CAF-derived LRRC15 orchestrates macrophage polarization and limits PD-1 immunotherapy efficacy in glioblastoma

Abstract

Background

The effectiveness of PD-1/PD-L1 immune checkpoint blockade therapy in glioblastoma (GBM) is limited due to the tumor immunosuppressive microenvironment (TIME). Therefore, strategies of reprogramming TIME to a pro-inflammatory state offer a promising therapeutic approach.

Methods

We applied bioinformatics analysis of single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (stRNA-seq) to identify a significant accumulation of cancer-associated fibroblasts (CAFs) subclusters with elevated LRRC15 expression in the nonresponders to anti-PD-1 therapy. The molecular mechanism of LRRC15 was functionally validated in vitro and in vivo.

Results

These CAFs subclusters drive the infiltration of macrophages (Mφ) into the tumor microenvironment and promote their polarization toward the M2 phenotype. Deletion of Lrrc15 in CAFs significantly restrained tumor growth and prolonged survival in mouse models. Mechanistically, LRRC15 in CAFs promotes IL8 expression by activating the downstream FAK/SRC/NF-κB pathways, leading to Mφ migration and M2-like polarization. In turn, M2-like Mφs secrete TGF-β, which induces LRRC15 expression in CAFs via SMAD2-dependent transcriptional activation. Targeting CAFs subcluster with elevated LRRC15 expression in combination with anti-PD-1 treatment enhanced antitumor efficacy.

Conclusions

Our findings suggest that targeting LRRC15 may provide a novel strategy to augment anti-PD-1 therapy and overcome immunotherapy resistance in GBM.

Graphical Abstract

Graphical Abstract.

Importance of the Study.

Many GBM patients exhibit primary resistance to anti-PD-1 treatment and the underlying mechanisms remain unclear. This underscores the urgent need to identify strategies to overcome immune checkpoint blockade (ICB) resistance. Additionally, the role of cancer-associated fibroblasts (CAFs) in GBM is not well understood. This study uncovers the pivotal role of a CAFs subcluster with elevated LRRC15 expression in creating an immunosuppressive tumor microenvironment in glioblastoma (GBM), which limits the effectiveness of PD-1 ICB therapy. By elucidating the mechanisms through which these CAFs drive macrophage (Mφ) infiltration and polarization towards the M2 phenotype, the research highlights LRRC15 in CAFs as a promising therapeutic target. Targeting LRRC15 may offer a novel strategy to enhance anti-PD-1 therapy and overcome immunotherapy resistance in GBM. These findings provide critical mechanistic insights that could inform the development of more effective therapeutic approaches for GBM patients.

Key Points.

• ScRNA-seq and stRNA-seq reveal a CAF subcluster with elevated LRRC15 expression accumulation in anti-PD-1 nonresponders.

• LRRC15⁺ CAFs promote Mφ M2-like polarization.

• LRRC15 in CAFs boosts IL8 expression by activating FAK/SRC/NF-κB pathways.

Glioblastoma (GBM) is the most common primary, aggressive, malignant brain tumor in adults. Patients with GBM face a grim median survival of about 15 months and a 5-year relative survival rate of only 6.8%.^1,2 While the current standard of care offers some improvement in survival, its impact is limited, underscoring the need for more durable therapeutic options.³ Recently, immune checkpoint blockade (ICB) has emerged as a promising approach for various cancers.⁴ However, despite the remarkable success of programmed cell death protein-1 (PD-1)/programmed cell death ligand 1 (PD-L1) checkpoint blockades in multiple cancer types,^5,6 GBM remains refractory to anti-PD-1/PD-L1 treatment patterns. Many patients exhibit either primary resistance or develop resistance after an initial response.^7–9 This highlights the necessity of identifying strategies to overcome ICB resistance.

The tumor immunosuppressive microenvironment (TIME) plays a crucial role in mediating resistance to immunotherapy. TIME comprises various immune cells, stromal cells, and extracellular matrix components, which interact to influence tumor progression and response to ICB treatment.¹⁰ Among these, macrophages (Mφ) and cancer-associated fibroblasts (CAFs) play pivotal roles in shaping the immune landscape in TIME.^11,12 Mφ, main immune cells in the GBM TIME, can polarize toward M1-like phenotypes in the presence of pro-inflammatory cytokines (eg, IFNg, LPS) or toward M2-like phenotypes in the presence of anti-inflammatory cytokines (eg, IL-4, IL-10, IL-13). High infiltration of M2-like Mφ in the tumor microenvironment (TME) are predictive of poor outcomes in GBM.^13,14

Previously, CAFs were presumed absent in GBM given the lack of fibroblasts in the central nervous system.¹⁵ However, mounting evidence suggests that CAFs exist in GBM due to advances in sequencing technology,^16–19 such as single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics sequencing (stRNA-seq), which have provided unprecedented insights into the cellular heterogeneity and spatial organization of tumors.²⁰

The abundance, origin, biomarkers, and biological contributions of CAFs in GBM are poorly understood. Moreover, CAFs are more abundant in the mesenchymal-like subtype than in other subtypes of GBM.¹⁸ Fibroblast activation protein, a surface marker of CAFs in various tumors, assists CAFs in suppressing antitumor immunity and is associated with poor prognosis in GBM.^21,22 Anti-PD-1 treatment leads to T-cell activation but also results in an increased presence of M2-like Mφ and CAFs, which contribute to the development of resistance to anti-PD-1 treatment.²³

By reanalyzing the scRNA-seq and stRNA-seq profiling of samples from 24 GBM patients in our previous study,²⁴ we discovered a distinct population of CAFs with upregulated leucine-rich repeat-containing protein 15 (LRRC15) in anti-PD-1 nonresponse GBM. LRRC15 has been reported to be upregulated in CAFs of various tumors and is directly expressed in mesenchymal tumors such as GBM, sarcomas, and melanoma.^25,26 However, the specific underlying mechanism by which LRRC15 functions in tumors remains unclear, particularly in GBM.

