SPP1 orchestrates development of an immunosuppressive tumor microenvironment that supports liver metastasis of colorectal cancer cells, offering insights into potential strategies for improving immunotherapy efficacy in liver metastatic colorectal cancer.
Abstract
Colorectal cancer remains a major cause of cancer-related morbidity and mortality globally, with 30% to 40% of cases developing metastasis, mainly to the liver. Although immunotherapy has shown promise for colorectal cancer treatment, patients with colorectal cancer liver metastasis (CRLM) experience limited therapeutic benefits, potentially because of an immunosuppressive tumor microenvironment. Thus, an urgent need exists to identify the key players that drive CRLM and potentiate immunotherapeutic resistance. In this study, we established liver metastatic cells through continuous passaging in vivo, allowing the screening of RNA expression profiles related to CRLM. A combination of spatial transcriptomic sequencing and single-cell analysis revealed a substantial upregulation of SPP1 expression and secretion in CRLM. SPP1 induced immunotherapeutic resistance by stimulating CXCL12 production by cancer-associated fibroblasts through activation of β-catenin/HIF1α-related transcription. CXCL12 promoted epithelial–mesenchymal transition of colorectal cancer cells but suppressed CD8+ T-cell infiltration. Treatment with a CXCL12 receptor antagonist or anti-SPP1 antibody markedly activated intratumoral CD8+ T-cell infiltration and enhanced the efficacy of anti–PD-1 antibody treatment. Elevated SPP1 and CXCL12 corresponded to immunotherapy resistance in patients with CRLM. Together, this study highlights the potential of the SPP1/CXCL12 axis as a target and a biomarker for precise cancer immunotherapy in CRLM. The intricate interactions within the tumor microenvironment offer promising avenues for improving therapeutic outcomes in patients with CRLM.
Significance:
SPP1 orchestrates development of an immunosuppressive tumor microenvironment that supports liver metastasis of colorectal cancer cells, offering insights into potential strategies for improving immunotherapy efficacy in liver metastatic colorectal cancer.
Introduction
Colorectal cancer is one of the most prevalent cancers globally (1). Although colorectal cancer can be effectively cured with surgical resection, approximately 20% to 25% of cases develop metastasis in the following years, resulting in poor survival rates and limited treatment options (2–4). Population-based data indicate that the liver is the most common site of colorectal cancer metastases (5–7). Advancements in immunotherapy have brought hope for treating multiple cancers; however, patients with colorectal cancer liver metastasis (CRLM) have restricted survival benefits (8, 9). This compromised therapeutic effectiveness is due to the formation of an immunosuppressive tumor microenvironment (TME), involving the differentiation, activation, and spatial distribution of lymphocytes, myeloid cells, and stromal cells (10). However, the factors that drive CRLM and elicit microenvironmental changes remain unclear. Thus, understanding the mechanisms underlying liver metastasis–associated resistance to immunotherapy is critical for the therapeutic optimization of colorectal cancer.
Secreted phosphoprotein 1 (SPP1; osteopontin) has emerged as a key player in the TME of different cancers because of its association with tumor-associated macrophages. SPP1-positive macrophages have been linked to immunosuppressive activity and unfavorable colorectal cancer prognosis (11). An increasing proportion of SPP1-positive macrophages contribute to immune suppression and desert via their inhibitory interactions with T cells in metastatic liver lesions (12). In hepatocellular carcinoma, the SPP1–CSF1/CSF1R axis is crucial in trafficking tumor-associated macrophages, and blocking CSF1/CSF1R enhances immunotherapy efficacy (13). However, the link between SPP1 and CRLM remains elusive.
Compared with myeloid cell–type macrophages, cancer-associated fibroblasts (CAF) are the most abundant stromal cell type in the TME and are the key modulators of cancer initiation, progression, and metastasis (14). By physically interacting with and adhering to cancer cells, CAFs facilitate tumor embolism formation and gastric cancer metastasis (15), whereas certain CAF subtypes have been identified in CRLM (16). As master producers of secreted molecules, CAFs contribute to the immunosuppressive TME by mediating the release of diverse growth factors, cytokines, or chemokines (17). Thus, CAFs repel cytotoxic lymphocytes or repress their immune efficacy while supporting cancer cell growth and metastasis (18). CAF activation facilitates immunotherapeutic resistance in gastrointestinal cancers, indicating the therapeutic potential of targeting CAFs in cancer management (19). Thus, deciphering the involvement of CAF-related CRLM in immunotherapeutic resistance may provide important clues for colorectal cancer treatment.
