Naïve T cells, isolated from mouse splenocytes, were seeded at a concentration of 1.5×10ˆ5 cells/well in 96-well round-bottom plates and activated with plate-bound anti-CD3 (3 µg/mL) and soluble anti-CD28 (5 µg/mL) in RPMI 1640 medium supplemented with 10% FBS, 1% P/S, and 0.1% β-mercaptoethanol. The cells were polarized into Th9 cells (3 ng/mL hTGFβ, 20 ng/mL IL-4 and 20 µg/mL anti-IFNγ), Treg (5 ng/mL hTGFβ and 100 U/mL IL-2), Th1 cells (10 ng/mL IL-12, 50 U/mL IL-2, 10 µg/mL anti-IL-4), Th2 cells (20 ng/mL IL-4, 50 U/mL IL-2, 10 µg/mL IL-12) or Th17 cells (1 ng/mL hTGFβ, 50 ng/mL IL-6, 10 ng/mL IL-1β, 10 µg/mL anti-IFNγ, 10 µg/mL anti-IL-4) for 48 hours, 72 hours or 5 days.

To analyze the proliferation of Th9 and Treg cells, freshly isolated naïve CD4⁺ T cells from mouse splenocytes were stained with 0.5 µM carboxyfluorescein succinimidyl ester (CFSE)-PB. CFSE-stained naïve CD4⁺ T cells were seeded at 1×10ˆ5 cells per well in 96-well flat-bottom plates and differentiated for 3 days into Th9 cells (plate-bound anti-CD3 (3 µg/mL), anti-CD28 (5 µg/mL), 3 ng/mL hTGFβ, 20 ng/mL IL-4 and 20 µg/mL anti-IFNγ) or Treg cells (plate-bound anti-CD3 (3 µg/mL), anti-CD28 (5 µg/mL), 5 ng/mL hTGFβ and 100 U/mL IL-2). The proliferation was quantified by analysis of CFSE dilution using flow cytometry.

In vitro differentiated CD4⁺ T cells, isolated tumor-infiltrating lymphocytes, and lymph node cells were stained with antibodies against CD3 (145–2 C11), CD4 (RM4-5), CD8α (53–6.7), CD11c (HL-3), TCR β chain (H57-597), TCRγδ (GL3), Foxp3 (FJK16s), CD45 (30-F11), IFNγ (XMG1.1), IL-13 (eBio13A), IL-9 (RM9A4), IL-17A (eBio17B7), IRF4 (IRF4.3E4), PU.1 (7C2C34), CD25 (PC61), CD44 (IM7) CD62L (MEL-14). Antibodies were purchased from eBioscience (San Diego, USA) or BioLegend (San Diego, USA). For intracellular cytokine and transcription factor analysis, cells were fixed with 2% paraformaldehyde for 20 min, permeabilized with 0.5% saponin in PBS/BSA, and stained for 30 min at room temperature in the dark. Populations were analyzed with a BD FACSCanto II flow cytometer, and data were analyzed using FlowJo software (TreeStar, Ashland, USA).

According to the manufacturer’s instructions, RNA from tumor tissue, isolated lymphocytes, or in vitro differentiated T cells was isolated using the Qiagen RNeasy kit or the Zymo Research quick-RNA miniprep kit. The RNA concentration was determined using the NanoDrop 2000 system. According to the manufacturer’s instructions, the first-strand cDNA was synthesized using a RevertAid first strand cDNA synthesis kit from Thermo Fisher Scientific or High capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems) on the Mastercycler RealPlex2 (Eppendorf) and the LightCycler 480 real-time PCR system (Roche Diagnostics). Primers were designed using the Universal Probe Library (Roche Applied Science, Mannheim, Germany) and were purchased from Invitrogen.

