR2/4/5. pDCs expressed relatively low levels of CCR4/5/6/7/10 and CXCR4. XCR1⁺ cDC clusters expressed a wide range of cytokine receptors, including IL2RG, IL4R, IL10RA/B, IL13A2, and IL17RA ( Figure 4D ). In contrast, cytokine gene expression in pDCs was restricted to IL2RG and IL10RB ( Figure 4D ).

TLRs: Both chicken pDCs and XCR1⁺ clusters expressed TLR7 (albeit weakly in XCR1⁺ cDCs) and TLR21 [a functional homologue of mammalian TLR9 (66)]. XCR1⁺ cDCs, but not pDCs expressed TLR15 ( Figure 4E ), a TLR unique to avian, non-teleost fish, and reptilian lineages that recognises fungal-derived protease agonists (67). TLR3 expression was restricted to pDCs (Cluster 9) and immature XCR1⁺ cDCs (Cluster 5; Figure 4D ). Immature XCR1⁺ cDCs also expressed weak levels of TLR1A.

Growth factors: pDCs expressed FLT3 and the gene annotated as CSF2RA (ENSGALT00000026942; Figure 4F ). CSF1R expression in XCR1⁺ cDC clusters was not detectable, despite the cells expressing low levels of CSF1R and the CSF1R-transgene (20). Non-proliferating XCR1⁺ cDC clusters expressed high levels of FLT3 and the CSF2RA paralog ENSGALG00000019147, but not CSF2RA (ENSGALT00000026942).

XCR1⁺ cDC Cluster 7 expressed the cell proliferation markers TOP2A, PCNA, MCM6, MKI67 and STMN1 ( Figure 5A ). This represented 8.6% of XCR1⁺ cDCs (666 of 7,734 total XCR1⁺ cDCs). To confirm that chicken splenic XCR1⁺ cDCs are proliferating in situ, we independently determined the percentage of proliferating cells by measuring the DNA content of chicken XCR1⁺ cDCs (68). Consistent with the scRNASeq data, ~7% of XCR1⁺ cDCs were in S/G2/M phases ( Supplementary Figure 4 ). Non-proliferating XCR1⁺ cDCs (Clusters 0, 5 and 8) could be differentiated from each other on basis of differential expression of genes associated with MHCII pathway antigen processing and presentation (e.g. BLB1, BLB2, CITTA, CD40, CD80, CD86, CD74, SCPEP1, NRP1, IFI30, CTSA and CTSS), or anti-viral activity (e.g. IFITM1, IFITM3, LY86, CG-1B and CD1c) ( Figures 5B, C ). We used PHATE to analyse the structure of the XCR1⁺ cDC population (Clusters 0, 5 and 8; Figure 5D ), and confirm the separate developmental origin for these cells and the pDC population (Cluster 9; Figure 5D ). Proliferating cells of different developmental origins share the expression of highly variable cell cycle genes. As this can confound trajectory analysis, we excluded proliferating XCR1⁺ cDCs (Cluster 7) from this analysis. All XCR1⁺ cDC clusters remained within the same cell trajectory, whereas pDCs were a separate population of cells. Non-proliferating XCR1⁺ cDC populations displayed a single cell trajectory with no obvious branching. The termini of the cell trajectory were represented by cells expressing high levels of genes related to MHCII antigen presentation (e.g. CD74 and BLB1) or IFITM3, CG-1B and CD1c ( Figure 5D ). The expression of CIITA, the main transcription factor controlling expression MHCII genes (i.e. BLB1 and BLB2 encode for MHCII beta chain in the chicken), was intermediate between BLB1 and IFITM3 expressing cells. These data suggested a single cell population undergoing local maturation within the spleen, with immature XCR1⁺ cDCs expressing IFITM1/3, CG-1B, LY86, CD1c and relatively low levels of MHCII related genes, and mature cells expressing high levels of MHCII, but not IFITM1/3, LY86, CG-1B or CD1c. As tools to stain for IFITM1/3, LY86, CG-1B were not available at this time, to confirm the substructure within the chicken splenic XCR1⁺ cDC population we used a monoclonal antibody to chicken CD1.1 (69), which detects chicken CD1c ( Figure 6 ). We found that chicken splenic XCR1⁺ cDCs could be partitioned into of CD1.1^LOW MHCII^HIGH, CD1.1^HIGH MHCII^HIGH or CD1.1^HIGH MHCII^LOW subsets ( Figure 6A ), with the CD1.1^HIGH MHCII^LOW and CD1.1^LOW MHCII^HIGH representing the least and most abundant subsets respectively. In contrast, the vast majority of blood XCR1⁺ cDCs are CD1.1^HIGH MHCII^LOW ( Figures 6A, B ). As maturation of XCR1⁺ cDCs is determined by cell intrinsic and local tissue conditions (reviewed by Roquilly et al., 2022) (70), we reasoned that recently hatched chicks would contain more immature XCR1⁺ cDCs than older birds. To test this hypothesis, we determined the proportion of CD1.1^HIGH MHCII^LOW and CD1.1^LOW MHCII^HIGH XCR1⁺ cDCs in 1-week-old chicks compared to chicks aged 2 and 12 weeks ( Figure 6C ). Week-old chicks were found to have approximately 3-fold more CD1.1^HIGH MHCII^LOW XCR1⁺ cDCs than older chicks ( Figure 6D ). Taken together, these data suggest that splenic CD1.1^HIGH MHCII^LOW XCR1⁺ cDCs are immature recently migrated from the blood; and furthermore, CD1.1 (and also IFITM1, IFITM3, LY86 and CG-1B) expression diminishes as MHCII expression increases in maturing splenic XCR1⁺ cDCs.

