Abstract
Purpose: Humanized mouse models using NOD/Shi-scid-IL2rγnull (NOG) and NOD/LtSz-scid IL2rγnull (NSG) mouse are associated with several limitations, such as long incubation time for stem cell engraftment and the development of xenograft versus host disease in mice injected with peripheral blood mononuclear cells (PBMCs). To solve problems, we used humanized major histocompatibility class I- and class II-deficient NOG mice (referred to as NOG-dKO) to evaluate the antitumor effect of anti-programmed death-1 (PD-1) antibody.
Experimental Design: Humanized NOG-dKO mice, in which human PBMCs and human lymphoma cell line SCC-3, or glioblastoma cell line U87 were transplanted, were used as an immunotherapy model to investigate the effect of anti-PD-1 antibody. A biosimilar anti-PD-1 mAb generated in our laboratory was administered to humanized NOG-dKO mice transplanted with tumors.
Results: Within 4 weeks after transplantation, human CD45+ cells in antibody-treated mice constituted approximately 70% of spleen cells. The injection of anti-PD-1 antibody reduced by more 50% the size of SCC-3 and U87 tumors. In addition, induction of CTLs against SCC-3 cells and upregulation of natural killer cell activity was observed in the antibody-treated group. Tumor-infiltrating lymphocyte profiling showed that more exhausted marker (PD1+TIM3+LAG3+) positive T cells maintained in anti-PD-1 antibody–treated tumor. A greater number of CD8+ and granzyme-producing T cells infiltrated the tumor in mice treated with the anti-PD-1 antibody.
Conclusions: These results suggest that NOG-dKO mice might serve as a good humanized immunotherapy model to evaluate the efficacy of anti-PD-1 antibody prior to the clinical treatment. Clin Cancer Res; 23(1); 149–58. ©2016 AACR.
Despite the clinical success with immune checkpoint antibody therapy against advanced cancers, there are still difficulties associated with predicting precisely the clinical responses prior to the treatment. Biomarker studies are also performed intensively; however, these are not potent enough for the precise prediction. In this study, we developed and used a novel NOG-MHC dKO mouse model. We found that this humanized mouse model exhibited no obvious signs of GVHD and that the injection of anti-PD-1 antibody inhibited PD-L1–positive SCC-3 tumor growth and induced tumor-specific immune responses. Therefore, in vivo investigation of anti-PD-1 antibody effect using humanized NOG-dKO mouse can contribute to the profiling of patients to predict the efficacy of anti-PD-1 antibody prior to the clinical treatment. These observations indicate that the NOG-dKO mouse might be a good tool giving a promising future to a translational research of immune checkpoint antibody therapy.
Introduction
With the recent success of immune checkpoint antibodies, such as ipilimumab and nivolumab, reported in patients with metastatic melanoma, many ongoing clinical trials are underway to evaluate their efficacy in various solid cancers other than melanomas (1–5). Specifically, a promising combination therapy of ipilimumab and nivolumab has demonstrated a very high response rate and long-term survival benefit in patients with advanced cancers, including non–small cell lung cancer (6–8).
Despite these promising results, the response rate associated with single-antibody treatment is approximately 20% to 40%. Furthermore, it is difficult to accurately predict the responders to antibody therapy based on the results of preclinical studies (9, 10).
Multiple types of humanized mice have been developed and used as a therapeutic model in preclinical studies of new cancer treatments. The severely immunodeficient mouse strains, such as NOD/Shi-scid-IL2rγnull (NOG; refs. 11, 12), NOD/LtSz-scid IL2rγnull (NSG; ref. 13), and BALB/c Rag2null IL2rγnull (14) are ideal in vivo platforms for reconstituting the human hemato-lymphoid system due to their lack of an endogenous mouse immune system. Several researchers demonstrated these humanized mouse models are capable of temporary antigen-specific immune responses, such as CTL activation (15) and human antibody production (16, 17).
In addition, humanized mouse models transplanted with both human peripheral blood mononuclear cells (PBMCs) and cancer cells have been used in some preclinical studies to evaluate antibody-based therapies (18–20). These mouse models are amenable to hematopoietic stem cell transplantation; however, they cannot undergo human PBMC transplantation due to a severe xenograft versus host disease (xeno-GVHD) response (21, 22).
