Abstract

Background

   High-grade serous ovarian cancer (HGSOC) is a highly aggressive
   malignancy marked by an immunosuppressive tumor microenvironment that
   hinders effective immune responses. A key feature of this environment
   is the extensive infiltration of myeloid cells, which contributes to
   immune evasion. This study explored how mesothelin (MSLN), a
   tumor-associated antigen, modulates the expression of CD24, an emerging
   target for immune modulation, and its role in promoting immune evasion
   in HGSOC. Understanding these underlying mechanisms is crucial for
   enhancing the efficacy of immune checkpoint blockade (ICB) therapies
   and improving outcomes in patients with HGSOC.

Methods

   We analyzed the expression of MSLN in HGSOC samples and examined its
   correlation with clinical outcome. In vitro and in vivo models were
   used to explore how MSLN influences CD24 expression and the
   polarization of tumor-associated macrophages (TAMs). We also
   investigated the role of MSLN in the activation of Wnt/β-catenin
   signaling and its impact on T-cell function and antitumor immunity. The
   effects of Msln knockdown on CD24 expression and the response to
   anti-programmed cell death protein-1 (PD-1) therapy were evaluated in
   syngeneic mouse models.

Results

   MSLN expression was found to be significantly elevated in HGSOC, with
   high MSLN levels correlating with poor prognosis and resistance to ICB.
   MSLN upregulated CD24 and promoted the protumorigenic polarization of
   TAMs, contributing to T-cell dysfunction. Mechanistically, MSLN
   activated Wnt/β-catenin signaling, which in turn enhanced CD24
   expression. This activation forms a positive feedback loop that further
   promotes MSLN transcription. In contrast, Msln knockdown reduced CD24
   expression, relieved cytotoxic T-cell suppression, and significantly
   improved the efficacy of anti-PD-1 therapy in syngeneic models.

Conclusions

   This study elucidates the critical role of MSLN in immune evasion in
   HGSOC and its underlying mechanisms. Targeting MSLN in combination with
   ICB is a promising strategy to enhance the efficacy of immunotherapy
   and improve patient outcomes in HGSOC.

   Keywords: Ovarian Cancer, Immune Checkpoint Inhibitor, Tumor
   microenvironment - TME, Macrophage
     __________________________________________________________________

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

     * This study provides novel insights into how MSLN contributes to
       immune evasion in HGSOC by modulating CD24 expression through
       activation of the Wnt/β-catenin signaling pathway. The results
       showed that MSLN promoted protumorigenic polarization of
       tumor-associated macrophages and induced T-cell dysfunction.
       Importantly, this study identified a positive feedback loop that
       enhances MSLN transcription, further promoting immunosuppression.
       Moreover, MSLN knockdown led to reduced CD24 expression, alleviated
       cytotoxic T-cell suppression, and improved the efficacy of
       anti-programmed cell death protein-1 therapy in preclinical models,
       offering potential strategies for enhancing immunotherapy.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

     * This study underscores the critical role of MSLN in immune evasion
       within the TME of HGSOC, thus, positioning MSLN as a promising
       therapeutic target. By targeting MSLN in combination with ICB, this
       approach could offer a more effective immunotherapy strategy for
       improving the clinical outcomes in patients with HGSOC. These
       findings could drive the development of novel combination therapies
       and pave the way for clinical trials aimed at improving the
       prognosis of patients with HGSOC.

Introduction

   Although most immune checkpoint blockades (ICBs) exhibit limited
   efficacy,[43]^1 mesothelin (MSLN) has emerged as a key
   immunosuppressive mediator.[44]^2 3 MSLN, a membrane glycoprotein
   overexpressed in mesothelioma, non-small cell lung cancer, pancreatic
   cancer, and ovarian cancer, is minimally expressed in a few selected
   normal tissues,[45]^4 5 making it an attractive target for
   immunotherapy. For example, the use of chimeric antigen receptor NK-92
   cells to target MSLN in ovarian cancer.[46]^6 Complementing these
   findings, Jiang et al observed that combining the MSLN-targeted
   immunotoxin LMB-100 with anti-programmed cell death protein-1 (PD-1)
   antibody significantly improved outcomes in patients with mesothelioma,
   underscoring the synergy between immune checkpoint inhibitors and
   targeted therapies.[47]^7 Additionally, in mouse tumors, LMB-100
   treatment induces the formation of tertiary lymphoid structures (TLS)
   by upregulating-related genes and activating antitumor immunity.[48]^8
   Despite these advances, challenges remain in identifying reliable
   biomarkers for therapeutic response. For instance, Katz et al found
   that while serum soluble MSLN-related protein and fibulin-3 correlated
   with tumor volumes in mesothelioma, they lacked predictive value in
   immunotherapy trials.[49]^9 Despite the progress in mesothelioma,[50]^5
   10 pancreatic[51]11,[52]13 and lung cancers,[53]^14 the role of
   MSLN-targeted therapies in high-grade serous ovarian cancer (HGSOC) and
   its impact on tumor microenvironment (TME) remodeling remains
   underexplored.

   The TME is dynamically shaped by the dysregulation of key signaling
   pathways, such as the Wnt/β-catenin pathway—a complex network of
   protein interactions closely linked to cancer progression.[54]^15 16
   This pathway is pivotal in sustaining cancer stem cell renewal, driving
   uncontrolled cell proliferation and differentiation, ultimately
   contributing to TME remodeling and impacting treatment response.[55]^17
   Mutational biases vary across cancer types, with APC (Adenomatous
   Polyposis Coli) mutations being particularly prevalent in ovarian
   cancer, underscoring the significance of Wnt-driven alterations in
   tumor development.[56]^18 In view of these challenges, elucidating the
   mechanisms by which Wnt-mediated reshape the TME is crucial for
   screening patients for ICB therapy and enhanced efficacy of ICB
   therapy.

   In this study, we present findings from HGSOC organoids and murine
   ovarian cancer models, demonstrating that MSLN upregulates CD24
   expression via the activation of the Wnt/β-catenin pathway. Targeting
   MSLN significantly enhanced the efficacy of anti-PD-1 treatment in
   HGSOC organoids and murine models by remodeling the TME, promoting a
   favorable tumor-associated macrophages (TAMs) phenotype, activating
   CD8^+ T cells, and further increasing the abundance of T-cell
   exhaustion progenitor cells (Tpex).

Materials and methods

Patient sample collection

   The Discovery cohort (n=120) was obtained from the Gynecology and
   Obstetrics Hospital of the Fudan University between March 2013 and
   November 2015. The Validation cohort (n=160) was obtained from Shanghai
   Cancer Center between January 2012 and October 2019. All enrolled
   participants (n=280) who underwent surgery completed follow-up, with no
   attrition observed (attrition rate=0%). This observational study did
   not involve randomization, and participants were consecutively
   recruited post-surgery and followed-up prospectively. Fresh tumor
   tissues from 76 patients with HGSOC were obtained from our hospital and
   subjected to flow cytometry analysis (n=48) and ex vivo intervention
   (n=28). All HGSOC specimens were surgically resected from the primary
   site and were histologically verified. Detailed clinicopathological
   characteristics of the patients are shown in [57]online supplemental
   table 1. The pathological staging and histological classification were
   based on the FIGO (International Federation of Gynecology and
   Obstetrics) criteria (2018). The overall survival (OS) was calculated
   from the date of surgery to the last follow-up or death. Follow-up for
   the Discovery cohort continued until March 2019, while the Validation
   cohort was monitored until September 2023. None of the patients with
   HGSOC received any therapeutic intervention before surgery.