The HCmel3 cell line was first generated from a primary murine melanoma in which the hepatocyte growth factor is overexpressed and the p16lnk4a-dependent cell cycle regulation is impaired due to an oncogenic CDK4 (R24C) germline mutation. Landsberg et al43 HCmel3 melanoma cells were routinely cultured in RPMI 1640 medium (Life Technologies) supplemented with 10% FCS (Biochrome), 2 mM L-glutamine, 10 mM non-essential amino acids, 1 mM HEPES (all from Life Technologies), 20 µM 2-mercaptoethanol (Sigma), 100 IU/mL penicillin and 100 µg/mL streptomycin (Invitrogen). B16-OVA cells were cultured in RPMI 1640 medium supplemented with 10% FCS, 1% penicillin–streptomycin, 0.4 mg/mL G418 (Sigma). Cells were cultured under sterile conditions at 37°C and 5% carbon dioxide in a water-saturated atmosphere.

For HCmel3 tumor transplantation experiments, 200 µg of InVivoMAb anti-mouse IL-9 (9C1) or InVivoMAb mouse IgG2a isotype control (both Bio X Cell) was intravenously administered to mice for three consecutive days starting from the day before tumor injection (day −1, 0, 1). HCmel3 cells (4×10ˆ5) were injected intracutaneously into the right flank on day 0. Anti-IL-9 or control IgG was administered intravenously to the respective mice every third day. Tumor size was measured two times per week using a caliper. Mice were sacrificed when the tumor volume reached 1 cm³ (0.5×length×width²) or when other criteria requiring euthanasia were reached. For B16-OVA tumor transplantation experiments, C57BL/6N mice were injected intracutaneously with 1×10ˆ5 B16-OVA cells in the right flank. On days 5 and 9 post tumor injection, in vitro generated Th9 cells (3×10ˆ6) from OT-II and OT-II*Cblb^−/− mice were injected intravenously. Tumor size was measured every other day using a caliper. Mice were sacrificed when the tumor volume reached 1 cm³ (0.5×length×width²) or when signs of illness were observed.

Cytokine levels were measured in cell culture supernatants that were either freshly processed or stored at −80°C prior to analysis. Cytokine measurements were performed using different BD Cytometric Bead Array Flex Sets against murine IFNγ, IL-13, IL-9, IL-2, or TNFα.

Th9 cells were generated in vitro from naïve CD4⁺ T cells isolated from the murine spleen using a naive CD4⁺ T cell isolation kit from Miltenyi Biotec. Cells from six samples in total (3×wildtype (WT), 3×Cblb^−/−) were polarized into Th9 cells (3 ng/mL hTGFβ, 20 ng/mL IL-4 and 20 µg/mL anti-IFNγ) for 48 hours, purified using CD4 (L3T4) MicroBeads from Miltenyi Biotec, and transferred to the Next Generation Sequencing Core Facility of the University Hospital Bonn for single-cell RNA (scRNA) sequencing. According to the manufacturer’s specifications, a library was prepared using the 10× Genomics Single Cell 3’ reagent Kits V.3. Sequencing was performed using an Illumina NovaSeq 6000 Sequencing System, which generated 26 nt barcode reads as well as 98 nt transcript reads (contrary to manufacturer’s recommendations of 28 nt and 91 nt, respectively), aiming for 7500 cells per sample with a sequencing depth of approximately 80,000 reads per cell. Data analysis and visualization were performed by FSD using multiple tools and R/Bioconductor packages.

Cells were harvested, washed once with cold PBS, (1 min, 9300 rcf, 4°C) resuspended in 90 µl Buffer A (10 mM HEPES pH 7.9; 10 mM KCl; 0.1 mM EDTA; 0.1 mM EGTA; 1 mM DTT; 0.5 mM PMSF) and kept on ice for 15 min. Next, 6 µl of 10% NP-40 (final concentration 0.63%) was added, and the pellet was vortexed thoroughly and centrifuged (5 min, 2300 rcf, 4°C). The supernatant was collected for the cytosolic protein fraction, and the remaining nuclear pellet was washed twice with 300 µl Buffer A (5 min, 2300 rcf, 4°C). The supernatant was discarded, 30 µl Buffer C (20 mM Hepes pH 7.9; 0.4 M NaCl; 1 mM EDTA; 1 mM EGTA; 1 mM DTT; 1 mM PMSF) was added to the pellet, and put on a shaker for 30 min at 4^◦C. After centrifugation (10 min, 13,000 rcf, 4°C) the supernatant was collected for the nuclear protein fraction. For whole-cell protein lysates, cells were collected, washed once with cold PBS (1 min, 9300 rcf, 4°C) and lysed in 30 µl lysis buffer (5 mM Na₄P₂O₇, 5 mM NaF, 1 mM Na₃VO₄, 5 mM EDTA, 150 mM NaCl, 50 mM Tris (pH 7.3), 1% NP-40, aprotinin, and leupeptin (50 µg/mL each)) for 30 min on ice. After centrifugation (15,000 rcf, 4°C), the supernatant was collected for protein lysates.