In this study, we used MHC gene-double knockout (dKO) NOG mice, deficient in both the murine MHC class I and class II genes. MHC dKO NOG mice were generated by KO of the gene encoding β2-microglobulin, a component of the MHC class I molecule, and the gene encoding IAβ, the light chain of the MHC class II molecule (referred to as NOG-dKO or NOG-β2m, IAβ dKO mice). Yaguchi and colleagues reported for the first time that NOG-dKO mice that had undergone transplantation with human PBMCs exhibited a much milder xeno-GVHD response, with less weight loss and a longer survival period, compared with control NOG mice (23).
Consistent with these results, we were able to successfully transplant PBMCs into NOG-dKO mice, whereas transplanted PBMCs were rejected by regular NOG mice. On the basis of these observations, we established humanized NOG-dKO mice transplanted with both PBMCs and human cancer cells as a model to evaluate cancer immunotherapy. We used this therapeutic in vivo model to evaluate antitumor activity of anti-PD-1 antibody with respect to the immunologic response of human origin.
Materials and Methods
Antibodies and flow cytometry
The following antibodies were used for a flow cytometric analysis. The anti-mouse CD45 antibody used to label mouse cells was purchased from BD Pharmingen. The anti-CD3-biotin (HIT3a), anti-CD4-PE (RPA-T4), anti-CD8-FITC (HIT8a), anti-CD11b-PE-Cy7 (ICRF44), anti-CD14-PerCP (MP9), anti-CD19-FITC (HIB19), anti-CD25-FITC (M-A251), anti-CD33-PE (WM53), anti-CD45-FITC (2D1), anti-CD45RA-FITC (HI100), anti-CD45RO-APC (UCHL1), anti-CD56-PE (B159), anti-CD127-PE-Cy7 (A019D5), anti-CD138-APC (MI15), anti-human IgD-PE (IA6-2), anti-human IgM-PE-Cy5 (G20-127), and anti-human IgG-PE (G18-145) used to label human cells were also purchased from BD Pharmingen. Anti-FoxP3-PE (hFOXY) antibody was purchased from eBioscience, Inc. Anti-TIM3-PE (F38-2E2) and anti-LAG3-FITC (17B4) antibodies were purchased from Miltenyi Biotec and Adipogen. Propidium iodide (PI) was purchased from Sigma-Aldrich Co. and used to distinguish living cells from those that had undergone cell death. The anti-PD-1-APC (EH12.2H7), anti-PD-L1-APC (29E.2A3), and anti-Ki67-PE-Cy7 (Ki-67) antibodies were purchased from BioLegend Inc.
Single-cell suspensions were obtained from mouse spleens and peripheral blood using ACK lysing buffer (Thermo Fisher Scientific). Tumor-infiltrating lymphocytes (TILs) were also separated from the control or anti-PD-1 antibody–treated tumors by anti-human CD45-microbeads (Miltenyi Biotec) using autoMACS system (Miltenyi Biotec). Cells were stained with primary antibodies for 15 minutes at 4°C and washed with cold PBS + 2% FBS. If applicable, cells were subsequently stained with the secondary antibodies for 15 minutes at 4°C. Cells were then washed, fixed with 0.5% paraform aldehyde–containing PBS (−), and analyzed on a FACSCanto II flow cytometer (BD Biosciences). Human cells were identified by gating the mouse CD45−PI− and human CD45+ fractions.
Production of the full-length anti-PD-1 mAb
We generated biosimilar of the anti-PD-1 mAb nivolumab (manuscript in submission). Briefly, the amino acid sequence of nivolumab was downloaded from the J-PlatPat data base from the National Center for Industrial Property Information and Training (INPIT; https://www.j-platpat.inpit.go.jp/web/tokujitsu/tkbs/TKBS_GM401_ToPDF.action). The anti-PD-1 mAb was produced using the expi293 expression system, purified with a protein A column, and used for the in vivo experiments.
Development of humanized NOG-dKO mice
Six-week-old NOG-dKO mice were kindly supplied from Dr. Mamoru Ito (The Central Institute for Experimental Animals, Kawasaki, Japan). All animals were cared for and treated humanely according to the Guidelines for the welfare and use of animals in cancer research, and the experimental procedures approved by the Animal Care and Use Committee of Shizuoka Cancer Center Research Institute.
