Differential T cell accumulation within intracranial and subcutaneous melanomas is associated with differences in intratumoral myeloid cells Authors: Katarzyna Stasiak, Aaron D Stevens, Ashley C Bolte, Colleen T Curley, Mirna Perusina Lanfranca, Robin S Lindsay, Ukpong B Eyo, John R Lukens, Richard J Price, Timothy N J Bullock, Victor H Engelhard Published: 2024 ABSTRACT AbstractPatients with metastatic brain melanomas (MBM) experience shorter-lasting survival than patients with extracranial metastases, and this is associated with a higher fraction of dysfunctional CD8 T cells. The goal of this study was to understand the underlying cause of T cell dysfunction in MBM. To accomplish this, we compared murine B16 melanomas implanted intracranially (IC) or subcutaneously (SC). CD8 T cell activation was not altered, but representation in IC tumors was lower. Transferred activated or naïve CD8 T cells accumulated in similar numbers in both tumors, suggesting that the vasculature does not differentially impair T cell presence. Surprisingly, we found no evidence for T cell activation in draining lymph nodes of SC or IC tumor-bearing mice, consistent with the fact that dendritic cells (DC) that had acquired tumor antigen showed an immature phenotype. Instead, T cell activation occurred within both tumors, where the majority of tumor antigen+ myeloid cells were found. While, the numbers of intratumoral DC were comparable, those in IC tumors acquired less tumor antigen, and were alternatively matured based on upregulation of MHCII without upregulation of CD86. Additionally, in IC tumors, the largest population of tumor antigen+ myeloid cells were microglia. However, their presence did not influence either antigen acquisition or the phenotype of other myeloid cell populations. Overall, our data suggest that diminished representation of CD8 T cells in IC tumors is a consequence of alternatively matured DC and/or microglia that induce distinctly activated T cells, which ultimately fail to continue to accumulate inside the tumor.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00262-024-03832-0.Keywords: T cells, Melanoma, Metastasis, Brain, Antigen presenting cells Introduction The majority of cancer-related deaths in solid tumors are associated with metastasis [1]. About 30% of melanoma patients develop metastases to skin, lungs, lymph nodes (LN), subcutaneous tissue, or the central nervous system (CNS) [2]. Melanoma is typically resistant to radiotherapy or chemotherapy but is one of the most immunogenic cancer types [3, 4] and responds well to immunotherapy [5]. In fact, checkpoint immunotherapy is currently a first-line treatment for patients with advanced-stage melanomas [6]. While all these patients exhibit initial response to checkpoint immunotherapy, patients with metastatic brain melanoma (MBM) have an overall survival of approximately one year [2] and tend to experience shorter-lasting benefits [7], whereas patients with metastatic extracranial melanoma (MEM) have an overall survival of nearly five years [2] yet the underlying causes of this difference remain unclear. One explanation might be differences in the quantity and quality of tumor infiltrating lymphocytes (TIL). There is a common perception that brain tumors have smaller T cell infiltrates compared to tumors in other locations, due to the “immune privileged” status of the CNS. But despite a scarcity of MBM patient samples available for analysis [8] and variations in TIL densities among patient cohorts from different clinical studies, the few studies that directly examined and compared immune infiltrates in different melanoma metastatic sites reported somewhat disparate findings. One study reported that MBM have fewer CD8 T cells than patient-matched MEM [9], while another showed that it depended on the location of the extracranial metastatic tumor [10]. Recent single-cell transcriptomics of samples from treatment-naïve patients revealed a higher frequency of dysfunctional CD8+TOX+ T cells in MBM compared to MEM [11]. Additionally, a study employing genomic and transcriptomic analysis found that MBM have a reduced T cell-inflamed score relative to MEM, with diminished inflammatory responses and dendritic cell (DC) maturation [12]. Finally, compared to MEM, MBM exhibit notable immunosuppression and reduced diversity in T cell repertoire [13]. These diminished aspects of T cell presence and functionality in MBM, together with the limited clinical response to immunotherapy, could be attributed to features of the CNS that have contributed to the perception that the brain is immunologically privileged. These include: (a) a unique vessel structure with constrained permeability [14]; (b) absence of professional antigen presenting cells (APC) in the brain parenchyma [15]. Instead, it contains microglia, brain resident macrophages that express MHC and costimulatory molecules but are poor APC [16], with DC found only at the CNS borders (meninges and choroid plexus) [17, 18]; and (c) lack of a conventional lymphatic system in the brain parenchyma, with lymphatic vessels present only in the meninges [19]. Understanding the site-specific differences that impact T cell presence and functionality could greatly improve the clinical effectiveness of available therapies. Several recent studies utilizing murine models have explored different aspects of immunity to metastatic melanomas and the efficiency of immunotherapy [20]. However, research directly comparing the immune response to melanoma grown in subcutaneous (SC) or intracranial (IC) compartments is limited. To address this gap, we compared the immune infiltrate of B16 melanomas grown in