In this study, through integrated scRNA-seq, stRNA-seq, in vitro, and in vivo experiments, we revealed that CAFs interacted with Mφ through a positive feedback loop to maintain the immune-suppressive microenvironment in GBM, thereby promoting tumor progression and resistance to anti-PD-1 treatment. Specifically, LRRC15 promotes the secretion of IL8 in CAFs via the FAK/SRC/NF-κB signaling axis, which subsequently promotes Mφ polarization towards the M2-like phenotype. Correspondingly, M2-like Mφ promote SMAD2 phosphorylation and induce LRRC15 translation in CAFs through secretion of TGF-β. These findings provide spatial insight into the niche-level regulation of tumor immunity through CAF-Mφ interactions and highlight LRRC15 in CAFs as a potential therapeutic target for cancer immunotherapy.

Methods

Mice

All animal studies were approved and conducted by the Animal Institute of Guangdong Provincial People’s Hospital, following protocols approved by the Medical Experimental Animal Care Commission of Guangdong Provincial People’s Hospital (KY-Q-2022-367). The Lrrc15 knockout (Lrrc15^−/−) C57BL/6j mouse line was obtained from GemPharmatech (Nanjing, China). All mice used in this study were between 6 and 12 weeks of age and were housed under specific-pathogen-free conditions at the animal facility of the Laboratory Animal Center of DKPharm.

Antibodies, Primers, Key Commercial Kit, and Software

Detailed information on the antibodies, primers, and key commercial kits used in this study is provided in the Supplementary material.

Primary Cell Culture and CAFs Isolation

Digestion and dissociation of tumor specimens were followed by the protocol previously reported^16,27 and details were shown in Supplementary Methods.

Single-Cell RNA Sequencing

Single-cell suspensions used for scRNA-seq were acquired from the Lrrc15^+/+ or Lrrc15^-/- CT2A orthotopic tumor and details were shown in Supplementary Methods.

Quantitative Real Time-Polymerase Chain Reaction Assays and Bulk RNA Sequencing

Quantitative real time-polymerase chain reaction (qRT-PCR) and bulk RNA-seq performance details were in Supplementary Methods.

Flow Cytometry

The cell suspension was prepared in advance and blocked with purified CD16/CD32 antibody (553141, BD Pharmingen, USA) and details were shown in Supplementary Methods.

Multiplex Immunofluorescence

The Multiplex immunofluorescence (mIF) was performed using a 7-color multiplex fluorescent immunohistochemical staining kit according to the manufacturer’s protocol (RS0039, ImmunoWay, TX, USA) and details were shown in Supplementary Methods.

Cytokine Secretome Arrays

The cytokine profiles of conditioned medium from different CAF subsets (CAF-sgLRRC15 and CAF-sgNC) were assessed using the Quantibody Human Cytokine Array C5 (RayBiotech, GA, USA) and details were shown in Supplementary Methods.

Western Blot and Co-Immunoprecipitation

Western blot (WB) analysis was performed using a standard protocol²⁵ and for Co-immunoprecipitation assays details were shown in Supplementary Methods.

LRRC15 Knockout Using CRISPR and LRRC15 Overexpression

A pool of 3 sgRNA plasmids targeting human LRRC15 was purchased from Obio Technology Corp. Ltd. (Shanghai, China) and performed according to the manufacturer’s instruction.

Immunohistochemistry Staining

Immunohistochemistry staining of paraffin-embedded human or mouse tumor sections was performed following standard protocols and details were shown in Supplementary Methods.

Statistical Analyses

All data were analyzed using GraphPad Prism 10 (GraphPad Software, La Jolla, CA, USA) or R v.4.0 packages. See Supplementary Methods for details.

Results

A Population of CAFs with Upregulated LRRC15 Expression Is Enriched in Anti-PD-1 Nonresponse GBM and Positively Correlated to Mφ Polarization

Previous research has highlighted the presence of CAFs in GBM and their critical role in the TME.¹⁶ We reanalyzed previously published scRNA-seq and stRNA-seq profiling data²⁴ (Supplementary Table 1) to identify CAF subtypes characterized by high expression of ACTA2 and COL1A1 in GBM (Figure 1A and Supplementary Figure 1A). Our analysis revealed that CAFs, Mφs, and microglia were predominantly present in the anti-PD-1 nonresponse group (Figure 1B). Clusters analysis identified 4 distinct fibroblast clusters: Myofibroblastic CAFs (C1: MyoCAFs), characterized by high ACTA2 and LRRC15 expression; Progenitor CAFs (C2: ProCAFs), marked by C7, BGN, and OGN expression; Matrix-Producing CAFs (C3: MatCAFs), distinguished by COL3A1 and POSTN expression; Inflammatory CAFs (C4: ICAFs), identified by CCL2 and CCL5 expression (Figure 1C–D). StRNA-seq profiling and multiplex immunofluorescence were used to validate the presence of these 4 CAF subtypes in GBM (Supplementary Figure 2B and C). Lineage and differentiation analysis revealed that MyoCAFs are late-stage GBM CAFs that are more prevalent in mesenchymal-like GBM subtypes,¹⁶ suggesting their immunosuppressive role. Consistent with this finding, our results showed that MyoCAFs were most abundant in the anti-PD-1 nonresponse group (Figure 1E). To characterize this immunosuppressive CAF subpopulation, we analyzed the top differentially expressed genes (DEG) in those tumor CAFs from anti-PD-1 response and nonresponse groups using scRNA-seq data, and identified elevated expression of LRRC15 in the nonresponse CAFs (Figure 1F).