Pierce Lane Marker Reducing Sample Buffer (3900) was added to the samples and western blotting analysis with antibodies against NFATc1 (Santacruz 7294; dilution 1:1000), Cbl-b (Abcam 54362; dilution 1:1000), PU.1 (Cell Signaling Technology 2266, 1:1000), IRF4 (Cell Signaling Technology 4964, 1:1000), GAPDH (Cell Signaling Technology 5174, 1:1000), actin (Santacruz 1615; dilution 1:2000) or lamin B1 (abcam16048; dilution 1:10,000) was performed. In addition, densitometric analysis was performed using the ImageJ software.

Sample size for animal experiments was determined with the G*Power software during initial study design. The number of performed experiments and used mice or biological replicates, which equals cells taken from one mouse, are stated in the figure legends. Statistical analysis was performed using the unpaired Student’s t-test if only two independent groups or two conditions in one experimental group were compared. One-way analysis of variance (ANOVA) was used to determine any statistical differences between three or more independent experimental groups. Two-way ANOVA was used to analyze the mean differences between groups that were split into two independent variables. All data are presented as mean±SEM unless stated otherwise in the figure legend. Significance was defined at *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001. All statistical analyses were conducted with the software GraphPad Prism.

OT-II and OT-II*Cblb^−/− naïve CD4⁺ T cells were polarized in vitro into Th9 cells for 48 hours. B16-OVA and B16-WT tumor cells were labeled with CFSE, mixed at a 1:1 ratio, and seeded into 96-well plates (50,000 cells/well; 25,000 CFSE low B16-OVA and 25,000 CFSE high B16-WT). Th9 cells (100,000) were added and incubated for 16 hours. The degree of dead/killed B16-OVA and B16-WT cells was determined through usage of flow cytometry and the Zombie Fixable Viability Kit from BioLegend, bringing living and dead cells into proportion.

Mouse tumor samples were immersed in a zinc-based fixative (BD, Franklin Lakes, USA). Tissues were embedded in paraffin and sections (4 µm) stained with H&E according to standard protocols. Immunohistochemistry was also performed on 4 µm paraffin sections. Melanomas were bleached before staining (20 min at 37°C in 30% H₂O₂ and 0.5% KOH, 20 s in 1% acetic acid and 5 min in Tris buffer). The sections were dried overnight at 37°C and rehydrated through a graded alcohol series. Sections were bleached for 20–40 min in 3% H₂O₂ with KOH and washed afterwards with TBS buffer. All sections were heat–pretreated in a citrate buffer (pH 6) for 10 min at 95°C–99°C to eliminate potential antigen blockings and washed afterwards with TBS buffer. The sections were incubated with anti CD45 mAb (30-F11) from BD Pharmingen at 25°C for 1 hour (1:500) in a staining chamber and a secondary polyclonal rabbit anti-Rat Immunoglobulins Biotinylated (Dako). The Dako Real Detection System (Dako, California, USA) was used for the detection of immunoreactivity.

All mice used for this study were housed under specific pathogen free conditions in the ‘Haus für Experimentelle Therapie’ at the University Hospital Bonn. All animal experiments performed in this study have been approved by the ‘Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen’ under the file reference number ‘81–02.04.2017.A418’. When planning and carrying out experiments, we strictly followed the 3R principle (replacement, reduction, refinement). Close monitoring of the mice with regular clinical scoring, observation of tumor growth and weight was performed. Animal treatment as well as termination criteria are in unison with the specifications of the Society of Laboratory Animal Science (GV-SOLAS).

