Ectopic integrin β1 overproduction by squamous carcinoma cells contributes to immune evasion-associated metastasis

Abstract

Distant metastasis often predicts poor survival for squamous cell carcinoma (SCC) patients. To identify potential targets for metastatic SCCs, we performed spatial profiling of SCC specimens from head and neck cancer patients to identify stage-associated markers. Laminin-binding integrins, normally restricted to basal keratinocytes, were enriched in Stage IV-A intratumoral regions vs Stage II/III SCCs, consistent with poor survival correlations of these integrins in TCGA. Due to the association of multiple laminin-binding integrins with poor prognosis, we assessed their collective impact on metastasis by knocking down integrin β1, a common partner of α-integrins, in mouse SCC cells derived from keratin 15 (K15)⁺ stem cells harboring a Kras^G12D mutation and Smad4 deletion (Smad4^−/−). SCC cell lines derived from primary tumors with or without spontaneous lung metastasis were established. Metastatic SCC cells had higher levels of laminin-binding integrins than non-metastatic SCC cells derived from this model. Knockdown of the integrin β1 gene (shITGB1) in metastatic SCC cells ablated lung metastases following implantation into immunocompetent but not immunocompromised hosts. RNA sequencing of control/shITGB1 murine SCC cells identified that integrin β1 was linked to pathways related to leukocyte recruitment. At the protein level, inflammatory cytokine arrays and immunostaining of metastatic murine tumors revealed that metastatic SCCs exhibited elevated myeloid chemotactic proteins, granulocyte infiltration, and low CD8 T cell presence. shITGB1 reversed these patterns. Conversely, CD8α depletion enabled shITGB1 SCC cells to reestablish lung metastasis in immunocompetent hosts. Our results reveal an unappreciated role of SCC cell-produced integrin β1 in driving immune suppression-associated metastasis.

1. Introduction

Squamous cell carcinomas (SCCs) arise from stratified epithelial tissues such as the skin and head and neck [1]. Distant metastasis, most frequently spreading to the lungs, often predicts therapeutic resistance and poor survival [2, 3]. Cancer cell distant metastasis is a multistep process of dissemination that includes intravasation into vessels, survival in circulation, extravasation from vessels into distant tissue, colonization of the distant site, and finally expansion [4]. These sequential steps must be carried out by cancer cells while evading immune cell-mediated killing.

The tumor microenvironment (TME) of solid tumors is a complex network of pro- and anti-tumor cell types, including infiltrating immune cells, that significantly contributes to or eliminates metastatic disease [5, 6]. The TME of metastatic cancers has been shown to be enriched in immunosuppressive myeloid cells, such as granulocytes, and depleted in cytotoxic CD8 T cells, the primary effectors of the targeted anti-tumor response [7, 8]. Remodeling the immune microenvironment by immune checkpoint inhibition (ICI), principally targeting the programmed cell death protein 1/programmed death-ligand 1 (PD-1/PD-L1) interactions, has been a lifesaving treatment for some patients with metastasis. However, only ~20% of head and neck SCCs [9] and ~15–30% of non-resectable/metastatic skin SCCs achieve durable or complete response to ICI [10, 11]. Thus, there remains a pressing need to identify targetable mechanisms of tumor cell immune evasion and SCC metastasis.

Integrins are transmembrane heterodimer receptors through which cells interact with the stroma. Integrin receptors consist of α- and β-subunits that are categorized as collagen-, laminin-, Arg-Gly-Asp (RGD)- and leukocyte-specific receptors [12]. Laminin-binding integrins, including α3, α6, α7, and β1 subunits, are normally restricted to the basal layer of the stratified epithelium and can be upregulated following wounding to support re-epithelialization [13, 14]. Overproduction of laminin-binding integrins has been reported in SCCs, but reported functions of individual integrin subunits vary significantly, potentially due to organ- or stage-specific effects [12]. Several hallmark cancer activities, including invasion and resistance to cell death, have been attributed to tumor cell-produced integrins [15]. Yet, attempts to target integrins overproduced by tumor cells have broadly failed in the clinic, indicating that crucial knowledge gaps in integrin functions in cancer remain. Although immune cell-produced integrins are known to impact the efficacy of the anti-tumor immune response [16], it is unknown whether SCC cell-produced integrins directly contribute to tumor cell immune evasion during metastasis.

A challenge for preclinical studies into biologic mechanisms that drive metastatic SCC is the shortage of immunocompetent murine models that develop spontaneous distant metastasis and accurately represent the metastatic cascade. Previously, we found that induction of a Kras^G12D mutation and Smad4 deletion (Kras^G12DSmad4^−/−) in keratin 15 expressing (K15⁺) epithelial stem cells generates spontaneous SCCs in mice [17]. We used this approach to produce SCCs in C57BL/6 mice, and derived K15.Kras^G12D.Smad4^–/– SCC cell lines from those tumors. Despite the same oncogenic driver mutations, when these lines are implanted into syngeneic, immunocompetent mice, SCCs metastasize to the lungs with variable frequencies [18]. These K15.Kras^G12D.Smad4^–/– SCC lines provide a novel model system to study the properties of metastasis-initiating cells (MICs) that are thought to be a subpopulation of cancer stem cells (CSCs) with unique characteristics to survive throughout the metastatic cascade [19]. Importantly, since these metastases develop spontaneously with a natural immune system, this model allows studying immune evasion mechanisms that are necessary for metastasis.

In this study we sought to determine how integrin overproduction by SCC cells affects MIC properties and immune evasion during metastasis. We uncovered that SCC cell-produced integrin β1 was necessary to regulate cell-autonomous functions associated with MICs and remodel the immune microenvironment to achieve SCC metastasis, implying both functional aspects need to be targeted for effectively treating SCC metastasis in the future.

2. Materials and Methods

2.1. Human Samples

Human cancer specimens were collected and archived at the UC Davis Medical Center via an IRB approved protocol (UC Davis IRB #1841486). Non- human papillomavirus (HPV)-associated SCC patient samples were identified from internal head and neck SCC databases including clinical data, cancer stage, pathologic findings, treatment, and survival as detailed in Supplementary Table 1. Primary tumor and surrounding TME were removed from formalin-fixed paraffin embedded (FPPE) blocks of head and neck SCC specimens with a scalpel and embedded into a new paraffin block to build tumor microarrays (TMAs) and sectioned for spatial transcriptomic analysis as performed by the UC Davis Center for Genomic Pathology Laboratory (CGPL).

