Seven-week-old female BALB/c, BALB/c-nu, and C57BL/6 mice were purchased from Japan Charles River. Pmel-1 (B6.Cg-Thy1a/Cy Tg(TcraTcrb)8Rest/J) mice with transgenic gp100 melanoma antigen-specified T cell receptor and Ly5.1 (B6.SJL-Ptprca Pepcb/BoyJ) mice were purchased from the Jackson Laboratory (ME, USA). Each experiment group contained 8 mice except where otherwise specified. All animal experiments were conducted in accordance with institutional guidelines with the approval of the Animal Care and Use Committee of the University of Tokyo and the Tokyo University of Science.

Colon26 cells were obtained from the Cell Resource Center for Biomedical Research (RRID: CVCL_0240; Institute of Development, Aging, and Cancer, Tohoku University, Japan). B16F10 cells were obtained from the American Type Culture Collection (RRID: CVCL_0159; ATCC, USA). Colon26 cells (2 × 10⁵ cells / mouse) and B16F10 cells (5 × 10⁵ cells / mouse) were inoculated subcutaneously into the right flank of BALB/c, BALB/c-nu or C57BL/6 mice. Tumor diameter was measured by twice weekly and used to calculate tumor volume (V, mm³) using the formula V = L × W × W/ 2 (where L is tumor length and W is tumor width).

Recombinant HMGN1 proteins were produced in E. coli and purified using sequential fractionation by heparin affinity column, ion exchange column and reverse-phase column. The final protein products had over 99% purity with under 1 endotoxin unit concentration per microgram protein, as assessed by SDS-PAGE and an Endospecy ES-50 M Kit (Seikagaku Corporation, Japan), respectively. Detail of the production and purification of mouse and human recombinant HMGN1 proteins is described in Additional file 1: Method S1, Figure S1 and Table S1.

HMGN1 protein (at a dose of 0.16 μg per mouse per injection, unless otherwise specified) was administered intraperitoneally on days 9, 14, 17, and 20 after tumor inoculation. Anti-CD4 depleting antibody (clone GK1.5; BioXcell, USA) was injected intraperitoneally on days 5 and 9 after tumor inoculation, at a dose of 200 μg per mouse per injection [2]. The optimized protocol for B16F10 tumor-bearing mice is described in Additional file 1: Figure S2.

Three minutes before collecting tissues, intravascular leukocytes were stained by intravenous injection of fluorescein isothiocyanate (FITC)-conjugated antibody (3 μg/mouse) against CD45 [12]. Single cell suspensions were prepared by enzymatic or mechanical dissociation of tissues with or without subsequent density separation, as described previously [13, 14]. Flow-Count fluorospheres (Beckman Coulter, USA) were used to determine cell numbers. Cells were pretreated with Fc block reagents (anti-mouse CD16/CD32 antibody, clone 2.4G2; BioXcell), then stained with a mix of fluorophore-conjugated anti-mouse antibodies as indicated in Additional file 1: Table S2. Data were acquired on a Gallios flow cytometer (Beckman Coulter) and analyzed by using FlowJo 10.5.3 software (FlowJo, LLC, USA). Nonviable cells were excluded from the analysis based on forward and side scatter profiles, and dead cells were excluded by propidium iodide (PI) staining. For intracellular cytokine detection, enriched tumor-infiltrating CD8⁺ T cells were re-stimulated with 1 μg/ml ionomycin (IM) and 25 ng/ml phorbol myristate acetate (PMA) in the presence of GolgiPlug (BD Biosciences, USA) for 4 h at 37 °C. The re-stimulated CD8⁺ T cells were stained with surface antigens, and these cells were stained for intracellular cytokines using a Cytofix/Cytoperm kit (BD Biosciences, USA), according to the manufacturer’s instructions. For the transcriptome analysis, CD8⁺ T cells from the tumor were sorted on FACSAria II Cell Sorter (BD Biosciences, USA).

Bone marrow cells were extracted from the femurs of Ly5.1 mice and hematopoietic progenitors were enriched by depleting lineage (CD3, B220, NK1.1, Ly-6G, Ter119) positive cells with magnetic beads (Miltenyi Biotec, Germany). Bone marrow-derived dendritic cells (BMDCs) were generated by culturing hematopoietic progenitors for 7 days in complete medium (RPMI 1640, 55 μM 2-mercaptoethanol, 1 mM sodium pyruvate, 10 mM HEPES, 100 U/mL Penicillin-Streptomycin, 0.1 mM non-essential amino acids, and 10% fetal bovine serum) with 20 ng/mL GM-CSF. After 7-days of culture, immature BMDCs were further cultured in maturation medium (complete medium with 10 ng/mL GM-CSF and 0.5 μg/mL lipopolysaccharide) for 24 h.

