Peripheral blood was taken by venipuncture of the vena cephalica antebrachii into heparinized collection tubes (BD Vacutainer^®, 10 ml, Li-Heparin 17 IU/ml Becton Dickinson, Heidelberg, Germany) from healthy experimental Beagle dogs (7 female, 4 male, age: 4-10 years) of the Faculty of Veterinary Medicine (Leipzig University, Leipzig, Germany). The number of dogs used for individual analyses is indicated in the figure legends. All dogs received routine vaccinations against canine distemper, rabies, canine infectious hepatitis, parvovirus infection, parainfluenza, and leptospirosis. The study was authorized by the Saxony State Office (Landesdirektion Sachsen) in Leipzig, Germany (approval numbers: DD24.1-5131/444/30; DD24.1-5131/468/16).

PBMC were isolated by density gradient centrifugation. Briefly, blood was diluted with phosphate buffered saline (PBS) at a ratio of 1:1, layered above Biocoll Separating Solution (density 1077 g/l, Biochrom AG, Berlin, Germafny) and centrifuged at 500g for 30 min at room temperature (minimal acceleration and braking). PBMC at the interphase were harvested into PBS and centrifuged at 500g for 10 min at RT. After another washing step with PBS, remaining erythrocytes were lysed by incubation in 150 mM NH₄Cl, 8 mM KHCO₃, 2 mM EDTA (pH 7.0) for 5 min at RT. The reaction was stopped by addition of PBS containing 3% fetal bovine serum (FBS, Thermo Fisher Scientific, Carlsbad, USA; and PAN-Biotech, Aidenbach, Germany). After washing with PBS, PBMC were counted in Trypan blue (Sigma-Aldrich, Taufkirchen, Germany) using a hemocytometer (Laboroptik, Lancing, UK).

All incubation steps performed to stain cells for fluorescence-activated cell sorting were performed for 15-20 min on ice in the dark. First, PBMC were incubated with the fixable viability dye eFluor 780 (Thermo Fisher Scientific, Carlsbad, USA) according to the manufacturer’s protocol to discriminate viable from dead cells. The subsequent staining steps were preceded by washing with PBS, 3% FCS. To block binding of Fc receptors, cells were preincubated for 5 min with a mixture of heat-inactivated normal serum derived from dog and rat (each 15% in PBS) before addition of the anti-canine TCRαβ (clone CA15.8G7) hybridoma supernatant ( Table 1 ) and further incubation. In the next incubation step, a PerCP/Cy5.5-conjugated goat-anti-mouse IgG secondary antibody (Biolegend, San Diego, USA) was used for detection of TCRαβ. Finally, PBMC were incubated with FITC-conjugated anti-canine CD4 (clone YKIX302.9) and APC-conjugated CD8α (clone YCATE55.9) antibodies ( Table 1 ). Stained cells were applied over a 30 µm filter (Sysmex, Norderstedt, Germany) and immediately sorted using a BD FACSAria™ III Cell Sorter (Becton Dickinson, Heidelberg, Germany). The sorting strategy is shown in Figure 1A . After exclusion of dead cells and doublets, TCRαβ⁺ T cells from the lymphocyte gate were selected and CD4⁺ single-positive (sp), CD8α⁺ sp, CD4^brightCD8α^bright double-positive (dp), and CD4^-CD8α^- double-negative (dn) T cell subpopulations were sorted with a purity >99% (Re-analysis using the FlowJo™10 software (Treestar Inc., Ashland, OR, USA).

