WHAT IS ALREADY KNOWN ON THIS TOPIC

In humans, a polymorphism in PTPN22 is highly associated with the onset of autoimmunity. Accordingly, PTPN22 has been established as a negative regulator of TCR signaling, and PTPN22 knockout (KO) mice are better at controlling syngeneic tumors. However, T cell-based therapies do not benefit from PTPN22 deletion, begging the question of what other immune compartments might regulated by PTPN22 and important in immune-mediated tumor control.

WHAT THIS STUDY ADDS

Here, we use a conditional KO mouse model to demonstrate that PTPN22 is a negative regulator of dendritic cells (DCs), resulting in improved DC functions and ultimately driving improved tumor control and responses to anti-PD-L1 therapy.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Broadly, our work highlights the potential to identify novel targets for immunotherapy from loss-of-function single-nucleotide polymorphisms associated with autoimmunity. Further, our study underscores the need to understand lineage-specific effects of targets of interest to better understand the underlying biological context and develop relevant clinical therapies. Lastly, we report a novel regulator of DC biology and demonstrate the potential of modulating the myeloid compartment to augment T cell-mediated antitumor immunity.

Introduction

In autoimmunity, self-tolerance is lost, and the immune system inflicts damage on healthy tissue, while in cancer, a lack of effective immunity against cancerous cells allows tumors to persist.1 This inverse relationship is suggestive that the mechanisms of pathology in one disease could serve as therapeutic interventions in the other. This is further highlighted by the observation that immune checkpoint blockade (ICB) treatment of cancer patients can result in toxicities termed immune-related adverse events (irAEs) that resemble autoimmunity.2 3 Importantly, across tumor types, the onset of irAEs is associated with increased survival in patients treated with ICB.4–6 Thus, understanding the development and consequences of autoimmune-like sequelae in patients treated with ICB can be informative.

Loss of function polymorphisms in the genes encoding PD-1 and CTLA-4 have been associated with increased incidence of a variety of autoimmune diseases.7 8 It is of interest that the proteins encoded by these autoimmunity risk genes have become drug targets for cancer immunotherapy. Broader single-nucleotide polymorphism (SNPs) identified from autoimmune genome-wide association studies (GWAS) have also been shown to be predictive of the development of irAEs and clinical response to ICB therapy.9–11 Taken together, these observations prompt the notion that mining autoimmune GWAS data to identify loss of function polymorphisms in immune regulatory genes might provide an opportunity to identify new therapeutic targets for cancer immunotherapy.

A germline SNP in the gene encoding the tyrosine-protein phosphatase nonreceptor type 22 (PTPN22) is attributed with the greatest risk for autoimmune disease outside polymorphisms in the human leucocyte antigen (HLA) locus.12–15 This SNP in PTPN22 has also been shown to be protective against the development of cancer, and patients with cancer bearing this SNP are more likely to respond to ICB therapy.16 17 Studies in PTPN22 global knockout (gKO) mice have demonstrated improved tumor control in a variety of models and have established PTPN22 as a negative regulator of T cell receptor (TCR) signaling.16–21 However, deletion of PTPN22 in T cells did not improve CAR T cell efficacy and has been shown to promote T cell exhaustion, suggesting that immune-mediated tumor control in PTPN22 deficient mice cannot be solely explained by a T cell-intrinsic effect.22 23

To be able to investigate a functional role of PTPN22 in distinct immune cell compartments, we developed a novel PTPN22 conditional KO (cKO) mouse model. Because of high expression of PTPN22 in various myeloid cells including dendritic cells (DCs) and given the critical role of DCs in the generation of tumor antigen-specific CD8+ T cell responses, we crossed PTPN22fl/fl mice to CD11c-Cre transgenic mice for DC-specific deletion.24–26 We showed that PTPN22 is a negative regulator of DC-mediated antitumor immunity. PTPN22 cKO mice showed increased tumor control, augmented tumor antigen-specific CD8+ T cell responses, and an expanded proliferating pool of CD103+ DCs. PTPN22 cKO DCs were more sensitive to Flt3L-mediated proliferation in vitro and demonstrated improved antigen processing and presentation. Together, these data highlight the therapeutic potential of targeting host PTPN22 as a potential cancer therapeutic.