Construction of stably transfected cell line

   ID8agg-luc cells and OV2944-HM1-luc cells were transduced with
   Msln-targeting short hairpin RNA (shRNA) lentivirus at a multiplicity
   of infection of 20. Stable cell lines were subsequently established by
   selection with hygromycin B (#10687010, Invitrogen, RRID: SCR_025910).
   The knockdown efficiency of each shRNA sequence was first measured by
   real-time PCR and flow cytometry, and only shRNAs showing at least 60%
   knockdown were chosen for further experiments. Full-length Cd24a was
   cloned into a pcDNA3.1 (+) vector. The plasmid was sequenced prior to
   transfection to confirm the correct nucleotide sequence. pcDNA3.1-Cd24a
   was transfected into ID8agg-luc cells using Lipofectamine 3000 reagent
   (#L300001, Invitrogen, RRID: SCR_025911), according to the
   manufacturer’s recommendations, and selected with puromycin to
   establish a stable cell line. The target sequences are listed in
   [58]online supplemental table 2.

Flow cytometry

   Fresh tissues from mouse and human tumors were collected, digested with
   a gentleMACS Dissociator (#130-093-235, Bio X Cell, RRID: SCR_025922),
   and passed through a Falcon 100 µm cell strainer (#352360, CORNING,
   RRID: SCR_025923). After washing with ice-cold phosphate-buffered
   saline and lysing for red blood cell removal with 1×RBC Lysis Buffer
   (#2881585, eBioscience, RRID: SCR_025960), the purified single-cell
   suspensions were blocked with Fc receptor blocking reagent (#156604,
   BioLegend, RRID: SCR_025924) to prevent non-specific binding.
   Fluorescence-conjugated antibodies ([59]online supplemental table 3)
   targeting cell surface markers were added and the cells were incubated
   in the dark at 4°C for 30 min. Before intracellular and intranuclear
   staining, samples were treated with FOXP3/Transcription Factor Staining
   Buffer Set. To control background staining, parallel cells were stained
   with isotype-matched antibodies. Viable cells were identified using the
   Zombie Aqua Fixable Viability Kit (#423102, BioLegend, RRID:
   SCR_025925). Fluorescence data were collected on a CytoFLEX flow
   cytometer (RRID: SCR_025926, Beckman Coulter) and analyzed using FlowJo
   software V.10.8.1 (RRID: SCR_008520, Tree Star). The gating strategies
   for flow cytometry analyses of HGSOC specimens and mouse samples are
   detailed in [60]online supplemental figures 1,2.

RNA extraction and quantitative reverse transcription PCR

   Total RNA was extracted using the FastPureRNA Isolation Kit V2
   (#RC112-01, Vazyme, RRID: SCR_02592). Total RNA was reverse-transcribed
   using the HiScript III RT SuperMix Kit (#R323-01, Vazyme, RRID:
   SCR_025929). Quantitative (qPCR) was performed using the Taq Pro
   Universal SYBR qPCR Master Mix kit (#Q712-02, Vazyme, RRID: SCR_025930)
   on an ABI QuantStudio 3 Flex instrument (Thermo Fisher Scientific,
   RRID: SCR_025959). Gene expression was normalized to that of GAPDH in
   each sample, using the comparative threshold cycle method. Primers used
   in this study are listed in [61]online supplemental table 4.

Western blotting

   Tumor cells were lysed in radioimmunoprecipitation assay buffer
   (#89900, Thermo Fisher Scientific, RRID: SCR_025931) with protease and
   phosphatase inhibitors (#87786, Thermo Fisher Scientific, RRID:
   SCR_025932). Protein quantification was performed using the Pierce
   Bicinchoninic Acid Protein Assay Kit (#A55860, Thermo Fisher
   Scientific, RRID: SCR_025933). The protein samples were loaded onto 10%
   sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels. After
   electrophoresis, proteins were transferred onto polyvinylidene
   difluoride (PVDF) membranes. After blocking, PVDF membranes were
   incubated with primary antibodies ([62]online supplemental table 5) at
   4°C overnight. After 1 hour of incubation with secondary antibodies at
   room temperature, protein bands were detected using an enhanced
   Chemiluminescence Detection Kit (#PA112, TIANGEN Biotech, RRID:
   SCR_025935) and an Alliance MINI HD 6 Analyzer (UVITEC, RRID:
   SCR_025936).

Differentially expressed genes and pathway enrichment analysis

   The R package “DESeq2” (V.1.42.1, RRID: SCR_015687) was used to
   identify differentially expressed genes (DEGs) between the two groups.
   Significantly enriched pathways were obtained through Gene Set
   Enrichment Analysis (GSEA, RRID: SCR_003199) using the Gene Ontology
   (RRID: SCR_002811), Kyoto Encyclopedia of Genes and Genomes (RRID:
   SCR_012773) and WikiPathways (RRID: SCR_002134) databases with the R
   package “clusterProfiler” (V.4.12.0, RRID: SCR_016884). GSEA gene sets
   are provided in [63]online supplemental table 6.

Statistical analysis

   Statistical analyses were performed using GraphPad Prism V.9.2.0 (RRID:
   SCR_002798) or MedCalc V.19.0.5 (RRID: SCR_015044). The D’Agostino and
   Pearson omnibus normality test was applied to assess the normality of
   the data in each comparison. For data sets that followed a Gaussian
   distribution, parametric tests were performed, including the two-tailed
   unpaired Student’s t-test, one-way or two-way analysis of variance
   (ANOVA) with Tukey’s post hoc test for multiple comparisons. In cases
   where the data did not follow a Gaussian distribution, non-parametric
   tests were applied, such as Mann-Whitney U test, one-way ANOVA with
   Kruskal-Wallis and Dunn’s post hoc test for multiple comparisons, or
   Spearman’s correlation. The χ^2 test was used to compare categorical
   variables between independent groups. The prognostic significance of
   the risk factors was assessed using Cox proportional hazards regression
   models for both univariate and multivariate analyses. Survival curves
   were generated using the Kaplan-Meier method, and survival times
   between subgroups were compared using the log-rank test. All
   experiments were performed with at least two independent replicates.
   Data are presented as mean±SD, and a two-tailed p value<0.05 was
   considered statistically significant.

Results

Enhanced MSLN expression is associated with poor prognosis in HGSOC and may
contribute to resistance against ICB therapy

   We first analyzed the relationship between MSLN expression and survival
   outcomes in three independent cohorts ([64]online supplemental figure
   3A–C). In all cohorts, higher MSLN expression was significantly
   associated with reduced OS. Further analysis demonstrated a link
   between elevated MSLN expression and advanced FIGO stage in both the
   Discovery and Validation cohorts ([65]online supplemental figure 3D–E).
   A pooled cohort analysis confirmed that MSLN expression was an
   independent prognostic factor for OS in patients with HGSOC ([66]online
   supplemental figure 3F). Given the role of MSLN in cancer progression,
   we explored the potential role of MSLN in immune evasion. The Tumor
   Immune Dysfunction and Exclusion (TIDE) score predicts immune evasion
   and ICB therapy resistance.[67]^19 In TCGA (The Cancer Genome Atlas)
   cohort, high MSLN expression was significantly associated with an
   increased TIDE score ([68]online supplemental figure 3G). These
   findings highlight the dual role of MSLN as both a prognostic biomarker
   and a critical target in HGSOC management.