Isolated TCRαβ⁺ T cell subpopulations or whole PBMC were seeded at a density of 5x10⁵ per well in IMDM medium with L-Glutamine and 25 mM HEPES (PAA Laboratories, Cölbe, Germany) supplemented with 10% FBS (Thermo Fisher Scientific, Carlsbad, USA), 100 U/ml penicillin, and 100 μg/ml streptomycin (both purchased from PAA Laboratories) into 96-well round bottom cell culture plates (Techno Plastic Products AG, Trasadingen, Switzerland) and rested over night at 37°C and 5% CO₂. Then, cells were polyclonally stimulated for 5 hours at 37°C with phorbol myristate acetate and Ionomycin (P/I) (both: 0.22 µg/ml; Sigma-Aldrich, Taufkirchen, Germany). Medium incubation (M) served as negative control. For subsequent intracellular cytokine staining, 5 µg/ml Brefeldin A (BrefA) (Enzo Life Sciences, New York, USA) was added to the culture.

Total RNA from sorted TCRαβ⁺ T cell subpopulations was isolated using 1 ml RNA-Solv^® Reagent (Omega Bio-tek, Norcross, USA) following manufacturer instructions. To increase the RNA yield, 2 µl glycogen (Thermo Fisher Scientific, Vilnius, Lithuania) were added per sample. The concentration of the RNA was determined spectrophotometrically (NanoDrop, PEQLAB, Erlangen). DNA was digested with one unit DNase (Thermo Fisher Scientific)/µg RNA for 30 min at 37°C. For isolation of poly(A)+ RNA, an additional ethanol precipitation was conducted with 1/8 volume RNase-free 3 M NaOAc, pH 5.7. Reverse transcription was performed using the ‘High Capacity cDNA Reverse Transcription Kit’ (Applied Biosystems, Foster City, USA), containing RiboLock RNase inhibitor (Thermo Fisher Scientific), and a mixture of oligo (dT)18 (Thermo Fisher Scientific) and random hexamer primers in a PTC-200 (MJ Research, Watertown, USA). Specific primers were designed using the primer-BLAST Software (National Center for Biotechnology Information, Bethesda, USA) and synthesized by Metabion (Planegg, Germany) ( Table 2 ). The primers for IL2RA (CD25), IL4, and IL5 were commercially available (Bio-Rad, Munich, Germany). Real-time PCR with SYBR^® Green detection was performed using the iTaq Universal SYBR^® Green Supermix (Bio-Rad, Munich, Germany) and a CFX Connect real time PCR detection system (Bio-Rad). Specificity was tested by melting curve analysis and an agarose gel electrophoresis (LE agarose, Biozym, Hess. Oldendorf; Germany). Only primers resulting in a single peak in melting curve analysis and a single band with the expected fragment size in gel electrophoresis were used. RT-qPCR efficiencies were calculated with the CFX Maestro Software version 2.2 (Bio-Rad) from standard curves generated by serial dilutions of appropriate cDNA samples (2-fold dilutions, 8 point-curve, conducted in duplicates). The efficiency of all reactions ranged between 90 and 110%. Beta-2-microglobulin (B2M) and succinate dehydrogenase subunit A (SDHA), were used as reference genes of different functional classes with stability verified using the CFX Maestro software based on the GeNorm algorithm (18) ( Supplementary Table 1 ). The same amount of cDNA was used for analysis of expression of the respective reference gene and target genes from the same sample. Relative quantification of the transcripts was done by the 2^(-ΔCt) method (19, 20). Results obtained with B2M as reference gene are shown in Figures 1 – 4 , those normalized to SDHA in Supplementary Figures 1 – 4 .

Viability-dye and surface staining of stimulated canine PBMC were essentially performed as described in paragraph 2.3, except for using flow cytometry buffer (3% FBS, 0.1% sodium azide in PBS) instead of PBS/3% FBS. Primary antibodies used are listed in Table 1 . For intracellular staining of IL-4, the FoxP3 staining kit (Thermo Fisher Scientific) was used according to the manufacturer’s protocol. For intracellular detection of IL-10, alternatively, cells were fixed with 2% paraformaldehyde (Serva, Heidelberg, Germany) for 15 minutes at 4°C and subsequently permeabilized using 0.5% saponin (w/v, Serva) in flow cytometry buffer. After permeabilization, blocking of Fc receptors was performed by incubation with a mixture of rat, mouse and dog serum (each 15% in PBS) for 10 minutes at room temperature. Cells were incubated for 30 minutes at room temperature with the monoclonal antibodies specific for the selected cytokines ( Table 1 ). Finally, cells were incubated for 20 minutes with PE-Cy7-conjugated streptavidin to detect IL-4 and IL-10, respectively. Data were acquired with a BD LSR Fortessa (Becton Dickinson, Heidelberg, Germany). FlowJo 10.7.1 software (FlowJo, LLC, Ashland, OR, USA) was used for analysis. Gates were set using appropriate controls: unstimulated controls, and fluorescence minus one (FMO) controls, which include all specific antibodies of the staining panel, except the one of interest, which is replaced by its isotype control.