Materials and methodsMouse models

All of the animal experiments were performed with mice on the C57BL/6 background. PTPN22fl/fl mice were generated exclusively for our laboratory by Cyagen. Briefly, loxP sites, along with self-deletion anchor sites, were inserted into the PTPN22 gene locus of mouse embryonic stem cells (ES), flanking exons 2–8. ES clones N-1H7 and N-2B5 were injected into C57BL/6 albino embryos, which were then reimplanted into immunocompromised CD-1 pseudo-pregnant females. Desired animals were identified by their coat color, and their germline transmission was confirmed by breeding with C57BL/6 females and subsequent genotyping of the offspring. These mice were subsequently bred with C57BL/6 mice over several generations to ensure the proper genetic background.

For conditional deletion in DCs, PTPN22fl/fl mice were bred with CD11c-Cre mice (Jackson Laboratory, Strain #: 008068). Mice breeding pairs always included a Cre+ (heterozygous) and a Cre− (wild type) mouse, both on the PTPN22fl/fl background, to ensure mixed litters for experiments. Littermates were used in experiments to reduce the impact of genetic drift and gut microbiome differences, as has been previously shown. Mice aged 6–12 weeks were used in all of the experiments, ensuring age-matching between experimental groups. Both male and female mice were used and indicated as necessary.

CD11c+ cell enrichment

Spleens collected from mice were minced in unsupplemented RPMI media, passed through a 70 µM cell strainer (Corning, 352350), and washed with ice-cold PBS. Following RBC lyse of single-cell suspensions (BD # 555899), CD11c+ cells were enriched using the CD11c MicroBeads kit from Miltenyi (130-125-835) per the manufacturer’s protocol.

qRT-PCR

To verify the cKO of PTPN22, CD11c+ cells were enriched utilizing CD11c MicroBeads for MACS purification. Following the manufacturer’s protocol, RNA was isolated using a Qiagen RNeasy Plus Micro Kit (74034) and subsequently quantified on a NanoDrop One (Fisher Scientific, 13-400-525). cDNA was generated from isolated RNA using the High-Capacity cDNA Reverse Transcription Kit (Fisher 4368814). The quantification of mRNA was conducted using the TaqMan gene expression master mix (Fisher, 4369514) on a StepOnePlus Real-Time PCR System (Applied Biosystems,4376600) in MicroAmp Fast Optical 96-Well reaction plates (Applied Biosystems, 434907). Each sample was run in triplicate wells, and the CT of those wells was averaged before expression levels were calculated by the following formula: DCT=CTGapdh– CTGene of interest; expression level= 2DCT. PTPN22 primers were designed through NCBI Primer-BLAST tool (forward: cattgtcatggcatgtatgga; reverse: ggatatagaaaaggggccaaa) and GADPH was used as a control (forward: agcttgtcatcaacgggaag; reverse: tttgatgttagtggggtctcg) and Roche’s Universal Probes were used (probe 2 and 9, respectively).

Syngeneic tumor models

Tumor models consisted of murine cancer cell lines syngeneic to the C57BL/6 background. B16F10 cells were engineered to express the model antigen SIYRYYGL (SIY) along with dsRed using retroviral transduction, and the murine colorectal cancer cell line MC38 was similarly transduced to express SIY along with GFP. These cell lines are referred to as B16.SIY and MC38.SIY, respectively. Using the respective fluorophores, these cell lines were routinely checked and sorted via fluorescence-activated cell sorting to ensure uniform expression of the model antigen to achieve desired immunogenicity.27 28

All cell lines were cultured in Dulbecco’s Modified Eagle Medium high in glucose and pyruvate (ThermoFisher, 11995073) supplemented with 5% FBS and 1% MOPS buffer, MEM Non-Essential Amino Acids (ThermoFisher, 11140050), and Penicillin-Streptomycin (ThermoFisher, 15140122). Cells were grown in incubators at 37°C and 8% CO2 and were never passaged more than twice prior to injection to minimize genetic drift. Cells were routinely checked for mycoplasma using ATCC Universal Mycoplasma Detection Kit (30–1012K) per the manufacturer’s protocol.