Herein, we induced and identified a liver metastatic LoVo cell subline of colorectal cancer cells by continuous passaging using in vivo and in vitro xenografts. Starting from screening alterations in RNA expression profiles related to CRLM, we comprehensively conducted studies based on models and human samples. The upregulated SPP1 expression and secretion via colorectal cancer cells strongly contributed to liver metastasis and the immunosuppressive microenvironment. SPP1 manipulated the CXCL12-dependent remodeling of CAFs in the colorectal cancer TME, ultimately contributing to immunotherapy resistance in patients with colorectal cancer. In addition to revealing the SPP1–CAF–CXCL12 axis’s role in mediating CRLM and therapeutic resistance, our work provides solid preclinical evidence to support SPP1’s potential as a biomarker and druggable target for cancer precision treatment.
Materials and Methods
Ethics statement
Experiments involving patient specimens were approved by the Institutional Ethics Committee of Peking University Cancer Hospital and Institute (approval number: 2024KT163). Written informed consent was obtained from all patients. The study was conducted in accordance with the Declaration of Helsinki. All clinical specimens from patients with colorectal cancer, including paraffin-embedded sections, freshly frozen tissues, blood samples, and associated prognostic data, were collected at Peking University Cancer Hospital and Institute in adherence to the approved protocol.
Cell lines
LoVo (RRID: CVCL_0399), HCT116 (RRID: CVCL_0291), and MC38 (RRID: CVCL_B288) cell lines were obtained from the Chinese Academy of Medical Sciences, whereas human embryonic kidney 293 T (HEK293T; RRID: CVCL_0063) cells were obtained from ATCC. All cell lines were cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco) and 100 U/mL penicillin–streptomycin at 37°C with 5% CO2. Cells were cultured to approximately 90% confluence, harvested by digestion, and resuspended in CELLSAVING (cat. #C40100, New Cell and Molecular Biotech). The cell suspension was aliquoted into cryovials (NEST Biotechnology Co. Ltd.) and stored at −80°C. All cell lines were screened for Mycoplasma and profiled with short tandem repeat profiling– upon receival in the laboratory. Cell lines were routinely screened for Mycoplasma using PCR testing monthly and were found to be free of Mycoplasma.
Animal use and care
All animal experiments were performed in accordance with the Institutional Animal Care and Use Committee guidelines of the Beijing Cancer Hospital (ethics approval number: EAEC 2023-19). BALB/c nude (RRID: IMSR_CRL:490), C57BL/6J (RRID: MGI:2159769), and NOD/Shi-scid IL2rγnull (NOG; RRID: IMSR_TAC:HUPBMC-NOG) mice were obtained from Beijing Vital River Laboratory Animal Technology. Mice were housed under specific pathogen-free conditions in ventilated cages with a controlled 12-hour light/dark cycle, temperature, and humidity, with ad libitum access to enriched food and water.
Antibodies and reagents
Antibodies against β-actin (cat. #HA722023, RRID: AB_3096833), GAPDH (cat. #ET1601-4, RRID: AB_3069615), β-tubulin (cat. #EM0103, RRID: AB_2819165), lamin B1 (cat. #ET1606-27, RRID: AB_3069729), SPP1 (cat. #0806-6, RRID: AB_3068666; cat. #ER1802-16, RRID: AB_3069113), E-cadherin (cat. #ET1607-75, RRID: AB_3069782), N-cadherin (cat. #ET1607-37, RRID: AB_3069761), Slug (cat. #HA722828), Snail (cat. #ER1706-22, RRID: AB_3069038), vimentin (cat. #ET1610-39, RRID: AB_3069923), FAP (cat. #ET1704-23, RRID: AB_3070474), αSMA (cat. #ET1607-53, RRID: AB_3069772), β-catenin (cat. #ET1601-5, RRID: AB_3069616), TGFβ1 (cat. #HA721143), hypoxia-inducible factor-1α (HIF1α; cat. #RT1278), YAP1 (cat. #ET1608-30), pYAP1 (cat. #ET1611-69, RRID: AB_3070048), NF-κB p65 (cat. #ET1603-12, RRID: AB_3069668), p-NF-κB p65 (cat. #ET1604-27, RRID: AB_3069692), AKT1 (cat. #ET1609-47, RRID: AB_3069857), pAKT1 (cat. #ET1607-73, RRID: AB_2940863), CDX2 (cat. #ET1605-4, RRID: AB_3069712), CK7 (cat. #ET1609-62, RRID: AB_3069871), CK20 (cat. #ET7110-54, RRID: AB_3071018), and Ki-67 (cat. #HA721115, RRID: AB_3072239) were obtained from HUABIO. The antibody against CD44 (cat. #15675-1-AP, RRID: AB_2076198) was purchased from Proteintech. Plasmids were constructed in all experiments using Gibson assembly cloning techniques as previously described (15). Full-length human SPP1 was amplified from LoVo cell cDNA with the forward primer 5′-ATGAGAATTGCAGTGATTTGCT-3′ and reverse primer 5′-ATTGACCTCAGAAGATGCACTA-3′. Full-length mouse Spp1 was amplified from the MC38 cell cDNA using the forward primer 5′-ATGAGGCTGCAGTTCTCCTGGC-3′ and reverse primer 5′-GTTGACCTCAGAAGATGAACTC-3′. Human recombinant SPP1 (HY-P70499) and CXCL12 (HY-P70469) proteins were obtained from MedChemExpress. Additionally, the β-catenin inhibitor MSAB (cat. #HY-120697) and the CAF inhibitor talabostat (cat. #HY-13233) were sourced from MedChemExpress. Matrigel was purchased from BioGenous (cat. #M315066) and Absin (cat. #abs9490).
