Kindlin-2-mediated hematopoiesis remodeling regulates triple-negative breast cancer immune evasion

Abstract

Triple-negative breast cancer (TNBC) presents significant clinical challenges due to its limited treatment options and aggressive behavior, often associated with poor prognosis. This study focuses on Kindlin-2, an adaptor protein, and its role in TNBC progression, particularly in hematopoiesis-mediated immune evasion. TNBC tumors expressing high levels of Kindlin-2 induce a notable reshaping of hematopoiesis, promoting expansion of myeloid cells in bone marrow (BM) and spleen. This shift correlated with increased levels of neutrophils and monocytes in tumor-bearing mice over time. Conversely, genetic knockout of Kindlin-2 mitigated this myeloid bias and fostered T cell infiltration within the tumor microenvironment, indicating Kindlin-2’s pivotal role in immune modulation. Further investigations revealed that Kindlin-2 deficiency led to reduced expression of PD-L1, a critical immune checkpoint inhibitor, in TNBC tumors. This molecular change sensitized Kindlin-2-deficient tumors to host anti-tumor immune responses, resulting in enhanced tumor suppression in immune-competent mouse models. Single-cell RNA sequencing, bulk RNA-seq, and immunohistochemistry data supported these findings by highlighting enriched immune-related pathways and increased infiltration of immune cells in Kindlin-2-deficient tumors. Therapeutically, targeting PD-L1 in Kindlin-2-expressing TNBC tumors effectively inhibited tumor growth, akin to the effects observed with genetic Kindlin-2 knockout or PD-L1-KO. Our data underscore Kindlin-2 as a promising therapeutic target in combination with immune checkpoint blockade to bolster anti-tumor immunity and counteract resistance mechanisms typical of TNBC and other immune evasive solid tumors.

Introduction

Metastatic breast cancer (BC) is responsible for the death of ~90% of BC patients (1). In fact, metastatic BC is the 2^nd leading cause of cancer-related deaths in women in the United States, annually accounting for more than 44,000 deaths and 281,000 new cases of BC (2). Once BC progresses to the metastatic stage it becomes clinically incurable and results in a mediocre median survival of only 1.5 to 3 years. Clinically, ~30% of BC patients diagnosed with early-stage, noninvasive disease will ultimately progress to late-stage metastatic disease, an event that severely limits treatment options and associates with dismal clinical outcomes (3). These limitations are further exacerbated by the fact that BCs are heterogeneous and comprised of at least 5 genetically distinct subtypes and more than ten molecular subtypes (4–6). Amongst individual BC subtypes, those classified as triple negative BCs (TNBCs) are especially lethal due to their highly metastatic behavior and propensity to recur rapidly. As a group, TNBCs lack expression of hormone receptors (ER-α and PR) and ErbB2/HER2 (7–10), which has prevented the development of FDA-approved targeted drug therapies effective against this BC subtype. Likewise, TNBCs frequently acquire resistance to standard-of-care therapies through mechanisms that remain incompletely understood.

Anti-tumor immune evasion is one of the major hallmarks of cancer (11–13), especially in the so-called immune cold tumors, such as TNBC, which are notorious for their lack of response to immunotherapy(14). The interaction between tumor cells and their microenvironment plays a major role in tumor progression and metastasis (15–17). For example, infiltration of solid tumors by myeloid cells is generally associated with poor patient prognosis and disease severity (14, 18–23). On the other hand, increased tumor-infiltering lymphocytes (TILs) is associated with positive disease outcome (17, 24–26), which determines the efficacy of anti-tumor response (24, 27) (28). Recent studies have now established a direct systemic effect of breast cancer tumors in inducing changes in hematopoiesis (23, 29) that favors expansion of myeloid cell populations involved in anti-tumor immune evasion (19, 23).

Kindlins are a small gene family (3 members) of FERM domain-containing adaptor proteins that function as essential drivers of integrin activation. Moreover, aberrant Kindlin expression and activity is associated with several human pathologies, including cancer (30–32). Kindlin-2 is the most widely expressed member of the Kindlin family; its homozygous deletion in mice is embryonic lethal, while mice heterozygous at the Kindlin-2 locus exhibit overtly normal phenotypes that give way to defects in angiogenesis, hemostasis, and the cytoskeletal architecture upon challenge (32–34). Similar to its family members, Kindlin-2 expression is also dysregulated in several human cancers, including those of the breast (30–32). Several published studies from our group and others have established Kindlin-2 as a major driver of the oncogenic behavior of TNBC tumors through the regulation of several hallmarks of cancer (31).

