Introduction

Melanoma is the most lethal form of skin cancer, accounting for approximately 75% of skin cancer–related deaths^1,2. It develops through the malignant transformation of melanocytes, which proliferate uncontrollably and acquire metastatic potential³. Whereas numerous targeted small molecule inhibitors and immune-activating therapeutics are FDA approved to treat metastatic disease, only 50% of patients achieve durable response⁴. Thus, it is critical to better understand the development of melanoma to identify improved methods for diagnosis and treatment. Among the major drivers of melanomagenesis are activating somatic mutations in NRAS and BRAF, which are often coupled with loss-of-function mutations in tumor suppressors like TP53 and PTEN^5,6. However, these same genetic alterations are found in benign melanocytic nevi^7–10, implying that factors other than oncogenic melanocyte mutations are also necessary for malignant progression.

The immune system represents an obvious source for melanocyte-extrinsic determinants of melanoma. Multiple immune cell types, including natural killer cells and CD8⁺ cytotoxic T lymphocytes, destroy cancer cells directly, and this anti-tumor activity is widely thought to eliminate many nascent tumors before they become clinically detectable^11,12. Anti-tumor immunity is particularly relevant for melanoma, which can be successfully treated with immune checkpoint blockade (ICB) antibodies that act by derepressing existing clones of tumor-specific T cells^13,14. Accordingly, the capacity of mutant melanocytes to overcome or evade anti-tumor lymphocytes would presumably be crucial for their outgrowth in the skin.

Regulatory T (T_reg) cells, which are defined by expression of the transcription factor Foxp3, function as critical suppressors of lymphocyte activation in both homeostatic and disease contexts¹⁵. As such, they represent an intriguing candidate immune evasion mechanism for melanoma. Indeed, multiple lines of evidence link T_reg cell activity to melanoma outgrowth. Intratumoral T_reg cell accumulation has been associated with disease progression in the clinic, and T_reg cell depletion inhibits melanoma outgrowth in transplantable models of the disease^16–19. However, the role of T_reg cells during the early, premalignant stage of melanocyte expansion has not been examined.

Beyond anti-tumor immunity, the immune system can also influence tumor progression via tissue inflammation^12,20. Inflammatory responses are typically triggered by the release of activating cytokines and chemokines in peripheral tissues^21,22. This drives the recruitment of a variety of innate immune cells from the blood, among them neutrophils and Ly6C⁺ monocytes, the latter of which rapidly differentiate into inflammatory macrophages. Infiltrating myeloid cells further amplify the inflammatory milieu and release factors that promote tissue remodeling. A key component of this remodeling response is angiogenesis, a process in which initial vascular destabilization is followed by the sprouting of new vessels²³. Importantly, myeloid-driven inflammation and angiogenesis have both been shown to support tumor progression by providing the space and nutrients necessary for cancer cell growth^20,23.

The link between inflammation and cancer may be particularly relevant to melanoma because ultraviolet (UV) light, the primary environmental driver of this disease, is known to elicit robust cutaneous inflammation and vascular remodeling^24–26. UV is canonically thought to drive melanomagenesis by generating oncogenic mutations in melanocyte DNA^6,27–30. While UV-induced mutations clearly contribute to disease progression, emerging evidence indicates that UV may also promote disease via non-mutagenic pathways^31–35. Mouse models of melanoma have shown that a single dose of UV is sufficient to induce melanomagenesis with no significant increase in oncogenic UV signature mutations^36,37. Furthermore, acute high doses of UV (e.g., blistering sunburn) are a more significant risk factor for melanoma than chronic UV exposure^31,38–41, despite presumably generating less oncogenic DNA damage. In light of these data, it is tempting to speculate that inflammatory tissue remodeling induced by UV light might create local conditions that are conducive to melanomagenesis. Intriguingly, UV-induced inflammation is also associated with cutaneous T_reg cell expansion and the suppression of adaptive immunity in the skin^42–44. Hence, UV could support melanoma outgrowth by driving both tissue remodeling and immune suppression.

To study the interplay between incipient melanoma and cutaneous inflammation, we subjected an autochthonous murine model of melanoma (LSL-Braf^V600E;Pten^fl/fl;Tyr::CreERT2 mice) to three distinct inflammatory immune perturbations: transient T_reg cell depletion, acute UV-B irradiation, and 2,4-dinitrofluorobenzene (DNFB)-induced contact hypersensitivity. Each of these perturbations accelerated premalignant melanocyte outgrowth, which was unexpected given that both T_reg cell depletion and DNFB treatment markedly enhanced conventional T (T_conv) cell infiltration into the skin. Detailed analysis of each inflammatory response revealed a shared cellular and molecular signature comprising myeloid infiltration, characteristic cytokines and tissue remodeling factors, and vascular permeability. Importantly, pharmacological suppression of this inflammatory signature mitigated melanocyte expansion, indicating that it plays a central role in early disease progression. These results are consistent with a model in which oncogenic mutation synergizes with tissue destabilization to drive melanomagenesis.

Results

Accumulation of T_reg cells in early melanoproliferative lesions

To explore early interactions between oncogene-containing melanocytes and the cutaneous immune system, we employed an established autochthonous model of melanoma: LSL-Braf^V600E;Pten^fl/fl;Tyr::CreERT2 (BPT) mice⁴⁵. Topical treatment of these animals with 4-hydroxytamoxifen (4-HT) induces the expression of oncogenic Braf^V600E concomitantly with the deletion of the tumor suppressor Pten in melanocytes, which proceed over the ensuing ∼8 weeks to form melanoma. We focused on the glabrous skin of the mouse ear because it is conducive to intravital microscopy and because it closely mimics the architecture of human skin.

Our initial studies utilized BPT mice bearing an LSL-TdTomato allele, which generate TdTomato⁺ melanocytes after 4-HT treatment. This labelling strategy was selected to facilitate the imaging of melanocytes in situ and to enable detection of melanocytic material uptake by immune cells⁴⁶. BPT;LSL-TdTomato mice, along with Tyr::CreER;LSL-TdTomato controls lacking tumorigenic alleles, were subjected to 4-HT ear painting, and immune cells in the ear skin and the draining lymph nodes (dLN) were analyzed by flow cytometry 35 days later (Fig. 1A, Fig. 1 - supplement 1A-B). At this time point, BPT mice exhibited substantial skin darkening due to melanocyte outgrowth, but they had not yet developed obvious tumors (Fig. 1A). Multiple CD45⁺ immune cell types were enriched in BPT ears relative to wild type controls, including T cells (CD3⁺), monocytes (CD11b⁺Ly6G^-F4/80⁺MHCII^-CCR2⁺), and macrophages (CD11b⁺Ly6G^-F4/80⁺) (Fig. 1B). Conventional type I (CD11c⁺MHCII⁺F4/80^-CD103⁺CD11b^-) and type II (CD11c⁺MHCII⁺F4/80^-CD103^-CD11b⁺) dendritic cells (cDC1 and cDC2) were also more abundant in BPT skin (Fig. 1C). Infiltrating T cells were predominantly CD4⁺, with only small numbers of CD8⁺ T cells detected (Fig. 1B). Notably, approximately one quarter of CD4⁺ T cells in BPT skin expressed Foxp3, indicative of a substantive T_reg cell response (Fig. 1B). Increased T_conv and T_reg cell numbers were also observed in the BPT dLN, but these changes were more modest and fell short of statistical significance (Fig. 1 - supplement 1C-D).