MSLN exhibits an inverse correlation with response to anti-PD-1 treatment in
HGSOC organoids

   The efficacy of ICB therapy in ovarian cancer remains limited, with
   response rates to anti-PD-1 monotherapy rarely exceeding 15%.[69]^20 21
   This underscores the urgent need for novel therapeutic targets to
   enhance ICB sensitivity. In mesothelioma, combining the MSLN-targeting
   immunotoxin LMB-100 with anti-PD-1 therapy significantly increased
   response rates from below 20% to nearly 40%, achieving substantial
   tumor regression.[70]^7 Building on this evidence, we explored the role
   of MSLN in modulating anti-PD-1 efficacy in HGSOC. Using organoids
   derived from tissues of 28 patients with HGSOC, none of whom had
   received any prior treatments, including immunotherapies such as
   anti-PD-1. Based on their ex vivo response to anti-PD-1, the organoids
   were classified into two subgroups: responders (n=6), which exhibited
   marked immune marker alterations, and non-responders (n=22), which
   showed minimal changes([71]figure 1A). Notably, non-responders had a
   significantly higher MSLN H-score compared with responders ([72]figure
   1B, [73]online supplemental figure 4A). During ex vivo experiments,
   anti-PD-1 was directly added to the organoid cultures. PD-1 blockade
   significantly promoted CD8^+ T-cell proliferation and cytotoxicity in
   organoids with low MSLN expression, whereas this effect was absent in
   those with high MSLN expression ([74]figure 1C–1D, [75]online
   supplemental figure 4B). These findings suggested that MSLN expression
   contributes to resistance to anti-PD-1 therapy. To further investigate
   this, we treated 22 non-responsive organoids with a combination of
   anti-PD-1 and LMB-100 antibodies. Positive responses were observed in
   12 cases, predominantly in samples with high MSLN expression
   ([76]figure 1E, [77]online supplemental figure 4C). Immune profiling
   revealed distinct differences in CD8^+ T cells and TAMs populations
   between responders and non-responders ([78]figure 1F, [79]online
   supplemental figure 4D). In responders, TAMs were located closer to
   tumor cells and exhibited antitumorigenic phenotypes, while
   non-responders showed a predominance of protumorigenic TAMs ([80]figure
   1G–1J). Combination treatment significantly enhanced effector CD8^+
   T-cell responses and promoted the polarization of antitumorigenic TAMs,
   characterized by high expression of HLA-DR, CD86 and CXCL9, while
   simultaneously reducing the prevalence of protumorigenic TAMs, defined
   by elevated levels of arginase-1 and CD206, as assessed by flow
   cytometry ([81]figure 1K–1L). However, this trend was not observed in
   non-responsive samples ([82]online supplemental figure 4E,F). These
   findings demonstrate that targeting MSLN, in conjunction with PD-1
   blockade, represents a promising strategy to modulate the TME and
   overcome ICB resistance in HGSOC with higher MSLN expression.

Figure 1. MSLN exhibits an inverse correlation with response to anti-PD-1
treatment in HGSOC organoids. (A) Heatmap clustering based on the relative
abundance of cleaved caspase-3^+ CD45^− cells, multiple subsets of CD8^+ T
cells, TH1 cytokines (IL-2, IFN-γ) and TH1 chemokines (CXCL9, CXCL10)
demonstrated anti-PD-1 responses in HGSOC organoids via flow cytometry
(n=28). Color codes represent response subgroups, MSLN subgroups, and Z-score
(log[2] intensity). Complete linkage and Euclidean distance were used for
clustering. (B) MSLN expression levels, as indicated by the H-score, in
anti-PD-1 responders (n=6) and non-responders (n=22) from HGSOC organoids.
(C–D) Proportion of Ki-67 (C) and TNF-α, IFN-γ, TNF-α/IFN-γ, granzyme B/IFN-γ
expressing CD8^+ T cells (D) in HGSOC organoids stratified by the median
H-score of MSLN with isotype or anti-PD-1 treatment (n=28). (E) Heatmap
clustering based on the relative abundance of cleaved caspase-3^+ CD45^−
cells and multiple subsets of CD8^+ T cells illustrating the response to the
combination of anti-PD-1 and LMB-100 versus anti-PD-1 alone in HGSOC
organoids, as determined by flow cytometry (n=22). Color codes represent
response subgroups, MSLN subgroups, and Z-score (log[2] intensity). Complete
linkage and Euclidean distance were used for clustering. (F) The proportion
of CD8^+ and CD68^+ cells among CD45^+ cells in HGSOC organoids between
responders (n=12) and non-responders (n=10) to the combination of anti-PD-1
and LMB-100 treatment. (G–H) Representative immunofluorescence images (G) and
contour maps (H) illustrating CD68, CD8a, and Pan keratin staining in
responders and non-responders to the combination of anti-PD-1 and LMB-100
treatment. (I) Quantification of the distance from CD8^+ T cells (left) and
CD68^+ macrophages (right) to the nearest Pan keratin^+ cell in responders
(n=12) and non-responders (n=10) to anti-PD-1+LMB-100 treatment. (J)
Proportion of arginase-1, CD206, HLA-DR, CD86, CXCL9 expressing CD68^+
macrophages in HGSOC organoids between responders (n=12) and non-responders
(n=10) to anti-PD-1+LMB-100 treatment. (K–L) Proportion of TNF-α, IFN-γ,
granzyme B, TNF-α/IFN-γ, TNF-α/granzyme B, and IFN-γ/granzyme B expressing
CD8^+ T cells (K) and arginase-1, CD206, HLA-DR, CD86, CXCL9 expressing
CD68^+ macrophages (L) in HGSOC organoids under the treatment with isotype
IgG, anti-PD-1, LMB-100 alone, or the combination of anti-PD-1 and LMB-100
for 48 hours (n=12), as assessed by flow cytometry. Data are presented as
mean±SD. Statistical significance was determined using a two-tailed unpaired
Student’s t-test or Mann-Whitney U test (B, F, I–J), two-sided paired
Student’s t-test (C–D), and two-way analysis of variance, followed by Tukey’s
post hoc test for multiple comparisons (K–L). Con, control; CXCL9, C-X-C
motif chemokine ligand 9; CXCL10, C-X-C motif chemokine ligand 10; HGSOC,
high-grade serous ovarian cancer; IFN-γ, interferon-γ; IL-2, interleukin-2;
MSLN, mesothelin; NS, no significant difference; Pan-CK, pan cytokeratin;
PD-1, programmed cell death protein-1; TNF-α, tumor necrosis factor-α.

   [83]Figure 1
   [84]Open in a new tab

Msln knockdown correlates with an antitumorigenic TME and confers protection
against tumor development

   We used single-cell RNA sequencing (scRNA-seq) data to analyze TME in
   HGSOC, with MSLN expression in the upper and lower quartiles. Tumors in
   the MSLN-low group exhibited a higher abundance of TAMs, with an
   enrichment of antitumorigenic subtypes and a reduced presence of
   protumorigenic subtypes ([85]online supplemental figure 5A–E).
   Additionally, analysis of CD8^+ T-cell subtypes in the MSLN-low
   subgroup revealed increased levels of effector CD8^+ T cells, along
   with a lower prevalence of exhausted CD8^+ T cells ([86]online
   supplemental figure 5F,G). Immunohistochemistry analysis of the
   Discovery cohort confirmed that low MSLN expression was associated with
   significantly higher levels of CD8^+ T cells and antitumorigenic TAMs
   (CXCL9^+ CD68^+), while showing lower expression of immune checkpoint
   molecules (cytotoxic T lymphocyte associate protein-4, T-cell
   immunoglobulin domain and mucin domain-3) and protumorigenic TAMs
   (SPP1^+ CD68^+) ([87]online supplemental figures 6,7A). These findings
   were further corroborated by flow cytometric analysis of fresh tumors
   from patients with HGSOC ([88]figure 2A).