In this study we investigated the systemic effect of Kindlin-2-expressing TNBC tumors on hematopoiesis remodeling and its subsequent effect on anti-tumor immune evasion. We used a combination of in vitro and in vivo assays to show that the Kindlin-2 is involved in anti-tumor immune evasion of TNBC tumors, through the remodeling of hematopoiesis in mice bearing Kindlin-2-expressing tumors. Specifically, we show that Kindlin-2 expression in TNBC tumors modifies hematopoiesis by increasing myelopoiesis in the bone marrow (BM) and spleen and altering the immune cell landscape, including heightened neutrophil counts, and changed T-cell distributions, while Kindlin-2 loss not only counteracts these changes but also activates immune checkpoint molecules, suggesting Kindlin-2’s critical influence on immune evasion and tumor progression. We also show that genetic targeting of Kindlin-2 in TNBC tumors resulted in a loss of expression of PD-L1, a major immune checkpoint inhibitor. As a therapeutic application of this finding, we show that treatment of tumor bearing mice with anti PD-L1 inhibited growth of TNBC tumors to levels comparable to those achieved by genetic targeting of Kindlin-2 or PD-L1.

Together, our data establishes a novel function of Kindlin-2 in the regulation of hematopoiesis and anti-tumor immune evasion, thereby supporting the establishment of Kindlin-2 is a potential therapeutic target for TNBC tumors.

Materials and Methods

Ethics Statement

All animal studies were performed under protocols approved by the Institutional Animal Care and Use Committee and conducted in accordance with the guidelines and regulations set and approved by the MetroHealth Medical Center, Case Western Reserve University, and NIH. For this study, we used six- to eight-week-old female BALB/c mice (Jackson Laboratory, Farmington, CT, USA).

Cell lines and reagents

4T1, and E0771 cells were procured from the American Type Culture Collection (ATCC Cat# CRL-2539, RRID: CVCL_0125 and Cat# CRL-3461, RRID: CVCL_GR23; Manassas, VA) and maintained in accordance with the manufacturer’s specified protocols. Although cell line authentication was not explicitly conducted, we relied on the manufacturer’s quality control assurances. Periodic testing for Mycoplasma contamination was performed every 9 to 12 months. All cells were cultured at early passages (no more than 15), and each culture was passaged no more than five times before introducing a fresh vial. Kindlin-2-deficient cells were generated through electroporation of cancer cells with a ribonucleoprotein mixture of guide RNAs (sgRNA) and Cas9 (Synthego), following the manufacturer’s instructions and as described previously (35). A pool of three verified sgRNAs was used for each human or mouse gene (Synthego), with scrambled sgRNAs serving as a negative control (Supplementary Table 1). Western blot (WB) analysis validated efficient and stable knockout (KO). In cases where knockdown efficiency was below 80%, a second round of sgRNA delivery was implemented. No antibiotic selection was required, as the knockdown efficiency was sustained throughout the cells’ utilization. Primary antibodies for Kindlin-2, 3 was sourced from EMD Millipore (clone 3A3, Millipore Cat# MAB2617, RRID: AB_10631873), and anti-PD-L1 (clone 1C10, Cat# SC-293425) from Santa Cruz Biotechnology was used for WB analyses (1:500 dilution). PE-conjugated anti-mouse PD-L1 (BD Biosciences Cat# 568085, RRID: AB_2916834), and the PE-mouse IgG1 K Isotype control (BD Biosciences Cat# 555749, RRID:AB_396091) were used for flow cytometry analyses. Anti-mouse PD-L1555 (clone B7-H1, Bio X Cell Cat# BE0101, RRID: AB_10949073) and it’s corresponding IgG2b Isotype (Bio X Cell Cat# BE0086, RRID: AB_1107791) were used for injection of tumor-bearing animals (50 μg/mouse). Goat horseradish peroxidase-conjugated anti-mouse IgG and goat horseradish peroxidase-conjugated anti-rabbit IgG were used for western blot (Bio-Rad). Gel electrophoresis reagents for protein and DNA were from Bio-Rad.

Human subjects:

Human tumor specimens used in this study are fully described in our previously published studies (36, 37). Briefly, 45 TNBC tumors were identified from the MetroHealth breast cancer tumor biobank of formalin fixed paraffin embedded (FFPE) tumors, that were confirmed for tumor content and disease pathology by a breast pathologist. Total RNA was isolated from the tumor sections using the Qiagen AllPrep FFPE DNA/RNA kit in accordance with manufacturer’s protocols. After RNA extraction, RNA purity was assessed using the 260/280 ratio, as well as the RNA integrity number (RIN), which all showed a RIN above 7 and were subsequently processed for qt-RT-PCR as previously described (35).

Complete Blood Count Analysis:

Peripheral blood was collected by submandibular bleeding into Microtainer EDTA tubes (Becton-Dickson, Cat #365974, RRID: SCR_008418). Blood counts were analyzed using a Hemavet 950 FS hematology analyzer (Drew Scientific, RRID:SCR_020016).