Western blotting
Cells or patient-derived organoids (PDO) were lysed in RIPA buffer (cat. #R0010, Solarbio) and prepared for Western blot analysis according to the standard protocols. For tumor tissue protein extraction, 100 mg of the tissue was combined with 1 mL of the RIPA buffer and homogenized using a tissue homogenizer. The homogenate was centrifuged at 12,000 rpm for 10 minutes, and the supernatant was collected, mixed with 5× loading buffer, and heated at 100°C for 10 minutes.
For nuclear–cytoplasmic separation, cytoplasmic and nuclear proteins were extracted using a Nuclear and Cytoplasmic Protein Extraction Kit (cat. #P0028, Beyotime) following the manufacturer’s protocol. For coimmunoprecipitation experiments, β-catenin (cat. #ET1601-5, RRID: AB_3069616) and HIF1α (cat. #RT1278) antibodies were used to enrich their respective proteins from the cells, which were then monitored using the corresponding antibodies. After SDS/PAGE separation, proteins were transferred to 0.45-μm polyvinylidene difluoride membranes and incubated overnight with primary antibodies. Subsequently, membranes were probed with horseradish peroxidase–conjugated secondary antibodies and visualized using enhanced chemiluminescence (cat. #P10300, New Cell and Molecular Biotech).
Cell migration and invasion assay
When the cells reached the logarithmic growth phase, they were harvested and resuspended in serum-free medium. Tumor cells were adjusted to a 1 × 106 cells/mL concentration, whereas CAFs were adjusted to 5 × 105 cells/mL. Transwell chambers were placed into a 24-well plate containing 700 μL of DMEM supplemented with 15% FBS in the lower chamber, and 200 μL of the cell suspension was added to the upper chamber. Chambers were incubated at 37°C for 24 hours for migration assays and 48 hours for invasion assays. After incubation, the chambers were fixed with 4% paraformaldehyde for 30 minutes and stained with crystal violet or DAPI. The cell counts on the undersurface of the PET membrane were photographed using a Leica microscope.
Wound healing assay
Colorectal cancer cells (1 × 106) and CAFs (5 × 105) were seeded in six-well plates to achieve approximately 80% confluence. Upon reaching approximately 95% confluence, three vertical scratches were made in each well using a sterile blue pipette tip. Subsequently, the plate was washed once with PBS and replenished with 2% FBS–supplemented DMEM. Scratches were visualized under a microscope, and images were captured at multiple time points (0, 12, 24, and 36 hours). ImageJ (RRID: SCR_003070) was used for quantitative analysis of the wound healing ratio at specified locations over time.
EdU assay
The EdU-594 Cell Proliferation Assay Kit (cat. #C0078S, Beyotime) was utilized to assess cell proliferation according to the manufacturer’s standard instructions. Briefly, CAFs (5 × 105 cells) were cultured overnight in six-well plates. The following day, cells were incubated with 2× EdU working solution for an additional 2 hours at 37°C. After EdU labeling, cells were fixed with 1 mL of 4% paraformaldehyde at room temperature for 15 minutes. The cells were subsequently washed three times and treated with 1 mL of permeabilization solution at room temperature for 15 minutes. After further washing, the cells were incubated with click reaction solution (0.5 mL) in the dark at room temperature for 30 minutes. After the removal of the click reaction solution, the cells were washed three times with washing buffer. Finally, the cells were assessed and imaged using a fluorescence microscope at an excitation wavelength of 594 nm.