Figure 1.

A population of CAFs with upregulated LRRC15 expression is enriched in anti-PD-1 nonresponse GBM and positively correlated to Mφ polarization. (A) UMAP plot displaying scRNA-seq data from 48,961 GBM-derived cells, illustrating 14 distinct cell type clusters. OPC: oligodendrocyte progenitor cells; CAFs: cancer-associated fibroblasts. (B) Proportion of each cell type in nonresponse (n = 5) versus response (n = 7). (C and D) CAFs color-coded into 4 subclusters: C1, C2, C3, and C4; volcano plot of DEGs across the 4 CAF subclusters. Selected marker genes for each CAF subcluster are indicated. (E) Bar plot showing the proportion of each CAF subcluster in PD-1 response (n = 5) and nonresponse (n = 4) groups. ICAF, MatCAF, ProCAF, and MyoCAF are shown as percentages of total CAFs in each group. (F) Volcano plot of DEGs between CAFs in PD-1 response and nonresponse groups. (G) Analysis of LRRC15 mRNA and protein expression in αPD-1 response and nonresponse GBM samples. (H) Analysis of LRRC15 expression in scRNA-seq data. (I) Bubble plot illustrating GO enrichment results for DEG in CAFs between PD-1 nonresponse and PD-1 response group. The x-axis represents the gene ratio, the y-axis represents the GO terms, the bubble size indicates the number of genes, and the bubble color represents the adjusted P-value. (J) the mIF analysis displaying the spatial distribution of LRRC15 or CD206 expression in GBM tissue sections from PD-1 response and nonresponse groups. Spatial distance analysis of cells from LRRC15⁺ cells to CD206⁺ cells based on mIF analysis. Data are presented as mean ± SEM. P-values were determined by a two-sided Student’s t-test (B) and (E). Box limits represent the interquartile range, middle line represents the median, and whiskers represent the 5th to 95th percentiles. Two-sided Wilcoxon rank-sum test (J) (*P < .05, **P < .01, ***P < .001).

We then assessed the mRNA and protein levels of LRRC15 and results showed that increased LRRC15 expression in the GBM anti-PD-1 nonresponse group (Figure 1G).

Consistent findings were observed across several ICB study datasets, including 4 cohorts from the ICBatlas database, one cohort from Zhao et al.’s study and one cohort from Yang et al.’s study²⁸ (Supplementary Figure 1D). ScRNA profiling revealed that LRRC15 is predominantly expressed in CAFs across various cancers (Figure 1H and Supplementary Figure 1E and F). Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis revealed that the DEGs were primarily involved in cytokine–cytokine receptor interactions. Furthermore, CellChat analysis indicated potential CXCL signaling interactions between CAFs and myeloid cells, suggesting that CAFs may communicate with Mφ to mediate resistance to anti-PD-1 therapy (Figure 1I and Supplementary Figure 2A). Correlation analysis based on TCGA data demonstrated a significant positive correlation between LRRC15 and M2-like Mφ as well as exhausted T-cell gene signatures (Supplementary Figure 2B and C). Furthermore, stRNA profiling showed that LRRC15 was closely associated with CD68⁺CD206⁺ Mφs (Supplementary Figure 2D), consistent with results from multiplex immunofluorescence staining (Figure 1J). Subsequently, we evaluated LRRC15 expression in glioma using a human glioma tissue microarray combined with immunohistochemical staining (Supplementary Figure 2E). Kaplan–Meier survival analysis demonstrated a significant association between high LRRC15 levels and poor overall survival, consistent with results obtained from the TCGA dataset (Supplementary Figure 2F–H). Together, these results identify a subpopulation of CAFs characterized by upregulated LRRC15 expression, which may drive immunosuppressive Mφ polarization in GBM.

Genetic Ablation of LRRC15 in CAFs Inhibits Mφs Immunosuppression In Vitro and In Vivo

