6–12-week-old female C57BL/6 (Charles River, Wilmington, MA) mice were challenged with a lethal dose of B16F10 melanoma (5×10⁴ cells, >95% viability) administered intradermally on the mouse flank on day 0 of therapy (10, 11). Tumor size was recorded by calipers every 1–3 days as square mm (L × W). Mice were monitored for signs of distress in accordance with IACUC protocols.

Vaccine antigen is the MIP-3α-Gp100-Trp2 (tyrosinase-related protein 2) DNA construct in the pCMVe mammalian expression plasmid published here (11). Vaccination-grade plasmids were extracted from E. coli DH5-α (Invitrogen™ ThermoFisher Scientific, Waltham, MA) using Qiagen^® (Germantown, MD) EndoFree^® Plasmid Kits and were diluted with endotoxin-free 1xPBS. Vaccine DNA preps were verified by insert sequencing (JHMI Synthesis and Sequencing Facility, Baltimore, MD), spectrophotometry, and gel electrophoresis, and then administered at 50 μg/dose into the gasctocnemius muscle followed by in vivo electroporation, pulsing the muscle with the ECM 830 Electro Square Porator with 2-Needle Array Electrode (BTX Harvard Apparatus; Holliston, MA) under the following parameters: 106 V; 20 ms pulse length; 200 ms pulse interval; 8 total pulses (10). 50 μg ODN2395 Type C CpG (Innaxon LPS Biosciences, Tewkesbury, UK) was administered intramuscularly 2 days post-vaccination into vaccinated muscle. Recombinant mouse interferon alpha-A (IFNα, R&D Systems, Inc. Minneapolis, MN) was administered intratumorally as a series of doses: one high dose (10,000 units) followed by 2-3 days of low doses (1000-2000 units). InSolution™ 5 Aza 2′-deoxycytidine (5Aza, CalBiochem^®, MilliporeSigma, Burlington, MA) was administered intraperitoneally at 1 mg/kg in 50 μl, at approximately 20 μg/mouse. Figure 1A outlines the therapy schedule.

Tumor cell suspensions were prepared as previously described, with or without the Lympholyte M (Cedarlane Labs, Burlington, NC) purification step (10, 11). Briefly, tissue was extracted, kept cold, ground through a filter, washed, either purified or processed to lyse red blood cells, and used for downstream applications. In one experiment, the tumor cells were cryopreserved in 90% FBS 10%DMSO freezing media using isopropanol baths at -80°C prior to moving the samples to -150°C. Cells were quick-thawed and allowed to rest for 4 hours at 37°C before proceeding. Cells undergoing a freeze-thaw were only utilized for T-cell subtype analyses. If fewer than 10 live CD8+ T cells were measured, the samples were not included due to poor cell viability. Results of T-cell subtypes from cryopreserved cells were not significantly different from the remainder of the datasets. Tumor cell suspensions were stained in a 96-well V-bottom plate (Sarstedt, Inc., Newton, NC), with combinations of the following anti-mouse mAbs: PercPCy5.5 conjugated anti-CD3, CD11b-APC, Ly6G-FITC, Ly6C-Percp-Cy5.5 (eBioscience, Inc., San Diego, CA), FITC-CD8, NK1.1-PE, Live/Dead Near-IR (Invitrogen by Thermo Fisher Scientific, Carlsbad, CA), CD8-AF700, CD62L-APC, CD44-AF700, NK1.1 AF700, CD11c-PE, and CCR7 PE-Cy7 (Biolegend, San Diego, CA). The Attune™ NxT (Thermo Fisher Scientific, Waltham, MA) flow cytometer was utilized. Flow data were analyzed by FlowJo Software (FlowJo, LLC Ashland, OR) or Attune NxT Software v3.2.1 (Thermo Fisher Scientific, Waltham, MA). Tumors smaller than 25mm² were not analyzed due to an insufficient amount of tissue. Figure 1B shows common initial gates: cells were gated on potential immune cells by FSCxSSC, screened out doublets and clumps, and selected for alive cells. %Alive refers to the percentage of all cells passing through these first three gates. Gates were formulated using full-minus-one (FMO) staining controls as reference.

Cross-sections of tumor weighing less than 100 mg were harvested. Tumor was minced as finely as possible, added to 1 ml Trizol^® (Ambion^® by Life Technologies, Carlsbad, Ca), and then homogenized by the Fisher Scientific™ PowerGen125 (Thermo Fisher Scientific, Waltham, MA). RNA was extracted according to the manufacturer’s protocol. The cDNA reverse transcription reaction utilized the SuperScript™ III First-Strand Synthesis System (Invitrogen ™, Waltham, MA), as per the manufacturer’s protocol. Real-time quantitative reverse transcription-PCR (qRT-PCR) was performed utilizing TaqMan^® Gene Expression Master Mix or Fast Advanced Master Mix and TaqMan^®Gene Expression Assays (Applied Biosystems™ by Thermo Fisher, Halethorpe, MD) with probes specific for mouse GAPDH, XCL1, and CXCL10, utilizing the manufacturer’s protocols. Ct threshold was standardized across experiments, and the Ct statistic equated to the average of triplicate technical replicates. For analysis, ΔCt is calculated by subtracting the Ct value of the housekeeping gene GAPDH from that of the gene of interest. qRT-PCR was performed utilizing the StepOnePlus™ machine and software (Applied Biosystems™ by Thermo Fisher, Halethorpe, MD).