TUNEL
The TUNEL assay was conducted on paraffin-embedded tumor sections using a TUNEL Assay Kit (Beijing Solarbio Science & Technology Co., Ltd., cat. #T2190). The apoptosis index was determined by calculating the mean intensity of the TUNEL+ cells in the tumor tissues, with the mean of the control group normalized to 1 for graphical representation.
ELISA
Approximately 100 mg of the tumor tissue from mice or patient-derived colorectal cancer tissues was subjected to five freeze–thaw cycles in liquid nitrogen. Subsequently, 1 mL of PBS was added per 100 mg of tissue, and the samples were homogenized. The homogenate was subsequently centrifuged at 8,000 rpm for 10 minutes, and the supernatant was collected for ELISA. Mouse granzyme B (cat. #SEKM-0088), IFNγ (cat. #SEKM-0031), and TGFβ1 (cat. #SEKM-0035) ELISA kits and human granzyme B (cat. #SEKH-0193), IFNγ (cat. #SEKH-0046), TGFβ1 (cat. #SEKH-0316), and CXCL12 (cat. #SEKH-0310) ELISA kits were obtained from Beijing Solarbio Science & Technology Co., Ltd. Human SPP1 (cat. #BSEH-122-96) ELISA kits were purchased from BioSharp.
qRT-PCR and mRNA sequencing
Total RNA was extracted using TRNzol Universal Reagent (cat. #AG21102, Accurate Biotechnology (Hunan) Co., Ltd) and subsequently reverse transcribed with the PrimeScript RT Reagent Kit (Takara, cat. #RR037Q). qRT-PCR was performed using Premix Ex Taq (probe qPCR; Takara, cat. #RR39LR). All experiments were conducted independently in triplicates and the results are presented as relative gene expression levels normalized to ACTB. Supplementary Table S1 lists the primer sequences used in this study. The mRNA libraries for sequencing were prepared using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina according to the manufacturer’s instructions. Gene expression levels were quantified as FPKMs using StringTie (RRID: SCR_016323). Differentially expressed genes were identified using the R package Limma (RRID: SCR_010943), and statistical significance was set at P < 0.05 and |log2 (fold change)| > 1.
RNAi
The siRNAs targeting human SPP1 and mouse Spp1 were obtained from GenePharma. For cell transfection assays, siRNAs were transfected into cells using GP-transfect-Mate (GenePharma) in a serum-free medium, following the manufacturer’s protocol. For intratumoral injection, cholesterol-modified siRNAs were dissolved in diethyl pyrocarbonate–treated water and injected into the tumors. Supplementary Table S2 lists the siRNA sequences.
CAF isolation and culture
CAFs were isolated and cultured from the primary tumor tissue and liver metastases of patients with colorectal cancer, as previously described (20). Briefly, fresh tumor tissue was washed three times with PBS, necrotic areas were removed, and the remaining tissue was cut into 2-mm pieces. CAFs were allowed to migrate in 20% FBS-DMEM medium at 37°C and 5% CO2 for several weeks. Once a sufficient number of CAFs were obtained, they were digested with trypsin and transferred to new culture dishes.
Patient-derived xenograft generation
This study used 4- to 6-week-old female NOG mice to establish a patient-derived xenograft (PDX) model. Fresh and surgically resected primary colorectal cancer and liver metastasis specimens (F0 tumors) were subcutaneously implanted into the dorsal hind flank of the NOG mice for tumor expansion (F1 tumors). Once the F1 tumors reached a size of 500 mm3, the mice were euthanized, and the subcutaneous tumors were excised for subsequent passaging (F2 tumors). During passaging, necrotic tissue was first removed, and the tumors were dissected into 1-mm3 tissue blocks using a surgical scalpel. Subsequently, the tissue blocks were implanted into the dorsal hind flank of a new NOG mouse. Tumor tissues from F3 generation and beyond were preserved in cryopreservation medium containing 90% FBS and 10% DMSO and stored in liquid nitrogen for long-term preservation.