Cell Collection:

Bone marrow cells were obtained by isolating the tibia and femur from mice, then flushing the bones with PBS (phosphate-buffered saline) + 2% FBS (fetal bovine serum) (FACS buffer) to collect the cells. Splenocytes were collected by removing the spleen, followed by mincing and passing through a 70 μm cell strainer to create a single-cell suspension.

Flow Cytometry:

Cells were resuspended in FACS buffer and stained with fluorescently labeled antibodies against CD45R/B220 (RA3–6B2, RRID:AB_312993), CD11b (M1/70, RRID:AB_312795), CD3e (17A2, RRID:AB_312661), CD4 (RM4–4, RRID:AB_11149680), CD8 (53–6.7, RRID:AB_893423) Ly-6G (1A8-Ly6G, RRID: AB_2802355) and Ly6C (HK1.4, RRID:AB_10640120), TER-119 (TER-119, RRID:AB_394986), Ly-6A/E (D7, RRID: AB_465333), CD117 (2B8, RRID: AB_313221), CD150 (TC15–12F12.2, RRID: AB_439797), and CD48 (HM48–1, RRID: AB_2075051) for 15 minutes. The cells were then fixed in 1% paraformaldehyde (PFA) for 15 minutes, washed with FACS buffer, and resuspended in 200 μL of FACS buffer. Data acquisition was performed on an LSRII flow cytometer (BD Biosciences, RRID: SCR_002159), and data analysis was conducted using FlowJo software (TreeStar, RRID: SCR_008520).

Single Cell RNA Sequencing

10× 3’ GEX Library Preparation Kit:

Cells were resuspended in PBS+ 0.04%BSA after tissue dissociation. The quantity and quality of the cells were accessed by Acridine Orange and Propidium iodide dye on a Cellometer Auto 2000 (Nexcelom, RRID:SCR_018656). 16000 cells with the viability higher than 80% were loaded on a 10x Chromium Controller using 10X Genomics 3’ GEX v3.1 Platform (10x Genomics). After cell partitioning and GEM generation, reverse transcription was performed, and cDNA was pooled and cleaned up by beads. cDNA was further amplified for 12 cycles and cleaned up by SPRI beads (Beckman). The quality of cDNA was assessed by High Sensitivity D5000 Tapestation (Agilent Technologies Inc., California, USA, RRID:SCR_014994) and quantified by Qubit 2.0 DNA HS assay (ThermoFisher, Massachusetts, USA, RRID:SCR_020553). The library was then constructed according to the 10X Genomics 3’ GEX v3.1 manual (10x Genomics).

Single Cell RNA Sequencing Analysis:

Single cells were captured and barcoded using the Chromium platform (10X Genomics). Libraries were prepped and sequenced on the Illumina HiSeq platform. Sequence quality was assessed using FastQC and low-quality reads and adapter sequences were trimmed using TrimGalore! (Babraham Institute). Reads were aligned to the mouse reference genome (mm10) using Cell Ranger v6.1.0 and visualized using Loupe Browser (10X Genomics). Differential expression was performed using DESeq2 with a significance cutoff of q<0.05 after multiple testing correction using Benjamini Hochberg. Significant genes were analyzed for gene set enrichment using the Molecular Signature Database (MSigDB) from the Broad Institute.

Animal work

BALB/c mice were purchased from Jackson and used for tumor growth studies as described in our published studies(36, 38–40). Parental 4T1 cells and their K2-KO derivatives, 5×10⁵ cells, were injected into mammary fat pads. For the scRNA-seq study, animals were sacrificed 2 weeks after tumor cell injections and tumors were excised and processed as described above. For the multiplex immunohistochemistry study, the experiment was conducted in collaboration with Akoya Biosciences who performed the tumor immunostaining using the Spatial Tissue Exploration Program (STEP) and the mouse multiplex panel (Supplementary Table 2). Tumors were excised 10 days after tumor cell injections, placed in the preservation buffer provided by Akoya and shipped to Akoya for further processing. For the therapeutic effect of ant-PD-L1 antibody, seven days after the injection of control 4T1 cells, mice were randomized into two groups of 5 mice each. Group 1 was treated with 3 injections (one injection every seven days) of anti-PD-L1 antibody, while group 2 received 3 injections of the control IgG. Tumor growth was monitored over a 30-day period, at the end of which, mice were sacrificed, and tumors or metastases were analyzed (36, 38–41).

Statistical analyses

Statistical analyses were performed using GraphPad Prism (version 8.0, RRID:SCR_002798) and SPSS (version 21.0, RRID:SCR_002865). All experiments were conducted in triplicate, and variables were expressed as mean ± SD. Student’s t-test was used, and significance was considered at p < 0.05.

Data and Materials Availability

The data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE288744.

Results

Loss of Kindlin-2 expression in TNBC tumors induces a systemic host anti-tumor immune response.