Figure 2. Msln knockdown correlates with an antitumorigenic TME and confers
protection against tumor development. (A) Dot plots demonstrating the
abundance of CD8^+ T-cell subsets and polarized TAMs between the MSLN^low
(n=24) and MSLN^high (n=24) subgroups in patients with HGSOC. Dot size
represents −log[10](p value), while the color scale indicates z-score
normalized immune cell abundance, as evaluated by flow cytometry. (B) Tumor
growth trajectories in immunocompetent C57BL/6 mice following subcutaneous
administration of 5×10^6 ID8agg-control cells or ID8agg-Msln^KD cells (n=6
per group). The dashed lines depict individual mice, whereas the solid line
represents the mean tumor volume. (C) Heatmap illustrating DEGs between
subcutaneous tumor tissues from control and Msln^KD groups, based on bulk RNA
sequencing analysis. The bars on the right represent the z-scores of TPM for
the corresponding genes. (D) GSEA (GO biological processes) on genes ranked
by log[2] (fold change) between control and Msln^KD group. Bars on the right
indicate the range of −log[10] (q value) of corresponding pathways. Dots on
the right represent the number of counts. Norminal q values were adjusted
using the Benjamini-Hochberg method. (E–F) GSEA revealed enrichment of the
Wnt pathway (E) and the T-cell infiltration gene expression profile (T cell
inf gene expression profile) pathway (F) on Msln knockdown, compared with the
control. The nominal p value was adjusted using the Benjamini-Hochberg
method. (G–I) For the advanced ovarian cancer model, ID8agg-luc cells with or
without Msln knockdown were implanted via intraperitoneal injection (n=6 per
group). The experiment was conducted with two independent biological
replicates. (G) IVIS bioluminescent images of peritoneal tumor burden from
week 2 to week 5 after control or Msln^KD ID8agg-luc cell injection. (H)
Quantification of the tumor burden in mice shown in (G) via radiant
efficiency. (I) Kaplan-Meier survival plot of mice corresponding to (G).
(J–K) Proportion of TNF-α, IFN-γ, IFN-γ/granzyme B expressing CD8^+ T cells
(J) and arginase-1, CD206, MHC-II, CD86, CXCL9 expressing F4/80^+ CD11b^+
cells (K) from ascites of mice corresponding to (G) (n=6 per group). (L–N) To
evaluate the role of CD8^+ T cells in advanced ovarian cancer, ID8agg-luc
cells with or without Msln knockdown were implanted via intraperitoneal
injection in mice with or without depletion of CD8^+ T cells (n=6 per group).
Tumor burden was quantified by evaluating fluorescence via radiant efficiency
(L), ascites volume (M), and metastatic nodules in the peritoneal wall,
omentum, mesentery, and diaphragm (N). (O–Q) To evaluate the role of
macrophages in advanced ovarian cancer, ID8agg-luc cells with or without Msln
knockdown were intraperitoneally injected into mice with or without
macrophage depletion (n=6 per group). Tumor burden was quantified by
evaluating fluorescence via radiant efficiency (O), ascites volume (P), and
metastatic nodules in the peritoneal wall, omentum, mesentery, and diaphragm
(Q). (R) Proportion of TNF-α/IFN-γ, TNF-α/granzyme B, and IFN-γ/granzyme B
expressing CD8^+ T cells from ascites of mice corresponding to (O). Data are
presented as mean±SD. Statistical significance was assessed using two-tailed
unpaired Student’s t-test or Mann-Whitney U test (J–K), log-rank test (I),
and two-way analysis of variance, followed by Tukey’s post hoc test for
multiple comparisons (H, L–R). CL, clodronate liposomes; Con, control;
CTLA-4, cytotoxic T lymphocyte associate protein-4; CXCL9, C-X-C motif
chemokine ligand 9; DEGs, differentially expressed genes; GO, Gene Ontology;
GSEA, Gene Set Enrichment Analysis; HGSOC, high-grade serous ovarian cancer;
IVIS, In Vivo Imaging System; IFN-γ, interferon-γ; IgG, immunoglobulin G; KD,
knockdown; LAG3, lymphocyte activation gene-3; Msln, mesothelin; NES,
normalized enrichment score; PD-1, programmed cell death-1; TAMs,
tumor-associated macrophages; TIM-3, T-cell immunoglobulin domain and mucin
domain-3; TME, tumor microenvironment; TNF-α, tumor necrosis factor-α; TPM,
transcripts per million.

   [89]Figure 2
   [90]Open in a new tab

   To explore the functional impact of Msln knockdown, we generated stable
   Msln knockdown cell lines (Msln^KD). The knockdown efficacy was
   confirmed by a significant reduction in MSLN expression ([91]online
   supplemental figure 7B–E). Colony formation and proliferation assays
   indicated that Msln knockdown significantly reduced proliferation, with
   a more pronounced effect observed in OV2944-HM1 cells ([92]online
   supplemental figure 8A–D). The result of the western blot further
   confirmed that the secretion levels of MSLN were also significantly
   reduced in knockdown sequences #1 and #2 ([93]online supplemental
   figure 8E). Initial in vivo studies demonstrated that stable Msln
   knockdown in OV2944-HM1 cells resulted in a complete loss of
   tumorigenic potential. However, due to the inherently low
   aggressiveness of OV2944-HM1 cells, we transitioned to using the ID8agg
   subline for subsequent in vivo experiments. The ID8agg subline was
   developed by serial passage of standard ID8 cells through
   immunocompetent wild-type hosts,[94]^22 resulting in a highly
   aggressive ovarian cancer model. Compared with the standard ID8 model,
   ID8agg exhibits significantly enhanced tumorigenic potential,
   characterized by accelerated tumor growth and increased metastatic
   capacity.[95]^23 Additionally, ID8agg reliably induces extensive
   peritoneal tumors resembling stage III and IV ovarian cancer,
   accompanied by a high incidence of hemorrhagic ascites, a hallmark of
   advanced HGSOC. These characteristics render ID8agg a more clinically
   relevant and robust platform for studying disease progression and
   evaluating therapeutic strategies. Using the ID8agg subcutaneous
   xenograft mouse model, we found that Msln knockdown (Msln^KD) cells
   significantly suppressed tumor growth in vivo ([96]figure 2B). RNA
   sequencing of tumor tissues further revealed downregulation of key
   protumorigenic pathways, including members of the Wnt and Tgf-β
   families ([97]figure 2C), suggesting that Msln plays a crucial role in
   driving tumor progression and maintaining a protumorigenic
   microenvironment. Additionally, protumorigenic pathways, such as Wnt,
   ERBB2-EGFR, Notch, and TGF-β, as well as pathways inhibiting T-cell
   proliferation, were significantly downregulated in Msln^KD tumors
   ([98]figure 2D–2E). In contrast, the T cell-inflamed gene expression
   profile was activated in Msln^KD tumors ([99]figure 2F).

   Using an orthotopic ovarian tumor model, we demonstrated that Msln
   knockdown significantly delayed tumor progression ([100]online
   supplemental figure 8F,G), inhibited peritoneal metastasis expansion
   ([101]figure 2G–2H), and extended OS in intraperitoneal exnograft mice
   ([102]figure 2I). Post-mortem analysis revealed reduced hemorrhagic
   ascites and fewer secondary tumor foci in the Msln^KD group
   ([103]online supplemental figure 8H,I). Furthermore, Msln knockdown
   enhanced effector CD8^+ T cells infiltration and antitumorigenic TAM
   populations while reducing protumorigenic TAMs ([104]figure 2J–2K,
   [105]online supplemental figure 8J–L). To confirm that the
   tumor-suppressive effects of Msln knockdown were mediated by CD8^+ T
   cells and TAMs, we performed depletion experiments in an
   intraperitoneal tumor-bearing mouse model ([106]online supplemental
   figures 9,10A–C). CD8^+ T cells depletion abolished the
   tumor-suppressive effects of Msln knockdown, leading to accelerated
   tumor growth, increased hemorrhagic ascites, and recurrence of
   metastatic lesions ([107]figure 2L–2N). Similarly, macrophage depletion
   nullified the inhibitory effects of Msln knockdown on tumor growth,
   ascitic fluid formation, and metastasis ([108]figure 2O–2Q). Flow
   cytometric analysis of the ascitic TME in macrophage-depleted mice
   revealed a significant reduction in effector CD8^+ T-cell infiltration
   ([109]figure 2R, [110]online supplemental figure 10D), underscoring the
   critical role of macrophages in enabling CD8^+ T cells mediated
   antitumor activity induced by Msln knockdown.