Figure 1.

To assess the antigen presentation pathway underlying the enhanced T cell infiltration observed in BPT skin, we quantified TdTomato fluorescence in patrolling DC subsets. Little to no TdTomato uptake was observed in control mice, despite the fact that the unmutated melanocytes in these animals were TdTomato⁺. By contrast, ∼50% of cutaneous cDC1s and ∼80% of cDC2s in BPT ears contained melanocyte material (Fig. 1C). A substantial number of TdTomato⁺ cDC1s and cDC2s were also observed in the dLNs of 4-HT-treated BPT mice (Fig. 1 - supplement 1D), implying that both subsets were potentially capable of presenting mutant melanocyte antigen to naive T cells. To explore a potential basis for melanocyte sampling by DCs, we reconstituted BPT;LSL-TdTomato mice with CD11c-YFP bone marrow, enabling simultaneous two-photon imaging of YFP⁺ DCs and TdTomato⁺ melanocytes in premalignant ears (Fig. 1D). In these experiments, DCs initiated contact with melanocytes within three days of 4-HT treatment, and their interaction frequency trended upward over time as the melanocytes expanded (Fig. 1E). Hence, cutaneous DCs take up material from mutant melanocytes and traffic it to the dLN. Taken together with the increased accumulation of cutaneous T_conv cells in premalignant BPT skin, these results suggest that melanocyte specific T_conv and T_reg cells may be primed in the early stages of mutant melanocyte outgrowth.

T_reg cells restrain oncogenic melanocyte outgrowth in the skin

The influx of both T_conv and T_reg cells into BPT skin led us to hypothesize that anti-melanocyte responses mediated by the former might be suppressed by the latter. To test this hypothesis, we prepared BPT mice containing the Foxp3-DTR allele, which drives diphtheria toxin (DT) receptor expression in T_reg cells and thereby sensitizes them to transient depletion by DT⁴⁷. In our experiments, T_reg cells were depleted with three injections of DT delivered at 48-hour intervals. This treatment scheme reduced circulating T_reg cells to background levels for at least 10 days (Fig. 2 - supplement 1A-B). Cohorts of BPT;Foxp3-DTR animals were subjected to three different T_reg cell depletion regimens, starting six days before (early), two days after (intermediate), or eight days after (late) 4-HT painting (Fig. 2A). Surprisingly, none of these regimens inhibited 4-HT-induced melanoproliferation. Instead, we found that T_reg cell depletion, and in particular early T_reg cell depletion, enhanced the outgrowth response, which we visualized photographically and quantified by qRT-PCR of the melanocyte-specific marker Tyrp1 (Fig. 2B-C). These results indicate that T_reg cells constrain premalignant melanocyte expansion during a temporal window closely aligned with the induction of tumorigenic mutations.

Figure 2.

In light of these unexpected findings, we re-examined the role of T_reg cells in transplantable models of tumor growth. Subcutaneous (s.c.) implantation of syngeneic B16F10 melanoma cells is widely used to study anti-tumor immunity, and it has been applied previously to interrogate the effects of T_reg cells on tumor suppression^16,18,19. To implement early stage T_reg cell depletion in the B16F10 system, Foxp3-DTR and wild type control mice were treated with the three dose DT regimen described above starting six days before B16F10 implantation (Fig. 2D). Transient T_reg cell deficiency significantly inhibited B16F10 tumor growth (Fig. 2E), confirming that T_reg cells constrain anti-tumor immunity in this model. In line with this interpretation, flow cytometric analysis of infiltrating immune cells in s.c. B16F10 tumors revealed a robust increase in both CD8⁺ and CD4⁺ T_conv cells in Foxp3-DTR animals (Fig. 2 - supplement 1C-D). Next, we performed an analogous set of experiments in mice receiving intravenous injections of B16F10, which leads to the formation of multifocal “metastatic” tumors in the lungs (Fig. 2F). T_reg cell depletion prior to B16F10 injection inhibited cancer outgrowth in this context (Fig. 2G), as well. We conclude that T_reg cells can both promote and antagonize incipient tumor formation in the skin, with the autochthonous BPT model revealing protective functions during early transformation that are either absent or substantially less important in transplantable B16F10 systems.

T_reg cell depletion dramatically alters the cutaneous T cell-DC axis

To investigate how T_reg cell depletion promotes mutant melanocyte outgrowth in skin, we subjected BPT;Foxp3-DTR;LSL-TdTomato mice and BPT;LSL-TdTomato controls to early DT treatment (days -6, -4, and -2) and then assessed leukocyte content in the ear and draining lymph node (dLN) eight days after 4-HT painting (Fig. 3A, Fig. 3 - supplement 1A). T_reg cell depletion triggered a dramatic influx of multiple immune cell types into the skin, including T_conv cells, neutrophils (CD11b⁺Ly6G⁺), monocytes, macrophages, and DCs (CD11c⁺MHCII⁺F4/80^-) (Fig. 3B-C). Skin T_conv cell expansion was apparent in both CD4⁺ and CD8⁺ compartments, but was much more obvious among CD4⁺ T_conv cells, which accounted for most of the cutaneous T cells at this time point (Fig. 3B). Among DCs, both cDC1 and cDC2 numbers increased, with cDC2s an order of magnitude more abundant than cDC1s in both T_reg cell deficient and T_reg cell sufficient skin (Fig. 3C). In the dLN, we observed expansion of all myeloid subsets, with more modest changes in T cell numbers (Fig. 3 - supplement 1B). T cell activation, however, as measured by upregulation of CD44 and downregulation of CD62L, was significantly enhanced in both CD4 and CD8 dLN subsets (Fig. 3 - supplement 1C). Upon restimulation with PMA/ionomycin in vitro, almost all dLN CD44⁺ CD4⁺ T_conv cells expressed IL-4, with little to no expression of IL-17 or IFNγ (Fig. 3 - supplement 1D). This observation is in line with prior work indicating that T_reg cells constrain an autoimmune T cell response with T_H2 characteristics in the skin⁴⁸.