Tumor cross-sections or whole tumors were fixed in 10% neutral buffered formalin. The samples were embedded in paraffin, cut in levels and adhered to slides, and then cuts from the same level were stained with hematoxylin and eosin (H&E) or labelled CD8 by immunohistochemistry (IHC) in parallel by the Sydney Kimmel Comprehensive Cancer Center Histology Core Facility (Baltimore, MD). All H&E and CD8 IHC cases were reviewed by a board-certified dermatopathologist (JCS) who was blinded during histologic scoring and evaluated for overall histologic appearance and degree of immune response. Immune infiltration was scored semi-quantitatively, with 0 for no inflammation, 1 for mild, 2 for moderate, and 3 for strong peripheral and intratumoral T-cell infiltration. Images were digitally brightened by 10%. Images are presented from the 10x objective, and zoomed images were digitally zoomed an additional 2x. Quantitative analysis of the infiltrating CD8+ cells was performed by two individuals blinded to the groups, each counting stained CD8+ cells across 10 random fields per sample (40x objective). The mean value of the 20 fields counted was utilized as the data point for analysis.

Tumor size, qRT-PCR, and flow cytometric analyses were statistically tested by one-way ANOVA with Tukey’s multiple comparison test if dataset distributions were approximately normal or with Dunn’s multiple comparison test if not. Normality was assessed by D’Agostino & Pearson test primarily or by Shapiro-Wilk test if the sample size was too small. Scatter plots were analyzed by simple linear regression with Spearman correlation coefficient test. Grouped experiments were analyzed by 2-way ANOVA with Sidak multiple comparison test. Microsoft^® Excel (Microsoft Corp, Redmond, WA) was used for database management. Prism 9 (GraphPad Software, Inc. San Diego, CA) was utilized for statistical analyses and figure creation. All error bars represent the estimation of the standard error of the mean, and all midlines represent the group mean. The significance level of α ≤ 0.05 was set for all experiments. Data provided in ( Supplementary file 1 ).

Figure 1A outlines the therapy schedule utilized both in this study and our prior work (10, 11). Importantly, the tumor growth phenotype remains consistent across experiments ( Figure 1C ), allowing the studies here to expand on previous work. Since our prior studies found that mice receiving no treatment did not survive to days 24-26 (10, 11), in the current study we decided to compare the three treatment groups to each other. At select time points, mouse weight was measured to ensure the therapeutic regimen did not induce excessive stress. A representative gating strategy indicating the initial steps of gating for samples is shown here as well ( Figure 1B ). All groups consistently gained weight over time, and there was no significant difference in weight change across the groups ( Figure 1D ).

Our previously published data showed significant upregulation of CCL19 in mice receiving combination therapy (11). Since CCL19 has been implicated in homing CCR7+ immune cells to the lymph nodes (12), additional chemokines associated with immune cell infiltration were chosen for qRT-PCR analysis. XCL1, essential in attracting cross-presenting DCs (13), showed higher expression in the vaccine alone (p = 0.0036) and combination (p = 0.0006) groups compared to IFN + 5Aza ( Figure 2A ). XCL1 expression was also correlated with trends of reduced tumor size in the combination group (p = 0.0591, R² = 0.321) ( Figure 2B ). CXCL10 is a primary recruiter of T cells (14), and its transcription levels were also significantly elevated in the combination group compared to both vaccine alone (p = 0.0117) and to IFN + 5Aza (p = 0.0054) ( Figure 2C ). Additionally, levels were correlated with trends of reduced tumor size in the combination group (p = 0.132; R² = 0.354) ( Figure 2D ). In both cases, the combination group is the only group with an R² above 0.3. Across all groups, XCL1 (R²: 0.53, p<0.001) and CXCL10 (R²: 0.584; p<0.001) also showed significant overall correlations.

Emerging evidence suggests that DC- and NK- cell infiltration into the TME aids in mounting an effective anti-tumor response (15, 16). To understand the cellular makeup of the TME, we performed flow cytometric analysis on tumor lysates. Figure 3A shows the gating strategy of representative samples. While there was too much variation to achieve statistical significance, a trend of increased intratumoral infiltration of CD3-CD11c+ DCs was observed between the combination group and the IFN + 5Aza (p = 0.1138) and vaccine alone (p = 0.0738) groups ( Figure 3B ). Interestingly, the percentage of CD8+ DCs, representing inflammatory and cross-presenting DCs (17), was modestly increased following vaccination alone (p = 0.0672), and significantly increased following combination treatment (p = 0.0012) relative to IFN+5Aza ( Figure 3C ). However, only the combination group had enhanced levels infiltrating the tumor compared to the IFN + 5Aza group (p = 0.0003; Figure 3D ). Correlation between XCL1 expression and the presence of CD8+ DCs was also highly significant (p = 0.0055, R2 = 0.443) ( Figure 3E ). Additionally, the upregulation of CCL19 and CCR7 (11) seen in the combination group, and the fact that CCL19 binds CCR7 directed us to investigate the percent of CCR7+ DCs present in the tumor lysate ( Figure 3F ), which were significantly higher in the combination group (p = 0.037) when compared to the vaccine group. NK cell numbers were also analyzed (representative gating in Figure 3G ) and were significantly higher in the combination therapy group when compared to vaccine alone ( Figure 3H , p = 0.0202).