We used GBM patient tumor-derived CAFs, which expressed LRRC15 and exhibited a transcriptomic profile similar to that of breast CAFs but distinct from pericytes (Supplementary Figure 3A–C), to investigate the role of LRRC15 in CAF in regulating Mφ polarization both in vitro and in vivo. CRISPR/single-guide RNA (sgRNA)-mediated knockout of LRRC15 inhibited CAF-induced immunosuppressive Mφ polarization, as evidenced by decreased expression of CD206 and other M2-like markers (Figure 2A and B). Conversely, Overexpression of LRRC15 in CAF-induced immunosuppressive Mφ polarization (Figure 2C and Supplementary Figure 4A). Moreover, CAFs with LRRC15 overexpression enhanced Mφ migration and vice versa (Supplementary Figure 4B). For in vivo analysis, we utilized Lrrc15 gene knockout (Lrrc15^−/−) C57BL/6J transgenic mice to investigate the role of Lrrc15 in CAF-mediated regulation of tumor immunity. GBM was subsequently induced in these mice (Figure 2D and Supplementary Figure 4C and D). Since Lrrc15 was predominantly expressed in CAFs but minimally in endothelial cells (Figure 1H, Supplementary Figure 1E and Supplementary Figure 5A–F), we mixed primary endothelial cells with GBM cells to assess their impact on tumor growth. The results confirmed that Lrrc15 knockout in endothelial cells did not significantly affect tumor growth or the TME (Supplementary Figure 6A–F). Subsequently, we found that Lrrc15 knockout augmented CD8⁺ T-cell infiltration and reduced T-cell exhaustion as well as M2-like Mφ infiltration in orthotopic GL261 or CT2A tumors, as evidenced by flow cytometry analysis, mRNA expression analyses, and multiplex immunofluorescence staining (Figure 2E and F and Supplementary Figure 7A–C). Lrrc15 deficiency in CAFs increasing the expression of T-cell activation markers, including IFNG and Ki67(Figure 2G). Furthermore, we performed scRNA-seq analysis and bulk transcriptomic analyses to evaluate the intracranial tumor microenvironment in Lrrc15^−/−mice (Figure 2H). Results showed that Mφs exhibited increased expression of antitumor response genes like Isg15, Cxcl9, and Ccl24 and T cells expressed higher levels of Ifng and Ifitm3 (Figure 2I). Lrrc15 deletion in CAFs led to immune response-activating signaling pathway and regulation of T-cell activation in Mφs and T cells (Supplementary Figure 8A–D). All in all, these findings suggest that LRRC15 in CAFs plays a critical role in Mφ-mediated immunosuppression in GBM.

Figure 2.

Genetic ablation of LRRC15 in CAFs inhibits Mφs immunosuppression in vitro and in vivo. (A) Protein levels of LRRC15 in different CAFs with LRRC15 overexpression or knockout were detected by WB. (B) Expression levels of M1-like and M2-like Mφ markers following treatment with conditioned media (CM) derived from different CAF populations. (C) CD206⁺ Mφs detected using flow cytometry (FCM) after treatment with different CAFs-CM. (D) Mice were orthotopically inoculated with GL261 or CT2A glioma cells, and in vivo bioluminescence imaging was performed on days 12 and 22. (E) Flow cytometry analysis of specific biomarkers on Mφs and CD8⁺ T cells from GL261 orthotopic tumors of WT or Lrrc15^−/− mice. (F) MIF images of orthotopic GL261 tumors in WT and Lrrc15^−/− mice showing staining for DAPI (nuclei), CD3 (T cells), CD8 (cytotoxic T cells), PD-1 (immune checkpoint), and CD206 (M2 Mφs) (up). Scale bar: 20 μm; quantification of immune cell infiltration (down). (G) Flow cytometry analysis of specific biomarkers on CD8⁺ T cells from GL261 orthotopic tumors of WT or Lrrc15^−/− mice. (H) UMAP plot showing scRNA-seq data from 15,206 CT2A orthotopic tumor-derived cells with 11 clusters of different cell types. (I) Heatmap showing the expression of selected genes in Mφs (left) and T cells (right) in WT or Lrrc15^-/- orthotopic tumors based on scRNA-seq. (J) Representative in vivo bioluminescence imaging of GL261 orthotopic tumor formation (up); quantitative analysis of bioluminescent signal intensity at days 12 and 22 (n = 10) mice per group, down; Kaplan–Meier survival curve showing the survival rate of mice with orthotopic tumors (right), with Log-rank test performed. Data from 3 independent experiments are summarized and displayed as mean ± SEM. Two-sided Student’s t-test was used for (B), (C), (E), (F), (G), (J) (left). (*P < .05, **P < .01, ***P < .001).

Deletion of LRRC15 in CAFs Impairs Tumor Growth and Improves Animal Survival

We next evaluated the impact of Lrrc15 knockout on tumor growth and animal survival. We found that deletion of Lrrc15 remarkably impeded the growth of GL261 (an immunogenic model) intracranial tumors compared to wild-type (WT) mice, thereby prolonging mouse survival (Figure 2J). Additionally, deletion of Lrrc15 impeded the growth of GL261 subcutaneous tumors (Supplementary Figure 9A). Similar results were observed with another relatively less immunogenic model, CT2A (Supplementary Figure 9B and C). To determine whether the inhibitory effect of LRRC15 deletion in CAFs on GBM growth is mediated by Mφ, we treated the mouse model with an anti-CSF1R (αCSF1R) antibody. F4/80⁺ Mφs were significantly depleted in GL261-bearing mice following αCSF1R treatment (Supplementary Figure 9D). Notably, depletion of CSF1R⁺ Mφs abrogated the tumor growth inhibition observed in Lrrc15-deficient mice (Supplementary Figure 9E and F), suggesting that tumor eradication in our murine Lrrc15 knockout GBM model indeed relies on Mφs. Collectively, these findings suggest a critical role of LRRC15 in CAFs in regulating GBM growth.