The clinical experiments using PBMCs derived from patients with glioma were approved by the Institutional Review Board of Shizuoka Cancer Center, Shizuoka, Japan. All patients provided written informed consent.
Eight-week-old NOG-dKO mice were irradiated with X-rays (2.5 Gy; HW-150; HITEX) on day 0, and 1 × 107 human PBMCs cells with the HLA-A*0201 genotype were intravenously administered to each mouse via the tail vein. For human cell monitoring, NOG-dKO mouse spleens and peripheral blood were obtained from sacrificed mice at 4, 6, and 8 weeks after transplantation. Engraftment of human immune cells was investigated using FACS analysis.
In the survival experiment, overall survival was compared in humanized NOG-dKO mice and control NOG mice that had undergone transplantation with human PBMCs. We set the natural death as an endpoint in this study.
The study design of the experiment evaluating mice treated with the anti-PD-1 antibody is shown in Fig. 1. Human PBMCs (HLA-A*0201 positive) were injected on day 0, and 2 × 105 SCC-3 human lymphoma cells (six mice) or U87 glioblastoma cell line (four mice) with HLA-A*0201 genotype were subcutaneously injected into the flank region in each group on day 1. SCC-3 cells expressed high levels of phosphorylated STAT3 and PD-L1 (Supplementary Fig. S1). However, U87 cells showed low expression levels of both markers.
Experimental design and treatment schedule of anti-PD-1 therapy against SCC-3 and U87 tumors. Human PBMCs (HLA-A*0201 positive) were injected on day 0, and SCC-3 or U87 cells with the HLA-A*0201 genotype were subcutaneously injected into six or four mice from each group on day 1. Beginning on day 28, anti-PD-1 was administered intraperitoneally biweekly until six injections had been administered. Mouse peripheral blood was obtained weekly following the initiation of anti-PD-1 injections. One week after the last injection of anti-PD-1 antibody, spleens and tumors were harvested from the control group and the anti-PD-1–treated group.
Experimental design and treatment schedule of anti-PD-1 therapy against SCC-3 and U87 tumors. Human PBMCs (HLA-A*0201 positive) were injected on day 0, and SCC-3 or U87 cells with the HLA-A*0201 genotype were subcutaneously injected into six or four mice from each group on day 1. Beginning on day 28, anti-PD-1 was administered intraperitoneally biweekly until six injections had been administered. Mouse peripheral blood was obtained weekly following the initiation of anti-PD-1 injections. One week after the last injection of anti-PD-1 antibody, spleens and tumors were harvested from the control group and the anti-PD-1–treated group.
Starting on day 28, the anti-PD-1 antibody (2 mg/kg) was administered intraperitoneally biweekly six times. The antitumor activity of the anti-PD-1 antibody was evaluated by measuring tumor volume. Tumor volume was calculated on the basis of the National Cancer Institute formula as follows: tumor volume (mm3) = length (mm) × [width (mm)]2 × 1/2. Mouse peripheral blood was collected from the retro-orbital venous plexus using heparinized pipettes on a weekly basis after anti-PD-1 antibody injection. One week after the last injection of anti-PD-1 antibody, spleens and tumors were harvested from the control group and anti-PD-1 antibody–treated group. Blood and spleen cells from one set of three mice were used for in vitro assays, including the CTL induction assay, the natural killer (NK) cell assay, and PCR analysis of cytokine expression. Tumors from the other set of three mice were primarily used for TILs analysis and IHC analysis.
CTL induction assay
CTL induction cultures were described previously (24). Briefly, spleen cells harvested from control and anti-PD-1 antibody-treated mice were restimulated with irradiated (180 Gy) SCC-3 cells at a ratio of 10:1 in the presence of IL2 for 7 days. Stimulated CTLs (1 × 105) and living SCC-3 cells (1 × 105) were coincubated in a round-bottom 96-well microculture plate for 24 hours. The supernatants were subsequently collected and IFNγ levels were measured using an ELISA Kit specific for human IFNγ (Biosource).
NK-cell assay
Harvested splenocytes were used as effector cells, and K562 cells, purchased from ATCC, were used as target cells. The effector:target (E:T) ratio ranged from 100:1 to 11:1. The cytotoxic activity of NK cells was measured using the DELFIA nonradioactive cytotoxicity assay (PerkinElmer Inc.) as reported previously (24). The percentage of specific lysis was determined by the following formula: percentage of specific lysis = (experimental release – spontaneous release)/(maximal release – spontaneous release) × 100.