Pmel-1 (CD90.1⁺) CD8⁺ T cells were enriched from spleen single cell suspensions by depleting the lineage (CD4⁺, CD11b⁺, CD11c⁺, B220⁺, NK1.1⁺, Ter119⁺) on an autoMACS cell separator (Miltenyi Biotec, Germany). Pmel-1 CD8⁺ T cells were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) at a final concentration of 2 μM/ 3 × 10⁶ cells/ml for 5 min at room temperature. In the DC-dependent assay, CFSE-labeled Pmel-1 CD8⁺ T cells were cultured with gp100-pulsed BMDCs (pre-stimulation with 1 μg/mL gp100 for 2 h) in complete medium with or without 100 ng/mL HMGN1 for 48 h. In the DC-independent assay, CFSE-labeled Pmel-1 CD8⁺ T cells were cultured in a dish pre-coated with anti-CD3/CD28 antibodies with complete medium with or without 100 ng/mL HMGN1 for 72 h. The proliferation of activated Pmel-1 CD8⁺ T cells (CD25⁺CD90.1⁺CD8⁺) was assessed by CFSE intensity using flow cytometry.

The whole transcripts were amplified from sorted CD8⁺ T cells and those transcripts were used to generate the 3’end Serial Analysis of Gene Expression (SAGE)-sequencing libraries (Additional file 1: Method S2). The sequencing was performed by using an Ion Hi-Q Chef kit, an Ion PI v3 Chip kit, and an Ion Proton Sequencer (Thermo Fisher Scientific) according to the manufacturer’s instructions except the input library concentration was 100 pM. Adapter trimming and quality filtering of sequencing data were performed by using Trimommatic-v0.36 [15] and PRINSEQ 0.20.4 [16]. The filtered reads were mapped on Refseq mm10 using Bowtie2–2.2.5 (parameters: -t -p 11 -N 1 -D 200 -R 20 -L 20 -i S,1,0.50 --norc). The reads unmapped to NlaIII cutting sites were removed, and the mapped reads per gene (raw tag counts) were be quantified as gene expression. Between-sample normalization of gene expression was performed against raw count data by using R 3.4.3 (https://cran.r-project.org/) with TCC [17], DESeq2 [18], and edgeR [19] packages. Genes with adjusted p value less than 0.05 and a fold-change of ≥ 2 between at least two samples were identified as statistically significant differentially expressed genes (DEGs). Raw data from the experiment have been deposited in the NCBI Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo); accession GSE113307.

Functional analysis of DEGs was performed by using Cytoscape 3.6.0 with ClueGO plugin (v2.5.0) [20, 21]. Significantly enriched Gene Ontology (GO) terms [22] (GO-biological process, GO levels: 3–8) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway terms [23] in DEGs were explored and grouped, and a term network was constructed based on the overlap of their elements (kappa score = 0.4). Leading terms within each group were defined as the most significantly enriched term in each group. Terms not connected with any other term were excluded. Significantly-enriched functional terms (adjusted p values < 0.05) for which genes from up- or down-regulated genes comprised over 60% of all genes are shown in the Additional file 1: Table S4. We used versions of the GO term database (Nov. 20, 2017) and KEGG pathway term database (Nov. 20, 2017).

The primers used in this study were designed using the University Probe Library Assay Design Center (Roche, Switzerland) and are listed in Additional file 1: Table S3. Quantitative real time PCR (qPCR) was performed using Thunderbird SYBR qPCR Mix (Toyobo, Japan) or Thunderbird Probe qPCR Mix (Toyobo) on an Applied Biosystem® 7500 qPCR system (Thermo Fisher Scientific, USA). Three independent biological replicates for each group and double technical replicates for each biological replicate were analyzed. Target gene expression was normalized to the expression of the internal control gene, Gapdh.

Unless otherwise noted, all experiments were performed at least three times. Data are presented as mean ± SEM. The significance of differential expression of genes in TCC-normalized 3′ SAGE-seq data was calculated by using TCC package (glmL RT formula in edgeR package) in R 3.4.2. The significance of GO term enrichment was calculated by using Cytoscape 3.6.0 with ClueGO plugin (v2.5.0). Correction for multiple comparisons was performed using the Benjamini-Hochberg method. For comparisons between groups in the in vivo study, we used one-way ANOVA with the Dunnett post hoc test in GraphPad Prism software 6.0 (GraphPad Software, USA). For comparisons between the means of two variables, we used two-tailed unpaired Student’s t-test in GraphPad Prism software 6.0. All statistical analyses were conducted with a significance level of α = 0.05 (P < 0.05).