It is well established that Cblb is expressed in and controls the activation threshold of CD4⁺ T cells.44 The expression levels of Cblb during the differentiation process of T cell subpopulations, particularly Treg and Th9 cells, have not been analyzed in detail. After 48 hours and 72 hours of differentiation, Th9 cells showed significantly lower Cblb messenger RNA (mRNA) expression levels than Th0 cells (figure 1A). This observation was even more impressive when analyzing protein levels, as Th0 cells and Treg cells showed strong expression of Cbl-b, while this was virtually undetectable in Th9 cells after 48 hours and 72 hours of differentiation (figure 1B). Thus, we next validated whether regulation of Cbl-b is also functionally involved in the regulation of T helper (Th) cell subset differentiation. Therefore, we generated Th1, Th2, Th9, Treg, and Th17 cells from naïve CD4⁺ T cells isolated from WT or Cblb^−/− mice, respectively. After 48 hours and 72 hours Th cells were analyzed by flow cytometry, with intracellular staining for the subset-typical hallmark cytokines IFNγ, IL-13, IL-9, IL-17 and the transcription factor forkhead box P3 (Foxp3) (online supplemental figures 1 and 2). A thorough analysis of the cytokine-positive cell subsets revealed considerable differences between the WT and Cblb^−/− Th cell subsets. Differentiation of Th1 and Th17 cells led to broadly similar frequencies of IFNγ and IL-17 positive cells, with more IFNγ⁺ cells in the absence of Cblb after 72 hours of Th1 cell differentiation (online supplemental figure 2C). As previously reported,42 loss of Cblb favored Th2 differentiation, mirrored by a significantly higher frequency of IL-13⁺ cells and higher Il13 mRNA expression levels (online supplemental figure 2B–D). However, the most significant differences between WT and Cblb^−/− mice could be observed in the differentiation of Treg and Th9 cells (online supplemental figure 2 and figure 1). Th9 cultures from Cblb^−/− mice displayed a significantly higher frequency of IL-9⁺ and IL-13⁺ cells, whereas the frequency of CD25⁺Foxp3⁺ Tregs was significantly lower when compared with Th9 cultures generated from WT T cells (figure 1C,D). This observation was slightly more pronounced at 48 hours than at 72 hours, indicating that knocking out Cblb might change the Th9 cell differentiation dynamics during the early differentiation phase (online supplemental figure 2). In line with the protein data, Cblb^−/− Th9 cells showed significantly higher Il9 mRNA expression levels (figure 1E and online supplemental figure 2C) and released more IL-9 and IL-13 into the cell culture supernatant (figure 1F and online supplemental figure 3). Besides Th9 cells, also Treg cell cultures from Cblb^−/− mice differed markedly from those of WT mice, as they contained a significantly lower frequency of CD25⁺Foxp3⁺ Treg, whereas the frequency of IFNγ⁺ and IL-9⁺ cells was significantly higher in these cultures, an effect that was visible after 48 hours and even more pronounced after 72 hours (figure 1 and online supplemental figure 2B). Lower Foxp3 mRNA expression levels were also measured in Cblb^−/− Treg cells collected after 72 hours of differentiation (online supplemental figure 2C). Cblb^−/− Tregs further released a diverse array of cytokines into the culture supernatant (IFNγ, IL-9, IL-2, and TNFα), which could not be observed for WT Treg cells, suggesting that these cells might be less suppressive compared with WT Treg cells (online supplemental figure 3A,B). Of note TNFα showed a generally higher concentration in all Th cell subsets generated from Cblb^−/− naïve T cells, although this difference was only significant under Th1 polarizing conditions (online supplemental figure 3C). Taken together, we observed that deletion of Cblb favored Th9 cell differentiation. This was paralleled by a clearly lower Treg frequency, indicating a potential role of Cblb as a switch in the polarization between Th9 and Treg cells.