RT-PCR analysis of cytokine expression
The RT-PCR analysis of cytokines, stem cell, and epithelial–mesenchymal transition (EMT) marker genes using the 7500 Real Time PCR System (Thermo Fisher Scientific) was performed as described previously. Briefly, all PCR primers [IL2, IL4, IL10, TNFα, IFNγ, IL12A, TGFβ1, GAPDH for cytokines, c-Myc, NANOG, NES, Oct3/4, SOX2 for stem cell markers; BIRC5 (Survivin), CCND1 (CyclinD1), FOXC2, MMP2, SMAD2, SNAIL1, SNAIL2, TCF4 and TWIST1 for EMT-associated genes] and TaqMan probes were purchased from Thermo Fisher Scientific. Total RNA was isolated from peripheral blood cells, spleen cells, and SCC-3 tumors using the NucleoSpin RNA Kit (Macherey-Nagel GmBH & Co.). Complementary DNA was synthesized using SuperScript III RTase and Oligo (dT)20 primer.
Immunohistochemistry
Anti-CD4 (4B12) and anti-CD8 (C8/144B) antibodies (Thermo Fisher Scientific), anti-granzyme B antibody (GrB-7; DAKO), anti-IL17 antibody (H-132; Santa Cruz Biotechnology Inc.), anti-FoxP3 antibody (236A/E7; Abcam), anti-CD204 antibody (SRA-C6; TransGenic Inc.), and anti-phospho-STAT3 antibody (D3A7; Cell Signaling Technology, Inc.) were purchased and used for immunohistochemical analysis. More than 10 areas of tumor at a high magnification (×200) in each section stained with various antibodies was calculated using image analysis software, Winroof (Mitani Corporation). TILs were analyzed using various antibodies on sections from control and anti-PD-1 antibody–treated groups.
Statistical analysis
Significant difference was analyzed using Student t test and Mann–Whitney U test. Values of P < 0.05 were considered statistically significant. Survival curves of NOG mice transplanted with human PBMCs were estimated using Kaplan–Meier method and log-rank test was used to compare the survival curves.
Results
PBMC transplantation in NOG-dKO mice
NOG-dKO mice received 2.5 Gy radiation and an intravenous injection of 1 × 107 PBMCs on day 0, as shown in Fig. 1. The engraftment of PBMCs was measured every 2 weeks for up to 8 weeks. The percentage of human CD45+ cells in the treated mice was approximately 30% in the blood and 60% in the spleens up to 6 to 8 weeks (Supplementary Table S1). Importantly, the total number of spleen cells increased severalfold compared with pretransplantation levels, which indicated injected human PBMC expanded similarly in mice. As to subpopulation of spleen cells, the ratio of human CD19+ cells in CD45+ human leukocytes reached over 30% at 4 weeks in spleens, but those cells rapidly diminished at 6 weeks. Interestingly, approximately more than half of CD19+ B cells were CD138+ during the entire period of monitoring. In contrast, CD3+ T cells reached more than 50% of human CD45+ cells at 4 weeks and increased up to 90% at 6 weeks. With regard to the subpopulations of CD3+ cells, CD8+ cells were dominant compared with CD4+ cells (78.7% vs. 19.8%).
Survival benefit from humanized NOG-dKO mice compared with humanized NOG mice
In another transplantation experiment using human PBMCs, the survival time between NOG-dKO and control NOG mice under the human PBMCs transplantation was compared. All five humanized NOG-dKO mice were alive 10 weeks after the transplantation. In contrast, all the control humanized NOG mice had died within 10 weeks after transplantation (Supplementary Fig. S2).
Antitumor effect of the anti-PD-1 antibody against SCC-3 and U87 tumors
After the initiation of treatment with the anti-PD-1 antibody, SCC-3 and U87 tumors decreased in size by more than 50% during six injection treatments on day 18 in SCC-3 tumor and on day 18 and 20 in U87 tumor between control and anti-PD-1 antibody–treated mice (Fig. 2). In addition, anti-PD-1 antibody treatment showed a tendency of preventing weight loss in the treated mice through the treatment period. However, antitumor effects of anti-PD-1 antibody against both tumors were not statistically significant compared with the control group. Of particular significance, anti-PD-1 antibody treatment did not significantly affect the engraftment of human PBMCs compared with the control group (Supplementary Table S2).