Our previously published studies (35) showed that Kindlin-2 that is expressed in TNBC tumors played a significant role in the regulation of the EGF-CSF1 paracrine and autocrine signaling loop to modulate macrophage polarization and infiltration into the tumors via a systemic route (35). To explore the potential systemic effect of TNBC tumors we, first sought to determine whether inhibiting Kindlin-2 expression in TNBC tumors triggers an anti-tumor response by the immunocompetent host mice, as a result of the immune landscape remodeling caused by loss of Kindlin-2. Our previously published studies have shown that tumors derived from TNBC cell lines that are deficient in Kindlin-2 expression grew much smaller and metastasized less than their Kindlin-2-expressing counterparts both in immune-deficient and in immune-competent mouse models (35, 40, 42). For validation purposes, we used three immune-competent syngeneic mouse models for TNBC tumors. Control 4T1, D2A1, or EO771 or their Kindlin-2-KO derivatives were injected in the mammary fat pads of their respective syngeneic mice (BALB/c mice for 4T1 and D2A1 cells, and Blk6 mice for EO771 cells). Tumor growth was then assessed for 25 to 30 days. Loss of expression of Kindlin-2 in either 4T1, D2A1 or EO771 cells resulted in a significant inhibition of tumor growth (Figure 1A, 1B and 1C, respectively), therefore validating our previous findings, and suggesting that loss of Kindlin-2 expression in the cancer cells may have activated a molecular pathway that allowed the host mice to mount an immune response against the Kindlin-2-deficient tumors.

Figure 1.

Kindlin-2-KO in 4T1 (A), D2A2 (B) or EO771 (C) cells inhibits tumor growth in their respective syngeneic immunocompetent mice. (D-G) Immune priming experiment. (D-E) Immune competent BALB/c mice primed with Kindlin-2-KO 4T1 cells (D) developed significantly smaller tumors when challenged with control 4T1 cells (E). (F-G) Conversely, mice primed with control 4T1(F) developed significantly larger tumors when challenged with Kindlin-2-KO 4T1 cells, compared to non-primed (PBS) mice (G). **,p<0.01; ***,p<0.001; Student-T test.

To confirm this supposition, we devised an immune priming experiment, where immune-competent mice were injected with either Kindlin-2-deficient 4T1 cells or PBS in the left side mammary fat pad. Five days later, when an innate immune response was triggered, the Kindlin-2-expressing 4T1 cells were injected in the contralateral right-side mammary fat pad (Figure 1D), and tumor growth was assessed for the following 2–3 weeks (Figure 1E). We found that priming mice with Kindlin-2-deficient 4T1 cells significantly inhibited the growth of Kindlin-2-expressing 4T1 tumor that were injected in the contralateral right-side mammary gland, as compared to the tumors grown in the mice primed with PBS (Figure 1E). The opposite effect was observed when mice were primed with Kindlin-2-expressing tumors cells (Figure 1F), where mice primed Kindlin-2-expressing 4T1 cells allowed the growth of Kindlin-2-deficient tumors that were significantly larger than mice primed with PBS (Figure 1G). Thus, these findings suggest that loss of Kindlin-2 may sensitize TNBC tumors to the anti-tumor immune response, via a systemic route.

TNBC induces alterations in hematopoietic output over time.

To explore the systemic effect of Kindlin-2 on hematopoietic output, we first analyzed blood and BM samples from naïve and 4T1 tumor-bearing mice at two- and five-weeks post-cell injection. Complete blood count analysis revealed distinct changes in peripheral blood cell populations over time (Figure 2A). At two weeks post-tumor cell injection, there was a significant increase in circulating neutrophils. Additionally, lymphocyte counts were significantly reduced at two weeks, suggesting an early immunosuppressive effect of the tumor, however the trend shifted towards increased lymphocytes by week five (Figure 2A, upper panels). Also, by five weeks, both neutrophil and monocyte levels were markedly elevated, indicating a progressive shift towards myelopoiesis as the tumor burden increased (Figure 2A, lower panels). These alterations in peripheral blood reflect the dynamic influence of TNBC tumors on hematopoietic output, promoting myeloid expansion as the disease progresses. Further, immunophenotypic analysis of BM and splenic preparations corroborated these findings, showing a notable expansion in myeloid populations, particularly in the spleen, which serves as an extramedullary hematopoietic site under stress conditions like tumor burden (Figure 2B). We additionally observed significant changes in B and T cells at the week five timepoint including increased BM CD4⁺ cells, decreased BM CD8⁺ cells, decreased spleen B cells, decreased total spleen T cells, and decreased spleen CD8⁺ cells. These findings suggest an evolving immunosuppressive environment as the tumor progresses (Supplementary Figure S1). BM populations enriched for short-term Lin(–) Sca1+/cKit+ (LSK) progenitors and long-term hematopoietic stem cells (LT-HSCs) and cells showed increased frequencies in the bone marrow and spleen, which may reflect the shift towards myelopoiesis (Figure 2C). These findings align with previous studies indicating that solid tumors can reprogram hematopoiesis to favor myeloid cell expansion (19, 23), thereby facilitating immune evasion and tumor progression. These findings were validated in a second model, the D2A1 cells (Supplementary Figure S2).