2.2. Nanostring GeoMx^® Digital Spatial Profiling

A semi-automated workflow was employed using the GeoMx^® Human Whole Transcriptome Atlas assay and performed by the UC Davis Genomics Shared Resource (GSR). After baking, slides underwent deparaffinization and antigen retrieval. Following in situ hybridization and blocking, the GeoMx^® Solid Tumor TME Morphology Kit was applied, using pan-cytokeratin (PanCK), CD45, and SYTO 13 as morphology markers. Slides were washed and loaded onto the GeoMx^® Digital Spatial Profiler (DSP) instrument for high-resolution, 20X fluorescent imaging. H&E images and morphology staining of a parallel serial section were used to guide the selection of regions of interest (ROIs). Each ROI was exposed to UV light, and library preparation was performed using GeoMx Seq Code Illumina-compatible primers. Validated libraries were sequenced on an Illumina NovaSeq X Plus Sequencing System. Further details of the GeoMx DSP workflow can be found in the Supplementary Materials and Methods.

2.3. Spatial Transcriptomic Data Analysis

FASTQ files generated by Illumina sequencing were processed using GeoMx^® NGS Pipeline Software for further downstream bioinformatic analysis. After quality control, filtered data were Q3 normalized. A summary schematic of the GeoMx DSP workflow is presented in Supplementary Fig. 1. A full description of data analysis performed can be found in the Supplementary Materials and Methods.

2.4. Analysis of Publicly Available Datasets

Data used to assess 5-year patient survival or associations between mRNA expression for genes of interest was retrieved from the Head and Neck Squamous Cell Carcinoma (HNSCC) Firehose Legacy dataset of The Cancer Genome Atlas (TCGA) [20]. Data for gene correlation analyses was accessed via cBioPortal [21–23] and through the Human Protein Atlas [24] (proteinatlas.org) for survival analyses. Cases across all stages were included for analysis, but samples were screened to only include patients confirmed negative for HPV infection. For survival comparisons, patients were stratified by quartile gene expression, with the highest (“High”) and lowest (“Low”) quartiles used for analysis. ITGB1 expression in non-pathologic human skin epidermis was originally produced by Solé-Boldo and colleagues [25] and retrieved from the Human Protein Atlas [24] (proteinatlas.org).

2.5. Cell Lines

Murine SCC cell lines A223, P029, and H1977m were derived from K15.Kras^G12D. Smad4^−/− mice backcrossed to C57BL/6J strain as previously described [17, 26]. Murine SCC cells were cultured in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12; Gibco, 11320082) containing 10% fetal bovine serum (FBS; ThermoFisher Scientific, A5670401) and 1% Primocin (InvivoGen, ant-pm-05). Cells were cultured at 37°C with 5% CO₂. Cell lines tested negative regularly for mycoplasma (Lonza, LT07–318). To generate shRNA knockdown lines, Sigma Mission shRNA vectors as detailed in Supplementary Table 2 were obtained from the University of Colorado Cancer Center Functional Genomics Shared Resource. Cells were transduced with lentivirus containing shRNA and stable knockdowns selected for by puromycin (Gibco, A1113803) at concentrations of 7 μg/mL for H1977m and 14 μg/mL for P029.

2.6. Immunoblotting

Cultured cell monolayers and tumor fragments were lysed with RIPA (Cell Signaling Technology, 9806) or Cell Lysis Buffer (Cell Signaling Technology, 9803) supplemented with protease/phosphatase inhibitors. Samples were prepared, run, and assessed according to standard procedures. A detailed method is in the Supplementary Materials and Methods; antibody details are provided in Supplementary Table 3.

2.7. Murine Models of the Metastatic Cascade

Mice were maintained in specific pathogen-free facilities at the University of Colorado, Anschutz or the University of California, Davis. Animals were handled with ethics guidelines and conditions according to protocols approved by each site’s Institutional Animal Care and Use Committee.

To establish subcutaneous tumors, male and female C57BL/6J mice (Jackson Laborites) and athymic nude mice (Charles River) aged 8–10 weeks old were injected subcutaneously (s.c.) with 1×10⁶ or 1×10⁵ H1977m (male) and P029 (female) SCC cells, respectively, suspended in 100 μl of 50% Matrigel (Corning, 356234)/50% PBS into the flanks of sex-matched hosts as performed previously [18]. Tumor volumes were measured with digital calipers and calculated as (length × width²)/2. Mice were euthanized at designated time points or when they reached humane endpoint.

As done previously [27], for SCC tail vein injections, 1×10⁵ SCC cells suspended in 100 μL of PBS were injected via the lateral tail vein of mice. Direct lung implants were performed as previously described [27]. Briefly, following incision of the skin and removal of subcutaneous fat tissue, 2,000 SCC cells suspended in a 40 μL solution of 15% Matrigel/85% PBS were injected into the exposed left lung. The skin incision was closed with vetbond tissue adhesive (Fisher, 50-282-8724) or Autoclip 9 mm staples (Braintree Scientific, ACS- KIT). Mice were administered Buprenorphine SR (Wedgewood) 1 mg/kg s.c. prior to surgery and 48 hours after surgery.

2.8. RNA Sequencing and Bioinformatics Analysis

RNA was extracted from cultured cell monolayers using the RNeasy Plus Mini kit (QIAGEN, 74134) following manufacturer protocol. After assessing RNA quantity and purity, samples were submitted to Novogene (Sacramento, USA) for library preparation, sequencing and normalization. Data was evaluated by gene set enrichment analysis (GSEA) and subsequent leading edge analysis against the Hallmark Gene Sets and Gene Ontology (GO) Biological Processes Gene Sets using GSEA 4.0.3 software. Additional details for RNA sequencing are reported in the Supplementary Materials and Methods.

2.9. Histology and Pathology

After tissue collection and fixation (24 hours), tissue was processed and sectioned and stained with H&E at the University of Colorado Pathology Shared Resource or the UC Davis CGPL. For immunohistochemistry (IHC), five fields of view per sample were captured at a 10x objective magnification. IHC staining was quantified in using QuPath software [28]. Data is presented as percent of positively stained cells over total number of cells in the field of view. A detailed description of immunostaining is reported in the Supplementary Materials and Methods. Antibodies utilized for IHC can be found in Supplementary Table 4.

2.10. Cell Expansion Assays

As performed previously [29], 2,000 cells/well were seeded in duplicate 96 well plates. After 12 or 72 hours, cells were fixed and assessed by SRB Assay. Absorbance was determined on a BioTek Synergy 2 microplate reading spectrophotometer (Aligent) at 530 nm. Relative cell densities were calculated by normalizing the absorbance of each well for plates at 72 hours to corresponding wells from plates harvested at 12 hours post-seeding.