PDO generation
Fresh tumor tissue was transported to the laboratory in a tissue preservation solution (cat. #K601005, BioGenous) after the surgical excision. The tissue was immediately washed several times with PBS containing 5% penicillin and streptomycin, and necrotic areas were excised. The remaining tissue was finely minced and digested in 5 mL digestion solution (cat. #K601003, BioGenous) at 37°C for 30 minutes. After digestion, 5 mL of 2% FBS was added to neutralize the solution, and the mixture was filtered through a 100-μm cell strainer. The filtrate was subsequently centrifuged at 300 × g for 5 minutes at 4 °C. The resulting cells were resuspended in 70% Matrigel (cat. #M315066, BioGenous) and seeded into two to four wells of a 24-well plate. After polymerization of the Matrigel, each well was supplemented with 500 μL medium of the colorectal cancer organoid kit (serum-free; cat. #K2103-CR, BioGenous) and incubated for further culturing. Furthermore, stable overexpression of vector/GFP-SPP1 in the PDOs was established via lentiviral infection, followed by selection with puromycin (2 μg/mL) to isolate PDOs with stable SPP1 expression. These PDOs were subsequently expanded, cultured, and cryopreserved for stable overexpression of the GFP-vector/GFP-SPP1 in PDOs, which was subsequently achieved via lentiviral infection.
Dual-luciferase reporter assay
CXCL12 luciferase reporter plasmids, including those with mutations in the HIF1α binding sites (HBS1 or HBS2), were constructed. Furthermore, a plasmid for HIF1α overexpression, lacking the oxygen-dependent degradation domain (HIF1α-ΔODD), was generated. HEK293T cells seeded in 24-well plates were transiently transfected with 50 ng of the luciferase reporter plasmid, together with 100 ng of either the HIF1α-ΔODD plasmid or control plasmids. As an internal control, 10 ng of the pRL-TK plasmid was co-transfected. Twenty-four hours after transfection, reporter gene activity was measured using the Dual-Luciferase Reporter Assay System (Promega) and a TD-20/20 luminometer (Turner Designs) according to the manufacturer’s protocol.
Collagen contraction assay
We prepared the collagen working solution by mixing 400 µL of DMEM with 200 µL of collagen type I stock solution (3 mg/mL). With 500 µL of the collagen working solution (1 mg/mL), 2 × 105 CAFs cells were combined and neutralized by adding 7 µL of 1 mol/L NaOH. The resulting mixture was poured into a six-well plate and incubated at 37°C for 30 minutes to allow collagen polymerization. After polymerization, the gels were gently detached from the edges of the wells using a sterile pipette tip. Subsequently, 2 mL of the culture medium was added to each well. The plate was placed in an incubator for culturing, and images were captured every 2 days. Collagen contraction was quantified by comparing the area of the collagen gel with the initial area. Data were analyzed using GraphPad Prism version 8 (RRID: SCR_002798).
T-cell barrier function and attraction assays
For barrier function and T-cell attraction assays, peripheral blood mononuclear cells (PBMC) were isolated from the matched patient samples using the Ficoll method. After isolation, the cells were stimulated with CD3/CD28/CD2 T-Cell Activator (STEMCELL Technologies, cat. #10970) for 3 days and subsequently cultured in IL2–supplemented RPMI-1640 medium (STEMCELL Technologies, cat. #78036) for an additional 4 days.
For the barrier function assay, 1 × 105 CAFs were seeded into 24-well transwells with an 8-µm pore size. After 24 hours, 3 × 104 of the carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled T cells were added to the top chamber containing the confluent fibroblast layer. After 12 hours, the T cells that migrated into the wells were counted under a microscope.
For the T-cell attraction assay, 1 × 105 of the CAFs were plated at the bottom of a 24-well plate 24 hours before the assay. Then, 3 × 104 of the T cells were placed in a 3-µm transwell chamber positioned above the CAF layer. After 12 hours, the attracted cells in the lower compartment were microscopically visualized, and the CD8+ T-cell proportion that migrated to the lower chamber was quantified via flow cytometry.
Autologous coculture of the PBMCs or CAFs with tumor organoids from patients with colorectal cancer
PBMCs, CAFs, and PDOs were simultaneously isolated and cultured from the same patient. We created an in vitro coculture system of T cells and PDOs by stimulating PBMCs with a CD3/CD28/CD2 T-Cell Activator for 3 days. The T cells were labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (Beyotime, cat. # C1036), whereas PDOs were stained with the green dye CFSE. The cells were subsequently mixed at a 5:1 effector-to-target ratio and seeded in a 24-well plate, with an equal number of CAFs placed in a 0.4-µm transwell chamber for indirect coculture. After 24 hours, T-cell infiltration into the PDOs was assessed using fluorescence microscopy, and the percentage of infiltrating T cells was quantified. We assessed T-cell cytotoxicity by pre-staining PDOs with CFSE, whereas the T cells remained unstained. Following the same mixing and seeding procedures, the CAFs were cocultured in a transwell chamber. After 24 hours, propidium iodide staining was used to determine PDO viability, and images were captured to quantify the proportion of propidium iodide–positive cells.