Inhibitory effect of anti-PD-1 on the growth of SCC-3 and U87 tumors in vivo. A, V/V0 values of anti-PD-1 antibody-treated SCC3 tumors (n = 6) are shown. The efficacy of the antibody treatment was expressed as the mean V/V0 value, where V is the tumor volume on the day of evaluation and V0 is the tumor volume on the day of treatment. B, The mean tumor volume of anti-PD-1 antibody–treated U87 tumors (n = 4) is shown. Body weight change in anti-PD-1–treated mice bearing SCC-3 tumors (C), and U87 tumors (D). () Control group, (
) Anti-PD-1–treated group. Anti-PD-1 was administered intraperitoneally six times. Each point represents the mean value derived from six mice in SCC-3 tumors or four mice in U87 tumors.
Inhibitory effect of anti-PD-1 on the growth of SCC-3 and U87 tumors in vivo. A, V/V0 values of anti-PD-1 antibody-treated SCC3 tumors (n = 6) are shown. The efficacy of the antibody treatment was expressed as the mean V/V0 value, where V is the tumor volume on the day of evaluation and V0 is the tumor volume on the day of treatment. B, The mean tumor volume of anti-PD-1 antibody–treated U87 tumors (n = 4) is shown. Body weight change in anti-PD-1–treated mice bearing SCC-3 tumors (C), and U87 tumors (D). () Control group, (
) Anti-PD-1–treated group. Anti-PD-1 was administered intraperitoneally six times. Each point represents the mean value derived from six mice in SCC-3 tumors or four mice in U87 tumors.
Flow cytometry analysis of TILs from SCC-3 tumors
The total number TILs from each tumor significantly increased in anti-PD-1 antibody–treated tumor compared with the control (Supplementary Table S3). Among CD45+ TILs in anti-PD-1 Ab-treated tumors, the frequency of CD3+CD8+ cells and PD-1+TIM3+LAG3+ cells increased in anti-PD-1 Ab–treated tumors. These results might suggest that anti-PD-1 Ab promoted the infiltration of CTLs and maintained exhausted T-cell populations. Meanwhile, activation marker positive cells (CD4+CD45RO+CD127+) showed a tendency of increase in anti-PD-1 Ab-treated group, but it was not significant (Fig. 3 and Supplementary Table S3). In addition, the frequency of CD4+Ki67+ and CD8+Ki67+ T cells was not different between the control and anti-PD-1 Ab–treated group.
The profiling of TILs using flow cytometry in the control and anti-PD-1 antibody–treated tumors. A, The frequency of CD4+ and CD8+ cells in CD3+-gated population, the frequency of activation T-cell marker (CD45RO+CD127+)-positive cells in CD4+-gated population, and the frequency of exhausted T-cell marker (TIM3+LAG3+)-positive cells in PD-1+-gated population were shown. B, The frequency of CD4+Ki67+ and CD8+Ki67+ cells in CD3+-gated population was shown. Representative data were shown in each cell population.
The profiling of TILs using flow cytometry in the control and anti-PD-1 antibody–treated tumors. A, The frequency of CD4+ and CD8+ cells in CD3+-gated population, the frequency of activation T-cell marker (CD45RO+CD127+)-positive cells in CD4+-gated population, and the frequency of exhausted T-cell marker (TIM3+LAG3+)-positive cells in PD-1+-gated population were shown. B, The frequency of CD4+Ki67+ and CD8+Ki67+ cells in CD3+-gated population was shown. Representative data were shown in each cell population.
Induction of CTL and NK activity in anti-PD-1–treated humanized mice
CTL activity against SCC-3 cells was identified using an IFNγ production assay in spleen cells derived from the anti-PD-1-treated humanized NOG-dKO (Fig. 4A and Supplementary Fig. S3). NK-cell cytotoxic activity targeting K562 cells was observed in all the anti-PD-1–treated mice with an E:T ratio of greater than 33 (Fig. 4B and Supplementary Fig. S3).