The priming of cytotoxic CD8⁺ T cells to exogenously-derived antigens is a process termed “cross-presentation”. Mammalian XCR1⁺ cDCs excel at cross-presentation of viral and tumour cell-associated antigens (9–13). As no chicken immortalised T-cell lines exist to directly test antigen cross-presentation function, we assessed the potential for antigen cross-presentation by chicken XCR1⁺ cDCs by examining the expression of genes involved in cross-presentation taken from previously published resources ( Supplementary Figure 5 ). XCR1⁺ cDC clusters exhibited a range of upregulated cross-presentation genes compared to pDC or macrophage clusters. These included genes with functions directly (WDFY4, CD74 and PPT1) or indirectly (CADM1, DNASE1L3, XCR1 and LY75) related to cross-presentation.

Previously we developed tools to identify chicken XCR1⁺ cDCs (20). However, these did not enable specific in situ visualisation nor in vivo manipulation of XCR1⁺ cDCs. Analysis of XCR1⁺ cDCs was problematic in many tissues, as markers such as FLT3 are lost during the enzymatic digestion of tissues required for cell extraction. Therefore, we aimed to produce a gene edited chicken line in which XCR1⁺ cDCs could be visualised and conditionally ablated. Inducible caspase-9 (iCaspase9), developed as a cellular suicide gene for human stem cell therapy, is an effective system for cellular ablation in chicken embryos (38). We used CRISPR/Cas9-mediated homology-directed repair (HDR) to replace the single exon of the XCR1 gene with an iCaspase9 construct in PGCs ( Figure 7A ). The iCaspase9 transgene was followed by a 2A self-cleaving peptide sequence then an enhanced red fluorescent protein (RFP) reporter gene to mark cellular expression. Transfected CSF1R-eGFP transgene positive PGCs were screened and clonally expanded. 50% (20/40) of selected clones were found to have bi-allelic edits ( Supplementary Figure 1A ). A single male PGC clone with biallelic edits was selected and injected into surrogate host embryos (38). One male founder was bred to wild type hens to produce G₁ offspring for breeding and analysis. All G₁ offspring were screened by PCR for the presence of iCaspase9-RFP transgene ( Supplementary Figure 1B ). The expression of RFP expression in splenic cells of G₁ offspring was analysed by flow cytometry ( Figure 7B ). The vast majority of RFP⁺ cells were found in the CSF1R-eGFP⁺ cell population, as expected for chicken XCR1⁺ cDCs (20). RFP⁺ cells were positive for the chicken cDC markers XCR1 and FLT3, expressed high levels of MHCII, but did not express markers for chicken macrophages (MRC1L-B), B-cells (Bu-1) nor T-cells (CD3), indicating the transgene was specifically expressed in chicken XCR1⁺ cDCs ( Figure 7B ). As the XCR1-iCaspase9-RFP transgene replaces the native XCR1 coding sequence, this has the potential to impact expression levels of XCR1. Previously we showed that XCR1 expression on chicken XCR1⁺ cDCs could be assessed by flow cytometry by measuring the binding of XCL1^AF647 (20). We found that in birds homozygous for the iCaspase9-RFP transgene (i.e. deficient in XCR1) there was a total lack of binding to XCL1^AF647; whereas heterozygous birds exhibited the same level of binding to XCL1^AF647 as wild-type (WT) birds ( Supplementary. Figure 6A ). Heterozygous birds are therefore functionally wild-type for XCR1 expression and are suitable for assessing XCR1⁺ cDC development and function. We assessed the impact of XCR1 deficiency on splenic immune cell populations ( Supplementary Figure 6B ). Apart from a slight, but significant, decrease in B-cells (Bu-1⁺ cells) in homozygous XCR1-iCaspase9-RFP transgenic (i.e. XCR1 deficient; “KO”) compared to heterozygous birds (“HET”), no impact of XCR1 deficiency was noted in the CD45⁺, CSF1R-eGFP⁺, MRC1L-B⁺ (macrophages), CD4⁺ or CD8⁺ T-cell populations ( Supplementary Figure 6B ). As males have approximately twice as many splenic XCR1⁺ cDCs than females ( Figure 1 ), we compared the number of XCR1⁺ cDCs between 20-week old CSF1R-eGFP XCR1-iCaspase9-RFP male and female chickens ( Supplementary Figure 7 ). While the proportion of CSF1R-eGFP⁺ cells expressed as a percentage of the CD45⁺ population did not differ between the sexes, males had approximately two-fold more XCR1-RFP⁺ cDCs than females, expressed as either a percentage of the CD45⁺ or CSF1R-eGFP⁺ cell populations ( Supplementary Figure 7 ).