For tumor inoculation, tumor cell lines were harvested and resuspended at a concentration of 10–20 million cells per mL in PBS. Each mouse was injected with 100 µL of cell suspension, resulting in the injection of 1–2 million cells, depending on the experiment. B16.SIY cells were injected intradermally or subcutaneously and MC38.SIY and MC38 cells were injected subcutaneously. Tumor measurements began on day 10 post-inoculation and were measured using calipers three times a week until the end of the experiment. Tumor volume was calculated using the following formula: width×height×length=volume (cm3). Age-matched littermates were used to minimize potential confounders for all experiments and no blinding was imposed for tumor growth curves.

Tumor processing

Tumors, spleens, and tumor-draining lymph nodes (tdLNs) were harvested at the indicated time points. Tissues from non-tumor-bearing mice were minced on ice and passed through 70 µM filters. Spleens were RBC-lysed with per manufacturer’s protocol (BD, 555899). For tumors and organs from tumor-bearing mice, tissues were minced on ice in digestion buffer containing 1 mg/mL Collagenase IV (Sigma, C5138-1G), 100 µg/mL Hyaluranidase V (Sigma, H6254-500MG), and 200 µg/mL DNAse I Type IV (Sigma, D5025-150KU) and incubated at 37°C while shaking for 30 min. Tumor immune infiltrates were isolated by Ficoll-Paque (Fisher, 17144003) density gradient centrifugation.

Flow cytometry

For SIY-pentamer staining, samples were first incubated with Fc Block (BioLegend, 101302) and SIY-pentamer (ProImmune, F1803-2B - 150 test R-PE) in Brilliant Stain Buffer Plus (BD, 566385) diluted 1:3 in PBS for 30 min in the dark. The master mix was then added to this solution, and the samples were incubated for an additional 30 min. After washing, cells were fixed and permeabilized for intracellular staining with the FOXP3 Transcription Factor Staining Set (ThermoFisher, 00-5523-00). Cells were then stained with the intracellular master mix in the dark, overnight, and at 4°C. The antibodies used are outlined in table 1. All spectral flow cytometry was run on 5L Cytek Aurora machines and analyzed in FlowJo version V.10.9.0, and gating strategies are shown in online supplemental figure 1.

Table 1

List of flow cytometry panels used

scRNAseq

Mice were injected subcutaneously with MC38.SIY-GFP cells (2×106), and tumors were harvested from two different mice at day 21. Single-cell suspensions of the tumor immune infiltrate were stained, and live CD45+ cells were sorted on a MACSQuant Tyto Cell Sorter using two regular speed cartridges. 10,000 cells from each biological replicate were loaded into the 10x Genomics Sequencer and sequencing was performed by the 10x Genomics Core at the University of Chicago. Reads were aligned, and feature-barcode matrices were generated using the 10x Genomics Cell Ranger software. Cells expressing a high percentage of mitochondrial RNA and lowly expressed genes were excluded. Data were normalized using Seurat’s SCTransform function. Principal component (PC) analysis was performed first, followed by dimensionality reduction using RunUMAP on the top 30 PCs. The k-nearest neighbors were calculated, and the SNN graph was constructed using Seurat’s FindNeighbors function. Pseudo-bulk differential expression via DESeq2-LRT method was performed to identify the top differentially expressed genes in each cluster. Based on the top differentially expressed genes, each cluster was annotated.

Statistical analysis

Data were managed and processed using Microsoft Excel or RStudio V.2022.07.0 and R V.4.2.1 (2022-06-23), and statistical analysis was performed using GraphPad Prism V.10. Tumor growth curves were analyzed by two-way analysis of variance (ANOVA), using Sidak’s multiple comparisons tests when appropriate, and are plotted as mean±SEM. Violin plots were similarly analyzed by two-way ANOVA and Sidak’s multiple comparisons test and by calculating the area under the curve and comparing it with an unpaired, two-tailed Student’s t-test. Bar graph comparisons were analyzed by unpaired, two-tailed Student’s t-test (parametric) or Wilcoxon-Mann-Whitney test (non-parametric) depending on normality and are plotted as mean±SD. Spearman’s correlations were used for multivariable analysis. Raw FCS files were unmixed using Cytek’s SpectroFlo software. For processed flow cytometry data, all counts, percentages, and mean fluorescence intensities (MFI) were calculated using FlowJo V.10.9.0.