Figure 2:

TNBC tumors induce alterations in hematopoietic output over time. (A) Complete blood count analysis of circulating cells at steady-state in naïve, or 4T1 injected mice, two- and five-weeks following cell infusion. (B) Immunophenotypic analysis of myeloid, B and T cells from bone marrow and splenic preps of the same mice in A. (C) Immunophenotypic analysis of LSK and LT-HSCs per hindlimb and spleen. N=5 BALB/c mice per group. Error bars represent SEM. Students two-tailed T test was used for statistical analysis.

Kindlin-2 knockout in TNBC tumors reduces myeloid skew and alters T cell populations in tumor-bearing mice.

To investigate the role of Kindlin-2 in TNBC-induced hematopoietic remodeling, we compared control and Kindlin-2 knockout (KO) 4T1 tumor-bearing mice at two- and five-weeks post cancer cell injection. Complete blood count analysis demonstrated distinct changes in circulating cell populations at both timepoints. At two weeks post-injection, Kindlin-2 KO tumor-bearing mice exhibited a significant reduction in circulating neutrophils with an insignificant trend towards decreased monocytes. By five weeks, both neutrophil and monocyte levels were significantly reduced in mice bearing Kindlin-2 KO tumors, implicating Kindlin-2 in promoting myelopoiesis under tumor conditions. Additionally, lymphocyte levels showed a significant decrease at five weeks, indicating a broader impact on the immune cell landscape (Figure 3A). Immunophenotypic analysis revealed that Kindlin-2 KO significantly mitigated the expansion of myeloid populations in both the BM and spleen. At two weeks, there was a significant reduction in BM myeloid cells and spleen myeloid cells, coupled with a trend toward increased BM CD8⁺ T cells and a significant increase in spleen CD8⁺ T cells. By five weeks, we continued to see reduced myeloid cells in both organs, reaching significance in the spleen. These findings suggest that the absence of Kindlin-2 limits myeloid expansion, contributing to a more balanced hematopoietic environment (Figure 3B). Additionally, there were trends towards altered B and T cell populations at the five-week timepoint, indicating a broader impact on the lymphoid landscape (Supplementary Figure S3). Analysis of hematopoietic stem and progenitor cells (HSPCs) further underscore the role of Kindlin-2 in hematopoiesis. At two weeks, we observed a reduction in both the LSK and LT-HSC populations in the spleen, suggesting reduced emergency hematopoiesis in the spleen. By five weeks, mice with Kindlin-2 KO tumors exhibited a trend towards decreased HSPCs in the bone marrow, while splenic frequencies returned to levels consistent with control tumors at this later timepoint (Figure 3C), which we also validated in the D2A model (Supplementary Figure S2).

Figure 3:

Loss of Kindlin-2 expression in the tumors reduces myeloid skew and alters T cell populations in TNBC mice. (A) Complete blood count analysis of circulating cells at steady-state in Control or Kindlin-2-KO 4T1 injected mice, two- and five-weeks following cell infusion. (B) Immunophenotypic analysis of myeloid, B and T cells from bone marrow and splenic preps of the same mice in A. (C) Immunophenotypic analysis of LSK and LT-HSCs per hindlimb and spleen. N=5 BALB/c mice per group. Error bars represent SEM. Students two-tailed T test was used for statistical analysis

Kindlin-2 knockout modulates the tumor microenvironment by altering myeloid and T cell populations.

To explore how Kindlin-2 knockout impacts the tumor microenvironment, we generated single-cell suspensions from harvested tumors and performed cytometric analysis at two- and five-weeks post-tumor injection (Figure 4A). At two weeks, there was a significant decrease in myeloid cells within the tumors of Kindlin-2 KO mice compared to controls, coupled with a trend towards increased pan T cell populations (Figure 4B). By five weeks, this pattern shifted, showing a trend towards increased myeloid cells within the tumors. Additionally, we observed significant increases in pan T cells and CD4⁺ T cells at this later time point (Figure 4C). These findings suggest that the loss of Kindlin-2 promotes the infiltration of immune cells into the tumors. The early decrease in tumor-associated myeloid cells, combined with increased T cell presence, indicates an initial immune activation response. By five weeks, the continued increase in T cell populations, particularly CD4⁺ T cells, alongside a trend towards increased myeloid cells, implies a dynamic modulation of the tumor microenvironment by Kindlin-2 loss, potentially enhancing anti-tumor immunity and altering the progression of the disease.