Two hundred μg anti-CD4 depleting antibody (αCD4) on day 5 and 9 [2], and various doses (0.0032 to 2 μg/ml) of murine HMGN1 (mH) on day 9, 14, 17, and 20 were administered intraperitoneally to Colon26-bearing mice (Fig. 1a). Monotherapy with αCD4 showed moderate tumor growth inhibition, whereas the combination of αCD4 with mH at doses higher than 0.08 μg per injection showed significant tumor growth inhibition in Colon26-bearing mice (p < 0.001) (Fig. 1b). We observed the highest tumor growth inhibition in the mice receiving aCD4 with an mH concentration of 0.08–0.4 μg per injection (Fig. 1b). Based on low doses results, we adopted a dose of 0.16 μg mH per injection as an effective dose in this study. In this setting, combination therapy of mH/αCD4 showed significant tumor growth inhibition relative to monotherapy with αCD4 on day 18 (p = 0.001) and day 24 (p < 0.001) (Fig. 1c). Similar anti-tumor effects were observed in the combination therapy of αCD4 with human HMGN1 in Colon26-bearing mice (Additional file 1: Figure S2A, B), suggesting that human HMGN1 might share the similar structure and function with murine HMGN1, and might have cross-species activity in its anti-tumor effects.Fig1Fig. 1

HMGN1/αCD4 treatment exerted robust anti-tumor effects in mice. a The optimized protocol for HMGN1/αCD4 treatment in Colon26 model. b The effective dose of murine HMGN1 (mH) in combination with anti-CD4 depleting antibody (αCD4) was screened within a range of 0.0032 to 2 μg/mL mH. #, P < 0.05, ##, P < 0.01, ###, P < 0.001 for a Student’s t-test comparing mH/αCD4 and αCD4-treated groups at day 29. c The tumor growth analysis during mH/αCD4 treatment, and the tumor volume of each mice on day 18. d The tumor progression analysis for 40 days. e Tumor rechallenge analysis. f The tumor growth analysis during mH/αCD4 treatment in Colon26-bearing BALB/c-nu model, and the tumor volume of each mice on day 18. Tumor growth is representative of three independent experiments with at least eight mice per group. Data are presented as mean ± SEM. *, P < 0.05, **, P < 0.01, ***, P < 0.001 for a dunnett’s post hoc test (compared with control); p values in the figure indicate Student’s t-test comparing mH/αCD4-treated and αCD4-treated groups

To investigate the nature of the T-cell mediated anti-tumor immunity during mH/αCD4 treatment, we prepared single cell suspensions from the implanted mouse tumor tissue and draining lymph node on day 13. Among tumor-infiltrating cells, we identified a population of lineage⁻ (CD11b⁻ CD11c⁻ CD49b⁻ Ly-6C⁻ Ly6G⁻ Ter119⁻) CD3⁺ CD4⁺ or CD3⁺ CD8⁺ T cells (Fig. 2a). Most of CD4⁺ T cells were depleted in the tumor and draining lymph nodes after mice were treated with αCD4 (Fig. 2b, Additional file 1: S4A).Fig2Fig. 2

HMGN1 expanded CD8⁺ T cells in the tumor. a Flow cytometry gating of tumor-infiltrating CD8⁺ T cell populations. b, c The compartment, frequency, and cell number of CD8⁺ T cells in the tumor by using flow cytometry. d, e The compartment, frequency, and cell number of CD44^hi PD-1⁺ CD8⁺ T cells in the tumor. f, g The compartment, frequency, and cell number of CD137⁺ PD-1⁺ CD8⁺ T cells in the tumor. Each result is representative of three independent experiments with at least four mice per group. Data are presented as mean ± SEM. *, P < 0.05, **, P < 0.01 for a dunnett’s post hoc test (compared with control); p values in the figures indicate Student’s t-test comparing mH/αCD4-treated and αCD4-treated groups