SC or IC compartments. We found that T cell accumulation in IC tumors is lower compared to SC. Our result established that at an early stage of tumor outgrowth, T cell activation occurs within IC and SC tumor masses, but not in tumor-draining LN (tdLN). Our data suggest that lower T cell accumulation in IC tumors is linked to the alternative maturation status of DC and/or activated microglia within tumors. Implementing this knowledge into current practice offers a significant potential for enhancing available immunotherapies by directly targeting APC within tumors. Materials and methods Mice C57BL/6 mice were from Charles River/NCI. OT-I transgenic (C57BL/6-Tg (TcraTcrb) 1100Mjb/J), Thy1.1 congenic (B6.PL-Thy1a/CyJ), and lymphotoxin-α knockout (LTαKO) (B6.129S2-Ltatm1Dch/J) mice were from Jackson Laboratories. Mice were bred and maintained in a pathogen-free facility at the University of Virginia. All procedures were approved by the University of Virginia Animal Care and Use Committee in accordance with the NIH Guide for Care and Use of Laboratory Animals. Reagents All reagents, media compositions and antibodies are shown in Supplementary Tables 1 and 2. Melanoma cell lines B16F1 expressing ovalbumin (B16cOVA) [21] and B16F10 expressing ZsGreen and minigenes encoding OVA257-264 and OVA323-336 epitopes (B16F10ZsGreenOVA, gift from Max Krummel, University of California San Francisco) [22] cells were cultured in Tumor Cell Medium (Supplementary Table 1) and implanted into mice within 2–8 passages. Blasticidin (10 µg/ml) was added to maintain ovalbumin expression in B16cOVA cells. Tumor implantation Mice were anesthetized by intraperitoneal (IP) injection of a mixture of Ketamine (40 mg/kg) and Dexdomitor (0.2 mg/kg) in 0.9% sterile saline. Buprenorphine (0.05 mg/kg) was administered subcutaneously. For SC tumor implantation, tumor cells were injected in the neck scruff in 200 µl of sterile saline. For IC tumor implantation, mice were injected over 4 min with 103, 105, or 4 × 105 cells in 2 µl as described before [23], and harvested on days 15, 10–11, or 7, respectively. Anesthesia was reversed with Antisedan. Mice were monitored daily for weight loss and signs of distress. Tumors, LN and/or spleens were harvested in Harvest medium and processed as shown below. Treatments of tumor-bearing mice Mice were injected IP with either: (a) anti-CD27 (100 µg), HMW poly (I:C) (75 µg), and ovalbumin (500 µg) at day 7 after tumor injection to activate T cells; or b) anti-CD40 (100 µg) and HMW poly(I:C) (75 µg) as a DC adjuvant at day 10 after tumor injection to activate/mature DC. To prevent T cell egress from the LN, FTY720 (25 µg) daily was administered daily IP beginning on day 3 or supplemented in drinking water (2 µg/ml) starting on day 8 for the duration of tumor growth. To inhibit protein export from the cell for intracellular cytokine staining, Brefeldin A (BFA) (250 µg) was administered IP 4 to 6 h prior harvest. For discrimination of intravascular T cells, fluorescently labelled CD8 antibody (3 µg) was administered intravenously (IV) 3 min prior harvest. Single cell isolation from tumors and tissues Tumors, LN, and spleens were harvested in Harvest Medium (Supplementary Table 1). LN and spleens were manually shredded using needles or cut with scissors. For immune cell isolation from tumors and APC isolation from LN, Harvest medium was supplemented with Liberase (76 µg/ml) and DNase (40 µg/ml) and samples were digested at 37 °C for 15 min (tumors) or 30 min (LN). For experiments investigating the effect of DC adjuvant treatment or utilizing LTαKO mice, no enzymatic digestion was performed on tumors or LN. All samples were manually homogenized and filtered through 70 µm mesh to prepare single cell suspensions, regardless of whether they were enzymatically digested. In vivo T cell activation and adoptive transfer of naïve or activated CD8 T cells Single cell suspensions of splenocytes from OT-IxThy1.1 mice were prepared as described above. To generate activated CD8 T cells, 5 × 104 splenocytes were adoptively transferred IV into C57BL/6 mice and rested overnight. Mice were vaccinated with a combination of anti-CD40 (50 µg), HMW poly(I:C) (100 µg), and ovalbumin (500 µg). After five days, splenocytes were isolated as described above and treated with Red Blood Cell lysis buffer. CD8 T cells were enriched with Ly-2 MicroBeads according to the manufacture’s protocol. Naïve CD8 T cells were enriched from single cell suspension of spleens from OT-IxThy1.1 mice by the same method. Isolated activated or naïve CD8 T cells were stained with CFSE or CTV, and 106 cells in phosphate-buffered saline (PBS) supplemented with 3,000U IL2 were injected IV into mice bearing 6- or 9 days old tumors. Tumors, spleens, and LN were harvested 1 or 6 days after adoptive transfer, processed to single cell suspensions as described above, and analyzed by flow cytometry as described below. Microglia depletion Adult mice were kept on chow containing PLX3397 (660 mg/kg), CSF1R inhibitor, or control chow starting four days prior tumor implantation. Subsequently, mice were implanted intracranially with B16F10ZsGreenOVA cells. Tumors were harvested seven days later, processed to single cell suspensions as described above, and analyzed by flow cytometry as described below. Flow cytometry Cells were treated with Red Blood Cell lysis buffer for 3 min at 37 °C, Fc blocked with anti-CD16/CD32 (1:1000) for 15 min at 4 °C, stained with Live/Dead Fixable Dye (1:500) in PBS for 20 min at 4 °C, stained with fluorescently-labeled antibodies in Awesome Macs buffer for 30 min at 4 °C, and then fixed in 2% paraformaldehyde (PFA) for 10 min