Figure 4:

Loss of Kindlin-2 expression modulates the tumor microenvironment by altering myeloid and T cell populations. (A) Experimental setup for generating single cell suspensions of harvested tumors for subsequent cytometric analysis. (B-C) Quantification of immune cell populations in the tumors at 2 weeks (B) and 5 weeks (C) post-tumor injection. N=3–5 BALB/c mice per group. Error bars represent SEM. Students two-tailed T test was used for statistical analysis.

Kindlin-2 is a major activator of anti-tumor immune evasion of TNBC tumors.

We further confirmed the involvement of Kindlin-2, or lack thereof, in activating the anti-tumor immune response in mice bearing Kindlin-2-deficient tumors. We used scRNA-seq analyses of tumors derived from control and Kindlin-2-deficient tumors (4T1 model) and found Kindlin-2-KO-derived tumors showed an enrichment of signaling pathways involved in both innate and adaptive immune responses, neutrophil degranulation, and cytokine signaling in immune system (Figure 5A and Supplementary Figure S4). Further analysis of the bulk RNA-seq data generated from Control and Kindlin-2-deficient tumors showed a significant enrichment in the K2-KO tumors in pathways associated with natural killer cell mediated cytotoxicity, Th1 and Th2 cell differentiation, and hematopoietic cell linage, all associated with immune response sensitization (Figure 5B). At the same time these analyses conformed the previously validated roles of Kindlin-2 in ECM-receptor interaction (43), and chemokine signaling and chemokine-cytokine receptor interaction (35). Gene Set Enrichment Analysis (GSEA) showed that Kindlin-2 expression levels correlate with enrichment of gene sets that are involved in the anti-tumor immune response (Figure 5C). We also used IHC analyses to show increased CD8⁺ T cells, CD11c cells and CD11b cells infiltration in tumors derived from Kindlin-2-KO 4T1 cells (Figure 5D and 5E), which is consistent with the data described in Figure 4. Tumor infiltration with immune cells was further validated in a cohort of human TNBC tumors where we established a significant negative correlation between TILs and Kindlin-2 expression (Figure 5F).

Figure 5:

(A) Pathway analyses of the scRNA-seq data showed a significant enrichment of genes involved in immune response, immune surveillance and activity in the Kindlin-2-KO tumors. (B) Pathway analyses of the bulck RNA-seq data showed a significant enrichment of genes involved in ECM interaction, chemokine signaling, chemokine-cytokine receptor interaction, hematopoietic remodeling, and immune response in the 4T1 Kindlin-2-KO derived tumors. (C) Gene Set Enrichment Analysis (GSEA) of the scRNA-seq data showed that Kindlin-2 expression levels correlate with enrichment of gene sets that are involved in the anti-tumor immune response. (D) IHC staining of control- and Kindlin-2-KO 4T1-derived tumors in the immunocompetent BALB/c mice using the Akoya STEP mouse antibody panel. The data shows as a significant increase of tumor infiltrating lymphocytes. (E) Quantification of the number of tumor infiltrating CD8, CD11b and CD11c cells. **,p<0.01; Student-T test. (F) Inverse correlation between K2 expression and tumor infiltrating lymphocytes (TILs) in a cohort of human primary TNBC tumors.

Loss of Kindlin-2 inhibits expression of PD-L1 immune checkpoint to sensitize TNBC tumors to the host anti-tumor immune response.

To determine the molecular underpinning of the Kindlin-2-mediated remodeling of the anti-tumor immune response, we further mined our bulk RNA-seq data for immune signatures that are impacted by the loss of Kindlin-2 and identified the immune checkpoint PD-L1 to be significantly underrepresented in tumor derived from the Kindlin-2-deficient 4T1 tumors (Figure 6A and Supplementary Figure S5). Analysis of the Breast Invasive Carcinoma (The Cancer Genome Atlas (TCGA), PanCancer Atlas) dataset (https://www.cancer.gov/ccg/research/genome-sequencing/tcga) further confirmed the positive correlation between Kindlin-2 and PD-L1 expression in human BC tumors (n=1082 tumors, p=2.21^e−14, Pearson:0.23, Supplementary Figure S6). Analysis of the same dataset showed that PD-L1 is preferentially overexpressed in BC tumors of basal subtype (Supplementary Figure S7). We use Western blot analyses to confirm loss of expression of PD-L1 in the Kindlin-2-defiencient 4T1 cells (Figure 6B), as well as flow cytometry analyses to show loss of cell surface expression of PD-L1 in 4T1-deficient cells (Figure 6C). Analysis of our scRNA-seq data also confirmed the significant loss of cell populations that express PD-L1 (Figure 6D and 6E).