Figure 3.

Next, we measured TdTomato fluorescence in each immune cell type to assess uptake of melanocyte antigen. Tumor sampling of this kind was largely restricted to macrophages, cDC1s, and cDC2s (Fig. 3C, Fig. 3 - supplement 1B). While T_reg cell depletion did not boost the fraction of TdTomato⁺ cells in each subset, their absolute numbers rose markedly due to increases in total macrophages and DCs. Notably, TdTomato⁺ DCs were observed in the dLN as well as the skin, indicative of successful antigen trafficking (Fig. 3 - supplement 1B). Taken together with the increased accumulation of cutaneous T_conv cells described above, these results are consistent with enhanced T_conv cell priming and activation in T_reg cell deficient skin, which is in line with prior reports^47–49. What is more surprising is that this immune response failed to antagonize mutant melanocyte outgrowth in the BPT model.

T_reg cell depletion induces the recruitment of tissue remodeling myeloid cells

To better understand how T_reg cells influence the skin microenvironment, we subjected Foxp3-DTR mice, along with C57BL/6J controls, to the three dose DT regimen and then profiled CD45⁺ cells in the ear by scRNA-seq eight days later (Fig. 4A). UMAP analysis of the resulting data revealed drastic differences in immune composition, encompassing both lymphoid and myeloid cells (Fig. 4B). T_reg cell depletion dramatically altered the cutaneous T cell compartment, as expected. Whereas control skin was largely devoid of T_conv cells, both CD4⁺ and CD8⁺ T_conv cells were readily apparent in T_reg cell-depleted skin (Fig. 4B, Fig. 4 - supplement 1A). Gzmb and Fasl were robustly expressed among CD8⁺ T_conv cells (Fig. 4 - supplement 1B), indicative of differentiation into armed cytotoxic T cells. CD4⁺ T_conv cells, for their part, exhibited high levels of Gata3 and Il4, but low expression of Rorc and Tbx21, indicative of a T_H2 response. Cd69, Cd44, and Icos were highly expressed across both subsets, consistent with ongoing T_conv cell activation. Collectively, these results suggest that T_reg cell depletion unleashes a strong T cell response in the skin, potentially targeting self-antigens or the cutaneous microbiota.

Figure 4.

Seurat reclustering of Immgen-defined DCs revealed five clusters of cells whose fractional representations did not change substantially between C57BL/6J and Foxp3-DTR mice (Fig. 4 - supplement 1C). Although detailed analysis of individual genes revealed small increases in the expression of activation markers (CD86 and CD40) and class II MHC (H2-Ab1) in certain DC clusters (Fig. 4 - supplement 1D), these transcriptional differences were quite modest overall. This was in sharp contrast to the marked increase in total cutaneous DC numbers we observed in T_reg cell deficient skin (Fig. 3). Taken together, these results indicate that T_reg cell depletion generates predominantly quantitative rather than to qualitative changes in the cutaneous DC compartment.

Conversely, the effects of T_reg cell deficiency among monocytes and macrophages were more striking, with several entirely new subsets appearing in the skin at the expense of other populations (Fig. 4C). UMAP replotting of the monocyte/macrophage clusters in the initial all-cell analysis identified two groups of resident cutaneous macrophages in C57BL/6J mice (Fig. 4C, Fig. 4 - supplement 2A). The first (DTR-M-cluster 3) expressed genes characteristic of resident macrophage identity (Gpr34, Ms4a7, Pf4), debris clearance (Stab1, Cd93), lipid processing (Apoe), and tissue maintenance (Igf1), implying a role in skin homeostasis. The second subset (DTR-M-cluster 5) exhibited indices of endothelial proximity (Lyve1, Colec12) and scavenging activity (Cd209f, Cd209g, Colec12), consistent with a resident perivascular macrophage identity.

T_reg cell depletion markedly diminished these subsets while inducing the emergence of three additional clusters. The first of these (DTR-M-cluster 4) bore transcriptional signatures characteristic of inflammatory monocytes (Ly6a, Plac8, Sell, Trem3, S100a8) and interferon signaling (Ifitm6, Vcan, Isg15, Gbp2, Gbp4). The second cluster (DTR-M-cluster 2) exhibited high levels of transcripts involved in macromolecular processing (Lgmn, Fabp5) and tissue remodeling (Mmp14, Arg1). This subset also expressed Ccl24, an established chemoattractant for eosinophils, which is consistent with the overall T_H2 character of T_conv cells in T_reg cell-depleted skin. Importantly, DTR-M-cluster 2 cells also expressed weak but detectable levels of the interferon signaling module seen in DTR-M-cluster 4. Finally, the third cluster (DTR-M-cluster 1) exhibited strong interferon module expression and more modest levels of the inflammatory monocyte and tissue remodeling modules. Taken together, these results were consistent with the ongoing recruitment of inflammatory monocytes (DTR-M-cluster 4) into T_reg cell depleted skin and their differentiation into tissue remodeling macrophages (DTR-M-cluster 2) via an inflammatory transitional macrophage subset (DTR-M-cluster 1). Indeed, pseudotime analysis identified a predominant differentiation trajectory starting at DTR-M-cluster 4 and moving sequentially through cluster 1 and then cluster 2 (Fig. 4 - supplement 2B-C). Interestingly, this trajectory then connected cluster 2 with DTR-M-cluster 0, a macrophage population expressing indices of tissue remodeling (Mmp12), lipid metabolism and alternative activation (Ramp3, Slc27a3), and immunoregulation (Cd226, Klrb1b). Cluster 0 also expressed skin epithelial markers (Skint3, Hepacam2), consistent with skin residence or the recent uptake of material from skin resident cells. Notably, cluster 0 was well represented in both Foxp3-DTR and C57BL/6J skin, implying that it could be a more long-lived, terminal differentiation state for monocyte-derived macrophages.

Further analysis of the scRNA-seq data revealed elevated expression of stromal remodeling genes, including the hypoxia-induced transcription factor Hif1a and the matrix metalloprotease Mmp14, by the three monocyte derived subsets (DTR-M-clusters 1, 2, and 4) that emerge in the context of T_reg cell deficiency (Fig. 4D). These subsets also displayed increased levels of the angiogenic factors Vegfa and Pgf, which were only weakly expressed by resident macrophages (DTR-M-clusters 3 and 5). The inflammatory cytokine Il1b was robustly expressed by the inflammatory subsets and also by the mixed identity macrophage cluster (DTR-M-cluster 0). This expression pattern would be expected to translate into higher overall Il1b production after T_reg cell depletion, given the dramatic quantitative increase in cutaneous monocytes and macrophages we observed under these conditions (Fig. 3).