Based on our hypothesis that Cbl-b acts as a specific switch in the regulation of Th9 versus Treg differentiation, we speculated that the elevated levels of IL-9 in the Cblb^−/− Treg cultures might be involved in the suppression of Foxp3 expression and consequently Treg differentiation. Therefore, we performed in vitro Treg differentiation for 72 hours in the presence of 5, 10, or 20 µg/mL neutralizing anti-IL-9 antibodies. Under normal circumstances, WT Treg cultures produced 80% CD25⁺Foxp3⁺ cells and Cblb^−/− cultures produced less than 40% CD25⁺Foxp3⁺ cells; blocking IL-9 increased the frequency of Treg cells in the Cblb^−/− cultures with increasing concentration of the neutralizing antibody (figure 2A,C). Blocking IL-9 in WT Treg cultures led to a slight but insignificant lowering of the frequency of Treg cells, while the frequency of Tregs in Cblb^−/− samples treated with 10 or 20 µg/mL anti-IL-9 antibody reached approximately the same level as in WT samples. Similarly, the frequency of Treg cells and the expression of Foxp3, which decreased in Cblb^−/− Tregs, could be restored to WT levels by neutralizing IL-9 (figure 2A,B). Neither knocking out Cblb nor blocking IL-9 had any influence on the expression of CD25 (figure 2C). We next wondered whether, besides influencing the frequency, knocking out Cblb or blocking IL-9 also leads to functional impairment of Treg cells. Therefore, naïve WT or Cblb^−/− CD4⁺ T cells were polarized into Treg cells in the presence or absence of a neutralizing anti-IL-9 antibody (20 µg/mL). Treg cells were then sorted (CD4⁺CD25^hiCD45RB^low) and co-cultured with CD3-depleted splenocytes and CFSE-stained WT CD4⁺ conventional T cells at a 1:1:1 ratio. While control cells showed robust proliferation (as indicated by CFSE dilution), WT, Cblb^−/− and anti-IL-9-treated Treg cells inhibited proliferation efficiently and showed no functional impairment (figure 2D,E). Taken together, these results indicate that Cbl-b suppresses the production and release of IL-9 by T cells, and that high levels of IL-9, as observed in Cblb^−/− cells, inhibit in trans Treg differentiation, without affecting their suppressive function.

Cblb^−/− T cells are hyper-responsive mirrored by a lower TCR activation threshold that does not require CD28 co-stimulation when TCR is ligated.44 The transcription factor family, known as the nuclear factor of activated T cells (NFATs), governs gene expression that regulates T cell development and differentiation. To assess whether enhanced Th9 cell differentiation could be due to a lower TCR activation threshold mediated by NFAT, we investigated the expression of NFAT proteins as well as the NFAT nuclear entry in Th9 cells and cells stimulated only with CD3/28 (Th0) from WT and Cblb^−/− mice. The expression of NFATc1, NFATc1 alphaP1, NFATc1 betaP2 and NFATc2 was generally higher in Th9 cells than in Th0 cells; however, their expression in Cblb^−/− Th9 samples was slightly but not significantly lower, except for NFATc1, which was significantly decreased in Cblb^−/− Th9 cells compared with WT Th9 cells (online supplemental figure 4A). In addition, we did not observe higher protein expression levels of NFATc1 in Cblb^−/− Th9 cells compared with WT Th9 cells, either in the nucleus or in the cytosol, as well as no higher NFATc1 levels in the nucleus compared with the cytosol, which would have indicated a higher degree of nuclear entry or nuclear activity (online supplemental figure 4B–E). This set of data collectively rules out that stronger Th9 cell differentiation of Cblb^−/− T cells is just a function of a lower TCR activation threshold mirrored by induced NFAT signaling.