2.11. Invasion Assays

Invasion Assays were performed as previously described [17]. Briefly, transwell Matrigel-coated invasion chambers with 8 μm pores (Corning, 3464) were prepared according to manufacturer’s instructions. 20,000 SCC cells were seeded in the upper chamber in serum free DMEM/F12 with 1% Primocin and 0.5 μg/mL Mitomycin C (Cayman Chemical, 11435). DMEM/F12 containing 10% FBS, 1% Primocin, and 0.5 μg/mL Mitomycin C was added in the lower well. After 48 hours, cells remaining in the upper chamber were removed and invaded cells below the membrane were fixed/stained. Excess stain was removed and whole well images were captured on an Olympus Stereo Microscope and analyzed using ImageJ (FIJI) [30].

2.12. Inflammatory Cytokine/Cytokine Array

For cytokine/chemokine evaluation, Proteome Profiler Mouse XL Cytokine Array (R&D Systems, ARY028) was used according to manufacturer’s instructions. In brief, membranes were blocked with blocking buffer at room temperature for 1 hour and incubated with tumor lysates (200 μg protein) at 4°C overnight. The next day, membranes were incubated in Detection Antibody Cocktail, Streptavidin-HRP and Chemi Reagent Mix separated by wash periods. Activity was detected on a ChemiDoc (BioRad) (10-minute exposure). Membrane images were analyzed using QuickSpots software for quantification.

2.13. In Vivo Drug Administration

To evaluate the functional role of immune cells in SCC metastasis, mice were randomized into treatment groups once tumors were established, ~50 mm³, 3 days post implant. Anti-mouse CD8α and isotype control antibodies were dosed at 10 mg/kg twice per week by intraperitoneal (i.p.) injections for the duration of depletion experiments as done previously [31]. As described, anti-mouse Ly6G/Ly6C (anti-Gr1) and isotype control antibodies were dosed at 200 μg/mouse three times per week by i.p. injections for the duration of depletion experiments [27]. Antibody details can be found in Supplementary Table 5.

2.14. Spectral Flow Cytometry

Blood was harvested and samples prepared according to standard protocols. Samples were stained Ghost Dye UV 450 (Tonbo, 13–0868-T100) to detect dead cells, underwent FC block (BD Biosciences, 553142), and surface proteins stained (Supplementary Table 6). Samples were run on a Cytek Aurora Instrument and data analyzed using FCS Express 7. Additional details of spectral flow cytometry methodology are reported in the Supplementary Materials and Methods.

2.15. Statistics

For single comparisons, two-tailed t test and non-parametric alternatives were performed. For gross metastatic incidence outcomes, Fisher’s exact test was performed. Multiple comparisons were made using a one-way ANOVA, two-way ANOVA, or mixed-effects model with Tukey’s post hoc test for mean differences. Survival comparisons were performed with the log-rank Mantel-Cox test. The strength of associations between the expression of genes was made using the Pearson correlation coefficient. Grubbs’ Test was used to determine if statistical outliers were present in data, with α = 0.05, and identified outliers were removed prior to analysis. Unless otherwise stated, error bars represent the mean ± the standard error of the mean (±SEM). P < 0.05 was considered statistically significant. All statistical analyses were performed in GraphPad PRISM.

3. Results

3.1. Ectopic SCC epithelial cell laminin-binding integrin overproduction correlated with disease progression and poor prognosis of SCC patients.

We examined intratumor epithelial cell-enriched regions of interest (ROIs) selected across fifteen primary SCC specimens from HPV negative head and neck SCC patients (Supplementary Table 1). Intratumor epithelial cell-rich ROIs were defined as regions dominated by PanCK staining with low CD45 cell infiltration (Fig. 1A). Following spatial transcriptomic analysis (Supplementary Fig. 1), we performed differential gene expression (DGE) analysis comparing intratumor ROIs from Stage IV-A patients, who often have or are at risk of progressing to distant metastatic disease [1], to Stage II/III patient intratumor ROIs. Gene symbols are italicized throughout, whereas protein names are written in plain text. Stage IV-A intratumor ROIs were enriched in expression of laminin-332 components, laminin α3 (LAMA3), β3 (LAMB3), and γ2 (LAMC2), as well as laminin-binding integrins α3 (ITGA3) and β1 (ITGB1) (Fig. 1B–D).

Figure 1. Ectopic SCC epithelial cell laminin-binding integrin overproduction correlated with disease progression and poor prognosis of SCC patients.

(A) Representative image of Intratumor (pan-cytokeratin dominant) region of interest (ROI) identified by staining for SYTO13 (nucleus, blue), pan-cytokeratin (PanCK, green) and CD45 (red). 21 Intratumor ROIs were selected for GeoMx Digital Spatial Profiling (DSP) from 15 FFPE HNC patient SCC specimens as outlined in Supplementary Table S1. Scale bar 200μm.

(B) Differential gene expression (DGE) analysis of high (Stage IV-A) vs lower (Stages II & III) Intratumor ROIs. Significant (p<0.05) DGEs are colored red.

(C-D) Expression of ITGA3 (integrin α3) (C) and ITGB1 (integrin β1) (D) as determined by GeoMx DSP in Stage II-III (n=4) and Stage IV-A (n=14) Intratumor ROIs.

(E-F) Survival rate of SCC patients with high (red) or low (blue) quartile expression of ITGA3 (E) and ITGB1 (F). Data retrieved from the HNSCC Firehose legacy TCGA patient cohort for non-HPV associated cases.

(G) Representative immunoblot showing relative levels of laminin-binding integrins in non-metastatic A223 and metastatic P029 and H1977m cell lines.

Significance was determined if the p-value was <0.05* and <0.0001**** by an unpaired two-tailed t test (C-D) and the log-rank Mantel-Cox Test for Survival comparisons (E-F). Error bars represent standard error of the mean (±SEM).

To assess if laminin-binding integrins enriched in Stage IV-A intratumors would also correlate with patient outcome, we evaluated the 5-year survival rates of SCC patients in The Cancer Genome Atlas (TCGA) [20]. We compared the highest and lowest quartile expression of laminin-binding integrins and found that high ITGA3 and ITGB1 correlated with worse 5-year survival rates independent of stage (Fig. 1E–F). We next evaluated protein levels of laminin-binding integrins in our K15.Kras^G12D.Smad4^–/– SCC lines derived from C57BL/6 mice. Compared to non-metastatic A223 cells, metastatic P029 and H1977m cells displayed ectopic overproduction of laminin-binding integrins α3 and β1, as well as α6 and α7 that also form heterodimer receptors with integrin β1 [12] (Fig. 1G). However, there was no difference in ITGA6 and ITGA7 expression levels in our 15 patient dataset comparing Stage IV-A to Stage II/III intratumor ROIs (Supplementary Fig. 2A–B). Within TCGA data [20], high ITGA6, but not high ITGA7, correlated with worse survival outcomes (Supplementary Fig. 2C–D).