at room temperature. For the studies investigating the effect of DC adjuvant treatment or utilizing the LTαKO mice, samples were additionally blocked with 2% Normal Mouse Serum and stained in FACS buffer followed by H-2 Kb/SIINFEKL dextramer staining to identify OVA257 + H-2 Kb (tumor Ag)-specific CD8 T cells. For intracellular staining, BD Cytofix/Cytoperm Staining Kit was used according to the manufacturer’s protocol. Data was acquired on Cytoflex (Beckman Coulter), Attune (BD Biosciences), or Aurora (Cytek) flow cytometers and analyzed using FlowJo software. Gating strategies are shown in Supplementary Figs. 1 and 2. Intra-cisterna magna bead injections and LN collection and clearance Mice were anaesthetized by IP injection of a mixed solution of ketamine (100 mg/kg) and xylazine (10 mg/kg) in sterile saline. Mice were prepared for the surgery as described before [24]. FluoSpheres carboxylate 0.5 µm-beads 505/515 (2 µl) in artificial CSF were injected at a rate of 2 µl/min. Subsequently, mice were injected SC with ketoprofen (1 mg/kg) and allowed to recover on a heating pad until fully awake. Mice were euthanized 2 h later with CO2 and then transcardially perfused with 20 ml cold 1xPBS. Deep cervical LN (dcLN) were dissected and fixed in 4%PFA for 2 h at 4 °C. dcLN tissue clearance was performed following the CUBIC protocol [25] with modifications described before [24]. dcLN were imaged with a Leica TCS SP8 confocal microscope using LAS AF software (Leica Microsystems) and analyzed using Fiji software. The percent volume of microbead coverage in cleared dcLN was assessed by creating a 3D reconstruction of the node and then calculating the volume covered by beads divided by the total volume of the node. Statistical analysis Data is displayed as mean with error bars representing standard deviation (SD). Groups were compared using a one-way ANOVA with Tukey’s multiple comparisons test, Mann–Whitney test, or a Students T test. Analyses were performed using PRISM software (Graphpad) or R Core Team (2023). R: A Language and Environment for Statistical Computing. R: Foundation for Statistical Computing, Vienna, Austria. < https://www.R-project.org/ > . Results T cell activation and accumulation in IC tumors is altered compared to SC tumors We have previously studied B16cOVA tumors grown SC or IP [26, 27], but to our knowledge, there has been no direct comparison of T cells in early-stage B16cOVA tumors grown IC and SC. Since, the brain environment is immunologically distinct from that of other organs [28], we hypothesized that IC tumors would have fewer T cells than SC tumors. Using flow cytometry of single cell suspensions (Supplemental Fig. 1), both CD8 and CD4 T cell numbers were significantly lower in IC B16cOVA tumors grown for 10–11 days than in comparable SC tumors (Fig. 1a). This difference was evident when measured either as cells per tumor, or normalized for tumor weight, although IC and SC tumors were similar in size. To understand what drove this difference, we looked at day 7, the earliest time point at which we could reliably detect tumors. B16cOVA tumors grown for 7 days in either location had similar numbers of both CD8 and CD4 T cells, whether normalized per tumor or per tumor gram (Fig. 1b). Consistent with earlier observations [27], the majority (~ 68%) of T cells inside day 7 SC tumors were antigen (Ag) experienced (CD44+), comprising effector cells (CD62Lneg) and central memory cells (CD62L+) (Fig. 1c). The remaining ~ 32% were either naïve (CD44neg CD62L+) or double negative (CD44neg CD62Lneg), the latter of which we have suggested elsewhere are at an early stage of activation [29]. However, the fraction of Ag-experienced cells in IC tumors was significantly lower, with correspondingly elevated fractions of both naïve and double negative CD8 T cells (Fig. 1c). Nevertheless, the expression of CD25, CD69, IFNγ, and TNFα did not differ on Ag-experienced cells inside tumors grown in either location (Fig. 1d). We suggest that the reduced fraction of Ag-experienced T cells in IC tumors on day 7 leads to the diminished accumulation of CD8 T cells in IC tumors on day 11. T cell activation and accumulation in IC tumors is altered compared to SC tumors. B16cOVA tumors were implanted SC or IC into C57BL/6 mice, harvested on Day 10–11 (A) or Day 7 (B–D), and single cell suspensions were analyzed by flow cytometry. For cytokine analysis (D), mice were treated with BFA 4–6 h prior to tumor harvest. Data points represent individual tumors. Data are from 2 (A), 2–5 (B), and 1–2 (C–D) independent experiments and n = 3–5 (A), n = 2–8 (B–C), n = 6–8 (D) mice/group. Statistics: (A) Unpaired Student T test; (B) One-Way ANOVA; (C–D) Mann–Whitney test. p values: ***, < 0.001; **, < 0.01; *, < 0.05; ns, > 0.05 IC tumors do not limit vaccine induced immunity to tumor antigen One hypothesis for the lower number of Ag-experienced CD8 T cells in IC tumors on day 7 is that a brain tumor leads to systemic immunosuppression, as it is often the case in glioblastoma patients [30, 31]. To test this, we vaccinated mice with anti-CD27, poly(I:C), and ovalbumin seven days after tumor implantation or sham injection of PBS. We found equivalent expansion of OVA257 + H-2 Kb (tumor Ag)-specific CD8 T cells 7 days later throughout the mice, including in the spleen, and inguinal (iLN), deep cervical (dcLN), and superficial cervical (sLN) LN (Supplementary Fig. 3a, b). Since, the OVA257 epitope is also expressed on B16cOVA, this shows that IC tumors do not limit the peripheral immune response to vaccination against tumor antigen. Tumor vasculature does not differentially impair T cell infiltration in early-stage