Previous studies have reported an increased proliferation rate of Cblb^−/− T cells due to a lower TCR activation threshold.36 Hence, we assessed whether the increased frequency of IL-9⁺ cells in Th9 cultures from Cblb^−/− mice resulted from hyperproliferation. In line with previous findings, Cblb^−/− Th9 cells showed higher proliferation rates. Strikingly, the CD4⁺IL-9⁺ population in WT and Cblb^−/− Th9 cells was found to proliferate at a similar rate, while CD4⁺IL-9^– cells from Cblb^−/− cultures showed a much higher degree of proliferation compared with CD4⁺IL-9^– cells from WT cultures (figure 3A). Strikingly, no differences in proliferation in Treg cultures, when looking seperately at CD25⁺Foxp3⁺ and Foxp3^– cells could be observed. Moreover, we observed that the mean fluorescence intensity levels of IL-9 in CD4⁺IL-9⁺ cells were significantly higher in Cblb^−/− mice than in WT mice (figure 3B), indicating a higher production of IL-9 in absence of Cbl-b. The transcription factors PU.1 and IRF4 have been described to be essential for Th9 development.3 4 Accordingly, we assessed the expression of IRF4 and PU.1 (SPI1) after 3, 6, 24 and 48 hours of Th9 cell differentiation (figure 3). WT and Cblb^−/− Th9 cells similarly upregulated IRF4 after 3 and 6 hours of differentiation, followed by a slow decline (figure 3C). SPI1 expression, which has been shown to be required for optimal IL-9 production in mouse T cells,3 was initially present in naïve T cells and after 3 hours of Th9 cell differentiation but declined over time (figure 3D). However, no significant differences were observed between WT and Cblb^−/− cells. We also investigated IRF4 and PU.1 protein levels through flow cytometry (FACS) and western blot (WB) analysis. While IRF4 was clearly detectable but equally expressed in WT and Cblb^−/− Th9 cells, PU.1 was not as strongly expressed and was consequently barely detectable by FACS and WB analysis (figure 3E,F). Nonetheless, the frequency of PU.1⁺ cells (figure 3E) and PU.1 protein levels (figure 3F) seemed to be slightly higher in Cblb^−/− Th9 cells, although this effect was not statistical significant. Because the results obtained for PU.1, especially the detection by WB, did not wholly satisfy us and might even be inadequate to explain the big differences in Th9 cell differentiation, we opted for a broader omics-based analysis at the single-cell level to better understand the observed changes during in vitro Th9 cell differentiation.

To understand the differences between WT and Cblb^−/− Th9 cells we performed a microarray and proteomic analyses of in vitro differentiated Th9 cells (data not shown). However, by virtue of these techniques, not discriminating between IL-9⁺ and other cells in the culture, we could not identify candidates that could potentially explain the phenotype of Cblb^−/− Th9 cells, presumably due to the relatively low frequency of IL-9 positive cells in the culture and the technique-inherent limitation of providing only integrals of protein and mRNA expression levels. Therefore, we performed single-cell mRNA sequencing (45,000 cells with a depth of 80,000 reads per cell) of in vitro differentiated Th9 cells in order to deconvolute the cellular complexity of Th9 cell cultures (figure 4A).