3.2. Ectopic SCC cell-produced integrin β1 drove metastasis in a lymphocyte-dependent manner.

To effectively evaluate the functional impact of laminin-binding integrins in metastatic SCC cells, where multiple laminin-binding α-integrins were elevated (Fig. 1G), we stably knocked down the mouse gene encoding integrin β1, Itgb1 (shITGB1), in H1977m and P029 SCC cells (Fig. 2A–B), which eliminated its complexes with all laminin-binding α-integrins. In normal human epidermis, laminin-binding integrin β1 production is largely restricted to basal keratinocytes rather than immune or stromal cells [25], and we found that most H1977m and P029 SCC cells ectopically overproduced integrin β1 with strong membrane staining at both primary and metastatic sites (Supplementary Fig. 3A–C).

Figure 2. Ectopic SCC cell-produced integrin β1 drove metastasis in a lymphocyte-dependent manner.

(A-B) Representative immunoblots showing relative levels of integrin β1 in H1977m (A) P029 (B) cells as a function of non-targeting control (shNC) or Itgb1 (shITGB1) shRNA knockdown in H1977m cells cultured in vitro. Results are representative n=3 independent biologic replicates.

(C) Average volumes of H1977m shNC and shITGB1 tumors implanted into the flanks of male C57BL/6 or nude (athymic) mice. For C57BL/6 tumor-bearing mice, n=15 shNC and n=7 shITGB1 tumors. For nude tumor bearing mice, n=10 shNC and n=10 shITGB1 tumors.

(D) Average volumes of P029 shNC and shITGB1 tumors implanted into the flanks of female C57BL/6 or nude mice. For C57BL/6 tumor bearing mice, n=12 shNC and n=13 shITGB1 tumors. For nude tumor bearing mice, n=9 shNC and n=10 shITGB1 tumors.

(E-F) Quantification of gross lung metastases from mice in (C-D).

(G-H) Representative H&E staining of the lungs from shNC and shITGB1 H1977m (G) and P029 (H) tumor bearing C57BL/6 mice from the cohort of mice described in (C-D). Arrowheads indicate examples of metastases.

Significance was determined if the p-value was <0.05*, <0.01**, <0.001***, and <0.0001**** by a mixed-effects model for tumor volume at selected time points (C-D) and Fisher’s exact test (E-F). Error bars represent standard error of the mean (±SEM).

We implanted shRNA non-targeting control knockdown (shNC) and shITGB1 SCC cells into flanks of C57BL/6 mice. Sustained Itgb1 knockdown in SCC cells in vivo (Supplementary Fig. 3C–E) produced a significant decrease in primary tumor volume relative to shNC tumors (Fig. 2C–D). To deduce whether the reduction in primary tumor volume was due to abrogation of SCC cell-autonomous signaling, or if it was a function of increased immunological killing, we also implanted shNC and shITGB1 H1977m and P029 SCC cells into athymic mouse hosts. As with primary SCCs in immunocompetent C57BL/6 hosts, shITGB1 significantly reduced primary tumor burden in nude mice (Fig. 2C–D).

We evaluated the necessity of SCC cell-produced integrin β1 for metastasis to the lungs. shITGB1 significantly reduced lung metastases in immunocompetent hosts, but not athymic hosts that lack mature T cells (Fig. 2E–H). Even when shITGB1 SCCs in immunocompetent hosts grew to tumor volumes equivalent to shNC SCC tumors, shITGB1 still abrogated lung metastasis (Supplementary Fig. 4A–B), with no significant correlation between primary tumor volume and distant metastatic burden for either SCC line (Supplementary Fig. 4C–D). Together, these results indicated that SCC cell-produced integrin β1 contributed to primary tumor growth, but the immune evasion-mediated metastasis by integrin β1 could only be identified in immunocompetent mice.

3.3. SCC cell-produced integrin β1 was necessary for SCC cell invasion and expansion.

We performed RNA sequencing (RNAseq) of shNC and shITGB1 H1977m and P029 SCC cells. GSEA of RNAseq gene expression data revealed that these two SCC lines shared top pathways downregulated by shITGB1, notably Hallmark Epithelial-Mesenchymal Transition (EMT) (Fig. 3A–B). Consistent with Hallmark EMT reduction by Itgb1 knockdown, shITGB1 SCC cells had less invaded cells through Matrigel-coated transwell chambers compared to shNC control SCC cells (Fig. 3C). In addition to EMT, pathways related to morphogenesis, associated with CSC pluripotency, were significantly downregulated in shITGB1 SCC cells (Fig. 3A–B). Because pluripotent CSCs are reported to initiate primary tumors [32], we evaluated if SCC cell-produced integrin β1 was necessary for clonal expansion of tumor cells. shITGB1 reduced SCC cell expansion in vitro compared to shNC (Fig. 3D–E). Consistently, IHC staining for Ki67 in SCCs from C57BL/6 mice revealed that relative to shNC tumors, shITGB1 tumors had significantly fewer proliferating cells (Supplementary Fig. 5).

Figure 3. SCC cell-produced integrin β1 was necessary for SCC cell invasion and expansion.

(A-B) Gene set enrichment analysis (GSEA) of RNAseq data comparing shITGB1 to shNC H1977m (A) and P029 (B) cells cultured in vitro. mRNA was harvested from n=3 biologic replicates/group.

(C) Representative images and percent invasive area reported from shNC and shITGB1 H1977m and P029 cells seeded in a transwell invasion assay. Each point represents one chamber (n=3 chambers per group). Representative results shown are from one of three independent biological replicates.

(D–E) Relative expansion of shNC and shITGB1 H1977m (D) and P029 (E) cells as quantified by SRB optical density (OD). Each point represents one well (n=6 wells per group). Representative results shown are from one of three independent biological replicates.

Significance was determined if the p-value was <0.05*, <0.01**, <0.001***, <0.0001**** by a One-Way ANOVA test with Tukey’s post hoc comparisons (C-E). Error bars represent standard error of the mean (±SEM).