MSLN augments protumorigenic macrophage polarization in a CD24-dependent
manner

   To explore the mechanisms by which MSLN mediates TAM’s polarization, we
   employed flow cytometry to analyze BMDMs co-cultured with tumor cells
   in both direct and indirect contact setups. The polarization of TAMs
   occurred exclusively in the direct co-culture system, where Msln^KD
   promoted a shift towards an antitumorigenic phenotype ([111]figure 3A,
   [112]online supplemental figure 11). Since this effect was observed
   only under direct cell–cell interactions, we focused on membrane
   surface molecules that potentially drive this process. Using
   overlapping analyses of DEGs from bulk RNA sequencing and scRNA-seq
   data sets ([113]figure 3B), we identified CD24 as the most
   significantly altered surface molecule associated with macrophage
   polarization.[114]^24 Msln knockdown reduced CD24 expression, an effect
   that could be reversed by treatment with recombinant MSLN (rMSLN;
   [115]figure 3C–3D). This positive correlation between MSLN and CD24
   expression was further validated in HGSOC cohorts ([116]figure 3E,
   [117]online supplemental figure 12A–C). Based on these observations, we
   hypothesized that CD24 mediates MSLN-induced TAM polarization. To test
   this, we generated a CD24 overexpressing ID8agg cell line ([118]figure
   3F–3G). Co-culture experiments demonstrated that CD24 overexpression
   reversed the antitumor polarization of TAMs induced by Msln^KD cells
   ([119]figure 3H). Additionally, CD24 overexpression suppressed the
   enhanced phagocytic capacity of TAMs observed in the presence of
   Msln^KD cells ([120]figure 3I and [121]online supplemental figure 12D).
   Notably, CD24 overexpression rescued the inhibitory effects of Msln^KD
   cells on CD8^+ T cells proliferation and cytotoxic activities
   ([122]figure 3J–3L, [123]online supplemental figure 12E,F). The gating
   strategies for flow cytometry analyses of in vitro macrophage
   phagocytosis of tumor cells and T-cell proliferation assay are detailed
   in [124]online supplemental figures 13,14. These results identify CD24
   as a critical mediator of MSLN-induced TAMs polarization. This
   previously unrecognized CD24-dependent mechanism highlights a pathway
   by which high MSLN expression fosters immune evasion of HGSOC.

Figure 3. MSLN augments protumorigenic macrophage polarization in a
CD24-dependent manner. (A) The proportion of F4/80^+ CD11b^+ cells expressing
Arginase-1, CD206, MHC-II, CD86, or CXCL9 in a direct co-culture system.
BMDMs were co-cultured with either control or Msln^KD ID8agg cells, with or
without supplementary rMSLN for 48 hours (n=6 per group). The experiment was
conducted with three independent biological replicates. (B) Venn diagram
summarizing the number of DEGs (p adj≤0.05). The left circle represents the
number of DEGs between MSLN^+ and MSLN^− cancer cells from scRNA-seq of HGSOC
in [125]GSE180661. The right circle represents the number of DEGs between
control and Msln^KD ID8agg cells. The top eight overlapping DEGs in each
analysis, ranked in descending order by log[2](FC), are enumerated on the
right. (C–D) Relative mRNA expression of Cd24a (C) and MFI of CD24 (D) in
control or Msln^KD ID8agg cells in the absence or presence of rMSLN for 24
hours (n=6 per group). The experiment was conducted with three independent
biological replicates. (E) Positive correlation between the IHC score of MSLN
and CD24 protein in HGSOC from the Discovery cohort (n=120). (F–G) Relative
mRNA expression of Cd24a (F) and MFI of CD24 (G) in control or Msln^KD ID8agg
cells with or without CD24 overexpression (n=6 per group). The experiment was
conducted with three independent biological replicates. (H–I) Proportion of
arginase-1, CD206, MHC-II, CD86, CXCL9 expressing F4/80^+ CD11b^+ (H), and
phagocytosis of BMDMs (I) in direct co-culture between BMDMs and control or
Msln^KD ID8agg cells with or without CD24 overexpression for 48 hours (n=6
per group). (J–L) The proliferation of CD8^+ T cells was assessed via CFSE
dilution (J), and the expression of TNF-α (K) and IFN-γ (L) in CD8^+ T cells
was evaluated by direct co-culture of CD8^+ T cells and BMDMs pre-cultured
with control or Msln^KD ID8agg cells in the absence or presence of CD24
overexpression for 48 hours (n=6 per group). Data are presented as mean±SD.
Statistical significance was determined by two-way analysis of variance
followed by Tukey’s post hoc test for multiple comparisons (A, C–D, J–L),
Spearman’s correlation (E), and the Kruskal-Wallis test with Dunn’s post hoc
test for multiple comparisons (F–I). BMDMs, bone marrow-derived macrophages;
CFSE, carboxyfluorescein succinimidyl ester; Con, control; CXCL9, C-X-C motif
chemokine ligand 9; DEGs, differentially expressed genes; FC, fold change;
HGSOC, high-grade serous ovarian cancer; IFN-γ, interferon-γ; IHC,
immunohistochemistry; KD, knockdown; MFI, median fluorescence intensity;
mRNA, messenger RNA; MSLN, mesothelin; OE, over expression; p adj, adjusted p
value; rMSLN, recombinant mesothelin; scRNA-seq, single-cell RNA sequencing;
TNF-α, tumor necrosis factor-α.

   [126]Figure 3
   [127]Open in a new tab

MSLN induces CD24 expression through Wnt signaling

   To elucidate the mechanisms by which MSLN modulates CD24 expression, we
   analyzed overlapping pathway activities, including Wnt and NF-κB
   signaling, using bulk RNA sequencing data from subcutaneous tumors and
   ID8agg cells with or without Msln knockdown, as well as scRNA-seq data
   from HGSOC samples with MSLN^+ cancer cells versus MSLN^− cancer cells
   ([128]figure 4A). To investigate the role of these pathways in
   MSLN-induced CD24 expression, we treated cells with Wnt signaling
   inhibitor ICG-001 and NF-κB inhibitor JSH-23. Notably, ICG-001
   significantly suppressed CD24 expression induced by rMSLN ([129]figure
   4B–4D). Consistently, activation of the Wnt pathway using the agonist
   BML-284 markedly enhanced CD24 expression in ID8agg cells ([130]figure
   4E–4G). As canonical Wnt signaling involves β-catenin nuclear
   translocation,[131]^25 we observed that rMSLN increased levels of
   active β-catenin ([132]figure 4H) and facilitated its nuclear
   translocation ([133]figure 4I–4J). Interestingly, our findings suggest
   that Wnt/β-catenin pathway activated by MSLN establishes a
   self-reinforcing loop that amplifies MSLN transcription ([134]figure
   4K). To further investigate the interaction between Wnt signaling and
   the CD24 promoter, luciferase reporter assays were conducted in ID8agg
   cells. Treatment with rMSLN rescued CD24 luciferase activity in Msln^KD
   cells, an effect that was inhibited by Wnt signaling blockade
   ([135]figure 4L). Sequence analysis of 2.0 kb upstream region of the
   transcriptional start site of the mouse Cd24a gene identified putative
   Tcf3 and Tcf7 binding motif. Truncation experiments using CD24 promoter
   deletion constructs revealed that deletion of the −1,169 to 0 bp region
   significantly reduced rMSLN-mediated luciferase activation, indicating
   that this segment, which includes the Tcf7 responsive element, is the
   primary site for Wnt signaling-mediated CD24 transcriptional
   activation. Subsequently, a Tcf7 point mutation was introduced into the
   CD24 plasmid, which also resulted in a significant reduction in
   rMSLN-mediated luciferase activity ([136]figure 4M).