The global volcano plot depicts genes with the highest log₂ fold change (FC) values and highlights that Il13, Il9, Gatm, Gm10093 and Lgals1 had the highest upregulation in Cblb^−/− compared with WT Th9 cells (online supplemental figure 5B). After 48 hours of Th9 cell differentiation, graph-based clustering led to the formation of six major and one minor clusters shown in two-dimensional t-distributed stochastic neighbor embedding (tSNE) space. All seven clusters were equally found in the WT and Cblb^−/− Th9 cultures (figure 4B and online supplemental figure 5A). Hence, knocking out Cbl-b shifts the frequency of the diverse subsets rather than generating novel T cell populations, with clusters 0 and 2 growing and clusters 1 and 3 shrinking in Cblb^−/− cultures (figure 4C). The calculated cluster identities corresponded well to the two-dimensional representation in tSNE and Uniform Manifold Approximation and Projection space (online supplemental figure 5A). The six major and one minor clusters express an array of marker genes, including cluster 2, which showed the highest level of Il9 expression and were allocated as Th9 cells (figure 4D). This cluster was about three times larger in cultures from Cblb^−/− mice than WT mice, which is in line with the FACS analysis of Th9 cultures shown in figure 1 (figure 4B,C). Besides Il9, cluster 2 showed high expression of Batf3, Lta, and Vim, all of which have previously been connected to Th9 cell biology.5 45 In order to understand what distinguishes the cluster 2 Th9 cells in WT vs Cblb^−/− cultures (besides sheer cluster size), we investigated differentially expressed genes (DEGs), with a particular look at the top 100 genes with the highest expression differences (log₂ FC). Beyond the genes with the highest significant upregulation in Cblb^−/− cells compared with WT cells, we identified Il13, Satb1, Cd53, Ntrk3, Ctla4, and others (figure 4E,F). The genes with the highest downregulation in Cblb^−/− cluster 2 cells are, for example, the unclassified genes Gm42031, Gm47283 and Ccl4, Irf8, or Coro2a. Gene Ontology (GO) analysis of all cluster 2 DEGs (P_adj <0.05) revealed that the genes were most significantly enriched for the Biological Processes ‘T cell activation’, ‘JAK-STAT cascade’, and ‘positive regulation of cytokine production’ (figure 4G). Highly enriched molecular functions were ‘cytokine receptor binding’ and ‘receptor regulatory activity’, enriched cellular components (CCs) are ‘mitochondrial outer membrane’, ‘external side of plasma membrane’ and ‘outer membrane’. In addition to GO analysis, we also looked at Enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The most highly upregulated pathways are the ‘JAK-STAT signaling pathway’, ‘Cytokine-cytokine receptor interaction’ and ‘TNF signaling pathway’ (figure 4H). Finally, we confirmed that Cbl-b and IL-9 were reversely regulated (online supplemental figure 6E). Similar Cblb expression levels were observed in all major clusters except for IL-9 positive cluster 2 (online supplemental figure 6B,D). The latter exhibited significantly lower Cblb expression levels than the remaining cell clusters (online supplemental figure 6B,C,F).

Taken together, the sequencing approach revealed that high Cblb expression limits the polarization of Th9 cells by limiting Il9 expression. Consequently, deficiency for Cblb in naïve T cells favors Th9 cell polarization and leads to the generation of Th9 cells that show a different gene expression profile and upregulation of pathways associated with T cell activation or cytokine stimulation. Therefore, we next wondered if Cblb^−/− Th9 cells might also be functionally different from WT Th9 cells.

Based on our in vitro results, we next investigated whether increased differentiation of Th9 cells from Cblb^−/− versus WT naïve T cells affected their ability to influence tumor growth. As a tumor-antigen cancer model, we crossed OT-II mice with Cblb^−/− mice to generate OVA-specific Cblb^−/− Th9 cells and injected C57BL/6 mice with the B16F10-OVA melanoma cell line. At days 5 and 10 of the tumor experiments, 3×10ˆ6 in vitro differentiated OT-II Th9 cells or OT-II*Cblb^−/− Th9 cells were intravenously injected into the tumor-bearing mice (figure 5C). Similar to WT and Cblb^−/− Th9 cells, naïve OT-II T cells generated from a Cblb^−/− background exhibited improved Th9 cell differentiation (when compared with WT OT-II Th9 cells), as evaluated by intracellular staining for IL-9 and subsequent flow cytometry analysis (figure 5A/B; after 48 hours of Th9 cell polarization, frequency of IL-9⁺ cells was almost doubled with 20% in OT-II Th9 cells vs 40% in OT-II*Cblb^−/− Th9 cells). Injection of WT OT-II Th9 cells slightly decreased tumor growth (at least in this dosing schedule) but did not prolong the survival of tumor-bearing mice (figure 5D,E). In contrast, Th9 cells generated from OT-II*Cblb^−/− mice showed significantly inhibited tumor growth when compared with the group that received WT OT-II Th9 cells (figure 5D). In line with their increased antitumor reactivity, mice receiving OT-II*Cblb^−/− Th9 cells had significantly improved survival compared with mice that received Cblb-proficient OT-II Th9 cells (figure 5E). In order to understand whether Th9 cells directly kill B16 tumor cells or if the effect is indirect through other immune cells, we co-cultured in vitro generated Th9 cells with a 1:1 mixture of B16-WT and B16-OVA tumor cells, labeled with different concentrations of CFSE. Using this approach, we could not detect any significant difference in the killing activities of either OT-II Th9 or OT-II*Cblb^−/− Th9 cells against B16-WT or B16-OVA cells (online supplemental figure 7), which indicates an indirect effect rather than direct tumor cell killing. Our data show that Cblb^−/− Th9 cells, in contrast to WT Th9 cells, can delay melanoma growth, presumably by acting on other immune cells. However, the exact role of IL-9 in this context remains unclear. Therefore, we next defined the role of IL-9 in the tumor rejection phenotype of Cblb^−/− animals.