3.4. Ectopic integrin β1 overproduction from SCC cells caused an immunosuppressive TME.

The cell autonomous effects of integrin β1 blocked by shITGB1 could not explain the differences in metastasis observed between nude and immunocompetent C57BL/6 hosts. GSEA of RNAseq data demonstrated that shITGB1 in SCC cells attenuated pathways related to angiogenesis, inflammatory response, and leukocyte migration/chemotaxis, even in the absence of stromal and immune cells (Fig. 3A–B). Consistent with the reduction in Hallmark Angiogenesis (Fig. 3A–B), the vascular area of SCC tumors, marked by CD31, was significantly reduced by shITGB1 (Supplementary Fig. 6). An inflammatory cytokine array of protein lysates from shNC and shITGB1 SCCs revealed that chemokines for granulocyte recruitment, CXCL1 and CXCL5 [33], were downregulated by shITGB1 (Fig. 4A–B). Analysis of SCC gene expression in the TCGA dataset [20] revealed that ITGB1 expression correlated with expression of CXCL5 and the granulocyte marker CD15 (FUT4) (Fig. 4C–D). IHC staining of Ly6G, marking granulocytes, demonstrated that shITGB1 SCCs had significantly fewer intratumoral granulocytes relative to shNC SCCs from C57BL/6 hosts (Fig. 4E). Furthermore, shITGB1 tumor-bearing hosts that lacked distant metastases had decreased granulocyte recruitment to the lungs compared to shNC tumor-bearing mice with distant metastases (Fig. 4F). These data suggested that ectopic SCC epithelial cell-produced integrin β1 promoted angiogenesis and granulocyte recruitment to the TME.

Figure 4. Ectopic integrin β1 overproduction from SCC cells caused an immunosuppressive TME.

(A-B) Representative cytokine array images (A) accompanied by quantification (B) for shNC (n=3) and shITGB1 (n=3) H1977m (left) and P029 (right) tumor lysate samples corresponding to tumors harvested from mice in Fig 2.

(C-D) Pearson Correlations between integrin β1 (ITGB1) and LIX (CXCL5, C) and CD15 (FUT4, D) in the HNSCC Firehose legacy TCGA patient cohort for non-HPV associated cases. n=462 patients.

(E-F) Representative IHC staining and quantification for Ly6G performed on shNC (n=5) and shITGB1 (n=5) H1977m (left) and P029 (right) tumors (E) and lungs of tumor bearing mice (F) from the cohort of mice described in Fig 2. Scale bars 60 μm. Arrowheads indicate cells positive for the specific IHC marker. The average of five images per tissue, each taken at 10X magnification, represents one sample; quantifications were performed with n = 5 samples per group.

(G-H) Quantification of gross lung metastases in sex-matched C57BL/6 mice bearing H1977m (G) and P029 (H) flank tumors and treated with Isotype or anti-GR1 (Ly6G/Ly6C, αGR1) depleting antibody. For mice with H1977m tumors, n=10 Isotype and n=5 αGR1 tumors were quantified (G). For mice with P029 tumors, n=10 Isotype and n=7 αGR1 tumors were quantified (H).

(I-J) Representative IHC staining and quantification for CD8α (I) and Granzyme B (GZMB; J) performed on H1977m (left) and P029 (right) tumors from mice treated with Isotype control (top) and αGR1 (bottom). Scale bars 200 μm. Arrowheads indicate cells positive for the specific IHC marker. For CD8 quantification, n = 10 Isotype and n = 10 αGR1 tumors across two experiments were quantified in P029 and H1977m tumors, respectively. (I). For GZMB IHC, n = 10 Isotype and n = 10 αGR1 tumors were quantified in H1977m tumors across two experiments; n = 10 control and n = 8 αGR1 tumors were quantified in P029 tumors across two experiments (J). The average of five images per sample, each taken at 10X magnification, represents one sample; quantifications were performed with numbers of tumors specified above.

Significance was determined if the p-value was <0.05* and <0.01** by an unpaired two-tailed t test (B, I), a Pearson Correlation (C-D), Fisher’s exact (G-H), and Mann-Whitney test (E-F, J). For correlations, R² value for simple linear regression, r-value and p-value are reported. Error bars represent standard error of the mean (±SEM).

We next assessed if SCC cell-produced integrin β1-induced granulocyte infiltration plays a causal role in producing an immunosuppressive TME in metastatic SCCs. After establishing subcutaneous SCCs in the flanks of C57BL/6 hosts, mice were treated with either isotype or anti-GR1 (Ly6G/Ly6C, αGR1) antibody to deplete granulocytes. Depleting granulocytes (CD45⁺, CD11b⁺, Ly6G⁺ cells; Supplementary Fig. 7) had no effect on primary tumor volume (Supplementary Fig. 8) but significantly reduced lung metastasis lesions in SCC-bearing mice (Fig. 4G–H). Further, granulocyte depletion produced a significant increase in tumor-infiltrated CD8⁺ cells (Fig. 4I) and cytotoxic cell infiltration, identified by positive granzyme B (GZMB) staining (Fig 4J).

3.5. CD8 T cell depletion attenuated the anti-metastasis effect of integrin β1 knockdown in metastatic SCC cells.

Similar to the increased intratumor CD8⁺ cell infiltration observed after granulocyte depletion, shITGB1 SCCs in C57BL/6 hosts also had an increased intratumoral lymphocyte infiltration (Fig. 5A), specifically elevated CD8 T cells (Fig. 5B), and increased GZMB⁺ cells in P029 tumors (Fig. 5C), compared to shNC SCCs. We therefore sought to determine if shITGB1-mediated anti-metastasis effects were CD8 T cell-dependent. After establishing shITGB1 SCCs in C57BL/6 hosts, we treated mice with isotype or anti-CD8α (αCD8) depleting antibody and confirmed decreased levels of CD8⁺ T cells (Supplementary Fig. 9). Compared to isotype treated mice, depletion of CD8 T cells attenuated shITGB1 effects on reducing primary SCC tumor volumes (Fig. 5D–E) and significantly increased metastasis to the lungs of C576BL/6 hosts (Fig. 5F–G). Relevant to human SCC patients, high ITGB1 transcription inversely correlated with CD3E, CD8A, and GZMB mRNA expression in TCGA cases [20] (Fig. 5H–J). These data suggest that SCC cell-produced integrin β1 overproduction is necessary to evade CD8 T cell-mediated destruction for metastasis to the lungs.

Figure 5. CD8 T cell depletion attenuated the anti-metastasis effect of integrin β1 knockdown in metastatic SCC cells.

(A-C) Representative IHC staining and quantification performed on H1977m (left) and P029 (right) shNC (n=5) and shITGB1 (n=5) tumors for CD3ε (A), CD8α (B), and Granzyme B (GZMB; C). Scale bars 60 μm. Arrowheads indicate cells positive for the specific IHC marker. The average of five images per sample, each taken at 10X magnification, represents one sample; quantifications were performed with numbers of tumors specified above.

(D-E) Average volumes of shITGB1 H1977m (D) and P029 (E) tumors implanted into the flanks of sex-matched C57BL/6 mice treated with Isotype or anti-CD8 (αCD8) depleting antibody.

(F-G) Quantification of gross lung metastases from mice in (D-E). n=8/group for shITGB1 H1977m tumor-bearing mice (D, F) and n=10/group for shITGB1 P029 tumor-bearing mice (E, G).

(H-J) Pearson Correlations between integrin β1 (ITGB1) and CD3E (H) CD8A (I), granzyme B (GZMB; J) in the HNSCC Firehose legacy TCGA patient cohort for non-HPV associated cases. n=462 patients.