To confirm and extend these results, we subjected Foxp3-DTR mice and C57BL/6J controls to the three dose DT regimen and then measured the expression of key inflammatory factors by qRT-PCR five days later (Fig. 4E). T_reg cell depletion triggered strong upregulation of both Il1b and a second inflammatory cytokine, Il6 (Fig. 4F). We also documented increased expression of Ang2, which encodes a secreted factor that destabilizes the vasculature (Fig. 4F). Taken together with the scRNA-seq data described above, these results document a broad and rapid myeloid inflammatory response in T_reg cell deficient skin.

T_reg cells maintain vascular homeostasis during premalignant melanomagenesis

The upregulation of angiogenic and blood vessel remodeling factors that we observed in T_reg cell deficient skin suggested that these cells might promote vascular stability. To evaluate this hypothesis, we used two photon microscopy to compare cutaneous blood vessel architecture in T_reg cell depleted mice with that of healthy controls five days after DT treatment (Fig. 5A). In normal C57BL/6 skin, i.v. injected high molecular weight fluorescent dextran was largely constrained within the vasculature. In Foxp3-DTR mice, however, a substantial amount of injected dextran leaked into the surrounding dermal interstitium (Fig. 5B), indicative of vascular permeability. To quantify this vessel destabilization phenotype, we calculated the fraction of dextran-positive area using Z-projection images of the skin. This approach revealed a nearly two-fold increase in the scope of dextran diffusion in T_reg cell deficient skin, consistent with a role for these cells in maintaining vascular integrity (Fig. 5C). As an alternative quantification approach, we i.v. injected the mice with the vital dye Evans Blue and then assessed its dissemination into extravascular ear tissue 24 hours later (Fig. 5D). We again found that T_reg cell depletion induced a significant, 2-3 fold increase in blood vessel permeability (Fig. 5E).

Figure 5.

Next, we examined the interplay between vascular integrity and T_reg cells in the context of early tumorigenesis. BPT;LSL-TdTomato;Foxp3-DTR mice and BPT;LSL-TdTomato controls were subjected to early DT treatment, 4-HT painting, and two-photon imaging with blood vessel tracing performed 30 days after oncogene induction (Figure 5G). TdTomato⁺ melanocyte expansion was readily apparent in both experimental groups, although more pronounced in BPT;LSL-TdTomato;Foxp3-DTR animals, as expected. Importantly, the increased melanocyte load in T_reg cell deficient skin was accompanied by severe dextran leakage, which obscured vessel architecture in large regions of tissue (Fig. 5G-H). These results strongly suggest that T_reg cells and the inflammatory responses they keep in check control vascular integrity in the premalignant microenvironment.

UVB radiation drives cutaneous inflammation and premalignant melanocyte expansion

The link between cutaneous inflammation and melanoproliferation that we observed in the context of T_reg cell deficiency prompted us to investigate whether other inflammatory stimuli might affect the premalignant microenvironment in similar ways. UV light is the major environmental trigger of human melanoma³⁸. Beyond causing DNA damage, UV exposure also induces a potent inflammatory response characterized by macrophage activation, the upregulation of pro-inflammatory mediators, such as IL-1β and TNF ^24,50–54, and VEGF production, which contributes to a pro-angiogenic environment^55,56. Given the similarities between these processes and the effects of T_reg cell depletion in the skin, we hypothesized that UV exposure would accelerate tumor progression via analogous mechanisms. To this end, we irradiated the ear skin of BPT mice with 2 kJ/m² UVB three days after 4-HT treatment (Fig. 6A). This dose is known to accelerate melanoma progression in autochthonous models, and while it does induce DNA mutations, driver mutations causative for the outgrowth phenotype have not been identified³⁷. UVB markedly enhanced the expansion of mutant melanocytes, leading to increased ear darkening and elevated Tyrp1 expression within a month of oncogene induction (Fig. 6B-C). This outgrowth essentially mirrored the consequences of T_reg cell deficiency.

Figure 6.

To better understand the microenvironmental effects of UVB in this context, we used qRT-PCR to quantify the expression of inflammatory and angiogenic factors in wild type skin after UV irradiation (Fig. 6D, Fig. 6 - supplement 1A). Within 24 hours of UVB treatment, we observed upregulation of both Il1b and Il6, along with increased expression of Vegfa and Hif1a (Fig. 6E). Expression levels of Il1b and Il6 remained high seven days post-UVB (Fig. 6 - supplement 1B). These results indicated that UVB induces a rapid inflammatory and tissue remodeling response in the skin with molecular properties that are similar to the effects of T_reg cell depletion. To investigate the cellular basis for this response in more detail, we profiled immune infiltrates in the skin by flow cytometry seven days after UVB treatment (Fig. 6 - supplement 1C). Neutrophil, monocyte, macrophage, and DC levels were all markedly enhanced by UVB (Fig. 6 - supplement 1D). CD4⁺ T_conv cell and T_reg cell numbers also increased, which was consistent with prior reports^42,43,57,58, while CD8⁺ T_conv cells remained unchanged (Fig. 6 - supplement 1E). Hence, whereas certain aspects of UVB-induced inflammation mirrored those of T_reg cell-depleted skin, other features, most notably the presence of T_reg cells themselves, were distinct.

Next, we performed scRNA-seq to profile the diversity of immune cells (CD45⁺) in the ear skin seven days after 2 kJ/m² UVB irradiation. These cells were predominantly of myeloid lineage, consistent with the flow cytometric results described above (Fig. 6 - supplement 2A-B). The most striking UV-induced changes manifested in the monocyte/macrophage compartment, where two entirely new subsets appeared in irradiated skin (Fig. 6F, Fig. 6 - supplement 2C). These subsets bore a striking resemblance to the monocyte derived populations observed in T_reg cell deficient skin (Fig. 4). The first (UV-M-cluster 2) strongly expressed transcripts indicative of inflammatory monocyte identity (Trem1, Plac8) and interferon signaling (Vcan, Ifitm1, Ifitm6) and more weakly expressed transcripts associated with tissue remodeling (Spp1, Gpnmb, Mmp14) and lipid handling (Fabp5, Apoc2, Hilpda). This pattern was reversed in the second subset (UV-M-cluster 3), which expressed high levels of the remodeling and lipid handling modules and lower levels of the monocyte and interferon signatures.