tumors Another hypothesis to explain the lower number of Ag-experienced CD8 T cells in IC tumors is that the unique structure of brain vessels and brain-tumor vasculature more strongly inhibits activated CD8 T cell infiltration. To test that, we transferred exogenously activated OT-I CD8 T cells, which recognize the OVA257 + H-2 Kb epitope, into mice bearing 6-day old IC or SC B16cOVA tumor. Tumors and spleens were harvested 24 h later, and the numbers of transferred cells that had infiltrated were quantified. Infiltration into SC and IC tumors was comparable, as was infiltration into the spleens of IC and SC tumor-bearing animals (Fig. 2a). We also tested the alternative hypothesis that tumor vasculature promotes entry of naïve T cells into IC tumors better than SC tumors. However, when naïve CD8 OT-I T cells were transferred into mice bearing 6-day old IC or SC B16cOVA tumors, infiltration in tumors 24 h later was again comparable (Fig. 2b). This demonstrates that brain and/or tumor vasculature does not differentially impair infiltration of CD8 T cells into IC tumors. Tumor vasculature does not differentially impair T cell infiltration in early-stage tumors. Flow cytometry analysis of single cell suspensions isolated from IC and SC day 7 B16cOVA tumors and spleens. Activated (A) or naïve (B) Thy1.1 + CD8 + OT-I T cells were injected IV 24 h prior to harvest. Data points represent individual tumors or spleens. Data are from 2 (A) and 1 (B) independent experiments and n = 3–5 (A) and n = 4 (B) mice/group. Statistics: Mann–Whitney test. p values: ns, > 0.05 No evidence for T cell activation in tumor-draining LN (tdLN) of early-stage IC and SC tumor-bearing mice We next addressed the hypothesis that the smaller fraction of Ag-experienced T cells inside IC tumors is a consequence of diminished T cell activation in tdLN. To test this, mice were implanted with SC or IC B16cOVA tumors and treated daily with FTY720 beginning on day 3 of tumor growth to prevent T cell egress from LN. Axillary LN (aLN) and dcLN, the draining LN for SC [32] and IC tumors [15], respectively, were harvested from all mice on day 7. While aLN of SC tumor-bearing mice had higher numbers of Ag-experienced (CD44+) CD8 T cells compared to dcLN of IC tumor-bearing mice (Fig. 3a), the numbers of CD25+ (Fig. 3b) and CD69+ (Fig. 3c) CD8 T cells were comparable. Importantly, there were no differences in these numbers when compared to the corresponding LN of FTY720-treated tumor-free mice. Because the total number of CD8 T cells is lower in dcLN compared to aLN (Fig. 3D), we also compared the fractions of Ag-experienced cells in these tdLN. The fractions of CD44+ CD8 T cells were comparable (Fig. 3e–g), but the fractions of CD25+ and CD69+ CD8 T cells were higher in dcLN. However, these fractions were again not different from those of the comparable LN of FTY720-treated tumor-free mice (Fig. 3e-g). These results demonstrate a lack of T cell activation in tdLN of day 7 SC and IC B16cOVA tumors. No evidence for T cell activation in tumor-draining LN (tdLN) of early-stage IC and SC tumor- bearing mice. B16cOVA tumors were implanted SC or IC into C57BL/6 mice. All tumor-bearing and control tumor-free mice were injected IP daily with FTY720 starting on day 3. The indicated LN were harvested on day 7 and single cell suspensions were analyzed by flow cytometry. Data points represent individual mice. Data are from 1 experiment and n = 4, 5 mice/group. Statistic: One-Way ANOVA. p values: ****, < 0.0001; ***, < 0.001; **, < 0.01; *, < 0.05. p-values greater than 0.05 are not marked Limits on antigen presenting cells in tdLN of IC and SC tumors One hypothesis to explain the lack of T cell activation in tdLN of SC and IC day 7 tumors is that they lack tumor Ag+ DC or that such DC have compromised immunogenicity. To test this, mice were implanted SC or IC with B16F10ZsGreenOVA cells, which express cytosolic ZsGreen as a tumor Ag proxy, allowing it to be tracked as it is phagocytosed or endocytosed by other cells, and LN were harvested on day 7. In SC tumor bearing mice, the average number of ZsGreen+ DC in tumor-draining aLN, as assessed by flow cytometry (Supplemental Fig. 2), was 100–350-fold higher than in non-tumor-draining dcLN, sLN or iLN (Fig. 4a left), although this difference was not statistically significant because of the wide variation. In contrast, there was no selective accumulation of ZsGreen+ DC in tumor-draining dcLN of IC tumor-bearing mice relative to non-tumor-draining aLN, sLN or iLN. However, the total number of DC in dcLN was significantly smaller than in other LN regardless of tumor location (Fig. 4a right). Taking this into account, the fraction of ZsGreen+ DC in dcLN of IC tumor-bearing mice was significantly higher than that of other non-draining LN, as was that of aLN in SC tumor-bearing mice (Fig. 4b). Thus, tdLN do contain a population of DC that have acquired tumor Ag. Limits on antigen presenting cells in tdLN of IC and SC tumors. C57BL/6 mice were implanted SC or IC with B16F10ZsGreenOVA tumors for analysis of DC (A–B, D–E), or implanted IC with B16cOVA tumors to analyze bead drainage (C) or the impact of DC adjuvant treatment (F–G). Flow cytometry analysis of single cell suspensions isolated from indicated LN (A–B, D–G) or immunofluorescent images acquired by confocal microscopy (C). The gating strategy is shown in Supplementary Fig. 2. A, B The numbers of ZsGreen+ DC were calculated by quantitating the number of ZsGreen+ DC in the indicated LN of individual mice bearing B16F10ZsGreenOVA tumors and subtracting the average background signal observed in the LN of B16cOVA tumor-bearing mice. C Mice were