Significance was determined if the p-value was <0.05*, <0.01**, <0.001***, <0.0001**** by a Mann-Whitney test (A-C), a mixed-effects model for tumor volume at selected time points (D-E), Fisher’s exact (F-G), and a Pearson Correlation (H-J). For correlations, R² value for simple linear regression, r-value and p-value are reported. Error bars represent standard error of the mean (±SEM).

3.6. SCC cell-produced integrin β1 was necessary to complete later steps of the metastatic cascade in immunocompetent hosts.

To determine if ectopic SCC cell-produced integrin β1 was necessary for immune evasion during the later steps of metastasis, we utilized in vivo experimental modeling of metastatic steps. We intravenously (i.v). injected shNC and shITGB1 SCC cells into C57BL/6 mice to mimic SCC cells traveling after dissemination from the primary site. While shNC SCC cells could metastasize to the lungs, shITGB1 in SCC cells largely abolished lung metastasis (Fig. 6A–B). Next, we directly implanted shNC and shITGB1 P029 SCC cells into the left lung lobe of C57BL/6 mice to assess SCC cell survival and immune evasion after extravasation. Compared to shNC SCC cells, shITGB1 SCC cells had significantly reduced expansion in the lungs (Fig. 6C). shITGB1 SCCs in the lungs of C57BL/6 hosts had reductions in proliferating SCC cells (Fig. 6D) and infiltrated Ly6G⁺ granulocytes (Fig. 6E) but increases in CD8⁺ cells (Fig. 6F). Taken together, our results demonstrate that ectopic SCC cell overproduction of integrin β1 is essential to regulate functions of MICs and evasion of CD8 T cell-mediated destruction throughout the metastatic journey (Fig. 6G).

Figure 6. SCC cell-produced integrin β1 was necessary to complete the later steps of the metastatic cascade in immunocompetent hosts.

(A-B) Quantification of gross lung metastases at 24 days post tail vein injection of shNC and shITGB1 H1977m (A) and P029 (B) cells into C57BL/6 mice.

(C-F) Representative H&E (C) and IHC staining (D-F) with quantification of shNC (n=5) and shITGB1 (n=4) P029 lung tumors. IHC staining was performed for Ki67 (D), Ly6G (E), and CD8α (F) of lung tumors. Unless otherwise indicated, scale bars 200 μm. Arrowheads indicate metastasis (C). The average of five images (or less if the entire tumor was imaged) per sample, each taken at 10X magnification, represents one sample; quantifications were performed with numbers of lungs specified above.

(G) Schematic of the multiple functions of SCC cell integrin β1 that drive metastasis to the lungs. Created with BioRender.com.

Significance was determined if the p-value was <0.05*, <0.01**, and <0.01*** by Fisher’s exact test (A-B) and the Mann-Whitney test (C-F). One data point is excluded from (F) as a significant outlier as determined by Grubb’s test (α = 0.05). Error bars represent standard error of the mean (±SEM).

4. Discussion

Ectopic SCC cell-produced integrin β1 is necessary to induce SCC metastasis by enhancing cell-autonomous MIC functions and immune suppression.

Our use of metastatic K15.Kras^G12D.Smad4^–/– C57BL/6 SCCs enabled us to dissect both cell-autonomous and immune-dependent mechanisms of metastasis regulated by SCC cell-produced integrin β1. GSEA of RNAseq performed on shNC and shITGB1 SCC cells along with functional in vitro assays indicated that SCC cell-produced integrin β1 drove expansion and invasion. These results are consistent with the reported functions of integrin β1 produced by non-SCC cancer cells [34]. These data explain how SCC cell-produced integrin β1 drove cell-autonomous functions that contributed to primary tumor growth in both C57BL/6 and nude mice, which were reduced by SCC cell-specific integrin β1 knockdown.

In addition to driving SCC cell invasion to support escape from the primary tumor, SCC cell-produced integrin β1 supported SCC expansion when implanted directly into the lungs of mice. Interestingly, although shITGB1 reduced SCC cell clonal expansion in vitro, proliferation in experimental lung metastases was more pronounced than at the primary site. These findings suggest that integrin β1 may enhance the ability of disseminated SCC cells to re-enter the cell cycle after adapting to the metastatic niche, in addition to supporting baseline clonal expansion. Consistent with these functional effects, downregulation of the EMT pathway coincided with abrogation of morphogenesis-related pathways in shITGB1 SCC cells. These results align with our previous findings that pluripotent and migratory CSCs are metastatic CSCs [17]. Together, our data suggest that SCC cell-produced integrin β1 regulates key cell-autonomous functions, namely invasion and expansion, associated with MIC behavior.

Knockdown of Itgb1 in SCC cells significantly reduced metastasis in C57BL/6 hosts but not in athymic hosts, suggesting that cell-autonomous effects of integrin b1 is insufficient to drive metastasis. In addition to the SCC cell-intrinsic effects of shITGB1 as determined by in vitro assays, we found that SCC cell-produced integrin β1 was also necessary for TME remodeling. shITGB1 in SCC cells reduced signaling factors for myeloid recruitment at both transcriptional and protein levels, though to different degrees in H1977m and P029. Therefore it appears that the contribution of SCC cell-produced integrin β1 to immune modulation is due to internal signaling downstream of integrin β1 rather than stabilization of extracellular signaling factors as has been previously shown [35]. Abrogation of myeloid chemokine production due to Itgb1 knockdown within SCC cells significantly reduced Ly6G⁺ myeloid cell recruitment and depletion of granulocytes in C57BL/6 mice bearing SCCs significantly reduced lung metastasis in immune competent background. These results are consistent with our previous report that MICs are reliant on granulocytes for successful colonization of the lungs [27].

Granulocytes can support MICs indirectly by suppressing the anti-tumor immune response [36]. Regarding CD8 T cells, granulocytes can impair homing to the tumor, CD8 T cell expansion, and cytotoxic functions [37]. In this study we found that granulocytes produced an immune suppressive TME, as demonstrated by increased CD8 T cells and GZMB⁺ cells in both SCC models following granulocyte depletion. GZMB is produced by cytotoxic T cells and natural killer (NK) cells; however, functional NK cells alone were insufficient to eradicate shITGB1 SCC distant metastasis in athymic mice. In contrast, depletion of CD8 T cells in C57BL/6 mice enabled shITGB1 SCC metastasis, indicating that cytotoxic CD8 T cells alone, or in combination with cytotoxic NK cells, are responsible for eliminating distant metastases. Future studies will further assess the role of cytotoxic NK cells alone or in combination with CD8 T cells in controlling SCC metastasis. In sum, our findings suggest that SCC cell-produced integrin β1 enables immune evasion, in part, through granulocyte recruitment.