Taken together, these results were consistent with UVB-induced recruitment of inflammatory monocytes from the blood, followed by their differentiation into inflammatory, tissue remodeling macrophages. Importantly, Il1b, Mmp14, Hif1a, Vegfa, and Pgf expression was strong across all UVB-induced subsets (Fig. 6G), suggestive of increased inflammatory and angiogenic activity. To assess this hypothesis directly, we performed intravital blood vessel tracing of ear skin and Evans Blue assays, eight and three days after UVB exposure, respectively (Fig. 6H). These experiments revealed significantly more blood vessel leakage in irradiated ear skin relative to mock-irradiated controls (Fig. 6I-J), consistent with the interpretation that UVB disrupts the cutaneous vasculature. Hence, while UVB-induced skin inflammation differs somewhat from the response to T_reg cell depletion, both feature the infiltration of tissue remodeling macrophages and vascular instability.

Contact dermatitis is sufficient to drive dysregulated melanocyte outgrowth

Having demonstrated that an environmental irritant with known immunomodulatory properties (UVB) could induce dysregulated melanocyte outgrowth, we next asked whether a topically applied chemical irritant would have similar effects. 2,4-dinitrofluorobenzene (DNFB) drives potent cutaneous inflammatory responses via covalent modification of skin proteins and other biomolecules, leading to oxidative stress, adaptive immune priming, and type 4 hypersensitivity⁵⁹. We were particularly interested in DNFB because of prior work indicating that it could strongly induce IL-1β and IL-6, a feature we had observed in both UVB-treated and T_reg cell deficient skin.

To induce DNFB dependent inflammation, we employed a classical approach in which mice are first sensitized with DNFB on their abdomen and then challenged on one of their ears five days later (Fig. 7A, Fig. 7 - supplement 1A). This protocol induced substantial swelling on the DNFB-treated but not the contralateral, vehicle-treated ear (Fig. 7 - supplement 1B), indicative of strong, DNFB-dependent type 4 hypersensitivity. Using qRT-PCR, we observed dramatic upregulation of Il1b, Il6, Hif1a, and Ang2 following DNFB elicitation (Fig. 7B). DNFB elicitation also induced marked vascular leakage, as measured by Evans Blue staining (Fig. 7C). Using flow cytometry, we documented a dramatic increase in immune cellularity within two days of DNFB elicitation, encompassing neutrophils, DCs, monocytes, macrophages, and both CD4⁺ and CD8⁺ T_conv cells (Fig. 7 - supplement 1C-D). Hence, DNFB-induced inflammation recapitulated key characteristics of the immunological responses seen in UVB-treated and T_reg cell-depleted skin.

Figure 7.

Next, we assessed the capacity of DNFB-induced type 4 inflammation to promote mutant melanocyte outgrowth. BPT mice were sensitized with DNFB, painted with 4-HT on both ears three days later, and then challenged with DNFB on one ear two days after that (Fig. 7D). DNFB-treated ears exhibited substantially more darkening and Tyrp1 expression than their contralateral counterparts (Fig. 7E), indicative of profoundly dysregulated melanocyte expansion. This DNFB induced effect was similar to but more robust than what we had observed in T_reg cell-depleted or UVB-treated mice, prompting us to speculate that it might be capable of promoting melanocyte outgrowth in non-oncogenic conditions. To test this hypothesis, we applied the same experimental protocol to LSL-Braf^V600E;Pten^fl/+;Tyr::CreERT2 (BT-Het) mice, which retain one copy of Pten (Fig. 7F). BT-Het mice fail to develop melanoma after 4-HT treatment, with melanocyte expansion arresting at the nevus stage⁴⁵. DNFB elicitation dramatically accelerated melanocyte outgrowth in these animals (Fig. 7G), suggesting that cutaneous inflammation can drive melanoproliferation even in the absence of tumorigenesis.

Finally, to determine whether cutaneous inflammation was the actual cause of DNFB elicitation-induced melanoproliferation, we treated 4-HT- and DNFB-painted BPT mice with dexamethasone (Dex), a powerful immunosuppressive anti-inflammatory drug. Dex was administered both by i.p. injection one hour before DNFB elicitation and in a topical manner concomitantly with DNFB (Fig. 7A, 7D, S7A). In the context of Dex treatment, we observed marked reductions in Il1b, Hif1a, and Ang2 expression in DNFB-painted ears (Fig. 7B), which were also less swollen than the corresponding ears on mice receiving no Dex (Fig. 7 - supplement 1B). Interestingly, Il6 levels were only modestly inhibited by Dex (Fig. 7B). Notwithstanding this mixed effect on inflammatory indices, we observed a substantial reduction in DNFB elicitation-induced melanocyte outgrowth in Dex-treated mice (Fig. 7E), which was accompanied by marked attenuation of the vascular leakage phenotype (Fig. 7C). Taken together, these data demonstrate that type 4 inflammation can drive robust outgrowth of melanocytes, and they also suggest that blood vessel remodeling plays a particularly important role in this process.

Discussion

The effects of acute inflammation on the premalignant phase of cancer progression are poorly understood, due in no small part to difficulties in studying early disease using standard transplantable tumor models. To address this issue, we combined the autochthonous BPT model with three inflammatory triggers: T_reg cell depletion, UVB irradiation, and type 4 hypersensitivity. Strikingly, all three perturbations markedly enhanced the outgrowth of mutant melanocytes, despite the fact that they are known to elicit cutaneous inflammation in mechanistically distinct ways. T_reg cell depletion drives systemic T_conv cell activation, leading to extensive infiltration of CD8⁺ T_conv cells, CD4⁺ T_H2 effector cells, and inflammatory myeloid cells into the skin^47,48,60. DNFB-induced hypersensitivity also drives T_conv cell and myeloid inflammation, but the response is more T_H1-skewed and constrained to the skin itself⁶¹. In contrast, UVB promotes robust cutaneous recruitment of myeloid cells along with CD4⁺ T_conv and T_reg cells, but not CD8⁺ T_conv cells^24,25,42,44. Although the distinct features of these three responses were readily apparent in our experiments, certain critical attributes were shared, including the robust accumulation of inflammatory monocytes and macrophages, increased expression of Il1b and Il6, and the destabilization of cutaneous vasculature, accompanied by the upregulation of tissue remodeling and angiogenesis programs. Hif1a, Vegfa, and Ang2 likely function together within this axis to drive vascular remodeling, promoting vessel destabilization, angiogenesis, and increased permeability. Importantly, reversal of inflammatory tissue remodeling with Dex inhibited DNFB-induced outgrowth of mutant melanocytes. Vascular destabilization and angiogenesis are thought to promote the progression of small tumors into more aggressive and metastatic states⁶². Our results indicate that this “angiogenic switch” may be initiated and/or activated at a much earlier, premalignant stage.