Establishment of colorectal cancer cell lines with high metastatic ability
Next, 2 × 105 of the LoVo or HCT116 cells were seeded into 8-µm transwell chambers. After 48 hours, migrating cells were collected and expanded in culture, followed by re-seeding into fresh 8-µm transwell chambers. This procedure was repeated three times to enrich LoVo or HCT116 cells with an enhanced migratory capacity. Subsequently, 1 × 106 of the enriched LoVo or HCT116 cells were injected into the spleen of NOG mice to establish a liver metastasis model. Liver metastatic foci developed over several weeks, allowing these lesions to isolate and expand. The isolated cells were reinjected into the spleens of the mice. This in vivo cycle of spleen–liver metastasis was conducted three times to select the LoVo or HCT116 cells with increased metastatic potential. Finally, three additional rounds of in vitro transwell selection yielded the most metastatic LoVo or HCT116 cells, designated LoVo-HM or HCT116-HM.
Establishment of an MC38 tumor model with acquired resistance to immunotherapy
Subcutaneous tumor models were generated using wild-type MC38 cells from the C57BL/6 mice to establish an MC38 cell line with acquired resistance to immunotherapy. When the tumor volumes reached 50 mm3, the mice received intraperitoneal injections of 2 mg/kg anti–PD-1 antibody (cat. #BE0146-100MG, RRID: AB_10949053) two times weekly. After 3 weeks, the largest tumors were selected for passaging, which was conducted five times. The dose of anti–PD-1 antibody was incrementally increased to 4, 6, 8, and 10 mg/kg during each passage. The resulting tumors, which exhibited resistance to immunotherapy, were designated as MC38-R.
Subsequently, subcutaneous tumor models were established in the C57BL/6 mice using MC38-R cells. MC38-R tumoral fragments were excised, cut into small pieces, and subcutaneously implanted into the right flank of the recipient mice. Tumors were measured using calipers, and tumor volume (mm3) was calculated using the formula tumor volume = (long axis) × (short axis)2 × 1/2. When the tumor volume reached 100 mm3, the mice were randomly divided into two groups (n = 5 per group) and administered intratumoral injections of either siRNA-NC or siRNA-SPP1 along with intraperitoneal injections of anti–PD-1 (10 mg/kg, two times weekly). After 3 weeks, the mice were euthanized for tumor volume measurements. The tumor immune microenvironment was analyzed using IHC and ELISA.
Mouse models of the colorectal cancer
Stable MC38 cell lines expressing either a vector or SPP1 were generated via lentiviral transduction to establish a subcutaneous tumor model in the C57BL/6 mice. The mice were subcutaneously injected with 1 × 106 MC38 cells. Upon reaching a tumor volume of 100 mm3, the mice were randomly assigned to the treatment groups. Mice received either intraperitoneal injections of anti–PD-1 (10 mg/kg, two times weekly) or oral administration of talabostat mesylate (10 µg/mouse, once daily). The tumor volume and body weight were monitored throughout the study. After 4 weeks, the mice were euthanized, and subcutaneous tumors were excised to measure the tumor volume and weight. The granzyme B, IFNγ, and TGFβ levels in the tumor tissues were quantified using ELISA, whereas IHC was used to assess the CD8, CD163, and αSMA expression in the tumors.
We established a liver metastasis model in the C57BL/6 mice by injecting 1 × 106 stable MC38 cell lines expressing either a vector or SPP1 into the spleen. After 3 weeks, the mice were euthanized, and the livers were dissected to quantify the number of metastatic lesions and measure the liver weight. For the peritoneal metastasis model in BALB/c nude mice, 1 × 106 stable LoVo cell lines expressing either a vector or SPP1 were injected into the peritoneal cavity. Six weeks later, in vivo imaging was performed, followed by euthanasia to evaluate the number of peritoneal metastatic lesions.
Establishment of mouse models with a humanized immune system
Using an established PDX library, we selected a case of PDX derived from primary colorectal cancer for subcutaneous implantation in the NOG mice. Each mouse received a tail vein injection of 2 × 106 healthy human-derived PBMCs once the tumor volume reached 100 mm3. When tumors reached a volume of 150 mm3, mice were randomized and administered intratumoral injections of either vector or SPP1 expression plasmid via in vivo-jetPEI (cat. #101000030, Polyplus). Additionally, according to the experimental groups, mice received intraperitoneal injections of anti–PD-1 (10 mg/kg, two times weekly) or oral administration of talabostat (10 µg/mouse, once daily). After 3 weeks, the mice were euthanized, and tumors were harvested to evaluate the tumor volume and weight. The SPP1, granzyme B, IFNγ, and TGFβ levels in the tumor tissues were quantified using ELISA, and IHC was used to evaluate the CD8 and αSMA expression.