Figure 6:

(A) Volcano plot of the differentially expressed genes between control and K2-KO 4T1-cells. The arrow point to CS274 (PD-L1) the expression of which is significantly downregulated in the K2-KO cells. (B-C) Kindlin-2-KO leads to loss of protein (B) and cell surface expression using flow cytometry analyses (C) of PDL-1. (D) UMAP plots of scRNA-seq data from 4T1-Kindlin-2-KO derived tumors confirmed loss of cell populations expressing CD274 (PD-L1) in the Kindlin-2-KO tumors. (E) Quantification of expression of CD274 (PD-L1) in the tumors derived from the parental 4T1 cells and their K2-KO counterparts. (F) Tumor volume quantification. Treatment of tumor-bearing mice with anti-PD-L1 antibody resulted in a significant inhibition of tumor growth to levels comparable to genetic inactivation of Kindlin-2 (K2-KO). 8 mice per group. **p<0,01; ANOVA. (G) Representative histograms of Flow cytometry analyses with anti-PD-L1 antibody for the quantification of PD-L1 cell surface expression levels in the parental 4T1 cells and their PD-L1-KO derivatives. (H) Tumor volume quantification. Genetic loss (KO) of either Kindlin-2 or PD-L1 resulted in a significant inhibition of tumor growth, 5 mice per group (10 tumors per group). **p<0,01, ***p<0.001; ANOVA.

As a therapeutic application of our findings, we assessed the use of anti-PD-L1 for in vivo studies and showed that treatment of tumor-bearing mice with anti-PD-L1 antibody resulted in a significant inhibition of growth of 4T1-derived tumors when compared to mice treated with the control isotype (IgG) (Figure 6F). In fact, tumor growth inhibition with anti-PD-L1 was close to that achieved with Kindlin-2-KO in 4T1 cells. Finally, we generated 4T1 cells that are deficient in PD-L1 (PD-L1-KO) and confirmed loss of cell surface expression of PD-L1 in these cell derivatives (Figure 6G). When these cells were injected in the mammary fat pads of syngeneic BALB/c female mice, the resultant tumors were significantly much smaller than those that resulted from the parental 4T1 cells or their Kindlin-2-deficient derivatives (Figure 6H). Thus confirming the role of the Kindlin-2-mediated regulation of PD-L1 in the modulation of the anti-tumor immune response.

Discussion

TNBC poses a significant clinical challenge due to its lack of hormone receptors and HER2 expression, rendering traditional targeted therapies ineffective. Despite recent advancements in immunotherapy, TNBC often exhibits a low response rate to immune checkpoint blockade therapy. This resistance is attributed to complex immune evasion mechanisms within TNBC tumors, including a heterogeneous tumor microenvironment and low levels of tumor-infiltrating lymphocytes, highlighting the urgent need for novel therapeutic strategies to improve treatment outcomes in this aggressive breast cancer subtype. In this study we investigated the role of Kindlin-2 in the regulation of hematopoiesis and its subsequent impact on anti-tumor immune evasion in TNBC. Our findings demonstrate that Kindlin-2 plays a critical role in remodeling the hematopoietic landscape, thereby promoting a tumor microenvironment conducive to immune evasion and tumor progression. Our data indicate that Kindlin-2 expression in TNBC tumors leads to significant alterations in hematopoietic output, particularly skewing differentiation towards myelopoiesis both in the bone marrow and spleen. This myeloid bias is evidenced by increased neutrophil and monocyte levels in the peripheral blood and spleen of tumor-bearing mice. Such changes align with previous studies suggesting that tumors can reprogram hematopoiesis to support myeloid cell expansion, thereby facilitating immune evasion and tumor growth.

Interestingly, genetic ablation of Kindlin-2 in TNBC tumors resulted in a marked reduction in myeloid cell populations and an increase in T cell infiltration within the tumor microenvironment. This shift suggests that Kindlin-2 loss disrupts the tumor’s ability to create an immunosuppressive milieu, thereby enhancing the host’s anti-tumor immune response. The observed increase in CD8⁺ T cells and activation of immune checkpoint molecules further supports the role of Kindlin-2 in modulating immune escape mechanisms.

Neutrophil traps, or NETs, play a critical role in tumor progression and metastasis by promoting inflammation and immune evasion (44). These structures, composed of DNA fibers and histones released by neutrophils, create a pro-tumorigenic microenvironment, aiding in tumor cell survival, migration, and the establishment of metastatic colonies in distant organs (45–48). Our data show that loss of Kindlin-2 in the tumors resulted in a significant decrease in tumor neutrophil infiltration, which may have led to inhibition of NET formation, and therefore, inhibition of tumor growth.