LRRC15 in CAFs Promotes Alternative Mφ Polarization via IL8

To gain molecular insights into how LRRC15 in CAFs regulates Mφ immunosuppression, we compared the cytokine secretion profiles between CAF#1 sgNC and CAF#1 sgLRRC15 using a cytokine array (Figure 3A). These findings were further validated by mRNA expression analyses (Figure 3B and Supplementary Figure 10A). Previous studies indicated that IL8 serves as a crucial mediator in M2-like Mφ polarization,²⁹ and we speculated that IL8 is an important downstream effector of LRRC15 in CAFs in regulating M2-like Mφ polarization. We further performed bulk RNA sequencing using CAFs transduced with either the vector or LRRC15 overexpression (OE) construct. DEG analysis revealed a significant upregulation of IL8 upon LRRC15 overexpression in CAFs, consistent with scRNA-seq data comparing between LRRC15^high CAFs and LRRC15^low CAF subsets (Figure 3C and Supplementary Figure 10B). Bioinformatic analysis showed a positive correlation between the mRNA levels of IL8 and LRRC15, as well as IL8 and CD206 (Figure 3D and Supplementary Figure 10C). Knocking out LRRC15 decreased IL8 secretion, while overexpression of LRRC15 upregulated IL8 expression (Figure 3E and Supplementary Figure 10D and E). Furthermore, rescue assay showed that IL8 knockdown significantly decreased LRRC15 overexpression-induced Mφ chemotaxis ability and CD206 expression (Figure 3F and G and Supplementary Figure 10F and G). Subsequently, exogenous recombinant IL8 (rIL8) stimulation increased Mφ migration ability and the expression of CD206, TGF-β, and STAT6 phosphorylation (Figure 3H and Supplementary Figure 10H). Overall, these results identify IL8 as a downstream target of LRRC15 in driving CAF-induced pro-tumorigenic Mφ polarization.

Figure 3.

LRRC15 in CAFs promotes alternative Mφ polarization via IL8. (A) Secretomics arrays (up) and statistics table (down) showing altered cytokines secreted by CAF-sgLRRC15 compared with CAF-sgNC. (B) QRT-PCR verifying the mRNA levels of different cytokines secreted by CAF-sgLRRC15 or CAF-sgNC. (C) Volcano plot showing the up- or downregulated genes in CAF-LRRC15 OE compared to vector from bulk RNA-seq. (D) Spearman correlation analysis between IL8 and LRRC15 based on CGGA-GBM dataset. (E) Protein levels of IL8 in different LRRC15 level CAF groups detected by WB. (F) CD206⁺ Mφs detected using flow cytometry after treatment with different CAFs-CM. (G) Migration ability of Mφs after co-culture with differently treated CAF cells. (H) WB was used to detect the protein levels of phosphorylated STAT6, CD206, and TGF-β1 in THP-1 cells after treatment with different concentrations of rIL8 for 24 h (Up) or 40 ng/ml of rIL8 and detected at different times (Down). Data from 3 independent experiments are summarized and displayed as mean ± s.e.m. Two-sided Student’s t-test was used for (A). The P-value in (D) was calculated using the Spearman correlation test. P-values in (B), (F), and (G) were calculated by one-way ANOVA followed by Dunnett’s tests. (n.s. no statistical significance, *P < .05, **P < .01, ***P < .001).

LRRC15 Regulates IL8 Expression in CAFs by Activating the FAK-SRC-NF-κB Pathway

To investigate how LRRC15 drives IL8 expression in GBM CAFs, we performed the signaling pathway activity analysis based on GBM scRNA-seq data, which showed activation of NF-κB signaling pathways in CAFs (Figure 4A). Gene Set Enrichment Analysis (GSEA) analysis based on TCGA-GBM data also showed that LRRC15 activated cytokine signaling and CD163 mediated an anti-inflammatory response (Figure 4B). Bulk RNA sequencing data showed that overexpression of LRRC15 resulted in the activation of immune response, integrin and NF-κB signaling pathways (Figure 4C and D). Subsequently, we detected the downstream of integrin, like phosphorylation of FAK, SRC, and P65, finding that knockout of LRRC15 decreased phosphorylation of FAK, SRC, and P65, while exogenous LRRC15 supplementation recovered their phosphorylation (Figure 4E). Additionally, we conducted a rescue assay using antagonists for FAK (PF573228), SRC (PP2 or Dasatinib) and NF-κB (CAPE). We observed a significant increase in IL8 levels in CAFs with LRRC15 overexpression, which was largely reversed by antagonists of FAK, SRC, and NF-κB (Figure 4F). We also observed that these treatments significantly reduced the LRRC15 overexpression-induced upregulation of M2-like markers and impaired Mφ chemotaxis ability (Figure 4G and H and Supplementary Figure 11A–C). Collectively, these results reveal an LRRC15-FAK-SRC-NF-κB axis regulating IL8 expression.

Figure 4.