Using the PDO platform established by our team, PDOs from the primary tumor and liver metastases of a patient with colorectal cancer were selected for subcutaneous implantation into NOG mice to create PDO-derived xenograft (PDOX) models. Peripheral blood was collected from the same patient to isolate PBMCs, which were cocultured with PDOs from either the primary tumor or liver metastasis, stimulated with CD3/CD28/CD2 T-Cell Activator for 7 days. When the subcutaneous tumor volume reached 50 mm3, each mouse received a tail vein injection of 5 × 106 PBMCs. Upon reaching a tumor volume of 150 mm3 in mice with primary tumor-derived PDOs, intratumoral injections of either a vector or SPP1 overexpression plasmid were administered, delivering 10 µg of plasmid via in vivo-jetPEI transfection reagent every 2 days. Mice with liver metastasis–derived PDOs received intratumoral injections of cholesterol-modified siRNA-NC or si-Spp1 (5 nmol administered every 2 days). After 2 weeks, blood was collected from the inner canthus of the mice for flow cytometric analysis of immune reconstitution. After 3 weeks, the mice were euthanized, and the tumors were excised for volume and weight measurements. Additionally, flow cytometry was used to evaluate the IFNγ+ CD8 T-cell proportion in the peripheral blood. Furthermore, ELISA was used to quantify SPP1, granzyme B, IFNγ, TGFβ, and CXCL12 levels in the tumor tissues.
Hematoxylin and eosin staining and IHC
Formalin-fixed, paraffin-embedded sections were initially heated at 65°C for 1 hour to facilitate deparaffinization using xylene, followed by rehydration through graded ethanol. The sections were stained with hematoxylin for 5 minutes, differentiated in 1% hydrochloric acid alcohol, and thoroughly rinsed in tap water until optimal blue coloration was achieved. Eosin was used as the counterstain. Then, the sections were dehydrated in ethanol, cleared in xylene, mounted with a neutral resin, and examined under a microscope.
For IHC analysis, paraffin-embedded samples were baked, deparaffinized, and rehydrated before antigen retrieval and blocking of nonspecific antibody binding. The sections were incubated overnight at 4°C with specific primary antibodies. After three 5-minute washes in PBS, secondary antibodies were applied for 30 minutes at room temperature. Following additional washes, visualization was achieved using an enzyme substrate, and the sections were counterstained with hematoxylin.
IHC quantification and immunofluorescence imaging were conducted using ImageJ to measure the average intensity of the specific proteins. For each section, three non-overlapping visual fields were randomly selected, and protein expression intensity was evaluated. The mean intensity of these fields was calculated for statistical analyses and that of the control group was normalized to 1 for graphical representation. All staining, imaging, and quantification procedures were conducted in a blinded manner to ensure sample identity and phenotypic integrity.
Establishment of cecal orthotopic liver metastasis with HCT116-HM
HCT116-HM-luci cells were diluted to a concentration of 4 × 107 cells/mL in 20% Matrigel (cat. #abs9490, Absin). After anesthetizing the NOG mice, a 2- to 3-cm incision was made along the midline of the lower abdomen, and the cecum was gently exteriorized. Using an insulin syringe, 25 μL of the cell suspension was injected into the cecal wall. The cecum was subsequently returned to the abdominal cavity and the incision was sutured. Liver metastasis was monitored in vivo at 4, 6, and 8 weeks postoperatively using an imaging system.
Macrophage depletion in vivo via clodronate liposomes
Before injection, clodronate liposomes (cat. #40337ES10, Yeasen) and control liposomes (cat. #40338ES10, Yeasen) were removed from the refrigerator and allowed to equilibrate at room temperature. The liposomes were gently mixed by inverting the vial eight to 10 times. Using a 1-mL syringe with a 26-gauge needle, 200 μL of either clodronate liposomes or control liposomes was administered via intraperitoneal injection to each mouse. Injections were administered every 3 days until the end of the experiment. At the end of the study, metastatic liver tissues were harvested for flow cytometry analysis to assess the efficacy of macrophage depletion.