Figure 4. MSLN induces CD24 expression through Wnt signaling. (A) Venn
diagram illustrating overlapping enrichment pathways between functional
enrichment analysis for GO, KEGG and Wiki pathways (with
Benjamini-Hochberg-adjusted p value≤0.05) using significantly upregulated
DEGs (p≤0.05, log[2] (fold change) ≥1) from the bulk transcriptomic
comparison of control and Msln^KD ID8agg cells, comparison of MSLN^+ with
MSLN^− cancer cells from scRNA-seq of HGSOC in [137]GSE180661 and comparison
of bulk transcriptomic comparison of control and Msln^KD subcutaneous tumor
tissues. Two pathways associated with CD24 are listed above. (B–C) Relative
mRNA expression of Cd24a (B, n=3 per group) and MFI of CD24 (C, n=6 per
group) in control or Msln^KD ID8agg cells in the absence or presence of
rMSLN, which were pretreated with NF-κB pathway inhibitor (JSH-23) or Wnt
pathway inhibitor (ICG-001) for 24 hours. The experiment was conducted with
three independent biological replicates. (D) Comparison of levels of CD24 in
control or Msln^KD ID8agg cells in the absence or presence of rMSLN, which
were pretreated with NF-κB pathway inhibitor (JSH-23) or Wnt pathway
inhibitor (ICG-001) for 24 hours. GAPDH served as an internal control. (E–F)
Relative mRNA expression of Cd24a (E, n=3 per group) and MFI of CD24 (F, n=6
per group) in control ID8agg cells with or without treatment with a Wnt
pathway agonist (BML-284) for 24 hours. The experiment was conducted with
three independent biological replicates. (G) Comparison of levels of CD24 in
control ID8agg cells with or without treatment with a Wnt pathway agonist
(BML-284) for 24 hours. GAPDH served as an internal control. (H) Comparison
of levels of dephosphorylated β-catenin (active β-catenin) and total
β-catenin in control and Msln^KD ID8agg cells with or without rMSLN exposure
for 24 hours. GAPDH served as an internal control. (I–J) Representative
immunofluorescence images (I) and quantification of nuclear/cytoplasmic
fluorescence ratio (J) illustrating the location of β-catenin (green) in
control or Msln^KD ID8agg cells in the absence or presence of rMSLN. (K) Bar
plot illustrating relative mRNA levels of Msln in ID8agg cells (n=4 per
group) with or without the addition of rMSLN, BML-284 and ICG-001. (L)
Luciferase reporter assays for control or Msln^KD ID8agg cells transfected
with full-length Cd24a promoter-containing reporter plasmids with or without
rMSLN and ICG-001 treatment (n=4 per group). (M) Luciferase reporter assays
for control or Msln^KD ID8agg cells transfected with specified reporter
plasmids containing either full-length, truncated Cd24a promoters or Tcf7
point mutated Cd24a promoters (n=4 per group). Data are shown as mean±SD.
Statistical significance was assessed using two-way analysis of variance,
followed by Tukey’s post hoc test for multiple comparisons (B–C, J–L),
two-tailed unpaired Mann-Whitney U test (E–F) and the Kruskal-Wallis test
with Dunn’s post hoc test for multiple comparisons (M). Con, control; DAPI,
4',6-diamidino-2-phenylindole; DEGs, differentially expressed genes; GO, Gene
Ontology; HGSOC, high-grade serous ovarian cancer; KD, knockdown; KEGG, Kyoto
Encyclopedia of Genes and Genomes; MFI, median fluorescence intensity; RNA,
messenger RNA; MSLN, mesothelin; NS, no significant difference; rMSLN,
recombinant mesothelin; scRNA-seq, single-cell RNA sequencing.

   [138]Figure 4
   [139]Open in a new tab

MSLN facilitates immune evasion in ovarian cancer through the CD24/Siglec-10
interaction

   The CD24/Siglec-10 signaling axis is a critical mechanism of immune
   evasion, facilitating communication between cancer cells and TAMs. This
   interaction suppresses TAMs-mediated phagocytosis of malignant cells,
   thereby promoting tumor progression.[140]^24 Analysis of scRNA-seq data
   from HGSOC revealed that SIGLEC-10^+ TAMs exhibit a protumorigenic
   profile, characterized by increased expression of markers such as SPP1,
   CD163, FCGR2A, FCGR2B, FN1 and TREM2 ([141]figure 5A). GSEA further
   demonstrated that SIGLEC-10^+ TAMs in the MSLN-Low expression group
   exhibited significantly enhanced phagocytic and antigen-presenting
   capabilities ([142]figure 5B). To validate these findings, flow
   cytometry analysis of Siglec-10^+ TAMs from 22 patients with HGSOC was
   performed. In HGSOC with high MSLN expression, Siglec-10^+ TAMs
   displayed a protumorigenic phenotype, marked by increased levels of
   arginase-1 and CD206, and reduced levels of HLA-DR, CD86, and CXCL9
   ([143]figure 5C). Co-culture experiments further revealed that CD24
   overexpression reversed the antitumorigenic polarization of TAMs
   induced by Msln knockdown ([144]figure 5D). Consistently, in an
   intraperitoneal ID8agg mouse model, CD24 overexpression restored tumor
   progression in Msln^KD cells, reduced survival times, and increased
   hemorrhagic ascites and secondary metastatic foci ([145]figure 5E–5H).
   Moreover, CD24 overexpression inhibited the cytotoxic activity of CD8^+
   T cells in ascites from the Msln^KD group ([146]figure 5I) and reversed
   the antitumorigenic polarization induced by Msln knockdown in TAMs
   ([147]figure 5J). Finally, the clinical significance of the
   co-expression of CD24 and MSLN was examined. Low CD24 expression
   effectively negated the poor prognosis associated with high MSLN
   levels, while patients with high CD24 and high MSLN expression
   demonstrated the lowest median survival ([148]figure 5K, log-rank p
   value=0.002, median survival 44 months vs 36 months). These findings
   underscore the role of CD24 in neutralizing the antitumor benefits of
   Msln knockdown and further highlight its potential as a therapeutic
   target for overcoming immune evasion mechanisms in HGSOC.