Induction of cytotoxic activity in humanized SCC3 tumor-bearing mice treated with anti-PD-1 antibody. A, CTL induction against SCC3 cells from three spleens of anti-PD-1–treated mice. Levels of IFNγ produced by CTL cells stimulated with irradiated SCC-3 cells were measured using an ELISA Kit. Open column, control group; closed column, anti-PD-1–treated group. B, NK-cell cytotoxic activity of three spleens from anti-PD-1–treated mice. Cytotoxic activity was measured using the Nonradioactive Cytotoxicity Assay Kit. C, control group; T, anti-PD-1–treated group. Each column represents the mean value of the triplicate experiments.
Induction of cytotoxic activity in humanized SCC3 tumor-bearing mice treated with anti-PD-1 antibody. A, CTL induction against SCC3 cells from three spleens of anti-PD-1–treated mice. Levels of IFNγ produced by CTL cells stimulated with irradiated SCC-3 cells were measured using an ELISA Kit. Open column, control group; closed column, anti-PD-1–treated group. B, NK-cell cytotoxic activity of three spleens from anti-PD-1–treated mice. Cytotoxic activity was measured using the Nonradioactive Cytotoxicity Assay Kit. C, control group; T, anti-PD-1–treated group. Each column represents the mean value of the triplicate experiments.
Changes in cytokine and other genes expression in anti-PD-1–treated mice
A RT-PCR analysis revealed that TGFβ1 gene expression was downregulated in PBMCs and SCC-3 tumor tissues in the anti-PD-1–treated group (Fig. 5A). In contrast, IFNγ gene expression on day 7 was upregulated in PBMCs in the anti-PD-1–treated group (Fig. 5B). Meanwhile, the expression of stem cell marker genes, such as nestin, Oct3/4, and SOX2 and EMT-associated genes, such as survivin, FOXC2, and MMP2 were suppressed by more than 90% in anti-PD-1–treated SCC3 tumor (Fig. 5C).
Analysis of various gene expression in peripheral blood cells and SCC-3 tumors using RT-PCR. A, Downregulation of TGFβ1 gene expression in PBMC and tumor tissue. Open column, control group; closed column, anti-PD-1–treated group. B, Upregulation IFNγ gene expression in PBMCs on day7. Open column, control group; shaded column, anti-PD-1–treated group. C, Stem cell marker and EMT-associated genes expression. Open column, control group; closed column, anti-PD-1–treated group. Each column represents the mean value of triplicate experiments. **, P < 0.01; *, P < 0.05, statistically significant.
Analysis of various gene expression in peripheral blood cells and SCC-3 tumors using RT-PCR. A, Downregulation of TGFβ1 gene expression in PBMC and tumor tissue. Open column, control group; closed column, anti-PD-1–treated group. B, Upregulation IFNγ gene expression in PBMCs on day7. Open column, control group; shaded column, anti-PD-1–treated group. C, Stem cell marker and EMT-associated genes expression. Open column, control group; closed column, anti-PD-1–treated group. Each column represents the mean value of triplicate experiments. **, P < 0.01; *, P < 0.05, statistically significant.
Immunohistochemical analysis of SCC-3 tumor tissues
Hematoxylin–eosin (H&E) stained tumor specimens did not exhibit obvious histologic differences in the anti-PD-1–treated group compared with the control group, with the exception of patchy lymphoid cell infiltrations observed in antibody-treated group (Fig. 6A). CD8+ and granzyme B+ lymphocytes were more frequently observed in the anti-PD-1–treated group; however, CD4+, FoxP3+ lymphocyte, and CD204+ immune cell levels did not significantly differ (Fig. 6B).
Effect of anti-PD-1 on immune cells infiltrating the SCC-3 tumors. A, Images of control and anti-PD-1–treated tumors stained with H&E, the anti-CD8, anti-granzyme B, and anti-CD204 antibodies. Magnification: ×400 for H&E, granzyme B, ×200 for CD8 and CD204 staining. B, Effect of anti-PD-1 on the number of infiltrating immune cells in SCC-3 tumors. More than 10 fields of view in sections of each tumor at a high magnification (×200). The sections were stained with various antibodies and the images were evaluated using image analysis software (Winroof). Positive cell count per field was compared between control and anti-PD-1–treated groups. **, P < 0.01, statistically significant.