RFP transgene expression was used to detect splenic XCR1⁺ cDCs in situ by immunofluorescence microscopy ( Figure 8 ). In the spleen, RFP⁺ cells were most abundant in the red pulp and located in the periarteriolar lymphoid sheaths (PALS), rarely located in the periellipsoid white pulp (PWP), and not detected in germinal centres (GC) ( Figure 8A ). XCR1-RFP⁺ cDCs were widely distributed in chicken tissues ( Supplementary Figure 8 ). In the small intestine (ileum and caecal tonsils, Supplementary Figures 8A, B ), XCR1-RFP⁺ cDCs were abundant in the lamina propria but excluded from germinal centres (GC) and the B-cell follicles of the bursa of Fabricius ( Supplementary Figure 8F ). XCR1-RFP⁺ cDCs were scattered through the parenchyma of the liver and lung ( Supplementary Figures 8D, E ). In the lung, XCR1-RFP⁺ cDCs were found within lymphocyte clusters adjacent to the parabronchi ( Supplementary Figure 8D ). In the thymus XCR1-RFP⁺ cDCs were abundant in the medulla and located in the septa but excluded from the cortex ( Supplementary Figure 8C ). Previously, two monoclonal antibodies, CVI-ChNL-68.1 and CVI-ChNL-74.2, were reported to stain red pulp macrophages (71). Due to the prominent location of XCR1-RFP⁺ cDCs in the chicken red pulp we investigated if either of these antibodies recognised XCR1⁺ cDCs. CVI-ChNL-68.1 marked XCR1-RFP⁺ cDCs in the red pulp and RFP^- cells in PWP ( Figure 8B ). CVI-ChNL-74.2 stained a ring of RFP^- cells surrounding the PWP and scattered RFP^- cells in the red pulp ( Figure 8C ). ChNL-74.2 did not stain RFP⁺ cells.

Our scRNA-Seq analysis suggests that like their mammalian counter-part, chicken XCR1⁺ cDCs can present antigens to CD4⁺ T-cells via MHCII pathway and drive cytotoxic T-cell (CTL) responses by presenting exogenous antigen to CD8⁺ T-cells (known as “cross-presentation”) (72, 73). We examined the distribution of these cells by confocal microscopy ( Figure 9 ). RFP⁺ cDCs were located in the RP and PALS, where they were intimately associated with CD4⁺ T^-cells ( Figures 9A, B ). Like mice, the adult chicken spleen contains TCRαβ CD8αβ (CTLs), but chickens also have a TCRγδ CD8αβ population of cells (74, 75). CD8⁺ CTLs and TCRγδ⁺ T^-cells can be distinguished by staining with antibodies to the chicken TCRγδ (75). Unfortunately, in our hands these antibodies did not work in the conditions required for detection of RFP. Therefore, we used a monoclonal antibody to chicken CD8β that detects both CTLs and TCRγδ⁺ CD8αβ. We found that RFP⁺cDCs were intimately associated with clusters of CD8β⁺ T-cells in the splenic RP ( Figures 9C, D ).