LRRC15 regulates IL8 expression in CAFs by activating the FAK-SRC-NF-κB pathway. (A) Signaling pathway activity analysis of different cell clusters based on scRNA-seq data. (B) GSEA of DEGs between LRRC15 high and LRRC15 low based on the TCGA-GBM dataset. (C) Heatmap showing DEGs between CAF#1-Vector and CAF#1-LRRC15 OE by bulk RNA-seq. Each sample has 3 replicates. (D) GSEA of DEGs between CAFs with LRRC15 overexpression (OE) and vector control, identifying several oncogenic signaling pathways active in LRRC15-OE CAFs compared to vector. (E) Protein levels of genes in the FAK/SRC/NF-κB signaling pathway in NC, sgLRRC15#1, sgLRRC15#2, or sgLRRC15#2 with added exogenous LRRC15 protein, analyzed by WB. (F) WB analysis of the indicated proteins in CAF#1 and CAF#2 cells with LRRC15 overexpression after treated with the indicated inhibitors (FAK inhibitor: PF573225 (10 μM), SRC inhibitor: PP2 (10 μM), SRC inhibitor: Dasatinib (0.3 μM) or NF-κB inhibitor: CAPE (10 μM)) for 24 h. (G) Expression of CD206 in THP-1 after treating with different CAFs-CM, determined by flow cytometry. (H) Migration ability of THP-1 after co-culture with different CAFs. Data from 3 independent experiments are summarized and displayed as mean ± s.e.m. P-values in (G) and (H) were calculated by one-way ANOVA followed by Dunnett’s tests. (n.s. no statistical significance, *P < .05, **P < .01, ***P < .001).

M2-Like Mφ-Derived TGF-β1 Promotes LRRC15 Expression in CAFs via SMAD2-Driven Transcriptional Activation

Since LRRC15 has been reported to be induced by TGF-β,³⁰ we analyzed the scRNA-seq data and found that Mφ have high TGFB1 expression (Supplementary Figure 12A). Our results showed that exogenous TGF-β1 stimulation induced LRRC15 expression in a dose-dependent manner (Figure 5A). We further utilized 3 databases (GTRD, hTFtarget, and JASPAR) to identify potential transcription factors (TFs) that could activate LRRC15 expression. The intersection of these 3 databases indicated that SMAD2, RUNX1, and ESR1 were potential TFs for LRRC15 (Figure 5B and Supplementary Figure 12B). However, treating CAF#1 with varying concentrations of TGF-β did not change the mRNA levels of SMAD2, RUNX1, and ESR1 (Supplementary Figure 12C). We further found that only siSMAD2 treatment downregulated LRRC15 expression (Supplementary Figure 12D–F). TCGA-GBM data also showed a positive correlation between LRRC15 and SMAD2 (Figure 5C). We hypothesized that TGF-β1 regulates LRRC15 levels by mediating SMAD2 phosphorylation and found that exogenous TGF-β1 stimulation indeed upregulated phosphorylation of SMAD2 in a time- and dose-dependent manner (Figure 5D and E). Bioinformatic analysis showed 3 SMAD2 binding sites on the LRRC15 DNA, with 2 located in the promoter region (Figure 5F and Supplementary Figure 12G). We performed the ChIP-qPCR assay and found that compared to the anti-IgG control, the enrichment of anti-p-SMAD2 at binding site 1 was significantly higher (Figure 5G, left). In CAF#1 cells, siSMAD2 treatment decreased the enrichment of anti-p-SMAD2 in the LRRC15 promoter region as shown (Figure 5G, right). We further designed 3 mutation sites based on SMAD2 DNA binding motifs. The WT SMAD2 binding site 1 (SMAD2 bs1) sequence and the 3 mutant sequences were cloned into the pGL3-luciferase reporter plasmid vector, respectively (Figure 5H). In the dual luciferase reporter assay, after TGF-β1 stimulation, the relative luciferase intensity of mutants 1 and 2 was significantly lower compared to the WT and mutant 3  (Figure 5I). In 2 other dual luciferase reporter experiments, after stimulation with TGF-β1, it was confirmed that in the presence of the DNA sequence ‘TGACTCA,’ p-SMAD2 can recognize and bind to the promoter region of LRRC15, thereby promoting the transcriptional activation of LRRC15 (Figure 5J–K). All in all, our data indicate that M2-like Mφ-produced TGF-β induces CAFs to express LRRC15 through SMAD2 transcriptional activation.

Figure 5.

M2-like Mφ-derived TGF-β1 promotes LRRC15 expression in CAFs via SMAD2-driven transcriptional activation. (A) MRNA levels of LRRC15 in CAFs treated with different doses of TGF-β1, measured by qRT-PCR. (B) Intersection of 3 transcription factor (TF) databases (GTRD, hTFtarget, and JASPAR) identifying SAMD2, RUNX1, and ESR1 as potential TFs regulating LRRC15 transcription. (C) Correlation analysis between LRRC15 and SMAD2 based on the TGGA-GBM dataset. (D and E). Protein levels of LRRC15 and phosphorylated-SMAD2 in CAFs treated with different doses of TGF-β1 (ng/ml) or 5 ng/ml TGF-β1 at various time points, detected by WB. (F) Diagram illustrating the binding sites of phosphorylated-SMAD2 in the promoter region of LRRC15. (G) ChIP-qPCR verifying the binding sites (bs) of phosphorylated-SMAD2 in the promoter region of LRRC15, with bs1 identified as the specific binding site for phosphorylated-SMAD2. (H) Diagram showing the construction of WT SMAD2 bs1 sequence and 3 mutant plasmids. (I–K) Dual luciferase assays demonstrating that TGF-β1 stimulation allows SMAD2 to activate LRRC15 transcription by binding to the “TGACTCA” consensus sequence in binding site 1 of the LRRC15 promoter region. Data from 3 independent experiments are summarized and displayed as mean ± SEM. Two-sided Student’s t-test was used for (G) (left) and (J). P-values in (A), (G) (right), (I), and (K) were calculated by one-way ANOVA followed by Dunnett’s tests. The P-value in (C) was calculated using the Spearman correlation test. (n.s. no statistical significance, *P < .05, **P < .01, ***P < .001).