injected with fluorescently labeled beads. dcLN were harvested 2 h later, tissue cleared, and the percent LN volume occupied by microbeads was assessed by 3D reconstruction of the node and bead volumes. Total LN volume was determined from Z-stack imaging of the DAPI signal, while bead volume was determined from the fluorescent signal in the same images. Each data point represents an average of the two dcLN from an individual mouse. F–G IC B16cOVA tumor-bearing mice were given drinking water supplemented with FTY720 starting at eight days after tumor injection and adoptive transferred naïve OT-I cells one day later. Mice were then untreated (Control) or injected IP with anti-CD40 and poly(I:C) as a DC adjuvant on day ten after tumor injection and LN were harvested on day fifteen. Data points represent individual mice. Data are from 3 (A, B, D–E) and 1 (C, F–G) independent experiments and n = 3–5 (A, B, D–G), n = 5 (D) mice/group. Statistic: (A–E) One-Way ANOVA, (F–G) Students T test. p values: ****, < 0.0001; ***, < 0.001; **, < 0.01; *, < 0.05. p-values greater than 0.05 are not marked The small representation of Ag+ DC in dcLN of IC tumor-bearing mice suggested that Ag or DC drainage was impaired by the presence of IC tumor or by differences in lymphatic anatomy of IC tumors. To test this, fluorescently labeled 0.5 μm carboxylate modified beads were injected into the cisterna magna of B16cOVA day 7 IC tumor-bearing mice and dcLN were harvested 2 h later. LN were optically cleared, stained with DAPI, and examined by confocal microscopy. There were no differences in bead volume normalized to total node volume between IC tumor-bearing and tumor-free sham or uninjected mice (Fig. 4c). Thus, day 7 IC tumors do not impair drainage from the meninges into the dcLN. Lastly, we evaluated whether there were differences in the quality of DC in SC and IC tdLN that might explain differences in the number of Ag-experienced T cells in day 7 tumors. There were no differences in the amount of Ag acquired by DC in tumor-draining aLN and dcLN as measured by ZsGreen mean fluorescence intensity (MFI) (Fig. 4d). There was also no difference in the MFI of MHCII, which is a marker of DC maturation (Fig. 4e). However, maturation as measured by MHCII expression on Ag+ DC did not differ from Agneg DC in tdLN or non-tumor-draining iLN of SC tumor-bearing mice and was lower than that of Agneg DC in tdLN of IC tumor-bearing mice. Overall, these data suggest that the lack of response to tumor Ag in tdLN of both IC and SC tumors is a consequence of the lack of maturity of Ag+ DC in these LN. We also considered whether the IC tumors provided sufficient inflammatory signals to generate a T cell response. We treated IC B16cOVA tumor bearing mice with anti-CD40 + poly(I:C), adjuvants known to drive DC maturation and T cell activation [33]. We adoptively transferred OT-I T cells as a reporter population and stopped cellular egress from the LN with FTY720 to determine, where they became activated. An increased proportion of dcLN and sLN DC expressed CD86, after anti-CD40 + poly(I:C) treatment (Fig. 4f). However, anti-CD40 + poly(I:C) failed to enable tumor Ag (OVA257) specific T cell activation in those LN of 4/5 mice (Fig. 4g). Taken together with the results of Supplementary Fig. 3, this suggests an insufficient number of tumor Ag+ DC, or another aspect of their maturation that is resistant to anti-CD40 + poly(I:C), limits induction of immunity to tumor Ag in tdLN of early-stage IC and SC tumor-bearing mice. Ag+ APC in IC and SC tumors are differentially matured Since, we were not able to detect T cell activation in tdLN, but we have shown before that established tumor masses can support activation of adoptively transferred naïve tumor specific CD8 T cells [26], we hypothesized that T cell activation in early-stage tumors is occurring in the tumor mass and that differences in Ag-experienced T cells in SC and IC tumors reflect differences in tumor-resident APC. Day 7 SC and IC B16F10ZsGreenOVA tumors had similar weights (Fig. 5a) and comparable weight-adjusted numbers of monocytes, but IC tumors had fewer macrophages and a trending diminution in DC (Fig. 5b). However, smaller fractions of DC, macrophages, and monocytes in IC tumors had captured tumor Ag (Fig. 5c) and those APC had captured less Ag per cell as measured by ZsGreen MFI (Fig. 5d). On a percentage basis, macrophages in SC tumors comprised ~ 40% of myeloid APC that had captured tumor Ag, while DC comprised ~ 35% and monocytes comprised ~ 21% (Fig. 5e). In contrast, macrophages in IC tumors comprised ~ 9% of the tumor Ag+ APC, while DC comprised ~ 20%, and monocytes ~ 19%. These results demonstrate multiple deficiencies in Ag acquisition by myeloid APC in IC tumors compared to their SC counterparts. Ag+ APC in IC and SC tumors are differentially matured. Flow cytometry analysis of single cell suspensions of cells isolated from B16F10ZsGreenOVA tumors (A-I) or LN (F-G) harvested on day 7 after tumor implantation. The gating strategy is shown in Supplementary Fig. 2. E Pie charts represent number of a given ZsGreen+ myeloid cell population per gram inside tumors, calculated as described in the legend to Fig. 4, as fraction of total number of ZsGreen+ intratumoral myeloid cells per gram. Data points represent individual tumors. Data from 3 independent experiments and n = 4–5 mice/group. Statistic: One-Way ANOVA. p values: ****, < 0.0001; ***, < 0.001; **, < 0.01; *, < 0.05. p values greater than 0.05 are not marked Interestingly, both ZsGreen+ and ZsGreenneg DC in both SC and IC tumors expressed higher levels of MHCII by MFI than the ZsGreenneg DC in non-tumor-draining