A category of innate immune cells known as MDSCs (myeloid derived suppressor cells) is often present in the TME and disables an effective anti-tumoral response by potentiating immunosuppressive activity (17). To analyze whether these cells were present in the TME, we performed flow cytometry on the tumor lysate; both classes of murine MDSCs were analyzed and included M-MDSC (monocytic MDSCs) and PMN-MDSC (polymorphonuclear or granulocytic MDSCs). Figure 4A shows the representative gating strategy. The percent of both PMN-MDSCs ( Figure 4B ) and M-MDSCs ( Figure 4C ) were reduced in the vaccine alone and combination therapy groups. The differences between mice receiving the vaccine alone compared to IFN + 5Aza were significant for PMN-MDSCs (p = 0.003) and M-MDSCs (p = 0.0379). When comparing combination therapy to IFN + 5Aza, there was significant reduction of PMN-MDSCs (p = 0.003) and a trend towards reduction of M-MDSCs (p = 0.0768).

Our previous data indicated increased levels of CD8+ T cells in the combination therapy group (10, 11), and findings from this study confirmed those results. Figure 5A shows the representative gating strategy used to stratify CD8+ T cells, and total CD8+ T cells were significantly enriched in the combination group compared to IFN + 5Aza ( Figure 5C , p = 0.0003) and trending towards significance in vaccine alone (p=0.1207). To understand differences in CD8+ T-cell composition, CD3+ CD8+ T cells were further gated on CD44 and CD62L to categorize effector (CD44+ CD62L-), naive (CD44- CD62L+), and central memory (CD44+ CD62L+) T cells ( Figure 5A ). When total CD8+ T cells were stratified based on percentage, naïve and double negative (CD44- CD62L-) T cells were qualitatively lower in the combination therapy group compared to mice receiving either IFN + 5Aza or vaccine alone, and the combination group was primarily composed of central memory and effector memory T cells ( Figure 5B ). Importantly, the combination therapy group showed significantly higher amounts of effector memory CD8+ T cells ( Figure 5E , p = 0.0068) compared to mice receiving IFN + 5Aza, and also showed increased numbers of central memory T cells relative to both mice receiving vaccine alone (p = 0.024) and IFN+ 5Aza (p = 0.0351) ( Figure 5D ). Furthermore, the correlations between decreased tumor size and increased CD8+ T effector memory cell or CD8+ T central memory cell infiltration into the TME in the combination treatment group (p = 0.0438, R² = 0.436; Figure 5F , p = 0.0029, R² = 0.73; Figure 5G respectively) were significant.

A representative sample of four tumors per group across two experiments were selected for microscopy analysis. Tumor cross-sections were mounted onto slides and stained for H&E and CD8 by immunohistochemical methods. Figure 6A shows representative images with selected clusters of CD8+ cells pointed to with arrows, with areas of interest outlined in boxes and zoomed an additional 2-fold to more clearly show the stained cells. In all the IFN + 5Aza samples and three out of four vaccine-only samples, CD8+ cells were visualized primarily around peritumoral vessels but not infiltrating the tumor mass (arrows, Figure 6A left and middle). Three out of four samples from the combination group showed substantial infiltration of CD8+ cells into the tumor mass (arrows, Figure 6A right). Figure 6B provides evidence that our sampling was representative, as the tumor size averages did not significantly differ from across group means.

Additionally, the samples were scored in a group-blinded fashion for level of inflammation ranging from 0-3 ( Figure 6C ). The IFN + 5Aza group showed little inflammation with one out of four samples scored as 0, and three with minimal inflammation, scored as 0+ to 1. The vaccine alone group showed higher variability, with one tumor lacking any inflammation (scored 0), two showing only peripheral CD8+ cell accumulation (scored 0+ to 1), and one with peripheral accumulation and intratumoral penetration (scored 2). The combination group resulted in the highest proportion of cases with both peripheral CD8+ cell accumulation and CD8+ cell penetration, with three out of four samples scored at 2.

To provide quantitative analysis of the infiltration of CD8+ cells into the tumor mass, twenty total randomly selected parenchymal 40X objective microscope fields per sample were counted by two group-blinded individuals. Figure 6D shows that the combination group had significantly more CD8+ cells that infiltrated the tumor parenchyma as compared to IFN+5Aza (p=0.0266) and to vaccine alone (p=0.0316).