Cblb^−/− mice spontaneously reject tumors, an observation mainly attributed to tumor-specific CD8⁺ T cells that are resistant to Treg-mediated suppression.39 Our data now highlight that in addition to CD8⁺ T cells, Th9 cells and, more specifically, IL-9 are involved in this tumor rejection phenotype. Therefore, we subcutaneously injected either WT or Cblb^−/− mice with the melanoma cell line HCmel3 and measured Il9 mRNA expression in the tumor tissue by quantitative PCR (qPCR) at day 30.43 Cblb^−/− mice showed significantly higher Il9 expression levels in the tumor tissue in vivo, further suggesting a potential role for IL-9 in suppressing tumor growth in Cblb^−/− mice (figure 6A). In order to confirm this hypothesis, we treated HCmel3 injected Cblb^−/− mice intravenously on days −1, 0, 1 and every following third day with a neutralizing anti-IL-9 antibody (figure 6B). As expected,39 Cblb^−/− mice injected with control IgG showed significantly decreased tumor growth (figure 6C) and smaller tumors (figure 6D,E). In contrast, blocking IL-9 re-established cancer growth, even though WT mice still had the most extensive tumors, indicating that IL-9-independent pathways are also involved in cancer growth suppression in Cblb^−/− animals (figure 6C–E). To understand differences in general tissue layout of the tumors in WT, Cblb^−/−, and anti-IL-9 treated Cblb^−/− mice, we performed a H&E stain and CD45 staining to visualize the tumor infiltrating leukocytes (figure 6F). While we observed big tumors with typical melanoma appearance in WT mice, in Cblb^−/− mice tumors were eradicated to a high degree and hence structurally different. Despite being not as small, tumors in anti-IL-9 treated Cblb^−/− mice structurally resembled tumors in Cblb^−/− mice. CD45 staining revealed a low degree of immune cell infiltrate in tumors of WT mice, while a high degree of CD45 expression in tumors of Cblb^−/− mice indicated a high degree of immune cells infiltration at this time point. CD45 expression in anti-IL-9 treated Cblb^−/− mice were intermediate.

Next, to characterize these infiltrates in more detail, we isolated tumor-infiltrating lymphocytes (TILs) and mapped their composition by flow cytometry (figure 6G). While there could not be observed any differences for the overall frequency of CD45⁺ immune cells infiltrating the tumors by FACS, the frequency of T cells was significantly higher in tumors isolated from Cblb^−/− mice compared with the one from WT mice (figure 6G). In-depth analysis of the infiltrating T cells revealed a higher CD8⁺/CD8^–Foxp3⁺ ratio and a lower CD4⁺/CD8⁺ ratio in Cblb^−/− mice than in WT mice, indicating the dominance of CD8⁺ T cells over Treg (Foxp3⁺) and CD4⁺ T cells within all CD3⁺ T cells (figure 6G). Blocking of IL-9 led to a partial reversal of these effects, lowering the CD8⁺/CD8^–Foxp3⁺ ratio and increasing the CD4⁺/CD8⁺ ratio, indicating an involvement of IL-9 in regulating the composition of T cell subsets by modulating the frequency of specific T cell subsets such as CD8⁺ T cells. Taken together, these data confirm the in vitro observation that knocking out Cbl-b leads to an increased expression of IL-9 in the tumor microenvironment and that IL-9 is at least partly responsible for the tumor rejection phenotype of Cblb^−/− mice.