ResultsConditional deletion of PTPN22 in DCs does not affect immune homeostasis but improves antitumor immunity

Previous work carried out in the PTPN22 gKO mouse model established PTPN22 as a negative regulator of antitumor immunity and immune cell activation. However, the improved antitumor immunity in these mice cannot be solely explained by the role of PTPN22 in T cells.16 17 21–23 We thus developed a PTPN22 cKO mouse model in order to address what other immune compartments might be driving improved tumor control under the regulation of PTPN22 and chose to focus on DCs. To this end, PTPN22fl/fl mice were bred with CD11c-Cre transgenic mice and designated PTPN22 cKO mice (online supplemental figure 2A). Successful and selective loss of PTPN22 in CD11c+ cells was confirmed by qPCR (online supplemental figure 2B).

We began by immune profiling naïve PTPN22 cKO mice to determine whether spontaneous immune cell activation might occur over time.18 Young, naïve PTPN22 cKO mice did not reveal any differences in T cell or DC phenotypes. Furthermore, no differences were observed when mice were aged 20 months (figure 1A and online supplemental figure 2C). In addition, mice remained healthy with comparable body weights to littermate control mice. These results argue that spontaneous autoimmunity did not occur with DC-intrinsic deletion of PTPN22.

Figure 1

Deletion of PTPN22 in DCs does not alter the immune profiles of naïve mice but drives increased spontaneous tumor control. (A) Schematic of PTPN22fl/fl CD11c-Cre+ and Cre− age-matched, littermate female mice aged for 20 months for analysis, including body weight, spleen weight, and percent of CD45+ and CD3+ T cells in spleens and lymph nodes (LNs). N≥7 with control Cre− mice in gray and cKO Cre+ mice in red. Data from spleens are represented as circles and from LNs as triangles. (B) Gating strategy for T cells and DCs. (C) Female mice were injected intradermally with 1 million B16.SIY (red) and male and female mice were injected subcutaneously with 1 million MC38.SIY (blue) cells and allowed to grow until endpoint. For B16.SIY, both N1H7 and N2B5 founder clones of the PTPN22fl/fl x CD11c-Cre+ mice were used with N≥15 per group and euthanized at day 27. For MC38.SIY, only the N1H7 clone of the PTPN22fl/fl x CD11c-Cre+ mice were used with N≥12 per group and euthanized at day 32. Pooled data from independent experiments are plotted for B16.SIY (N=3) and MC38.SIY (N=2). Control PTPN22fl/fl x Cre− controls always presented as black/gray. Bar graph comparisons were analyzed by unpaired, two-tailed Student’s t-test and plotted as mean±SD. Tumor growth curves were analyzed by two-way ANOVA, using Sidak’s multiple comparisons tests when appropriate, and are plotted as mean±SE. Survival curves were analyzed by Kaplan-Meier survival analysis with p values for Gehan-Brewslow-Wilcoxon test reported. Significance is defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; cKO, conditional knockout; DCs, dendritic cells.

We next investigated whether PTPN22 cKO mice displayed improved spontaneous tumor control when challenged with transplantable tumor models in vivo. In order to be able to measure the effects on host CD8+ T cell responses, we used the B16 and MC38 cell lines transduced to express the model antigen SIY.28 Importantly, PTPN22 cKO mice displayed markedly diminished tumor growth, resulting in improved mouse survival (figure 1C). These data suggest that deletion of PTPN22 in the DC compartment is sufficient to drive improved immune-mediated tumor control in vivo.