Effect of anti-PD-1 on immune cells infiltrating the SCC-3 tumors. A, Images of control and anti-PD-1–treated tumors stained with H&E, the anti-CD8, anti-granzyme B, and anti-CD204 antibodies. Magnification: ×400 for H&E, granzyme B, ×200 for CD8 and CD204 staining. B, Effect of anti-PD-1 on the number of infiltrating immune cells in SCC-3 tumors. More than 10 fields of view in sections of each tumor at a high magnification (×200). The sections were stained with various antibodies and the images were evaluated using image analysis software (Winroof). Positive cell count per field was compared between control and anti-PD-1–treated groups. **, P < 0.01, statistically significant.
Discussion
Cancer immunotherapy studies using immunocompetent mice serve as an excellent platform for the development of new therapeutic strategies and for providing insight into the functions of the immune system. However, these studies can be associated with significant limitations, as a mouse model is not capable of precisely predicting the clinical outcomes of patients with cancer. Inevitably, the mouse immune system elicits different responses to allogenic human tumor cells, some of which do not resemble the antitumor reaction in the human immune system at all.
In the past decade, severely immunodeficient mouse strains such as NOG, NSG, and BALB/c Rag2null IL2rγnull have been developed and used to establish humanized immunodeficient mouse models. However, allogenic PBMC transplantation in these mice is known to induce a strong xeno-GVHD response that results in lethality. Therefore, allogeneic transplantation studies in these models must be completed before the onset of xeno-GVHD, hereby limiting the time period available to conduct these experiments to 3 to 4 weeks after transplantation (21, 22). Sanmamed and colleagues (20) reported that a combination of the anti-human CD137 antibody (urelumab) and the anti-PD-1 antibody (nivolumab) inhibited tumor growth and increased IFNγ production in conjunction with allogeneic (HT29) or autologous (from a gastric cancer patient) stem cell transplantation in the Rag2−/−IL2Rγnull humanized mouse model of gastric cancer. Interestingly, they also reported that tumor growth was slowed in a strictly autologous transplantation model using PBMCs and surgically resected tumor tissue from the same patient. They concluded that these results validated the significance of this humanized mouse model for evaluating the human immune response; however, the experiments were still limited to a time period of 3 to 4 weeks posttransplantation due to the issue of xeno-GVHD onset.
In this study, we developed and used a novel NOG-dKO mouse model. We found that this humanized mouse model exhibited no obvious signs of xeno-GVHD and survived up to 12 weeks longer than control NOG mice. This observation suggests that the NOG-dKO mouse merits further investigations to evaluate the response of the human immune system to antibodies targeting immune checkpoint molecules.
In our humanized NOG-dKO mice transplanted with SCC-3 or U87 tumor cells, and treated with the anti-PD-1 antibody, tumor growth was inhibited. Furthermore, this antitumor response was not observed in humanized tumor-bearing mice that had not received human PBMC transplantation (data not shown), indicating that anti-PD-1 antibody-induced tumor growth was mediated by the engrafted PBMCs.
With respect to immunologic responses, the following four responses were observed in this study: (i) an SCC-3–specific CTL activity induction, (ii) upregulation of NK-cell activity, (iii) downregulation of TGFβ1 expression in PBMCs and SCC-3 tumors, and (iv) an increase in CD8+ and granzyme B+ T cells and maintenance of exhausted marker (PD-1+TIM3+LAG3+)-positive T cells in the tumor.
The induction of CTL activity targeting tumor cells and intratumoral infiltration of immune effector cells are responses frequently observed in patients with cancer treated with antibodies targeting immune checkpoint molecules (25–29). Although an upregulation of NK-cell activity is not a common observation of clinical trials with anti-PD-1 therapy, this response might depend on the time point at which tumors are harvest and the number of antibody injection. In general, activated NK cells express PD-1 on the cell surface, and IFNγ secretion from effector T cells strongly upregulates the expression of PD-L1 on tumor cells. These events might contribute to the increased resistance of tumor cells to NK-cell lysis (30). Anti-PD-1 treatment has been reported by Benson and colleagues to enhance NK-cell killing activity against multiple myeloma cells by blocking PD-1/PD-L1 signaling (31). Moreover, Das and colleagues demonstrated that an upregulation of genes involved in cytokine and NK-cell functions was observed in patients with cancer given a combination of anti-PD-1 and anti-CTLA-4 treatment (32). This observation suggests that NK-cell activation occurs early in the course of antibody treatment.