To understand why the mH/αCD4 treatment synergistically increases CD8⁺ T cells in the tumor, we focused on the C-C chemokine receptor (CCR) 7⁺ migratory DCs. Previous reports described that CCR7 is essential for the DC trafficking from the tumor to draining lymph nodes. Those CCR7⁺ migratory DCs can support priming and expansion of CD8⁺ T cells [29–31]. Here we defined the CCR7⁺ migratory DCs as lineage⁻ (CD3⁻ B220⁻ Ly6C⁻ Ly6G⁻) MHC class II⁺ CD11c⁺ CD11b⁺ CCR7⁺ cells (Fig. 3a, Additional file 1: S5). On day 13, we found a higher percentage and an increased number of CCR7⁺ migratory DCs in the tumor of mice treated with mH or mH/αCD4 (p < 0.05), relative to the mice without mH treatment (Fig. 3b, c). Surprisingly, those increased number of CCR7⁺ migratory DCs with an activated phenotype of CD80^hi CD86^hi MHC class II^hi were only observed in mice received mH/αCD4 treatment (p < 0.02) (Fig. 3d), suggesting that those activated CCR7⁺ migratory DCs might drive CD8⁺ T expansion.Fig3Fig. 3

HMGN1/αCD4 treatment increased CCR7⁺CD80^hiCD86^hi migratory DCs in the tumor. a Flow cytometry gating of CCR7⁺ migratory DCs. b, c The frequency and number of CCR7⁺ migratory DCs in the tumor. d The expression of CD80, CD86, and I-A/I-E (also known as MHC class II) on CCR7⁺ migratory DCs. Each result is representative of three independent experiments with at least four mice per group. Data are presented as mean ± SEM. *, P < 0.05, **, P < 0.01, ***, P < 0.001 for a dunnett’s post hoc test (compared with control); p values in the figure indicate Student’s t-test comparing mH/αCD4-treated and αCD4-treated groups. Mean fluorescence intensity (MFI)

To imitate the in vivo CD8⁺ T cell expansion, we first generated a CD8⁺ T cell expansion assay by co-culturing Pmel-1 CD8⁺ T cells with gp100-pulsed BMDCs in vitro (Fig. 4a). After 48-h co-culture, we observed a 13% of proliferating Pmel-1 CD8⁺ T cells (CD25⁺ CD90.1⁺ CD8⁺ T cells) in control group and a 20.9% of proliferating Pmel-1 CD8⁺ T cells in mH-added group (Fig. 4b). There was an increased number of proliferating Pmel-1 CD8⁺ T cells in mH-added group relative to unadded control (p = 0.003) (Fig. 4c), suggesting that HMGN1 might enhance the CD8⁺ T cell proliferation in the presence of DCs.Fig4Fig. 4

HMGN1 promoted DC-dependent CD8⁺ T cell expansion ex vivo. a The experimental design of CD8⁺ T cell expansion through CD3/CD28 stimulation or DC co-culture. b A CD8⁺ T cell expansion assay of co-culturing naïve Pmel-1 CD8⁺ T cells with Ly5.1⁺ gp100-specified BMDCs. After 48-h co-culture, the 13% increase in number of generation (G) ≥ 2, CD25⁺ CD90.1⁺ CD8⁺ T cells in control and the 20.9% increase in mH-added group. c The cell number of CD25⁺ CD8⁺ T cells in G ≥ 2 cell generation. d The CD8⁺ T cell expansion assay of stimulating naïve Pmel-1 CD8⁺ T cells by using anti-CD3 antibody and anti-CD28 antibody. After 72-h stimulation, the 55.4% increase in number of G ≥ 2, CD25⁺ CD90.1⁺ CD8⁺ T cells in control and the 53.4% increase in mH-added group. Each result is representative of three independent experiments. Data are presented as mean ± SEM. A Student’s t-test is used to compare the statistically significant difference between the mH-added and mH-unadded groups

Following antigen exposure, CD8⁺ T cells proliferate and differentiate into effector cells, then most of effector cells die off, leaving a small fraction of these cells to become memory cells. However, long-term antigen exposure converts CD8⁺ T cells into exhausted cells in tumor microenvironment. These exhausted CD8⁺ T cells express co-inhibitory molecules (e.g., PD-1, LAG-3, TIM-3, TIGIT) and lose the ability to produce multiple cytokines such as interferon (IFN)-γ, tumor necrosis factor (TNF)-α, and interleukin (IL)-2, namely they lose anti-tumor activities [32–35].