To begin to evaluate potential mechanisms, given the targeted deletion of PTPN22 in DCs, we began by analyzing tumor-infiltrating DCs. Tumors from PTPN22 cKO mice had an overall increase in the number of infiltrating CD45+ immune cells. In both the B16.SIY and MC38.SIY models, an increased number of DCs per gram of tumor was also observed (figure 2A and online supplemental figure 3A). Both the CD103+ and CD11b+ DC subsets trended toward an increase in numbers, with the strongest statistical difference observed with MC38.SIY tumors (online supplemental figure 3A). As a first measure of DC activation status, we assessed the level of expression of MHC II per DC. Interestingly, only CD103+ DCs had higher levels of MHCII in PTPN22 cKO DCs, with no significant differences observed in CD11b+ DCs (figure 2B).

Figure 2

Tumors from PTPN22 cKO mice have increased activated DCs and T cells, and tumor control is dependent on CD8+ T cells. B16.SIY tumor immune infiltrates collected at day 27 from PTPN22fl/fl CD11c-Cre+ and Cre− age-matched, littermate female mice showing (A) Dendritic cells (DCs) and CD103+ and CD11b+ DCs per gram of tumor. (B) The mean fluorescence intensity (MFI) of MHCII and percentage of CD103+ and CD11b+ DCs positive for Ki67. (C) Tumor immune infiltrating lymphocytes (TILS), CD8+ T cells, and antigen-specific CD8+ T cells (CD8+SIY+) per gram of tumor, along with representative gating strategies. (D) Activation and proliferation status are also shown as the percentage of CD8+SIY+ T cells positive for Ki67 and CD69. Lastly, the ratio of the number of CD8+ SIY+ T cells and regulatory T cells per gram of tumor is also shown. (E) Multivariable analysis showing the number of CD8+SIY+ T cells per gram of tumor by the number of CD103+ DCs per gram of tumor. Circle size represents tumor weight and end point, with red circles representing cKO Cre+ mice and black circles representing control Cre− mice. Spearman R correlation values are reported. Pooled data from independent experiments are plotted for B16.SIY (N=3) with N≥15 per group. (F) Mice were injected weekly with 200 ng of Anti-CD8β (aCD8β) antibody or PBS. Control mice treated with PBS are represented by filled circles and solid lines, and aCD8β treated mice are represented as open triangles with dashed lines with N≥5 per group. Bar graph comparisons were analyzed using unpaired, two-tailed Student’s t-tests and plotted as mean±SD. Tumor growth curves were analyzed by two-way ANOVA, using Sidak’s multiple comparisons tests when appropriate and are plotted as mean±SE. Significance is defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; cKO, conditional knockout; DC, dendritic cell.

With the increased number and activation status of intratumoral DCs, we investigated whether this led to an increased number of tumor antigen-specific CD8+ T cells. This was evaluated by assessing CD8+ T cells positive for the SIY-Kb pentamer. In fact, the number of CD8+SIY+ T cells per gram of tumor was increased in PTPN22 cKO mice, and these T cells also showed markers of activation and proliferation (figure 2C). To gain additional insights, we analyzed the ratio of the number per gram of tumor of CD8+SIY+ T cells to Tregs, which provide a counter-regulation to control ongoing immune responses.29–31 In accordance with the improved tumor control observed in PTPN22 cKO mice, the CD8+SIY+ T cell/Treg ratio was also increased in these animals, indicative of a shift toward an antitumor immune response in tumor microenvironment (TME) of PTPN22 cKO mice.

CD8+ T cell responses are particularly dependent on CD103+ DCs for activation and support. To understand the relationship between these two cell types in our model, we performed a multivariable analysis that revealed a strong correlation between the number of CD8+SIY+ T cells per gram of tumor and the number of CD103+ DCs per gram of tumor, which further correlated with decreased tumor volumes enriched in PTPN22 cKO mice (figure 2E). In order to determine whether CD8+ T cells were required for the improved tumor control seen in PTPN22 cKO mice, we depleted CD8+ T cells by administrating anti-CD8β mAb. This resulted in a complete loss of tumor control in both PTPN22 wild-type (WT) and PTPN22 cKO mice (figure 2F). Similar results for all of these experiments were seen in PTPN22 cKO mice challenged with MC38.SIY tumors (online supplemental figure 3). These results establish CD8+ T cells as the dominant effector cell type mediating improved tumor control when PTPN22 is deleted from DCs.