LRRC15 Blockade Sensitizes GBM to PD-1 Immunotherapy

The observed increase in T cell infiltration and reduction in immunosuppressive signaling following LRRC15 deletion in CAFs suggests that pharmacological inhibition of LRRC15 could serve as a promising strategy to enhance adoptive T cell therapy. We utilized an anti-LRRC15 monoclonal antibody (αLRRC15) to assess its effect on Mφ polarization in vitro. Bone marrow–derived macrophages were treated with L929-CM, L929-Lrrc15 OE-CM or L929-Lrrc15 OE-CM supplemented with αLRRC15. Flow cytometry and qRT-PCR analysis showed that αLRRC15 treatment significantly decreased M2-like Mφs polarization (Supplementary Figure 13A and B). We further administered αLRRC15 to tumor-bearing mice. Treatment with αLRRC15 significantly inhibited GL261 tumor growth in C57BL/6J mice and was associated with improved survival outcomes (Figure 6A and B and Supplementary Figure 13C). The levels of infiltrating M2-like Mφs (F4/80⁺CD206⁺) and CD8⁺ T-cell as well as the expression of CD8⁺ T-cell exhaustion markers were significantly reduced in the αLRRC15-treated group, whereas CD8⁺ T-cell proliferation (Ki67) and effector function (IFNG) were markedly enhanced, as evidenced by flow cytometry analysis, mRNA expression analyses and multiplex immunofluorescence staining (Supplementary Figure 13D–J). We further test whether LRRC15 deletion-induced inflamed tumor microenvironment could synergize with ICB therapy. We found that combining Lrrc15 deletion with αPD-1 treatment significantly inhibited GL261 orthotopic and subcutaneous tumor growth and extended mouse survival compared to no treatment, Lrrc15 deletion alone or PD-1 blockade alone (Supplementary Figure 14A and B). Consistent results were observed in another model using CT2A cells (Supplementary Figure 14C and D). Next, we treated C57BL/6J mice bearing orthotopic or subcutaneous GL261 or CT2A tumors with either αLRRC15, αPD-1 or a combination of both. Combination therapy resulted in significantly prolonged survival compared to monotherapy with either αLRRC15 or αPD-1 (Figure 6C and D and Supplementary Figure 14E–G), and markedly enhanced CD8⁺ T-cell infiltration while reducing M2-like macrophage accumulation and T-cell exhaustion (Figure 6E). Together, our data confirm that targeted LRRC15 treatment enhances the efficacy of PD-1 blockade.

Figure 6.

LRRC15 blockade sensitizes GBM to PD-1 immunotherapy. (A) Schematic illustration of LRRC15 blocking antibody (αLRRC15) therapy in WT mice. Mice were orthotopically inoculated with GL261 glioma cells and administered intraperitoneally (i.p.) with PBS, isotype control (IgG), antimouse LRRC15, or left untreated on days 6, 9, 12, and 15. In vivo bioluminescence imaging of GL261 orthotopic tumor formation with different treatments (up); quantitative analysis of bioluminescent signal intensity at days 12 and 22 (n = 5 mice per group, down); (B) Kaplan–Meier survival curve showing the survival rate of mice with orthotopic tumors (right), with Log-rank test performed. (C) Schematic illustration of αLRRC15 and/or αPD-1 therapy in WT mice orthotopically inoculated with GL261 or CT2A. In vivo bioluminescence imaging of GL261 orthotopic tumor formation with different treatments (up); quantitative analysis of bioluminescent signal intensity at days 12 and 22 (n = 8 mice per group, down); (D) Kaplan–Meier survival curve showing the survival rate of mice with orthotopic tumors (right), with Log-rank test performed. (E) MIF images of orthotopic GL261 tumors in WT mice with different treatments (up). Statistical analysis of immune cell infiltration based on the mIF images (down). (F) Schematic illustration showing how CAFs with elevated LRRC15 expression shape the immunosuppressive tumor microenvironment by interacting with Mφ, thereby promoting tumor progression and resistance to anti-PD-1 therapy. Specifically, LRRC15 in CAFs promotes the secretion of IL8 via the FAK/SRC/NF-κB signaling axis, which subsequently promotes Mφ polarization towards the M2 phenotype. Correspondingly, M2-like Mφs induce LRRC15 translation and activation in CAFs through the secretion of TGF-β and promote SMAD2 phosphorylation. Data from 3 independent experiments are summarized and displayed as mean ± SEM. P-values in (A) (down), (C) (down), and (E) were calculated by one-way ANOVA followed by Dunnett’s tests. The P-values in (B) and (D) were calculated using the Log-rank test. (n.s. no statistical significance, *P < .05, **P < .01, ***P < .001).