Figure 5. MSLN facilitates immune evasion in ovarian cancer through the
CD24/Siglec-10 interaction. (A) Quadrant plot depicting DEGs identified in
SIGLEC-10^+ TAMs and other cells (FDR q<0.05, log[2](fold change) ≥1). (B)
Heatmap showing enrichment scores (−log[10](adjusted p values),
Benjamini-Hochberg corrected FDR) of functional pathways in GO, KEGG and Wiki
pathway databases using significant DEGs of SIGLEC-10^+ TAMs between high and
low MSLN subgroups. (C) Relative abundance of arginase-1, CD206, HLA-DR, CD86
and CXCL9 expressing Siglec-10^+ cells within CD68^+ cells in single-cell
suspension of tumor tissues from patients with HGSOC corresponding to
[149]figure 1E (n=22). (D) Relative abundance of arginase-1, CD206, MHC-II,
CD86, and CXCL9 expressing Siglec-G^+ cells within F4/80^+ CD11b^+ cells,
following 48-hour co-culture of BMDMs with control or Msln^KD ID8agg cells,
with or without CD24 overexpression (n=4 per group). (E–F) Body weight
trajectories (E) and Kaplan-Meier survival curves (F) of advanced ovarian
cancer models intraperitoneally injected with control or Msln^KD ID8agg cells
with or without CD24 overexpression (n=6 per group). (G–H) Quantification of
ascites volume (G) and metastatic nodules (H) in the ovarian cancer models in
(E) at the endpoint (n=6 per group). (I–J) Proportion of TNF-α/IFN-γ,
IFN-γ/granzyme B and TNF-α/granzyme B expressing CD8^+ T cells (I) and
arginase-1, CD206, MHC-II, CD86 and CXCL9 expressing Siglec-G^+ cells in
F4/80^+ CD11b^+ cells (J) in ascites from ovarian cancer models in (E). (K)
Kaplan-Meier plots depicting OS rates of patients with HGSOC from the Pooled
cohort (n=280), stratified into three groups based on combined MSLN and CD24
expression: MSLN-low, MSLN-high/CD24-low, and MSLN-high/CD24-high. Data are
shown as mean±SD. Statistical significance was determined by two-tailed
unpaired Mann-Whitney U test (C), Kruskal-Wallis test with Dunn’s post hoc
test (D, G–J), two-way analysis of variance followed by Tukey’s post hoc test
for multiple comparisons (E), and log-rank test (F, K). BMDMs, bone
marrow-derived macrophages; Con, control; CXCL9, C-X-C motif chemokine ligand
9; DEGs, differentially expressed genes; FDR, false discovery rate; GO, Gene
Ontology; HGSOC, high-grade serous ovarian cancer; IFN-γ, interferon-γ; i.p.,
intraperitoneal injection; KD, knockdown; KEGG, Kyoto Encyclopedia of Genes
and Genomes; MSLN, mesothelin; NS, no significant difference; OE,
overexpression; OS, overall survival; Siglec-G, sialic acid-binding
immunoglobulin-like lectin G; Siglec-10, sialic acid-binding
immunoglobulin-like lectin 10; TAMs, tumor-associated macrophages; TNF-α,
tumor necrosis factor-α.

   [150]Figure 5
   [151]Open in a new tab

Msln knockdown synergize with anti-PD-1 immunotherapy in vivo

   Given that Msln knockdown reprograms the immunosuppressive TME, we
   hypothesized that it would enhance the efficacy of anti-PD-1
   immunotherapy in vivo. The workflow of the in vivo experiment is
   illustrated in [152]online supplemental figure 15A. In an
   intraperitoneal mouse model, Msln knockdown overcame anti-PD-1
   resistance, resulting in substantial reductions in tumor growth
   ([153]figure 6A–6B) and prolonged survival ([154]figure 6C).
   Furthermore, the combination of Msln knockdown and anti-PD-1 treatment
   inhibited ascetic fluid formation and secondary metastases ([155]figure
   6D, [156]online supplemental figure 15B,C). Next, we examined the
   effects of Msln knockdown combined with anti-PD-1 treatment on the
   tumor immune microenvironment. This combination led to enhanced
   activation of cytotoxic CD8^+ T cells and a decreased abundance of
   Siglec-G^+ TAMs, alongside polarization of antitumorigenic TAMs
   ([157]figure 6E–6G, [158]online supplemental figure 15D–F).
   Specifically, we observed an increase in Tpex, which proliferates and
   differentiates into intermediate (Tex-int) and terminally exhausted
   (Tex-term) cells, contributing to the ICB response.[159]^26 27
   Consistently, combination treatment increased the proportion of Tpex
   and Tex-int cells while reducing the fraction of Tex-term cells
   ([160]figure 6H). We further assessed the effect of anti-PD-1 and
   LMB-100 monotherapies, as well as their combination, in HGSOC
   organoids. These treatments progressively reduced CD24 expression in
   tumor cells ([161]figure 6I) and polarized antitumorigenic TAMs
   ([162]figure 6J). The induction of antitumor immune memory is crucial
   for enabling the immune system to mount a swift and effective response
   against previously encountered tumor antigens.[163]^28 To evaluate
   immune memory, we performed a rechallenge experiment using Msln^KD
   ID8agg cells with anti-PD-1 treatment. 60 days after the final IVIS
   assessment, the rechallenged mice showed minimal tumor progression and
   extended survival ([164]figure 6K–6L). Flow cytometry confirmed that
   this combination treatment induced antitumor immune memory in a
   peritoneal mouse model ([165]figure 6M). Notably, previous studies have
   shown that MSLN targeting can facilitate the formation of TLS in a
   humanized murine model.[166]^8 In our study, the combination treatment
   significantly enhanced TLS formation, which was closely correlated with
   increased Tpex cell abundance ([167]online supplemental figure 15G–I).
   These findings suggest that MSLN targeting, in combination with
   anti-PD-1 treatment, can improve responsiveness to anti-PD-1
   immunotherapy by promoting antitumorigenic TAMs polarization, enhancing
   cytotoxic T-cell activity, and fostering TLS formation and Tpex cell
   proliferation.

Figure 6. Msln knockdown synergize with anti-PD-1 immunotherapy in vivo.
(A–B) Representative IVIS bioluminescent images (A) and mean photon flux (B)
of the peritoneal tumor burden at weeks 1, 3, and 5 following the injection
of control or Msln^KD ID8agg-luc cells with anti-PD-1 treatment (n=6 per
group). (C) Kaplan-Meier survival plot of mice corresponding to (A). (D)
Quantification of ascites volume (left) and metastatic nodules (right) in the
peritoneum, omentum, mesentery, and diaphragm of mice in (A). (E–G)
Proportion of IFN-γ, TNF-α, granzyme B, TNF-α/IFN-γ, TNF-α/granzyme B,
IFN-γ/granzyme B expressing CD8^+ T cells (E), Siglec-G expressing F4/80^+
CD11b^+ cells (F) and arginase-1, CD206, MHC-II, CD86 and CXCL9 expressing
F4/80^+ CD11b^+ cells (G) in ascites of mice corresponding to (A). (H)
Percentage of PD-1^+ TCF-1^+, PD-1^+ TCF-1^− CD69^−, PD-1^+ TCF-1^− CD69^+
CD8^+T cells in ascites of mice corresponding to (A). (I–J) Percentage of
CD24^+ cells in CD45^− cells (I) and arginase-1, CD206, HLA-DR, CD86, CXCL9
expressing Siglec-10^+ macrophages (J) in HGSOC organoids treated with IgG
control, anti-PD-1, LMB-100 alone, or the combination of anti-PD-1 and
LMB-100. (K–L) Representative bioluminescence images (K) and survival curve
(L) of tumor cell rechallenge in mice treated and cured with the combination
of Msln^KD cells and anti-PD-1 treatment. A minimum of 60 days following the
disappearance of tumors in peritoneal tumor mice injected with Msln^KD,
ID8agg-luc cells treated with anti-PD-1 therapy were rechallenged
intraperitoneally with ID8agg-luc cells. As a control, naïve female C57BL/6
mice were intraperitoneally injected with ID8agg-luc cells (n=6 per group).
(M) Percentage of CD69^− CD103^−, CD69^+ CD103^−, and CD69^+ CD103^+ cells
among CD44^+ CD8^+ cells in ascites from mice, corresponding to (K). Data are
shown as mean±SD. Statistical significance was assessed by two-way analysis
of variance, followed by Tukey’s post hoc test for multiple comparisons (B,
D–J), log-rank test (C, L), and two-tailed unpaired Mann-Whitney U test (M).
aPD1, anti-programmed cell death-1; Con, control; CXCL9, C-X-C motif
chemokine ligand 9; HGSOC, high-grade serous ovarian cancer; IFN-γ,
interferon-γ; IgG, immunoglobulin G; IVIS, In Vivo Imaging System; KD,
knockdown; Msln, mesothelin; NS, no significant difference; PD-1, programmed
cell death-1; Siglec-G, sialic acid-binding immunoglobulin-like lectin G;
Siglec-10, sialic acid-binding immunoglobulin-like lectin 10; TCF-1,
transcription factor T-cell factor-1; TNF-α, tumor necrosis factor-α.