PTPN22 conditional deletion in DCs leads to improved early priming of tumor antigen-specific CD8+ T cells

Since PTPN22 cKO mice showed an increased number of tumor antigen-specific CD8+ T cells accumulated in the TME, we asked whether early T cell priming within the tdLN was also augmented in PTPN22 cKO mice. The tdLNs were collected on day 7 after B16.SIY tumor implantation intradermally, which is before tumors become palpable. Analysis of the tdLNs of PTPN22 cKO mice showed an increase in the overall number of DCs per LN, which was specifically driven by an expansion of CD103+ DCs but not CD11b+ or CD8α+ DCs (figure 3A). Further, only CD103+ DCs had an increase in the percentage of cells positive for the costimulatory markers CD80 and CD86 (figure 3B and online supplemental figure 4A). These results highlight a particular activation of CD103+ DCs on deletion of PTPN22, which represents the DC subset optimized for CD8+ T cell activation. We then analyzed CD8+SIY+ T cells, which indeed revealed an increase in this population per tdLN in PTPN22 cKO mice (figure 3C), and these T cells also expressed higher levels of proliferation and activation markers (online supplemental figure 4B). To quantify the functionality of primed CD8+SIY+ T cells in these mice, we performed IFN-γ ELISpot as well as an IL-2 ELISA from splenocytes, representing circulating SIY-specific CD8+ T cells following priming in LNs. These analyses showed that splenocytes from day 7 tumor-bearing PTPN22 cKO mice secreted more IFN-γ and IL-2 when exposed to their cognate peptide SIY ex vivo (figure 3C). Together, these results suggest that the increased number and activation status of CD103+ DCs led to a greater activation and expansion of CD8+SIY+ T cells.

Figure 3

PTPN22 cKO mice have improved priming evident in the tdLNs at day 7 post tumor inoculation. PTPN22fl/fl x Cre+ and Cre− aged-matched littermates were injected intradermally with 1 million B16.SIY cells. Spleens and tdLNs were harvested on day 7. (A) DCs in the tdLN were quantified as the total number per tdLN and subtyped as CD103+CD11b−, CD8α+CD11b−, and CD11b+CD103−. (B) Activation and proliferation status of CD103+ DCs showing the percentage of CD80, CD86, and Ki67 positive cells along with representative flow plots. (C) Representative flow plot showing identification of CD8+SIY+ T cells along with the number of CD8+SIY+ T cells per tdLN. Splenocytes were cultured overnight to quantify IFN-γ production via ELISpot and IL-2 production via ELISA of conditioned media. Pooled data are plotted from independent experiments (N=2) with N≥9 per group. Bar graph comparisons were analyzed by unpaired, two-tailed Student’s t-test and plotted as mean±SD. Significance is defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. DC, dendritic cell; tdLNs, tumor-draining lymph nodes.

PTPN22 cKO mice have an increase in proliferating, antigen-loaded intratumoral DCs

We next investigated aspects of DC biology that could have been potentiated on deletion of PTPN22. In order to have enough relevant DCs for analysis, we focused on the TME at a relatively late time point: day 20 following tumor implantation. Given the increased number of DCs observed, we stained for intracellular Ki67 expression as a marker of proliferation. Unexpectedly, we observed a large fraction of DCs that were proliferating within the TME. Moreover, the increase in overall DCs seemed to be primarily driven by an expansion of CD103+ DCs, as evidenced by an increase in the overall number per gram of tumor as well as an increase in CD103+ Ki67+ DCs (figure 4A). We also wanted to see if these DCs were taking up tumor-derived material that would include antigen. In the B16.SIY model, the model antigen SIY is coexpressed with dsRed. This system allowed us to analyze the presence of tumor-derived material acquired by DCs using dsRed fluorescence. Staining for dsRed showed an increase in the number of dsRed positive CD103+ and CD11b+ DCs (figure 4B). However, we failed to see an increase in the amount of dsRed per cell as indicated by MFI, suggesting no differences in the ability of PTPN22 WT and cKO DCs in antigen uptake. However, CD103+dsRed+ DCs had a more robust activated profile as compared with CD11b+dsRed+ DCs, as evidenced by CD80 and CD86 staining, arguing that the process of tumor antigen uptake was linked to DC activation (figure 4C). These trends were also observed in overall CD103+ and CD11b+ without taking dsRed status into account (online supplemental figure 5A).