Statistical analysis was performed using Graph Pad Prism 10.2.0 software (San Diego, CA, USA). Depending on normal distribution determined using the Shapiro-Wilk test, data sets are presented with the mean or median. Statistical analysis was performed using One-way ANOVA with Tukey’s post-test or Kruskal-Wallis test with Dunn’s post-test to compare for differences between T cell subpopulations. Additionally, significance of stimulation-induced effects was calculated for each subpopulation by direct comparison of medium (M) versus the PMA/ionomycin-stimulated (P/I) equivalent using the unpaired t test or Mann-Whitney test (two-tailed). The level of confidence for significance is shown in figure legends.

For a detailed investigation, non-conventional TCRαβ⁺ CD4⁺CD8α⁺ double-positive (dp) and CD4^-CD8α^- double-negative (dn) T cells were compared to conventional CD4⁺ and CD8α⁺ single-positive (sp) T cells. All four subpopulations were isolated with high purity (>99%) from peripheral blood of healthy Beagle dogs by fluorescence-activated cell sorting ( Figure 1A ) and gene expression was analyzed by reverse transcription-quantitative PCR (RT-qPCR). Noteworthy, a very high transcriptional activity of the dp T cell subset was observed. Despite a >20x lower cell yield of the dp T cell subset compared to conventional T cell subpopulations, the resulting amount of RNA was similar in all groups ( Table 3 ), which allowed comparison of gene expression between the populations.

The transcription of selected key markers was analyzed ex vivo ( Figures 1B–D ; Supplementary Figures 1B–D ) and after stimulation with PMA/Ionomycin (P/I) ( Figures 1E–G ; Supplementary Figures 1E–G ). In general, gene expression was normalized to the expression levels of two stable reference genes ( Supplementary Table 1 ) of different functional classes, B2M ( Figures 1 – 4 ) and SDHA ( Supplementary Figures 1 – 4 ), yielding comparable results. The expression of CD4 and CD8A mRNA ex vivo confirmed the sp/dp/dn phenotype of the sorted T-cell fractions ( Figure 1B ; Supplementary Figure 1B ). Increased IL2RA (CD25) transcription levels ex vivo support the high constitutive activation and/or the IL-2 dependence of both non-conventional T cell subpopulations ( Figure 1C ; Supplementary Figure 1C ) (5, 13). Upregulation of IL2RA (CD25) mRNA expression following P/I incubation confirmed efficient stimulation of each T cell subpopulation ( Figure 1F ; Supplementary Figure 1F ).

In accordance with previous flow cytometric results (13), dp T cells differ from their CD8α⁺ sp counterparts by only weak constitutive transcription of CD8B ( Figure 1C ; Supplementary Figure 1C ), indicating preferential expression of the CD8αα homodimer. After P/I stimulation, CD8B was almost not detectable in dp T cells, whereas in CD8α⁺ sp T cells a downregulation was observed ( Figure 1F ; Supplementary Figure 1F ).