Discussion

The advent of PD-1/PD-L1 immune checkpoint inhibitors has revolutionized the treatment of many solid tumors, marking a new era in cancer therapy.³¹ However, the clinical efficacy of ICB in GBM remains very limited. Recent studies have shown that GBM patients do not significantly benefit from PD-1/PD-L1 checkpoint blockade therapy.^8,9,32 Conversely, 2 other clinical trials demonstrated that neoadjuvant anti-PD-1 therapy (neo-aPD1) was associated with improved overall and progression-free survival in a subset of GBM patients.^33,34 Thus, there is a significant unmet clinical need to understand the mechanisms behind immunotherapy resistance in GBM patients. This study reveals that a subpopulation of CAFs with upregulated LRRC15 interacts with Mφs through a positive feedback loop, driving an immunosuppressive microenvironment in GBM and promoting tumor progression and anti-PD-1 resistance. LRRC15 enhances IL8 secretion in CAFs via the FAK/SRC/NF-κB axis, which induces Mφ polarization to the M2-like phenotype. In turn, M2-like Mφs promote SMAD2 phosphorylation and LRRC15 expression in CAFs through TGF-β secretion. Combining LRRC15 targeting with anti-PD-1 treatment significantly improves antitumor efficacy (summarized in Figure 6F).

GBM has a complex TIME, which reduces the effectiveness of immunotherapy.³⁵ Among the immune cell components in the TME of GBM, Mφs are the most abundant, constituting up to half of the total cells.³⁶ Mφs are predominantly biased toward immunosuppressive M2 polarization, wherein M2-like Mφs degrade pro-inflammatory factors and cytokines, contributing to the exhaustion of CD8⁺ T cells in GBM.^37,38 However, the mechanisms regulating Mφs tumor infiltration remain unclear. CAFs have been reported to play a significant role in regulating Mφs infiltration.³⁹ The interaction between CAFs and Mφs is of widespread concern in other tumors, but not in GBM.³⁹ Recent studies, facilitated by advances in sequencing technology like scRNA-seq and stRNA-seq, have identified CAFs in GBM, particularly in the mesenchymal-like state or subtype.^16,18 However, the function of CAFs in GBM immune regulation has received little attention until now.

In this study, we identified a CAFs subcluster with upregulated LRRC15 enriched in PD-1 nonresponse GBM samples based on scRNA-seq and stRNA-seq analysis. Functional experiments demonstrated that CAFs promote TAMs infiltration and M2-like Mφs polarization via FAK-SRC-NF-κB signaling pathway, which promotes IL8 secretion. This suggests that a CAFs subcluster with upregulated LRRC15 plays a critical role in shaping the TIME by interacting with Mφs and subsequently contributing to resistance to anti-PD-1 therapy in GBM. This finding aligns with other reports that IL8 promotes Mφs M2 polarization.^36,40 Therefore, targeting this CAFs subcluster and repolarizing immunosuppressive Mφs into immunostimulatory Mφs is a promising research direction.

Furthermore, we introduced an Lrrc15^−/− C57BL/6j transgenic mouse model. Compared to WT mice, we observed increased M1-like Mφs markers and decreased M2-like Mφs markers as well as markers of T-cell exhaustion in Lrrc15^−/− mice with CT2A orthotopic tumor. The CSF1R blocking experiment abolished the protective phenotype observed in Lrrc15^−/− mice. We hypothesize that the TME in tumor-bearing Lrrc15^−/− mice is enriched with M1-like Mφs and depleted of M2-like Mφs, creating an immune-activated tumor microenvironment that promotes tumor cell elimination. However, when Mφs were depleted using CSF1R antibodies, this antitumor immune microenvironment was disrupted, indicating that Mφs are critical downstream effectors through which CAFs subcluster exert their effects.

Our results showed that combined treatment with anti-LRRC15 and anti-PD-1 effectively suppressed GBM progression in mouse models. These findings suggest that co-targeting LRRC15⁺ CAFs and PD-1 may offer a promising strategy to overcome immunotherapy resistance. This study investigates how a CAF subcluster with elevated LRRC15 expression promotes M2-like Mφs polarization and how these Mφs, in turn, activate CAFs via TGF-β secretion. TGF-β stimulation increases LRRC15 expression in CAFs.³⁰ We observed that phosphorylated SMAD2-mediated regulation of LRRC15 transcription is TGF-β-driven. This positive feedback loop contributes to a more TIME in GBM.

In conclusion, our findings showed that a CAFs subcluster with upregulated LRRC15 interacted with Mφs through a positive feedback loop to maintain the immune-suppressive microenvironment in GBM. Targeting LRRC15 in CAFs may offer a viable strategy to improve the responsiveness to anti-PD-1 immunotherapy in GBM.

Conflict of interest statement

None declared.

Funding

The study design, data collection, data analysis, manuscript preparation, and publication decisions of this work were supported by NSFC Incubation Project of Guangdong Provincial People’s Hospital (KY0120220043), National Natural Science Foundation of China (No. 82203081), Basic and Applied Basic Research Foundation of Guangzhou, China (Grant No. 2023A04J0523) to H.P.C.; Natural Science Foundation of Guangdong Province of China (2022A1515012540) to D.Z.; Guangzhou Municipal Science and Technology Bureau (202002030128) to D.X.Z.

Author Contributions

F.F.L., D.Z., and H.P.C. conceived and coordinated the project. F.F.L. carried out most of the mouse experiments. S.Y.X. prepared the patient samples. H.P.C. performed bioinformatics analysis. Y.M., Y.W.L., J.H.Y., E.N.C., and C.S. helped with animal experiments. F.F.L., Y.M., H.P.C., and J.H.Y. performed the functional assay and molecular biology experiment. H.P.C. and F.F.L. assisted with manuscript and figure preparation. H.P.C., P.W., and F.F.L. interpreted the data. H.P.C. and F.F.L. wrote the manuscript.

Data Availability

The data of this study are available within the Supplementary materials. Further inquiries can be directed to contact the corresponding authors.