   [168]Figure 6
   [169]Open in a new tab

Discussion

   This study elucidated the significant role of MSLN, a tumor-specific
   cell surface glycoprotein, in HGSOC (graphic abstract). The
   investigation revealed that MSLN, which is abundantly expressed in
   HGSOC, enhances CD24 expression through the activation of the
   Wnt/β-catenin signaling cascade. Furthermore, this study demonstrated
   that Msln-mediated activation of Wnt/β-catenin signaling creates a
   self-reinforcing loop that amplifies Msln transcription. Consequently,
   this process triggers the CD24/Siglec-10 axis, resulting in the
   inhibition of effector functions in both macrophages and CD8^+ T cells
   within the TME. This mechanism ultimately facilitates immune evasion by
   ovarian cancer cells. These results suggest multiple avenues for
   potential translational applications. First, the expression levels of
   MSLN may serve as reliable indicators of poor prognosis in patients
   with HGSOC. Second, MSLN could serve as a valuable biomarker for
   predicting patient responses to ICB therapy and improving personalized
   treatment strategies. Ultimately, targeting MSLN could offer a
   promising strategy to enhance the responsiveness of patients with HGSOC
   to ICB therapy.

   Previous studies have established that MSLN facilitates tumor cell
   proliferation, invasion, and migration through the aberrant activation
   of several signaling pathways.[170]10,[171]12 Consistent with previous
   findings, we used proliferation assays to demonstrate that knockdown
   Msln significantly impaired tumor cell proliferation in murine ovarian
   cancer cell lines ([172]online supplemental figure 8A–D), and this
   effect was more pronounced in OV2944-HM1 cells. Previous studies have
   suggested that MSLN is involved in reshaping the highly fibrotic TME in
   pancreatic ductal adenocarcinoma, thereby affecting immune cell
   infiltration and drug metabolism.[173]^29 Based on these observations,
   we hypothesized that MSLN influences TME remodeling. However, the
   cellular and molecular mechanisms by which MSLN induces an
   immunosuppressive TME remain unclear. Unlike previous studies on
   colorectal cancer, which found that MSLN expression affected the
   infiltration of various immune cells, including macrophages,
   neutrophils, B cells, CD4^+, and CD8^+T cells,[174]^30 our findings
   demonstrated a high degree of similarity in immunosuppressive patterns
   between patients with HGSOC and ovarian mouse models. Specifically, we
   observed an accumulation of protumorigenic TAMs and depletion of CD8^+
   T cells in both MSLN-high HGSOC tissues and ascites from control tumor
   cell-bearing mice. Msln knockdown reversed this immunosuppressive TME
   and significantly extended the survival of tumor-bearing mice. Multiple
   potential mechanisms within malignant cells may lead to the
   polarization of macrophages towards a protumorigenic phenotype. For the
   first time, we identified and validated a direct ligand-receptor
   interaction between CD24 in MSLN^+ cancer cells and Siglec-10 in TAMs
   through in vivo and in vitro experiments. This interaction drives TAMs
   polarization towards a protumorigenic subtype, suppressing CD8^+ T-cell
   proliferation and activation. MSLN-targeting therapies have been
   reported to promote intratumoral CD8^+ T-cell infiltration.[175]^7 Our
   study elucidated the detailed cellular and molecular mechanisms
   underlying this phenomenon.

   This study is the first to propose that MSLN elevates CD24 expression
   by activating the Wnt/β-catenin signaling pathway. The Wnt/β-catenin
   pathway regulates the oncogenic processes of various solid tumors,
   including the promotion of cell proliferation, epithelial-mesenchymal
   transition, and maintenance of cancer stem cell
   self-renewal.[176]31,[177]36 In breast and colorectal cancers, studies
   have reported that mutations in key components of the Wnt/β-catenin
   pathway, dysregulation of Wnt ligands and receptors, and epigenetic
   modifications contribute to tumorigenesis.[178]36,[179]38 However, the
   role of the Wnt pathway in modulating TME in HGSOC remains unclear. Our
   study addresses this gap by providing preliminary insights into the
   cellular and molecular mechanisms by which the Wnt pathway influences
   TME alterations by promoting the expression of CD24 in HGSOC.

   ICB therapy has revolutionized the treatment landscape for solid
   tumors; however, it has shown limited efficacy for ovarian
   cancer.[180]^39 Targeting MSLN is considered an effective strategy to
   enhance antitumor immune responses and sensitize tumors to ICB
   treatment. The administration of LMB-100 to patients with mesothelioma
   has been observed to substantially enhance the effectiveness of
   anti-PD-1 therapy. This therapeutic approach has demonstrated an
   increase in treatment efficacy from below 20% to 40%, consequently
   impeding tumor progression.[181]^7 In this study, we demonstrated that
   targeting MSLN could shift the immunosuppressive, non-responsive
   microenvironment to an immunoreactive, responsive one. Msln knockdown,
   in combination with anti-PD-1 treatment, significantly prolonged the
   survival of mice bearing intraperitoneal ID8agg tumors and induced the
   formation of tumor-specific immune memory. Based on these findings, we
   propose that targeting MSLN is a promising strategy to enhance the
   effectiveness of ICB therapy against HGSOC.

   Our study had several notable limitations. First, although we observed
   significant tumor growth inhibition with the combination of Msln
   knockdown in a cancer cell line and anti-PD-1 treatment in
   tumor-bearing mice, further investigation is required to determine
   whether LMB-100 combined with anti-PD-1 can achieve comparable results
   given that the antigen pool formed by MPF (Megakaryocyte Potentiating
   Factor) in the serum may result in antigen sink and off-target effects.
   Second, HGSOC exhibits site-specific TMEs, and immune cell composition
   varies across different anatomical sites in individual patients. In
   this study, we did not fully explore the differential impact of MSLN on
   the TME of primary and metastatic tumors. Large-scale clinical studies
   are required to validate the role of MSLN in enhancing ICB therapy
   efficacy.

   In summary, the integration of computational analysis, animal studies,
   omics data, and functional data highlighted the role of MSLN as a key
   mediator of immune evasion. This study not only links MSLN to poor
   prognosis and immunosuppressive TME in HGSOC, but also provides
   evidence supporting MSLN as a promising target to enhance ICB efficacy,
   offering actionable insights for clinical decision-making.

supplementary material

   online supplemental file 1
   [182]jitc-13-2-s001.pdf^ (2.3MB, pdf)
   DOI: 10.1136/jitc-2024-011230
   online supplemental file 2
   [183]jitc-13-2-s002.jpg^ (473.3KB, jpg)
   DOI: 10.1136/jitc-2024-011230

Acknowledgements