Downregulation of TGFβ1 mediated by PD-1/PD-L1 blockade has not been commonly observed in clinical trials and other studies of the immune system. TGFβ1 is a blood marker associated with a poor prognosis and this molecule mediates immunosuppressive activity in the tumor microenvironment via the mechanistic PD-L1 upregulation and activation of STAT3 signaling (33–35). In addition, TGFβ1 is an important factor for inducing regulatory T cells or IL17-producing helper T cells (Th17) in association with IL23 (36). Therefore, it is reasonable to hypothesize that TGFβ1 downregulation contributes to the accumulation of CD8+ and granzyme+ CTLs in the tumor of humanized mice treated with anti-PD-1.
Furthermore, TGFβ1 is also known to be a promoting factor to mediate EMT and maintain stemness, which prefer to cancer progression and metastasis. A RT-PCR study revealed the downregulation of the expression level of several stem cell marker and EMT-associated genes including FOXC2, nestin, Oct4, and SOX2. Therefore, the association of TGFβ1 with these genes in cancer-favoring signal cascade can be indicated (37, 38), which might be blocked by anti-PD-1 antibody treatment. This observation should be the novel result in the study investigating the therapeutic mechanism of anti-PD-1 antibody. Importantly, to verify TGFβ1 downregulation, TGFβ1 protein levels in the tumor should be assessed using IHC.
Immunohistochemical analysis in this study revealed an increase in CD8+ and granzyme+ T-cell infiltration of tumors. In clinical trials evaluating anti-PD-1, CD8+ TILs appear to be primarily observed in antibody-treated patients. In contrast, in patients treated with anti-CTLA-4 antibody alone or a combination of anti-CTLA-4 and anti-PD-1, both an increase in CD8+ T cells and a decrease in regulatory T cells are frequently observed (25, 29). In addition, TIL profiling showed an increase of exhausted marker (PD-1+TIM3+LAG3+)-positive T cell frequency in anti-PD-1 antibody-treated mice, which was consistent with the observation that anti-PD-1 antibody promoted exhausted marker-positive T-cell expansion and survival (39).
Pham and colleagues reported that TAM, referred to as CD11b+F4/80+ cells, significantly increased in murine spontaneously arising medulloblastoma tumors as demonstrated by IHC; however, the effect of anti-PD-1 on TAM infiltration was not well characterized (40). There have been very few clinical studies specifically evaluating the role of TAM in immune checkpoint antibody therapy. However, in this study, CD204+ cell level did not significantly differ in anti-PD-1 antibody–treated tumors. The TIL profiling might depend on not only the strength of immune response induced by single or combination antibody therapy, but the immunologic status of the tumor in individual patient.
The significant advantage that our humanized NOG-dKO mice can provide to the development of novel human immunotherapy because they evade the limitations imposed by lethal xeno-GVHD should be stressed. The utilization of the humanized NOG-dKO mouse model for in vivo experiments will enhance the value of immunotherapeutic studies, such as those evaluating novel immune checkpoint antibody treatment and combinations of tumor-based vaccines or other small molecules. In the near future, we plan to use the humanized NOG-dKO mouse system to identify novel biomarkers that can serve as prognostic indicators of the antitumor response to antibodies targeting immune checkpoint molecules, such as PD-1 and CTLA-4, prior to the initiation of treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: Y. Akiyama
Development of methodology: T. Ashizawa
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Ashizawa, A. Iizuka, C. Nonomura, R. Kondou, C. Maeda, H. Miyata, T. Sugino, K. Mitsuya, Y. Nakasu
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Iizuka, R. Kondou, N. Hayashi
Writing, review, and/or revision of the manuscript: Y. Akiyama
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Maruyama, I. Katano, M. Ito, Y. Akiyama
Study supervision: K. Yamaguchi, Y. Akiyama
Other (provided recombinant antibodies): A. Iizuka
Acknowledgments
We thank Mr. Koji Takahashi for his excellent assistance in maintaining NOG-dKO mice in the animal facility.
Grant Support
This work was supported by a grant from JSPS KAKENHI, Japan (grant no. 26430178; to A. Iizuka).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.