To determine whether mH/αCD4 treatment is able to restore the effector functions of those exhausted CD8⁺ T cells, we analyzed the expression of co-inhibitory molecules and the ability of cytokine production from these PD-1^hi CD8⁺ tumor-infiltrating T cells on day17 after tumor inoculation (Fig. 5a). We observed that the expression of LAG-3 and TIM-3 on the PD-1^hi CD8⁺ T cells significantly decreased in the mH/αCD4-treated mice compared to the untreated control mice (p < 0.001) and the αCD4-treated mice (p = 0.002), whereas no significant difference was observed in the TIGIT expression among treatment groups (Fig. 5b). The number and percentage of LAG-3⁺ PD-1⁺ CD8⁺ T cells decreased in the mH/αCD4 treated mice relative to that of the αCD4 treated mice (p = 0.04) (S6A, B), whereas the percentage of TIM-3⁺ PD-1⁺ or TIGIT⁺ PD-1⁺ was equivalent among treatment groups (Additional file 1: Figure S6C). In addition, the proportion of the IFN-γ-, TNF-α-, and IL-2-producing multifunctional population within the PD-1^hi CD8⁺ T cells was higher in the mH/αCD4-treated mice than in the αCD4 treated (P < 0.001) or untreated control (P < 0.001) mice (Fig. 5c, d). Collectively, these results suggested that the HMGN1/αCD4 treatment reduces the level of co-inhibitory molecules in these exhausted CD8⁺ T cells, increases their effector functions, and decreases the proportion of exhausted CD8⁺ T cells in the tumor.Fig5Fig. 5

HMGN1/αCD4 treatment synergistically reduced exhausted CD8⁺ T cells in the tumor. a The compartment of PD-1^hi, PD-1^lo, PD-1⁻ CD8⁺ T cells in the tumor. b The expression level of LAG-3, TIM-3, and TIGIT on PD-1^hi CD8⁺ T cells. c The compartment and frequency of IFN-γ⁺ among in PD-1⁺ CD8⁺ T cells. d The compartment and frequency of IL-2⁺ TNF-α⁺ among in IFN-γ⁺ PD-1⁺ CD8⁺ T cells. Each result is representative of three independent experiments with at least four mice per group. Data are presented as mean ± SEM. *, P < 0.05, **, P < 0.01, ***, P < 0.001 for a dunnett’s post hoc test (compared with control); p values in the figure indicate Student’s t-test comparing mH/αCD4-treated and αCD4-treated groups. Geometric mean fluorescence intensity (GMFI)

To gain further insight into the phenotypic changes of CD8⁺ T cells during HMGN1/αCD4 treatment, we performed a transcriptome analysis on tumor-infiltrating CD8⁺ T cells on day 16. We identified 517 differentially expressed genes (DEGs) with adjusted p value less than 0.05 and a fold-change of ≥2 between at least two samples. In comparison with αCD4 treatment group, CD8⁺ T cells had 153 up-regulated and 364 down-regulated DEGs in mH/αCD4 treatment group (Fig. 6a). GO analysis revealed that these 153 up-regulated genes significantly included T cell activation-, chemokine production-, responses to interferons-, and ribosome related translation processes-related genes (Additional file 1: Figure S7). Particularly, these genes included cytotoxic function-related (Gzmm), chemotaxis-related (Ccl5), interferon-related (Ifitm1, Ifitm2, Ifitm3), T cell survival and memory-related (Rora, Rara, Irf1), and T cell-T cell interaction-related (Icam1) genes (Fig. 6b). In contrast, those 364 down-regulated genes significantly included cell cycle-, cell division-, telomere maintenance- and chromosome organization processes-related genes (Additional file 1:Figure S7). These genes included co-inhibitory-related (Pdcd1, Havcr2, Cd226), glycolysis-related (Pgam1), telomere-related (Nek2, Hnrnpa1, Hnrnpc, Dkc1, Cct2), and cell cycle/ cell division-related (Atr, Bub1b, Ccnb2, Ccnd3, Ccng2, Cdc27, Cdkn1b, Chek2, Mcm6, Pcna) genes (Fig. 6b).Fig6Fig. 6

CD8⁺ T cell transcriptome analysis revealed the promotion of activation after HMGN1/αCD4 treatment. a Analysis scheme of transcriptome analysis of CD8⁺ T cells in tumor-bearing mice. b Heatmap representation of DEGs that associated with T cell function-related significantly-enriched GO groups genes. Each column represents group, whereas each row represents an individual gene. Z-score of module groups is shown on the left bottom of the heatmap. c qPCR analysis of expression changes of PD-1(Pdcd1), TIM-3 (Havcr2), LAG-3 (Lag3), and CTLA-4(Ctla4). The expression levels of all target mRNAs were normalized against the expression level of Gapdh in each sample. Data are presented as mean ± SEM. A Student’s t-test is used to compare the statistically significant difference between mH/αCD4-treated and αCD4-treated groups