Consistent with their CD4⁺CD8A⁺ phenotype ( Figure 1B ; Supplementary Figure 1B ), canine peripheral blood dp T cells were characterized by constitutive mRNA expression of genes encoding the transcription factors ThPOK [known to be associated with the CD4 T helper cell gene program (21)] and Runx3 [establishing the gene program characteristic of CD8 cytotoxic T cells (22)] ( Figure 1D ). Albeit showing broader variation among the five dogs studied, ThPOK and Runx3 expression was significantly elevated in dp in comparison to the CD8α⁺ sp and CD4⁺ sp T cell subpopulation, respectively. Dn T cells showed a comparable transcription level of ZBTB7b (encoding ThPOK) and RUNX3 like their CD4⁺ sp counterparts ( Figure 1D ; Supplementary Figure 1D ). The expected downregulation of CD4 or CD8A mRNA expression in sp and dp populations upon P/I stimulation (5) ( Figure 1E ; Supplementary Figure 1E ) was accompanied by a downregulation of ZBTB7b/RUNX3 ( Figure 1G ; Supplementary Figure 1G ). Notably, the CD4^-CD8α^- phenotype of the double-negative T cell population was unaltered upon stimulation ( Figure 1E ; Supplementary Figure 1E ).

Overall, RT-qPCR analysis of sort-purified canine TCRαβ⁺ T cell subpopulations ex vivo and after stimulation confirmed key features of non-conventional dp and dn T cells identified by flow cytometry in previous studies and revealed stability of the dn phenotype upon stimulation in vitro.

Expression of characteristic markers of T helper (Th)1 and Th17 cells was analyzed by RT-qPCR in sorted non-conventional and conventional TCRαβ⁺ T cell subpopulations. Consistent with previous flow cytometric results showing expression of the transcription factor T-bet only in a small proportion of dp T cells with a CD4^bright/CD8α^bright phenotype (7), mRNA levels of the corresponding gene TBX21 was significantly lower in this population as compared to CD8α⁺ sp T cells ( Figure 2A ; Supplementary Figure 2A ). Upon stimulation, high transcription of IFNG (encoding the effector cytokine IFN-γ) was confirmed in both populations ( Figure 2C ; Supplementary Figure 2C ) (5, 13). Albeit to a lesser extent than in CD8α⁺ sp and dp T cells, a stimulation-dependent increase in transcription of IFNG was also observed in CD4⁺ sp and dn T cells ( Figure 2C ; Supplementary Figure 2C ), consistent with previous flow cytometric results (5). Expression of RORC, which encodes the lineage-defining transcription factor of Th17 cells RORγt (23, 24), was hardly detectable in any of the sorted populations ex vivo ( Figure 2B ; Supplementary Figure 2B ) and upon stimulation (data not shown). Nevertheless, transcription of the gene encoding the Th17 effector cytokine IL-17A was highly upregulated in CD4⁺ sp T cells and in dn T cells upon PMA/Iono stimulation ( Figure 2D ; Supplementary Figure 2D ). This is in line with previous flow cytometric findings (5). Interestingly, high stimulation-induced upregulation of IL17A mRNA expression was also found in dp T cells ( Figure 2D ; Supplementary Figure 2D ) suggesting additional Th17 properties in this population.

To test for a potential role of non-conventional dn and dp T cells in type 2 immune reactions, characteristic features of conventional Th2 cells, including mRNA expression of genes encoding the master transcription factor GATA-3, the key surface marker IL-33Rα as well as the effector cytokines IL-4, IL-5 and IL-13, were analyzed. In line with previous flow cytometric results (5), CD4⁺ sp T cells constitutively expressed GATA3 and minimal expression was also detected in CD8α sp T cells ( Figure 3A ; Supplementary Figure 3A ). Furthermore, both, dp and dn T cells constitutively transcribed GATA3 ( Figure 3A ; Supplementary Figure 3A ). For dn T cells, this is consistent with previously published flow cytometric results showing an even higher percentage of GATA-3⁺ cells in the dn than in the CD4⁺ sp subpopulation (5). For dp T cells, this is the first indication of a role in type 2 immunity. Indeed, dp T cells showed a high capacity for transcription of genes encoding the Th2-associated cytokines IL-4 and IL-13 comparable to their CD4⁺ sp counterparts ( Figure 3B ; Supplementary Figure 3B ). However, in dp T cells and their CD4⁺ sp counterparts, stimulation-induced transcription levels of the gene encoding IL-5 were not increased similarly as in CD8α⁺ sp T cells ( Figure 3B ; Supplementary Figure 3B ). Strikingly, highest mRNA expression levels of all three type 2 effector cytokines IL-4, IL-5, and IL-13 were detected in dn T cells upon PMA/ionomycin stimulation ( Figure 3B ; Supplementary Figure 3B ). Constitutive transcription of IL1RL1 (encoding IL33Rα) in this population further supports the Th2-like phenotype ( Figures 3C, D ; Supplementary Figures 3C, D ).