T_reg cells are widely thought to facilitate tumor growth by restraining T_conv cell responses against tumor-specific antigens⁶³. Our findings challenge this view by showing that T_reg cell depletion can enhance the outgrowth of mutant melanocytes during the early, premalignant phase of disease progression. This unexpected phenotype is likely caused by dysregulated T_conv cell responses against non-melanocytic self-antigens and/or cutaneous microbiota, leading to tissue damage, inflammatory myeloid infiltration, and the tissue remodeling response described above. Additional studies will be required to establish the precise chain of causation. At this stage, however, it seems fair to conclude that T_reg cells can exert both pro- and anti-tumorigenic effects on melanoma, depending on the stage of disease progression and on specific microenvironmental features that emerge through the course of tumor growth. In light of our findings, immunotherapeutic efforts centered on the modulation of T_reg cell activity should be evaluated with this more holistic conception of T_reg cell function in mind.

The inhibition of melanocyte outgrowth by T_reg cells in the BPT model is particularly striking given prior results from other labs (and our own data) showing that T_reg cells play a protumorigenic role in transplantable models of disease^16–19. Although transplantable systems are easier to implement and quantify, the injection of a bolus of cultured cancer cells, often with additional components like matrigel, disrupts the tissue interactions and stromal-immune crosstalk that define the microenvironment of nascent tumors. The capacity of autochthonous models to recapitulate these microenvironmental features is therefore critical for accurately representing the early, premalignant stage of disease.

It is becoming increasingly clear that UV light can promote melanocyte proliferation and melanoma independently of its capacity to induce oncogenic mutations. We and others have shown that a single dose of UVB is sufficient to promote melanomagenesis and that timing (UVB delivered immediately before or during the formation of driver oncogenes but not after) is critical for generating melanomagenic effects^{31,37,39–41}. Taken together, these findings suggest that UV enhances melanoma outgrowth in part by altering the local tumor microenvironment. Our present results lend further support to this idea by closely tying UVB-induced melanocyte expansion to an inflammatory immune response in the skin, which is consistent with prior work showing that acute UVB irradiation promotes the expansion of both oncogenically mutated and unmutated melanocytes in an inflammation dependent manner^64–66. Interestingly, two of these studies directly implicated inflammatory macrophages in the melanocyte proliferation response^65,66. Identifying the UVB-elicited factors underlying myeloid recruitment and also the molecular pathways by which these cells activate melanocytes could reveal new translational avenues for mitigating the oncogenic effects of UV. Notably, UVB-induced inflammation also leads to local T_reg cell expansion in the skin, which could facilitate premalignant outgrowth by restraining melanocyte-reactive T_conv cell activity. Although the relative importance of myeloid cells vis-a-vis T_reg cells in this context remains to be determined, our observations that T_reg cell depletion and DNFB-induced hypersensitivity both enhance early melanocyte expansion while simultaneously promoting T_conv cell activation and infiltration implies a more critical role for myeloid cells in this context.

Regular aspirin use has been associated with reduced melanoma incidence in postmenopausal women⁶⁷, implying that anti-inflammatory drugs could potentially be used as a chemoprevention strategy for this disease. While clinical findings like this must be interpreted cautiously, they are consistent with the causal role for inflammatory signaling in early melanomagenesis that our results support. That Dex suppresses DNFB-induced melanocyte outgrowth in our model provides a direct experimental correlate and motivates more targeted investigation of specific anti-inflammatory mediators as potential preventive agents in individuals with elevated melanoma risk.

Our findings that contact hypersensitivity can promote the expansion of non-malignant BT-Het melanocytes have interesting implications for benign hyperpigmentation conditions, such as post-inflammatory hyperpigmentation, Riehl’s melanosis, and melasma^68–74. These disorders can lead to significant psychosocial stress, and billions of dollars are spent annually on treatments to reduce excess pigmentation. Post-inflammatory hyperpigmentation and Riehl’s melanosis usually occur following T_conv cell dependent inflammatory conditions such as atopic dermatitis or allergic contact dermatitis, whereas melasma is associated with exuberant mast cell and T_H2 inflammation. In each case, the condition is defined by increased proliferation of melanocytes leading to increased epidermal pigmentation. In light of our results, it is tempting to speculate that a better understanding of BT-Het melanocyte outgrowth in the context of cutaneous inflammation could inform the treatment of both disorders.

While chronic inflammation is an established hallmark of cancer, acute inflammation is often presented as a critical component of anti-tumor immunity. Our results indicate that the cancer cell extrinsic consequences of acute inflammation can have the opposite effect by laying a stromal foundation for premalignant expansion. Whether pro- or anti-tumoral effects predominate in a given situation will likely depend on contextual features such as the structure of the tissue in question, the size and stage of the tumor, and the tone of local immune cells. Identifying these inflammatory determinants could facilitate improved cancer prevention and immunotherapy.

Methods

Mice

The animal protocols used in this study were approved by the Institutional Animal Care and Use Committee of Memorial Sloan Kettering Cancer Center. Unless specified, experimental mice were 8-10 weeks old. All comparative analyses were performed on age- and sex-matched cohorts, which were either male or female. Foxp3-DTR mice were kindly provided by A. Y. Rudensky. BPT (B6.Cg-Tg(Tyr-cre/ERT2)13Bos;Braf-tm1Mmcm;Pten-tm1Hwu/BosJ), C57BL/6J, Cd11c-yfp (B6.Cg-Tg(Itgax-Venus)1Mnz/J), and LSL-TdTomato (B6.Cg-Gt(ROSA)26Sor-tm9(CAG-tdTomato)Hze/J) mice were purchased from the Jackson Laboratory. All reported strains and combinations thereof were bred and maintained at Memorial Sloan-Kettering Cancer Center (MSKCC).

T_reg Cell Depletion

To achieve sustained T_reg cell depletion, a total of three intraperitoneal (i.p.) DT (25 mg/kg) injections were administered to Foxp3-DTR mice at 48-hour intervals. T_reg cell clearance was verified by flow cytometric quantification of Foxp3⁺ T cells in the blood.

UVB irradiation

Mice were anesthetized with isoflurane and laid prone on a warming blanket. They were then irradiated with UVB using a Daavlin hand-held broad band (280-320nm filter with sharp cutoff) lamp (5.2 mW/s). The lamp was metered before each irradiation to ensure consistency and distance from mice was maintained at 0 cm. Mock irradiation was conducted using a standard incandescent desk lamp which was confirmed to not emit UVB.

