Janus kinase inhibitor overcomes resistance to immune checkpoint inhibitor treatment in peritoneal dissemination of gastric cancer in C57BL/6 J mice

Background Cancer immunotherapy aims to unleash the immune system’s potential against cancer cells, providing sustained relief for tumors responsive to immune checkpoint inhibitors (ICIs). While promising in gastric cancer (GC) trials, the efficacy of ICIs diminishes in the context of peritoneal dissemination. Our objective is to identify strategies to enhance the impact of ICI treatment specifically for cases involving peritoneal dissemination in GC. Methods The therapeutic efficacy of anti-PD1, CTLA4 treatment alone, or in combination was assessed using the YTN16 peritoneal dissemination tumor model. Peritoneum and peritoneal exudate cells were collected for subsequent analysis. Immunohistochemical staining, flow cytometry, and bulk RNA-sequence analyses were conducted to evaluate the tumor microenvironment (TME). A Janus kinase inhibitor (JAKi) was introduced based on the pathway analysis results. Results Anti-PD1 and anti-CTLA4 combination treatment (dual ICI treatment) demonstrated therapeutic efficacy in certain mice, primarily mediated by CD8 + T cells. However, in mice resistant to dual ICI treatment, even with CD8 + T cell infiltration, most of the T cells exhibited an exhaustion phenotype. Notably, resistant tumors displayed abnormal activation of the Janus Kinase-Signal Transducer and Activator of Transcription (JAK-STAT) pathway compared to the untreated group, with observed infiltration of macrophages, neutrophils, and Tregs in the TME. The concurrent administration of JAKi rescued CD8 + T cells function and reshaped the immunosuppressive TME, resulting in enhanced efficacy of the dual ICI treatment. Conclusion Dual ICI treatment exerts its anti-tumor effects by increasing tumor-specific CD8 + T cell infiltration, and the addition of JAKi further improves ICI resistance by reshaping the immunosuppressive TME. Supplementary Information The online version contains supplementary material available at 10.1007/s10120-024-01514-5.

cells were derived from minced peritoneal tumor tissue.Both cell types were then incubated in RPMI-1640 (Nacalai Tesque, Japan) supplemented with 1% FBS, 10 mM HEPES, 0.2% collagenase (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) and 2 KU/mL DNase I (Sigma-Aldrich, St. Louis, Missouri, USA) at 37℃ for 40 min.All material was passed through a 70 μm cell strainer to obtain single cell suspensions.After staining dead cells using the Zombie Yellow Fixable Viability Kit (BioLegend) and blocking of Fc receptors with anti-CD16/32 mAb (2.4G2, Bio X Cell), the cells were stained with the fluorescently conjugated mAbs for cell surface antigens.Stained cells were acquired on a CytoFLEX S flow cytometer (Beckman Coulter, Atlanta, Georgia, USA) and analyzed using FlowJo software V.10.6.2 (BD Biosciences).

Bulk RNA-sequence
Total RNA was extracted from the stripped peritoneum with or without disseminated nodules, using ISOGEN (Nippon Gene Co., Ltd., Japan) according to the manufacturer's protocols.For analysis by RNA-sequence (RNA-seq), we ensured that RNA integrity was confirmed by Tape Station software, with the RNA integrity (RIN) score>8 (Agilent Technologies).Library preparation and sequencing (150 bp, pairedend reads) was performed as Illumina HiSeq standard protocols.The sequence reads were aligned to the mm10 reference genome using STAR V.2.5.2b.Mapped reads were counted by HTSeq V.0.6.1.Raw counts were normalized and differentially expressed genes (DEGs) were calculated using R software version 4.1.1 with DEseq2 and dplyr packages.For downstream analyses, we used STRING version 11.5 (https://www.string-db.org/),gene set enrichment analysis (GSEA, V.4.2.1) and TIMER2.0platform (http://timer.cistrome.org/).The murine Microenvironment Cell Populations counter (mMCP-counter) algorithm was employed to estimate the relative abundance of different immune cell types within individual samples, enabling crosssample comparisons.Additionally, we used the Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORT) algorithm to assess the relative proportions of various immune cell types within each individual sample.While CIBERSORT focuses on estimating immune cell proportions within individual samples, mMCP-counter facilitates horizontal comparisons of immune cell composition across different samples.We used the OmicStudio tools (https://www.omicstudio.cn/tool)for visualization purposes, including the generation of heatmaps, volcano plots and bubble plots.

Immunohistochemistry staining and quantification
For immunohistochemistry, 4μm paraffin sections were deparaffinized, rehydrated, and antigen-retrieved using Immunosaver (Nishin EM, Tokyo, Japan) in a microwave.Sections were then treated with 3% H2O2/methanol to quench endogenous peroxidase activity and blocked with serum.Overnight incubation with primary antibodies (CD8α, Granzyme B, TBR2/Eomes, Foxp3, CD68, Ly6G, CD206, and p-STAT3; supplementary Table 1) was followed by visualization using VECTASTAIN Elite ABC HRP Kit and DAB Substrate Solution (VECTOR).Nuclei were stained with hematoxylin.The quantitative results of immunohistochemical staining for CD8, GZMB, Ly6G, Foxp3 were represented as the number of positive cells per tumor area.
However, for CD68, CD206, and p-STAT3, counting was challenging due to the high number of stained cells, which could overlap.Therefore, we opted to express these results as the ratio of positive area to tumor area.Both counting methods were analyzed using ImageJ software across five high-power fields per sample.

Statistical analysis
For parametric data, the means among three or more groups were compared using one-way ANOVA followed by Tukey's multiple comparisons test.For non-parametric data or when the assumptions of normality and homogeneity of variances were not met, the Kruskal-Wallis test was used, followed by Dunn's multiple comparison test for post hoc analysis.All statistical analyses were performed using R software (version 4.1.1,R Foundation for Statistical Computing, Vienna, Austria) or Prism software version 9.0.0 (GraphPad Software, LLC, San Diego, CA, USA) at a significance level of α=0.05.
The data are presented as mean ± standard deviation (SD).
Schematic diagram of mouse treatment grouping.B Changes in body weight of mice during treatment.C Macroscopic image of the mouse that died before autopsy in the anti-CTLA4 group.D, E Representative macroscopic images of peritoneal dissemination of mice in the dual ICI-uncured group and in the dual ICI+anti-CD8 group (jaundice).ns: not significant, one-way ANOVA with Tukey's multiple comparisons test Representative images and quantitative analysis of IHC staining of CD68+ Macrophages, Ly6G+ Neutrophils and Foxp3+ Tregs on day 35.Scale bar: 50μm; *p<0.05,**p<0.01,***p<0.001,****p<0.0001,one-way ANOVA with Tukey's multiple comparisons test pie chart format depicted the immune cell composition 3 weeks after treatment, as analyzed using CIBERSORT.The dual ICI treatment group exhibited an increased relative proportion of CD8+ T cells, whereas in the untreated group, neutrophils accounted for a higher proportion.CIBERSORT: Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (A specialized computational method for identifying and estimating the relative proportions of different cell types within complex tissues or cell mixtures) Representative images of quantification of IHC staining of Ly6G+