Taken together, canine non-conventional dp and especially dn T cells show characteristics of conventional Th2 cells and their high ability to transcribe key effector cytokines points to an important involvement in type 2 immune responses.

In a previous study, we identified expression of the master transcription factor of conventional CD4⁺ regulatory T (Treg) cells FoxP3 in a subset of dn T cells (5). Here, we further assessed the potential immunoregulatory phenotype of canine non-conventional dn and dp T cells. Comparable mRNA expression levels of FOXP3 and the inhibitory receptor CTLA4 were observed ex vivo in sorted dn, dp, and conventional CD4⁺ sp T cells ( Figures 4A, B ; Supplementary Figures 4A, B ). Interestingly, the gene encoding the immunosuppressive cytokine IL-10 was constitutively transcribed by dn and CD4⁺ sp but not dp T cells ( Figure 4C ; Supplementary Figure 4C ). High up-regulation of IL10 and CTLA4 mRNA expression upon stimulation was detected in CD4⁺ sp but not CD8α⁺ sp and dp T cells ( Figures 4D, E ; Supplementary Figures 4D, E ). Intriguingly, dn T cells showed highest stimulation-induced expression levels of both inhibitory molecules, which even exceeded those of conventional CD4⁺ T cells ( Figures 4D, E ; Supplementary Figures 4D, E ). Taken together, these data point to a putative regulatory role of canine non-conventional dn T cells.

The capacity of dp and dn T cells to produce IL-4 and IL-10 was analyzed at the protein level by flow cytometry. Analysis of IFN-γ by intracellular cytokine staining (ICS) served as a positive control. As expected (7), IFN-γ was detected in all TCRαβ⁺ T cell populations analyzed after 5 h of PMA/Iono stimulation, with highest frequencies in dp and CD8α⁺ T cells, followed by CD4⁺ sp and dn T cells ( Figure 5A ).

Specific detection of IL-4 and IL-10 by ICS upon PMA/Iono stimulation was confirmed using several controls, including unstimulated and fluorescence minus one (FMO) controls. CD8α⁺ sp T cells showed almost no IL-4/IL-10 expression after stimulation ( Figures 5 , 6 ).

As expected, the Th1 and Th2 effector cytokines, IFN-γ and IL-4, are barely co-produced ( Figure 5A ). Polyclonal activation-induced IL-4 expression was detected in CD4⁺ sp T cells, and especially in dn T cells. Of note, the frequency of IL-4⁺ cells was higher in the dn than in the CD4⁺ sp population ( Figures 5A, C ). Furthermore, IL-4⁺ dn and CD4⁺ sp T cells exhibit an activated phenotype, indicated by CD25 expression ( Figures 5B, D ).

Production of the immunosuppressive cytokine IL-10 was significantly induced in dn T cells and CD4⁺ sp (but not in CD8α⁺ sp and dp T cells) in response to PMA/Ionomycin stimulation ( Figures 6A–C ). As expected, the majority of dn T cells producing IL-10 are CD25⁺ reminiscent of a T regulatory phenotype ( Figure 6D ).

Taken together, ICS analysis confirmed the high ability of dn T cells to produce IL-4 and IL-10, further supporting a potential role of these non-conventional T cells in type 2 and regulatory immune responses.