Subcutaneous B16F10 Implantation

500 B16F10 cells were resuspended in RPMI and mixed in a 3:1 ratio with growth factor–reduced matrigel. The sample was then subcutaneously (s.c.) injected in the flank of recipient mice. Subsequent primary tumor outgrowth was monitored by measuring tumor length (L) and width (W). Volume was then calculated as πLW2/6.

Cell Implantation for Metastasis

200,000 of luciferase expressing B16F10 cells were injected intravenously (i.v., tail vein) into recipient mice. Tumor growth was monitored weekly thereafter by bioluminescent (IVIS) imaging.

Prior to imaging, mice were anesthetized with isoflurane and their thoracoabdominal fur shaved using an electric clipper. D-luciferin (150 mg/kg) was injected retro-orbitally into each mouse, and bioluminescence recorded using an IVIS Spectrum Xenogen instrument (Caliper Life Sciences) equipped with Living Image Software v.2.50. A head-to-tail-base ROI (region of interest) of uniform size was applied to all subjects to measure the total flux (photons/second) of metastatic cancer cells.

4-HT Application

Mice were placed under isoflurane anesthesia, and 32 μg of 4-Hydroxytamoxifen (Sigma, H6278) in DMSO was delivered via micropipette to each ear. Isoflurane was maintained for 15-20 minutes to ensure unhindered absorption of the drug.

DNFB Sensitization and Elicitation

Mice were placed under isoflurane anesthesia, and 25 μL of 0.5% DNFB (1-fluoro-2,4-dinitrobenzene) in 1:4 olive oil:acetone was applied via micropipette to the shaved abdomen. The treated area was allowed to dry for 4 minutes. Hindleg nails were clipped to prevent subsequent scratching of the sensitized site (abdomen). 5 days later, elicitation was performed by applying 10 μL of 0.5% DNFB (1-fluoro-2,4-dinitrobenzene) in 1:4 olive oil:acetone to the right ear and the vehicle mixture to the left ear of each mouse, under isoflurane anesthesia. The treated areas were allowed to dry for 4 minutes before the mice were revived. Hindleg nails were clipped to prevent subsequent scratching of the elicited site (ear).

Dex Treatment of DNFB-Elicited Mice

One hour prior to elicitation, mice were administered Dex by i.p. injection (2.8 mg/kg, with 3:7 PEG400:PBS as vehicle). 0.45% Dex was also combined with the 0.5% DNFB solution (1:4 olive oil:acetone) applied topically to induce elicitation. Control mice received i.p. injection of the vehicle alone and were treated topically with 0.5% DNFB solution in 1:4 olive oil:acetone.

Evans Blue Assay

Isoflurane anesthetized mice were i.v. injected (retro-orbital) with 100 μL of 1% Evans Blue dye in PBS, filtered through 0.45 μm strainer. The mice were euthanized 1.5 hours later, at which point their ears were harvested and placed into 1 mL of formamide for 2 days at 56 °C. Evans Blue levels were then quantified as the difference between absorbance at 620 nm and 720 nm.

Ear Thickness Measurements

Two days post challenge, DNFB-elicited mice were anaesthetized with isoflurane, and their ear thickness was measured using the Dial Thickness Gauge (Mitutoyo, Manufacturer Part Number: 7300A).

Chemical Depilation

Ear skin was depilated 10-14 days prior to the initiation of any experimental protocol. Mice were first anesthetized with isoflurane. Mouse ears were coated with a thin layer of Nair^TM and rinsed with PBS following a 4-minute incubation. Ear skin was then dried using a kimwipe. Hind leg nails were clipped to prevent subsequent scratching of the chemically depilated site.

Whole tissue RNA isolation and Quantitative RT-PCR

Ears were harvested from euthanized mice and mechanically homogenized in TRIzol using razor blades, followed by bead lysis. RNA was isolated from the TRIzol mixture in accordance with the TRIzol Reagent User guide (Doc. Part No.15596026.PPS, Pub. No.MAN0001271). The resulting RNA was then purified using the Zymo Research RNA Clean & Concentrator kit per the manufacturer’s instructions. RNA purity and concentration were determined using a NanoDrop spectrophotometer. cDNA was synthesized from total RNA and then used as a template for qRT-PCR using the LunaScript® Universal One Step Kit on a StepOnePlus device (Applied Biosystems). mRNA expression levels were normalized to Gapdh expression and relative gene expression between samples was calculated using the -ΔΔCt method. Primer sequences are listed in Table 1.

Table 1.

Macroscopic Photography

A NikonZ6 camera fitted with Nikon 1.5 mm macro lens was used to capture all macroscopic images of mouse ears. Photographs were processed using Adobe Lightroom Classic (v14.1.1) and Microsoft PowerPoint. All compared images received identical processing.

Generation of bone marrow (BM) chimeras

BPT;LSL-TdTomato mice were lethally irradiated (900 cGy) using a cesium source and reconstituted with 4 × 10⁶ bone marrow cells derived from Cd11c-yfp donor mice. Reconstitution of the hematopoietic compartment was monitored in the blood starting 6 weeks after reconstitution.

Intravital Imaging of Murine Ears

For intravital imaging studies, mice were anaesthetized and maintained on 1.5% isoflurane in 100% oxygen for up to 2 hours. Each mouse was restrained on a stage warmer at 37°C (BioTherm Micro S37; Biogenics) and the ear was mounted onto a custom-made stage for imaging⁷⁵. Two-photon microscopy was performed on an Olympus FVMPE- RS system equipped with a 25× 1.05 NA water immersion objective and a tunable InSight laser. Images were acquired with Fluoview software (FVS31). For experiments using BPT-LSL-TdTomato::Cd11c-yfp mice, YFP and RFP were detected with 1020 nm excitation using the Yellow/Red (515–565 nm) filter, and second harmonics (collagen) were detected with 1020 nm excitation using the Blue/Green (460–500 nm) filters, separated by an LP 520 dichroic mirror. Overlapping images were assembled using the same acquisition software. For blood vessel tracing experiments using BPT-LSL-TdTomato mice, 950 nm excitation was used. FITC and second harmonics (collagen) was detected using the Blue/Green (460–500 nm) filter, and TdTomato was detected using the Green/Red (495–500 nm) filter, separated by an LP 570 dichroic mirror. We typically acquired 150-200 μm-deep Z-stacks with 4 μm Z resolution, 12 tiles (4×3), with an X-Y resolution of 800 × 800 pixels per tile, and 0.6364 μm per pixel. Post-production processing of images was done using Fiji (v2.9.0), including tile stitching and quantification of GFP signal. FITC-Dextran Signal Area was calculated as the percentage of total GFP pixels above threshold in a maximum Z-projected image.