Figure 4

Tumors in PTPN22 cKO mice have more proliferating and activated, antigen-loaded DCs. Intratumoral DCs from day 20 tumor-bearing, PTPN22fl/fl x Cre+ and Cre− aged-matched littermate mice were quantified. (A) Number of DCs, CD103+ DCs, CD11b+ DCs, and CD103+ Ki67+ and CD11b+ Ki67+ DCs per gram of tumor. (B) Representative flow plot showing identification of dsRed+ DCs. Quantification of the number and percentage of dsRed+ CD103+ DCs and CD11b+ DCs, respectively. (C) Activation profile of dsRed+ DCs measured as percentage of CD80 and CD86 positive cells. (D) Representative flow plots of CD103+ dsRed+ DCs positive for Ki67 and activated-Caspase3 (aCasp3) and the ratio of the number of Ki67 positive cells per gram of tumor over the number of aCasp3 positive cells per gram of tumor for CD103+dsRed+ and CD103+dsRed− DCs, respectively. (E) Correlation plots showing the percentage of CD80 and CD86 double-positive cells of CD103+ DCs by the percentage of dsRed positive (left) and dsRed negative (right) cells of CD103+ DCs, with Cre+ represented in red and Cre− represented in gray. R2 and p value for non-zero slope from a simple linear regression model is reported with 95% CI shading in gray within dotted lines. Pooled data from three independent experiments are plotted (N≥16 per group). Bar graph comparisons were analyzed by unpaired, two-tailed Student’s t-test and plotted as mean±SD. Significance is defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. cKO, conditional knockout; DC, dendritic cell.

We then sought to determine whether the DCs containing tumor-derived material were those that were proliferating as reflected by Ki67 staining. To control for the potential impact of cell death on the population, we additionally performed intracellular staining for activated-Caspase 3 (aCasp3) as a measure of apoptosis. Despite no significant changes in the percentage of aCasp3+ cells, we looked at the ratio of the number of Ki67+ cells over the number of aCasp3+ cells per gram of tumor. This ratio revealed that CD103+ dsRed+ DCs, but not CD103+ dsRed− DCs, were preferentially proliferating, resulting in the expansion of this population (figure 4D). In this case, we did not observe any changes in CD11b+ DCs regardless of dsRed status (online supplemental figure 5C). Moreover, similar trends were observed in overall CD103+ and CD11b+ DCs (online supplemental figure 5A). We then examined whether the acquisition of tumor-derived material was related to the expression of costimulatory molecules. Interestingly, the coexpression of the CD80 and CD86 was correlated with the presence of dsRed in CD103+ DCs only in PTPN22 cKO DCs, while activation status was correlated with dsRed acquisition in both PTPN22 cKO and WT CD11b+ DCs (figure 4E and online supplemental figure 5C). Together, these data show that the increased number of activated and antigen-loaded DCs in PTPN22 cKO mice is a result of their preferential proliferation within the TME.

PTPN22 cKO DCs show heightened sensitivity to Flt3L and display augmented antigen processing and presentation

Since our tumor data pointed toward an expansion of CD103+ DCs, we wondered whether these cells were more sensitive to proliferation signals. The Flt3–Flt3L interaction is crucial for the development and expansion of CD103+ DCs in vivo and has been studied as a method to increase intratumoral CD103+ DCs in various model systems. We first used single-cell RNAseq (scRNAseq) data from MC38.SIY tumor-bearing wild-type mice to probe whether Flt3 and Flt3L are expressed by cells within the TME. Interestingly, the CD103+ DC cluster had the highest expression of Flt3, with pDCs also expressing Flt3 at high levels (figure 5A). Moreover, CD8+ T cells were among the cells with the highest expression of FLT3L (figure 5B). This is in keeping with recent work from our laboratory showing that CD8+ T cells and Batf3-lineage DCs preferentially cluster in the TME.32