iLN (Fig. 5f). However, ZsGreen+ and ZsGreenneg DC in IC tumors expressed a significantly higher MHCII MFI compared to their counterparts in SC tumors. On the other hand, the MFI of CD86 was elevated over that of DC in non-tumor-draining iLN only in ZsGreen+ DC in SC tumors. Based on elevated expression of MHCII in the absence of a change in CD86, these results suggest that DC in IC tumors are alternatively matured [27] compared to those in SC tumors. We observed similar phenotypes in tumor-associated macrophages and monocytes. ZsGreen+ and ZsGreenneg macrophages in both SC and IC tumors expressed a higher MHCII MFI compared to those in iLN, while those in IC tumors expressed a higher MHCII MFI than those in SC tumors (Fig. 5g). However, ZsGreen+ macrophages in both SC and IC tumors expressed elevated CD86 compared to ZsGreenneg macrophages in either tumors or iLN. Tumor associated ZsGreen+ monocytes also expressed higher CD86 than ZsGreenneg monocytes and those in iLN, but in this case, monocytes in IC tumors expressed significantly more CD86 than those in SC tumors (Fig. 5h). These results suggest that macrophages and monocytes in IC tumors are more activated compared to those in SC tumors, but the differences in the extent to which MHCII and CD86 are elevated may again be suggestive of distinct activation states. Microglia do not influence the ability of DC or macrophages to capture tumor Ag nor their activation/maturation status B16F10ZsGreenOVA IC tumors contained a population of microglia that was significantly more abundant than the populations of DC, monocytes or macrophages (Fig. 5b). While the fraction of microglia that captured ZsGreen was similar to the fractions of ZsGreen+ DC, macrophages, and monocytes in IC tumors (Fig. 5c), the amount of ZsGreen captured per cell by microglia was significantly higher (Fig. 5d). Thus, microglia comprised half of the ZsGreen+ myeloid APC in IC tumors, and also acquired the highest level of tumor Ag (Fig. 5e). ZsGreen+ microglia were also more activated than ZsGreenneg microglia as measured by MHCII MFI and CD86 MFI (Fig. 5i). Thus, microglia have the potential to mediate and/or alter immunity in IC tumors. One hypothesis to explain why APC in IC tumors capture less tumor Ag and are alternatively activated/matured is that microglia: (1) limit tumor Ag availability and compete for tumor Ag with other APC; and/or (2) negatively influence maturation/activation of other APC. To test that, we depleted microglia by putting mice on chow containing PLX3397 (a CSF1R inhibitor), IC inoculated B16F10ZsGreenOVA tumor cells four days later, and harvested tumors seven days later. The PLX3397 diet had no influence on tumor growth (Fig. 6a) but resulted in an ~ 80% decrease in intratumoral microglia, a ~ 70% decrease in intratumoral DC, a ~ 60% decrease in intratumoral macrophages, and no change in monocytes (Fig. 6b). However, there were no differences in the fractions of these cells that captured tumor Ag, nor the amount of tumor Ag acquired (Fig. 6c–e). PLX3397 treatment did not augment ZsGreen+ DC maturation or ZsGreen+ macrophage activation as measured by MHCII MFI or CD86 MFI but resulted in lower expression of CD86 on monocytes. Overall, these results demonstrate that only monocyte activation is diminished by microglia presence in IC tumors. However, Ag uptake by, and activation/maturation of intratumoral DC and macrophages is influenced by other factors in the IC tumor microenvironment. Microglia do not influence the ability of DC or macrophages to capture tumor Ag nor their activation/maturation status. C57BL/6 mice were put on PLX3397 chow for 4 days to deplete microglia, or control chow, and then implanted with B16cOVAZsGreen tumors. Flow cytometry analysis of single cell suspensions from day 7 tumors. Data points represent individual tumors and error bars SD. Data are from 1 experiment and n = 5 mice/group. Statistic: Unpaired t Test. p values: ****, < 0.0001; ***, < 0.001; **, < 0.01; *, < 0.05; ns, > 0.05 T cells become activated within early-stage tumors Given that we did not see T cell activation in early-stage tdLN, and the abundance of APC in tumor masses, we hypothesized that T cells are becoming activated in the tumor site. To test this, we implanted B16cOVA tumors IC or SC, let them establish for three days, treated the mice with FTY720 daily to prevent T cell egress from LN, and harvested tumors on day 7. Compared to T cells in tdLN (Fig. 3), the fraction of tumor infiltrating CD8 T cells expressing CD44, CD25, and CD69 was elevated, suggesting that these cells became activated inside these early-stage tumors (Fig. 7a–c). However, at this time, we saw no differences in the fractions of CD8 T cells from SC and IC tumors that expressed these activation markers (Fig. 7d). To gain additional confidence that this activation was occurring within the IC tumors, we injected B16cOVA intracranially into LTαKO mice, which lack all LN. Tumors grew to similar sizes in wild type and knockout mice (Fig. 7e). Despite the absence of LN, tumors in knockout mice contained more Ag-experienced CD8 and a trending increase in Ag-experienced CD4 T cells as well as more naïve CD8 and CD4 T cells (Fig. 7f). We also observed a trending increase in the number of OVA257 specific CD8 T cells in tumors from knockout mice. This data shows that LN are not required to establish the IC T cell response. T cells become activated within early-stage tumors. B16cOVA tumors were implanted SC or IC into C57BL/6 mice (A–D) or IC into LTαKO mice (E, F) and tumors were harvested on Day 7. Starting on Day 3 mice were supplemented daily with IP injection of FTY720 (A–D). LN