Immune Cell Isolation

Lymph nodes were mechanically disrupted and filtered through a 70-μm strainer to create single cell suspensions in cold complete RPMI (cRPMI) (RPMI 1640 media supplemented with 2 μM glutamine, 100 U/ml penicillin/streptomycin, and 10% FBS). Ears were mechanically dissociated and then digested in 100 μg/mL Liberase, 500 μg/mL Hyaluronidase, 10 mM HEPES, in up to 5 mL serum-free DMEM, incubated for 1 hour at 37°C, and filtered over 70 μm nylon mesh to create single cell suspensions in cold cRPMI. Isolated B16F10 tumors were mechanically dissociated and then digested in 1 mg/mL Collagenase III for 1 hour at 37°C, and then filtered through a 70-μm strainer. Tumor homogenate was then centrifuged over a 40/80% Percoll gradient to isolate immune cells.

Intracellular cytokine and transcription factor staining

Intracellular cytokine staining was performed using the Cytofix/Cytoperm PlusKit per manufacturer’s instructions. In brief, single-cell suspensions from freshly harvested dLN were stimulated with PMA (20 ng/mL) and ionomycin (1 μg/mL) for 4 hours at 37 °C in the presence of GolgiPlug (brefeldin A). After staining for cell-surface molecules, the cells were fixed, permeabilized, and stained with antibodies to IFNγ, IL-17a, and IL-4. Flow cytometry antibodies are listed in Table 2.

Table 2.

Flow cytometry

Single cell suspensions were washed and resuspended with FACS buffer (PBS, 2% FBS, 0.02% Sodium Azide) containing purified rat anti-mouse CD16/CD32 (Fc Block). Cells were then stained with antibody cocktails for 20 minutes at 4°C. When necessary, intracellular Foxp3 staining was performed using a Foxp3 mouse T_reg cell staining kit (eBioscience). Cells were washed with FACS buffer after staining, resuspended in FACS buffer, and analyzed using a Cytek Biosciences Aurora Flow Cytometer. Flow cytometry data was analyzed with FlowJo (v10.0) software (BD Biosciences).

Quantification and statistical analysis

Unless otherwise stated, data in graphs are shown as mean ± standard deviation (SD). Statistical analyses were performed using Prism 10 (GraphPad v10.1.1). Statistical tests (e.g. one-way ANOVA, paired t-test, etc.), sample size (n) and what n represents (e.g. number of mice) are documented in the figure legends.

Single-Cell RNA Sequencing

Single-cell RNA sequencing was performed on FACS-sorted cell populations using the Chromium instrument (10x Genomics) following the manufacturer’s protocol for 3′ v3 chemistry. Briefly, single-cell suspensions were washed and resuspended in FACS buffer containing antibodies against rat anti-mouse CD45 and/or rat anti-mouse Ly6G, followed by staining for 30 minutes at 4°C. Cells were then washed, resuspended in FACS buffer, and sorted using an Aria II cytometer (BD Biosciences). Sorted cells were subsequently stained with TotalSeqTM-B barcoded hashing antibodies for sample multiplexing (Table 2). Immune cells from each condition were pooled at equal ratios, washed twice with PBS containing 0.05% bovine serum albumin (BSA), and resuspended in PBS with 0.04% BSA to a final concentration of 700–1,300 cells per μL. Cell viability was confirmed to be above 70% using 0.2% (w/v) Trypan Blue staining (Countess II). Library preparation of single-cell suspensions containing hash-tagged cells was performed by the MSKCC Integrated Genomics Operation (IGO) Core Facility. Subsequent cell capture, reverse transcription, cDNA amplification, and library construction were carried out according to the standard 10x Genomics protocol. Raw sequencing data were processed using the 10x Genomics

Cell Ranger pipeline. The Cell Ranger mkfastq tool was used to demultiplex raw sequencing files (BCL format) into FASTQ files. Alignment of gene expression libraries to the GRCm39 reference transcriptome, generation of count matrices, and processing of feature barcode and VDJ library reads, were performed using the Cell Ranger (v9.0) default pipeline. For downstream analysis, the Seurat package (v5.2.0) in R (v4.2.3) was used. Data from all scRNA-seq samples were merged into a single aggregated Seurat object using IntegrateData()⁷⁶. Hashtags were demultiplexed using HTODemux()⁷⁷. The standard Seurat workflow was followed, including quality control, normalization, feature selection, dimensionality reduction, clustering, and differential expression analysis. Immune cell subsets were annotated against the Immgen database using the SingleR R package⁷⁸. Cell lineage trajectories and pseudotime values were inferred using Slingshot⁷⁹.

Figure supplements

Figure 1- figure supplement 1.

Figure 2 - supplement 1.

Figure 3 - supplement 1.

Figure 4 - supplement 1.

Figure 4 - supplement 2.

Figure 6 - supplement 1.

Figure 6 - supplement 2.

Figure 7 - supplement 1.

Data availability

scRNA-seq data will be deposited in GEO. Code will deposited with github. All other source data will be available upon request.

Acknowledgements

We thank A. Y. Rudensky, J. Wolchok, and T. Merghoub for mice; A. Schietinger and M. O. Li for conceptual guidance; G. Diehl for critical reading of the manuscript; the MSKCC Flow Cytometry Core Facility for FACS purification; the MSKCC Integrated Genomics Operation for library preparation and sequencing; and members of the M. Huse lab for advice. Supported in part by the US National Institutes of Health (R01-AI087644 to M. H., R01-CA286566 to M. H., R01-AR077664 to J. H. Z., F31-CA294974 to Y. A. E., P30-CA008748 to MSKCC), the Ludwig Center for Cancer Immunotherapy (M. H.), the Vision of Children Foundation (J. H. Z.), and the National Organization for Albinism and Hypopigmentation (J. H. Z.).

Additional information

Funding

HHS | National Institutes of Health (NIH) (R01-AI087644)

• Morgan Huse

HHS | National Institutes of Health (NIH) (R01-CA286566)

• Morgan Huse

HHS | National Institutes of Health (NIH) (R01-AR077664)

• Jonathan H Zippin

HHS | National Institutes of Health (NIH) (F31-CA294974)

• Yassmin A Elbanna

HHS | National Institutes of Health (NIH) (P30-CA008748)

• Morgan Huse

Ludwig Center for Cancer Immunotherapy

• Morgan Huse

Vision of Children Foundation (VOC)

• Jonathan H Zippin

National Organization for Albinism and Hypopigmentation

• Jonathan H Zippin