Depletion of CD8 T cells in shITGB1 SCC-bearing C57BL/6 mice revealed that suppression of anti-tumor CD8 T cells by SCC cell-produced integrin β1 was essential for lung metastasis. Interestingly, primary shITGB1 SCC growth was not reversed to the same extent. These data suggest that multiple mechanisms, both dependent and independent of CD8 T cell exclusion, support primary tumor growth, while distant lung metastasis is principally dependent on avoiding CD8 T cell-mediated killing. Further investigation into the distinct roles of CD8 T cells in local versus distant SCC outgrowth is warranted.

Analysis of TCGA SCC cases [20] showed a positive association between ITGB1 and CD3E, CD8A, as well as GZMB expression, suggesting that this mechanism of SCC cell-produced integrin β1-mediated CD8 T cell suppression is apparent in human patients. It remains to be determined what specific roles different T cell subpopulations play controlling SCC distant metastasis. Our findings highlight that, among the mechanisms regulated by SCC cell-produced integrin β1, immune evasion is perhaps the most important for SCC metastases.

SCC cell-overproduced integrin β1 is necessary for immune evasion throughout the metastatic cascade.

Because knockdown of Itgb1 in SCC cells significantly reduced spontaneous lung metastases from the primary tumor in C57BL/6 mice, we evaluated the necessity of SCC cell-produced integrin β1 to complete the later steps of the metastatic cascade. Compared with shNC SCC cells, shITGB1 SCC cells i.v. injected into C57BL/6 mice produced significantly fewer lung metastases. Similarly, when we directly implanted SCC cells into the lungs of C57BL/6 mice, shITGB1 SCCs had reduced expansion relative to shNC SCCs. H&E and IHC staining of experimental lung metastases showed that the decrease in the shITGB1 SCC metastatic area was associated with reduced cell proliferation and loss of an immunosuppressive microenvironment, mirroring the primary tumor. Given the increased presence of CD8 T cells in experimental lung metastases, i.v. injected shITGB1 SCC cells likely experienced immunogenic killing upon extravasation into the lungs. Future investigation will determine if SCC cell-produced integrin β1 also plays a similar function in escaping immunogenic destruction during circulation.

Ectopic SCC epithelial cell integrin β1 overproduction has potential as a biomarker for metastatic SCCs and targeted therapeutic interventions.

Overall survival of SCC patients drops significantly upon development of distant metastases. A reliable biomarker for early detection of metastatic SCCs is essential to inform an appropriate treatment regimen for patients. Ectopic overproduction of integrin β1 in metastatic SCC patients has been reported previously [38, 39] and is consistent with our GeoMx transcriptomic analysis of Stage IV-A SCC intratumor ROIs. In addition to these clinical correlations, tumor cell- produced integrin β1 has been reported to increase MIC properties in multiple cancers [40], similar to our findings here. To our knowledge, ectopic SCC cell overproduction of integrin β1 has not previously been demonstrated to play a causal role in CD8 T cell evasion in metastasis, reinforcing its importance in metastatic SCC.

Several α integrins, including laminin-binding integrins α3 and α6, have previously been proposed as biomarkers for MICs [40]. Here, we found that laminin-binding α-integrins were similarly upregulated in metastatic murine SCC cells. Expression of ITGA3 and ITGA6 each correlated with worse SCC patient survival, and ITGA3, but not ITGA6, was enriched in Stage IV-A SCC intratumor ROIs. These observations are consistent with our finding that integrin α6 has a causal role in CSC expansion for SCC tumor initiation but does not correlate with SCC metastasis [17]. Integrin β1 forms heterodimeric receptors with all laminin-binding α-integrins, including integrins α3 and α6 [12, 41]; each α-integrin likely plays a distinct role in metastasis. Accordingly, the use of any single α-integrin as a MIC biomarker may be unreliable. Furthermore, the multiplicity of α-subunits allows formation of more integrin β1 complexes, increases the maintenance of integrin β1 stability, and can compensate for the loss of any individual α-integrin to maintain integrin β1 signaling [42, 43]. Together, these data provide a rationale for the use of ectopic SCC cell integrin β1 overproduction as a biomarker to predict patients at risk of distant metastasis.

At present, patients with metastatic SCCs receive chemotherapy and/or ICI targeting the anti-PD-1/PD-L1 axis, yet efficacy for ICI in metastatic SCC remains limited [9, 44]. The recent KEYNOTE-689 clinical trial (NCT03765918) led to FDA approval for neoadjuvant and adjuvant ICI treatment in resectable head and neck SCCs. Although ICI can generate a more inflamed TME, KEYNOTE-689 demonstrated only modest survival benefits and no benefit for local control [45], suggesting that effective therapy requires targeting both cell-autonomous activity and remodeling the immune microenvironment. While ICI is unable to control primary tumor growth and eliminate metastasis in most SCC patients, we demonstrate here that knocking down integrin β1 in SCC cells that do not respond to ICI, specifically P029 [18], reduces primary tumor growth and abrogates distant metastasis in an immunocompetent setting. These results suggest that ectopic overproduction of integrin β1 by SCC cells may serve both to reliably identify MICs and as a biomarker to design treatment modalities that engage complementary SCC cell-intrinsic and immune-dependent antitumor mechanisms.

An array of approaches has been developed to target integrins in cancer yet, to date, all such attempts have failed in clinical trials [46]. These failures are attributed to limited efficacy and unacceptable levels of adverse events. Ectopic SCC cell overproduction of integrin β1 provides an opportunity for improved drug delivery utilizing nanoparticle systems [47], or directed immunological killing directed by bispecific antibodies engineered against integrin β1. Future research efforts should be directed toward defining internal signaling mechanisms downstream of integrin β1 on SCC cells. Such knowledge could inform combinatorial strategies that have improved efficacy over direct inhibition of SCC cell-overproduced integrin β1. In sum, our findings support the design of therapeutic interventions based on SCC cell overproduction of integrin β1 as a biomarker to abrogate the cell-autonomous signaling necessary for MIC characteristics and immune evasion in SCC metastasis.

Supplementary Material

Supplementary Figure 1

Supplementary Figure 1. GeoMx DSP provided whole transcriptome analysis of SCC Intratumor ROIs. (A) Schematic of GeoMx regional transcriptomic profiling procedure for SCC patient specimens. Created with biorender.com.

(B) SCC patient specimens stained for SYTO13 (blue), pan-cytokeratin (green) and CD45 (red) used for GeoMx regional spatial transcriptomics.