data (A-C) and tumor data without FTY720 treatment (D) are shown here from Fig. 3 and Fig. 1 respectively to enable direct comparison. Single cell suspensions were prepared as described in Methods and analyzed by flow cytometry. Data points represent individual tumors and error bars SD. Data are from 2 (A–D), 1 (E, F) independent experiment and n = 5–8- (A–C), n = 2–8 (D) and n = 9–10 (E, F) mice/group. Statistics: Mann–Whitney test. p values: **, < 0.01; *, < 0.05 Discussion In melanoma, patients with MBMs experience shorter-lasting benefits (survival and response to immunotherapy) than patients with MEM, and this is associated with a higher fraction of dysfunctional CD8 T cells. To understand the basis for this, we analyzed murine B16 melanoma implanted SC or IC. We found that day 7 IC tumors had a lower fraction of Ag-experienced CD8 T cells than SC tumors, although the fractions of those cells expressing activation markers were not significantly different. By day 11, there was also a significantly lower representation of CD8 T cells in IC tumors. Infiltration of both naïve and activated CD8 T cells into IC and SC tumors on day 7 was comparable, suggesting that vasculature did not differentially impair T cell presence. Thus, the lower representation of CD8 T cells in day 11 IC tumors suggests impaired proliferation or persistence. Surprisingly, we did not observe T cell activation in tdLN at early times, likely due to a small representation of tumor Ag+ DC and/or their immature phenotype. Instead, our data indicated that T cell activation occurred within the tumor mass, where the majority of Ag+ myeloid APC were found. Compared to DC in SC tumors, DC in IC tumors acquired less tumor Ag and exhibited an altered maturation phenotype. Also, microglia were the predominant tumor Ag+ myeloid APC population in IC tumors and they have previously been shown to have the potential to interact with T cells [34, 35]. We propose that the altered maturation state of intratumoral DC and/or the presence of microglia in IC tumors leads to less robust T cell proliferation, anergy or exhaustion, ultimately leading to a lower accumulation of T cells in IC tumors over time. We were surprised not to see differences in T cell trafficking into day 7 IC and SC tumors. Normally, brain blood vessels are formed by specialized and tightly connected endothelial cells, wrapped by pericytes and covered by astrocytic end-feet [36]. This complex vessel anatomy forms the blood–brain barrier (BBB), which limits immune cell access. However, IC melanoma cells have been shown to grow immediately adjacent to blood vessels, displacing astrocytes, and thus disrupting the BBB [37]. The lack of difference in T cell trafficking into IC and SC tumors may reflect this. While, it is generally acknowledged that infiltration of adoptively transferred T cells into solid tumors is limited, our data suggest that the IC tumor vasculature phenotype does not impose an additional obstruction for T cell to infiltrate MBM. In fact, the strategies targeting vasculature by, for example, increasing the expression of homing receptor ligands, or normalizing tumor vasculature by VEGF treatment, could synergize with adoptive T cell therapies and enhance their efficiency. Alternatively, it has been demonstrated in the context of experimental autoimmune encephalomyelitis that engagement with APC in perivascular spaces enables activated Ag-specific T cells to cross the astrocyte layer and access brain parenchyma [38]. Regardless, our result strongly suggests that the diminished representation of T cells in long-term IC and SC tumors is due to a process other than effector cell entry. TdLN play a critical role in mediating antitumor immunity. They serve as the primary sites, where tumor-specific immune responses are generated, making them potential targets for successful immunotherapy [39, 40]. It has been shown that tumor-specific CD8 T cell activation occurs in tdLN, where tumor Ag is found [21, 41, 42]. Surprisingly, using highly sensitive TCR transgenic T cells, we did not detect a T cell response in tdLN in our tumors, which were evaluated at a very early stage. One possibility suggested by our data is that T cells are not activated because the Ag+ DC in tdLN are still immature. It is also possible that, due to the early stage of tumor growth examined, Ag+ DC in tdLN are too infrequent to activate a sufficient number of T cells or have only just reached the tdLN and have not yet activated T cells. It is notable that a DC adjuvant treatment does not promote immunogenicity, and that DC in tumors appear more mature than those in LN. Our previous studies indicated that tumor specific CD8 T cell activation can occur within late-stage SC tumors [26], and here we demonstrated that this is the predominant site of activation at earlier time points. We also provide evidence that IC tumor masses can be a site for CD8 T cell responses. Surprisingly, early stage CD8 T cell activation in IC and SC tumors is similar, suggesting that the selective deficit in T cell accumulation in IC tumors arises over time due to differences in T cell proliferation or survival. This difference in CD8 T cell accumulation could be because IC intratumoral DC are alternatively matured based on upregulation of MHCII but no change in CD86, and they have acquired less tumor Ag. We have previously shown that intratumoral DC are alternatively matured in the absence of NK cells, and they can drive T cell anergy [27]. Another study showed that DC from IC and SC gliomas exhibited comparable maturation phenotypes, but IC tumor-infiltrating DC were unable to stimulate T cell proliferation [43]. Collectively, these findings and those of the present study suggest that DC in IC tumors are not fully immunocompetent, resulting in compromised IC intratumoral immunity. One possibility is that factors in the IC tumor microenvironment induce the expression of inhibitory receptors on DC. For example, Dendritic Cell Immunoreceptor (DCIR) can lead to diminished DC activation, while Immunoglobulin-like Transcript 3 (ILT3) can lead to a more tolerogenic DC phenotype, inhibiting T cell activation [44]. Regardless, these results suggest that when designing a comprehensive and multitargeted immunotherapy, it is important to consider targeting intratumoral DC to induce their proper maturation. This could be achieved, for example, by using Talimogene laherparepvec (T-VEC), an approved treatment utilizing oncolytic herpes simplex virus engineered to target cancer cells and produce granulocyte–macrophage colony-stimulating factor (GM-CSF), or by various approaches targeting Toll-like receptors (TLRs), which are currently in clinical trials [45]. It is also possible that the poor accumulation of T cells in IC tumors is a direct consequence of their interaction with microglia. Activated microglia express MHC and costimulatory molecules and have been shown to present Ag to T cells [34, 35, 46]. They have been shown to express proinflammatory cytokines and chemokines that enhance T cell recruitment and activation [35, 47–50], but they also have been shown to be capable of diminishing effector responses and induction of cell death via FAS mediated apoptosis or NO production [51] or T cell anergy [16]. Although our results show that microglia do not diminish DC capacity to capture tumor Ag or their maturation, and it is still unclear what drives the distinct DC phenotype in the IC tumor, we suggest that microglia may induce T cell anergy or death in IC tumors, resulting in reduced T cell representation at later times. Our research highlights the importance of understanding the impact of the brain tumor microenvironment to elucidate the underlying factors driving altered DC maturation and contribute to poor T cell accumulation and/or persistence in IC tumors. The distinct features of the brain and its environment might result in a failure (short-term patient benefit) of many currently available immunotherapies. Our work suggests that DC vaccines or immunotherapies aimed at inducing appropriate DC maturation within tumors could synergize with other approaches when used in appropriate sequence and be therapeutically efficacious. Targeting microglia and reprograming their metabolism [52] shows promise in glioma models and understanding the dynamic interplay between microglia and MBM holds immense promise for advancing our understanding and treatment of brain tumors. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (JPG 586 KB) Supplementary file2 (JPG 477 KB) Supplementary file3 (JPG 340 KB) Supplementary file4 (JPG 349 KB) Supplementary file5 (JPG 448 KB) Supplementary file6 (JPG 139 KB) Supplementary file7 (PDF 156 KB) Acknowledgements We thank Dr. Max Krummel for sharing B16F10ZsGreenOVA cell line, Drs. Michael Kovacs and Tajie Harris for their input, and Marieke Jones for statistical consultation. Author contributions KS involved in Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing; ADS involved in Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft; ACB involved in Data curation, Formal analysis, Investigation, Methodology, Validation, and Visualization; CTC involved in Conceptualization, Data curation, Investigation, Methodology; MPL involved in Conceptualization, Data curation, Investigation, Methodology, Writing – review and editing; RSL involved in Conceptualization, Data curation, Investigation; UBE involved in Funding acquisition, Resources; JRL involved in Funding acquisition, Resources, Supervision; RJP involved in Conceptualization, Funding acquisition, Resources, Supervision; TNJB involved in conceptualization, funding acquisition, resources, supervision; VHE involved in conceptualization, funding acquisition, project administration, resources, supervision, writing – review and editing. Funding This work was supported by United States Public Health Service Grants R01 CA181794 (to VHE), R01 CA197111 (to VHE, TNJB, and RJP), R21 CA226607 (to TNJB and RJP), RJP and CTC were supported by R01EB030409 (to RJP), R21 NS118278 (to RJP), P30 CA44579, and Melanoma Research Alliance Grant #348727 (to TNJB and RJP). ADS, ACB, MPL, and RSL were supported by United States Public Health Service Training Grant T32 AI007496. ACB was also supported by a Wagner Fellowship from the University of Virginia, and United States Public Health Service Grant F30 AG069396. KS was the recipient of a Parsons-Weber-Parsons Fellowship in Oncology Research from the University of Virginia. Data availability The data that support the findings of this study are available from the corresponding author [VHE] upon reasonable request. Declarations Conflict of interest The authors declare no potential conflicts of interest. Ethics approval All procedures were approved by the University of Virginia Animal Care and Use Committee in accordance with the NIH Guide for Care and Use of Laboratory Animals. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary file1 (JPG 586 KB) Supplementary file2 (JPG 477 KB) Supplementary file3 (JPG 340 KB) Supplementary file4 (JPG 349 KB) Supplementary file5 (JPG 448 KB) Supplementary file6 (JPG 139 KB) Supplementary file7 (PDF 156 KB) Data Availability Statement The data that support the findings of this study are available from the corresponding author [VHE] upon reasonable request.