The presence of tumor infiltrating lymphocytes serves as a crucial indicator of treatment efficacy in various cancers (24, 26–28). Higher TIL levels often correlate with improved responses to therapy, reflecting enhanced anti-tumor immunity (28). This correlation underscores TILs’ potential as biomarkers for predicting and monitoring therapeutic outcomes in cancer patients (27). Our data show a significant increase in TILs in the tumors derived from the Kindlin-2-deficient 4T1 cells, clearly supporting the role Kindlin-2 in the regulation of the lymphocyte infiltration process and therefore, dictating response to therapy. With respect to the myeloid-derived suppressor cells (MDSCs). These cells, which are abundant in the tumor microenvironment, are potent immune regulators. They promote immune suppression through various mechanisms, including inhibiting T cell function and promoting regulatory T cell expansion. MDSCs play a critical role in tumor immune evasion and are implicated in limiting the efficacy of immunotherapy in cancer treatment (49). Here again, our results show that loss of Kindlin-2 inhibits tumor enrichment in MDSCs, therefore mitigating their pro-tumorigenic affect, resulting in reduced tumor growth. Together, our data point to Kindlin-2 as a key player in the hematopoiesis remodeling, favoring and an immune-evasive microenvironment by promoting a decrease in TILs, which is concomitant to an enrichment in pro-tumorigenic neutrophils and MSDCs in the tumors.

Our findings also highlight the therapeutic potential of targeting Kindlin-2 in TNBC. The reduction in PD-L1 expression upon Kindlin-2 knockout identifies a novel mechanism by which Kindlin-2 promotes immune evasion. The effectiveness of anti-PD-L1 therapy in Kindlin-2-deficient tumors further suggests that combining Kindlin-2 inhibition with immune checkpoint blockade could be a promising strategy for TNBC treatment. This combined approach may enhance the overall anti-tumor immune response, potentially overcoming the inherent resistance of TNBC to current immunotherapies. Moreover, the use of syngeneic mouse models to validate our findings ensures the relevance of our results to the immune-competent context. The significant inhibition of tumor growth in Kindlin-2-deficient TNBC models indicates that Kindlin-2 is a critical regulator of tumor-immune interactions and a viable target for therapeutic intervention.

Published studies (50, 51) have shown that Kindlin-2 expression is associated inflammatory cytokine production, which could be responsible for expansion in the myeloid compartment and recruitment of immunosuppressive cells. In fact, GSEA analysis of RNA-seq data from our group (52) showed that Kindlin-2 expression levels correlate with gene signatures involved in chemokine signaling, cytokine receptor signaling, immune cytokine signaling, and immune system reactome. The same data also showed that K2-deficiency elicits significant downregulation of several immunomodulatory cytokines and chemokines in TNBCs. Kindlin-2 deficiency elicits significant downregulation of several immunomodulatory cytokines and chemokines in TNBCs (Supplementary Figure S8). These data provide further information supporting the fact that Kindlin-2 expression is associated inflammatory cytokine production, which could be responsible for expansion in the myeloid compartment and recruitment of immunosuppressive cells.

The molecular mechanisms whereby Kindlin-2 regulate hematopoiesis and PD-L1 expression still need to be thoroughly investigated. However, based on our previously published studies (35), we can speculate that Kindlin-2 expressed in the tumor cells may be exerting a systemic effect that leads to the remodeling of hematopoiesis, therefore leading to increased tumor infiltration by type 2 polarized oncogenic macrophages, neutrophils and MDSCs, concomitant to decreased infiltration of immunogenic anti-tumor lymphocytes. This is exciting and plausible hypothesis is supported by our observation that small extracellular vesicles (sEVs) secreted by TNBC cell lines are enriched in Kindlin-2 and other oncogenic molecules, such as integrins, cytokines and chemokines; loss of expression of Kindlin-2 in these TNBC cells, and therefore, in the derived sEVs, results in a significant inhibition of the oncogenic behavior of recipient cancer cells incubated with Kindlin-2-deficient sEVs. We are actively pursuing this hypothesis in ongoing studies. As for the Kindlin-2-mediated regulation of PD-L1 expression, we also believe that this takes place at the transcription level, since our published studies (52) have shown Kindlin-2 to be translocated to the nucleus where it modulates the p53-mediated regulation of expression of cancer senescence genes.

In summary, our study establishes Kindlin-2 as a pivotal factor in the regulation of hematopoiesis and anti-tumor immune evasion in TNBC. The data support the potential of Kindlin-2 as a therapeutic target, particularly in combination with immune checkpoint inhibitors. While these findings are significant in the context of breast cancer, where reactivating the immune system remains a critical challenge, the implications of Kindlin-2-mediated immune modulation extend beyond TNBC, suggesting potential therapeutic strategies for other solid tumors characterized by immune evasion. Future research should focus on elucidating the precise molecular mechanisms underlying Kindlin-2’s role in immune modulation and exploring the clinical applicability of Kindlin-2-targeted therapies across a broader spectrum of solid tumors.

Implications:

Kindlin-2 regulates tumor immune evasion through the systemic modulation of hematopoiesis and PD-L1 expression, which warrants therapeutic targeting of Kindlin-2 in TNBC patients.

Data Availability Statement

All data are contained within the article. Requests for reagents should be addressed to K Sossey-Alaoui (kxs586@case.edu).