Figure 5

PTPN22 cKO DCs are more responsive to Flt3L-mediated proliferation and process and present more antigen in vitro. scRNAseq data were used to test for the expression Flt3 and Flt3L in MC38.SIY tumors. (A) A UMAP of cell clusters and (B) expression of Flt3 and Flt3L across all populations. (C) Splenic and lymph node-derived DCs were treated with 100 ng/mL of recombinant Flt3L (rFLT3L) for 48 hours, and Ki67 staining was assessed. Separately, splenic DCs were cultured in vitro and exposed to multiple concentrations of DQ-OVA substrate for 4 hours. (D) Antigen processing was assessed in CD103+ DCs, represented by MFI of hydrolyzed DQ-OVA normalized to cells not exposed to any substrate. Area under the curve (AUC) for the entire serial dilution was calculated. (E) A monoclonal antibody against OVA-loaded MHCI was used to calculate the percentage of cells positive for H2-Kb-OVA for CD103+ DCs. (F) The percent of H2-Kb-OVA positive cells was also compared in PTPN22 cKO CD103+ DCs and CD11b+ DCs and the correlation of DQ-OVA MFI by the percent of H2-Kb-OVA positive CD103+ DCs. R2 and p value for non-zero slope are reported with a 95% CI in gray within dashed lines. Experiments are representative of 3–4 independent experiments performed with both female and male mice with N=7 per group. Violin plots were similarly analyzed by two-way ANOVA and Sidak’s multiple comparisons test and by calculating the AUC and comparing it with an unpaired, two-tailed Student’s t-test. Significance is defined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; DC, dendritic cell; MFI, mean fluorescence intensities.

Given the presence of Flt3L within the TME and the expression of Flt3 by intratumoral DCs, we examined whether DCs from PTPN22 cKO mice showed improved sensitivity to Flt3L-induced proliferation. In order to obtain enough DCs for these experiments, we pooled splenic and lymph-node-derived CD11c+cells from PTPN22 WT and PTPN22 cKO mice. On stimulation in vitro with Flt3L, we saw an increase in the percentage of cells positive for Ki67 in DCs from PTPN22 cKO mice (figure 5C). These data suggest that PTPN22 negatively regulates proliferation signals in response to Flt3L in DCs.

While DCs within the TME of PTPN22 cKO mice did not show greater overall uptake of tumor-derived material, we wondered whether they would nonetheless show improved antigen processing and presentation. To assess these functions, we turned to the OVA (SIINFEKL) system in order to probe antigen uptake, processing, and presentation in vitro. Briefly, CD11c+ cells were enriched from the spleens of naïve mice and cultured in vitro with different OVA substrates. We first tested whether DCs in vitro uptake more antigen by culturing cells with OVA peptide conjugated to Alexa Fluor 647 (OVA-AF647). However, similar to what we observed in intratumoral DCs and dsRed uptake, we failed to see any differences in the intensity of OVA-AF647 between PTPN22 WT and cKO DCs, suggesting equal antigen uptake (online supplemental figure 6A). Next, we used the DQ-OVA substrate that only fluoresces on proteolytic degradation. Thus, this substrate can be used to study antigen processing with fluorescence representing proteolytic degradation in lysosomes following uptake. In this case, PTPN22 cKO DCs showed a greater intensity of DQ-OVA proteolytic degradation across a range of doses, suggesting increased antigen processing. Of note, this difference was observed in both CD103+ and CD11b+ DCs (online supplemental figure 6B and figure 5D). We also took advantage of the ability to stain for OVA peptide-loaded MHC I molecules with the use of monoclonal antibody specific for OVA-peptide bound H2-Kb. This revealed that PTPN22 cKO CD103+ DCs presented more antigen on MHCI compared with PTPN22 WT DCs (figure 5E). Importantly, this increase was only observed with CD103+ DCs and not CD11b+ DCs (figure 5F). These experiments indicate that the improved antigen-specific CD8+ T cell response in PTPN22 cKO mice is additionally facilitated by increased antigen processing and presentation by CD103+ DCs, leading to increased CD8+ T cell priming and activation.

Loss of PTPN22 in DCs improves response to anti-PD-L1 treatment

Given the augmented spontaneous antitumor T cell responses in PTPN22 cKO mice, we reasoned that these mice might show improved efficacy of checkpoint bloc