iCaspase9 has previously been used to ablate cell lineages in early chicken embryos (38). Its suitability as a system for ablating specific cell populations in post-hatch chickens was unknown. We tested the efficacy of the iCaspase9 gene under control of the chicken XCR1 promoter for the specific ablation of chicken XCR1⁺ cDCs. Groups of six birds heterozygeous for XCR1-iCaspase9-RFP transgene were intravenously injected with 0.5, 1.0 or 2.0 mg/kg of the B/B dimerization drug or carrier alone ( Figure 10A ). All dosage levels were found to specifically ablate XCR1⁺ cDCs (FLT3^HIGH XCR1⁺ XCR1-RFP⁺ cells) at an ablation rate of 94-96% ( Figure 10A ; Supplementary Figure 9A ). No effects were noted on macrophage, granulocyte, T or B-cell populations ( Supplementary Figure 9B ). Next we assessed the repopulation of the spleen by XCR1⁺ cDCs after ablation. Groups of six chickens were intravenously injected with B/B dimerization drug or carrier alone ( Figure 10B ) at a dose rate of 0.5mg/kg. Chickens were culled at 1, 2, 4 or 7 days post-injection ( Figure 10B ). After initial ablation, the spleen was rapidly repopulated by XCR1⁺ cDCs with the levels of splenic XCR1⁺ cDCs returning to the same level as in control birds by day 4-post injection. However, it was noted that the repopulating XCR1⁺ cDCs expressed lower levels of the RFP transgenic reporter, therefore despite the numbers of XCR1⁺ cDCs being normal by day 4, the mean fluorescence intensity (MFI) of the RFP expression did not return to the same level as control birds until day 7 post-injection ( Figure 10B ). We hypothesised that the reduced RFP levels observed in repopulating XCR1⁺ cDCs after ablation was due to the repopulation of spleen by immature XCR1⁺ cDCs. To test this hypothesis, we measured the levels of CD1.1 expression on XCR1⁺ cDCs after B/B reagent induced ablation ( Figure 11 ). We observed that the proportion of CD1.1^HIGH XCR1⁺ immature cDCs increased significantly at two days post- ablation and then returned to the same levels as observed in control birds by day 4 post-ablation ( Figure 11 ). In contrast, the MFI of RFP transgene expression did not return to the same levels as observed in control birds until day 7 post-ablation. ( Figure 10B ). Taken together, these data suggest that after ablation the spleen is initially repopulated by immature XCR1-RFP^LOW CD1.1^HIGH XCR1⁺ cDCs emigrating from the blood. Once recruited to the spleen these cells mature in situ to a mature CD1.1^LOW XCR1-RFP^HIGH phenotype, with the two markers showing different expression dynamics.

In mice, the chemokine receptors XCR1 and CCR7 jointly control XCR1⁺ cDC migration to CD8⁺ CTLs in T-cell areas within the splenic PALS (26). Chicken XCR1⁺ cDCs did not express CCR7, indeed, with the exception of weak expression of CCR4, CCR5 and CX3CR1 in immature XCR1⁺ cDCs, XCR1 itself was the only chemokine receptor expressed at high levels in these cells ( Figure 4 ). The organisation of immune cell compartments of the murine spleen differs significantly from that of chickens and other vertebrates (21). Chicken CD8⁺ CTLs are predominately found in the RP (27), so we examined the impact of XCR1 deficiency on distribution of RFP⁺ cDCs in the chicken spleen. B-cells are the major immune cell population of the PWP, with a ring of CVI-ChNL-74.2⁺ macrophages demarking the boundary between the PWP and RP ( Supplementary Figures 10 , 11 ). In birds heterozygous for the XCR1-iCaspase9-RFP transgene (wild-type (WT) for XCR1 expression) RFP⁺ cDCs were largely confined to the splenic RP ( Figure 12 , Supplementary Figures 10 , 11 ). WT RFP⁺ cDCs were rarely observed within the PWP, with the exception of the PALS. In contrast, in birds homozygous for XCR1-iCaspase9-RFP transgene (deficient for XCR1 expression) RFP⁺ cDCs were only occasionally located in the RP, being mostly concentrated within the PWP, intermingled with BAFFR⁺ B-cells and CVI-ChNL-74.2⁺ macrophages ( Supplementary Figures 10 , 11 ). In XCR1 deficient birds, but not WT birds, RFP⁺ cDCs were also observed within the splenic ellipsoid ( Supplementary Figure 10 ). We found that XCR1 deficiency results in a significant reduction in the number of RFP⁺ cDCs associated with CD8⁺ T-cell areas in the RP ( Figure 12 ). Taken together, these data indicate that while XCR1 is not required for recruitment of XCR1⁺ cDCs to the chicken spleen, it is the main chemokine receptor orchestrating movement of XCR1⁺ cDCs to the CD8⁺ T-cell areas in the splenic RP.