Supplementary Figure 2

Supplementary Figure 2: Laminin-binding integrin α6 correlated with worse SCC patient survival but was not enriched in Stage IV Intratumor ROIs. (A-B) Expression of α integrins 6 (ITGA6) (A) and 7 (ITGA7) (B) as determined by GeoMx DSP in Stage II-III (n=4) and Stage IV-A (n=14) Intratumor ROIs.

(C-D) Survival rate of SCC patients from the HNSCC Firehose legacy TCGA patient cohort for non-HPV associated cases with high (red) or low (blue) quartile expression of ITGA6 (C) and ITGA7 (D).

Significance was determined if the p-value was <0.05 by an unpaired two-tailed t test (A-B) and the long-rank Mantel-Cox Test for Survival comparisons (C-D). Error bars represent standard error of the mean (±SEM).

Supplementary Figure 3

Supplementary Figure 3: Knockdown of ectopically overproduced integrin β1 in metastatic murine SCC cells was maintained in vivo. (A) Average ITGB1 expression in scRNAseq clusters for human epithelial skin tissue as determined by Solé-Boldo and colleagues [24].

(B) Representative IHC staining of C57BL/6 murine skin for integrin β1. Scale bar 60 μm.

(C) Representative IHC images of shNC and shITGB1 H1977m (top) and P029 (bottom) tumors and metastases stained for integrin β1 from the cohort of C57BL/6 mice described in Fig. 2. shITGB1 SCCs did not have metastases for both lines to stain for and are therefore not included. Scale bars 60 μm.

(D-E) Representative immunoblots showing relative levels of Integrin β1 in H1977m (D) and P029 (E) shNC and shITGB1 tumors.

Supplementary Figure 4

Supplementary Figure 4: Ectopic SCC cell integrin β1 played a causal role for metastasis C57BL/6 hosts independent of primary tumor volume. (A) Individual flank tumor volumes of H1977m shNC (n=12) and shITGB1 (n=8) tumors implanted into the flanks of male C57BL/6 mice.

(B) Quantification of gross lung metastases from mice in (A) and Fig. 2.

(C-D) Pearson Correlations between primary tumor volume and micrometastatic burden of H1977m (C) and P029 (D) tumor-bearing mice implanted into sex-matched C57BL/6 mice from the cohort of mice described in Fig. 2.

Significance was determined if the p-value was <0.05 and <0.01 by Fisher’s exact test and a Pearson Correlation. For correlations, R² value for simple linear regression, r-value and p-value are reported.

Supplementary Figure 5

Supplementary Figure 5: Knockdown of integrin β1 in SCC cells reduced SCC cell proliferation in vivo. (A) Representative IHC staining and quantification for Ki67 performed on shNC (n=5) and shITGB1 (n=5) H1977m (left) and P029 (right) tumors from the cohort of mice described in Fig 2. Scale bars 200 μm. Quantification represents the average of 5 images for each tissue taken at 10X magnification.

Significance was determined if the p-value was <0.05* or <0.001**** by an unpaired two-tailed t test. Error bars represent standard error of the mean (±SEM).

Supplementary Figure 6

Supplementary Figure 6: Knockdown of integrin β1 in SCC cells reduced CD31⁺ vascular area in tumors. (A) Representative IHC staining and quantification for CD31 performed on shNC (n=5) and shITGB1 (n=5) H1977m (left) and P029 (right) tumors from the cohort of mice described in Fig. 2. Scale bars 200 μm. Quantification represents the average of 5 images for each tissue taken at 10X magnification.

Significance was determined if the p-value was <0.05* by an unpaired two-tailed t test. Error bars represent standard error of the mean (±SEM).

Supplementary Figure 7

Supplementary Figure 7: Granulocytes were depleted by anti-GR1 (Ly6G/Ly6C) antibodies in immunocompetent hosts. (A) Representative flow gating strategy for immune cells of interest.

(B-C) Abundance of granulocytes in whole blood of sex-matched C57BL/6 mice bearing H1977m (B) and P029 (C) tumors. Blood was collected at study endpoint, 28 days post SCC implant from mice treated with Isotype or anti-GR1 (αGR1) depleting antibody from the cohort of mice described in Fig. 4G–J.

Significance was determined if the p-value was <0.01** by an unpaired two-tailed t test (B-C). Error bars represent standard error of the mean (±SEM).

Supplementary Figure 8

Supplementary Figure 8: Granulocyte depletion does not significantly reduce primary tumor growth. (A-B) Average volumes of H1977m (A) and P029 (B) tumors implanted into the flanks of sex-matched C57BL/6 mice treated with Isotype or anti-GR1 (Ly6G/Ly6C, αGR1) depleting antibody. For mice with H1977m tumors, n=10 Isotype and n=5 αGR1 treatment (A). For mice with P029 tumors, n=10 Isotype and n=7 αGR1 treatment (B).

Significance was determined if the p-value was <0.05 by a mixed-effects model for at selected time points. Error bars represent standard error of the mean (±SEM).

Supplementary Figure 9

Supplementary Figure 9: CD8 T Cells were depleted by anti-CD8α antibodies in immunocompetent hosts. (A) Representative flow gating strategy for immune cells of interest.

(B-C) Abundance of CD8 T cells in whole blood of sex-matched C57BL/6 mice bearing shITGB1 H1977m (B) and P029 (C) tumors. Blood was collected at study endpoint (28 days post implant for shITGB1 H1977m bearing mice and 24 days post implant for shITGB1 P029 bearing mice) from mice treated with Isotype or anti-CD8α (αCD8) depleting antibody from the cohort of mice described in Fig. 5D–G.

Significance was determined if the p-value was <0.0001**** by an unpaired two-tailed t test (B-C). Error bars represent standard error of the mean (±SEM).

Glossary

CSC

cancer stem cell

DGE

differential gene expression

DSP

Digital Spatial Profiler

EMT

epithelial-mesenchymal transition

FPPE

formalin-fixed paraffin embedded

GSEA

gene set enrichment analysis

GZMB

granzyme B

HNSCC

head and neck squamous cell carcinoma

HPV

human papillomavirus

ICI

immune-checkpoint inhibitor

IHC

immunohistochemistry

i.p.

intraperitoneal

i.v.

intravenous

K15

keratin 15

MIC

metastasis-initiating cell

NK

natural killer

PanCK

pan-cytokeratin

PD-1

programmed cell death protein 1

PD-L1

programmed death-ligand 1

RGD

Arg-Gly-Asp

RNAseq

RNA sequencing

ROI

region of interest

s.c.

subcutaneous

SCC

squamous cell carcinoma

shITGB1

shRNA Itgb1 knockdown

shNC

shRNA non-targeting control knockdown

TCGA

The Cancer Genome Atlas

TMA

tumor microarray

TME

tumor microenvironment

Availability of data and material

Data will be made available in compliance with the NIH’s data sharing policy.