Intratumoral delivery of the plasmids and siRNA
For intratumoral plasmid injection, 40 µg of DNA was diluted in 100 µL of 5% glucose solution, gently vortexed, and spun down. Similarly, 5 µL of in vivo-jetPEI was diluted in 100 µL of 5% glucose solution, vortexed gently, and spin down. The diluted in vivo-jetPEI was subsequently added to the diluted DNA solution, briefly vortexed, and spun down. The mixture was then incubated at room temperature for 15 minutes. For intratumoral siRNA injection, cholesterol-modified SPP1 siRNA or control siRNA was diluted to 200 nmol/mL in diethyl pyrocarbonate–treated water. Using an insulin syringe, 25 µL of the plasmid mixture or siRNA solution was intratumorally injected into each mouse. Injections were repeated every 2 days according to the experimental timeline until the end of the study, with tumor gene expression monitored throughout to assess the treatment effects.
Spatial transcriptomics
We performed 10× Genomics Visium spatial transcriptomics on colorectal cancer samples with high RNA quality capable of distinguishing tissue pathologic states. Sequencing was performed using an Illumina NovaSeq 6000 platform (Illumina). Raw sequencing data were processed for alignment and quantification using the Spaceranger workflow (version 2.0.0). The resulting count matrix was then imported into Seurat and RCTD for data filtering, normalization, dimensionality reduction, and visualization. Single-cell RNA sequencing (RNA-seq) data were annotated in Scanpy to enable the spatial deconvolution of tissue locations, and cell type deconvolution and visualization were conducted using cell2location.
Mass spectrometry analysis
Serum-free medium was precipitated with a 6× volume of acetone, reduced with 20 mmol/L TCEP, and alkylated with 40 mmol/L IAA (Sigma-Aldrich). The mixture was digested with trypsin protease at a 1/100 (w/w) trypsin protease-to-protein ratio at 37°C overnight. The lysate was desalted using a Monospin C18 column (GL Sciences) and vacuum centrifuged to dryness. The dried peptides were redissolved in 0.1% formic acid and spiked with iRT peptides before DIA analysis.
A direct DIA Analysis built into Spectronaut (version 18.0) was used for data analysis. The MS/MS spectra were matched against the human Uniprot database (20,656 entries downloaded in October 2024). The DIA files were processed in default mode. Briefly, carbamidomethylation was set as a fixed modification, with acetylation of the N-terminus of the protein and oxidation of methionines set as variable modifications. The trypsin/P proteolytic cleavage rule was used, permitting a maximum of two missed cleavages, a minimum peptide length of seven amino acids, and a maximum peptide length of 52 amino acids. The FDR for PSM and protein quantification was set to 0.01.
Tissue dissociation and flow cytometric analysis
Tumor samples from liver metastases were minced into small pieces using a scalpel, dissociated using a Tumor Dissociation Kit (cat. #130-096-730, Miltenyi Biotec), and incubated at 37°C with gentle rotation for 30 minutes. The dissociated cells were filtered through 70-μm nylon cell strainers and washed with FACS buffer (PBS supplemented with 2% FBS). Subsequently, the cells were resuspended in FACS buffer, stained with Fixable Viability Stain 780 for 15 minutes at room temperature, washed, and then incubated with anti–mouse CD16/CD32 antibodies (BD Biosciences, cat. #553141, RRID: AB_394656) for 10 minutes at 4°C. After washing, the cells were stained with anti–mouse CD45 (BD Biosciences, cat. #563891, RRID: AB_2734134), CD3 (BD Biosciences, cat. #561798, RRID: AB_10898341), and CD8 (BD Biosciences, cat. #563898, RRID: AB_2738474) antibodies for 30 minutes at 4°C. Following fixation and permeabilization, the cells were washed with 1× intracellular staining perm and wash buffer and incubated with anti–mouse IFNγ (BD Biosciences, cat. #557649, RRID: AB_396766) and GZMB (BioLegend, cat. #372204, RRID: AB_2687028) antibodies overnight at 4°C. After the final wash, the cells were analyzed via flow cytometry, and the data were processed using FlowJo 10.9 (RRID: SCR_008520).
Statistical analysis
All bar graphs were analyzed using GraphPad Prism version 8 (RRID: SCR_002798). The data presented in this study represent the mean ± SEM from triplicate experiments. Two-tailed Student t tests were used to compare two variables. Kaplan–Meier curves were compared using the log-rank test, and Spearman correlation analysis was used to examine the correlations between two variables. Statistical significance was designated as *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; or nonsignificant (ns; P > 0.05).
Results
High SPP1 levels are related to CRLM and immunotherapy resistance
We designed a workflow for induction and screening by combining an in vitro transwell model with an in vivo intrasplenic injection model to explore the molecular changes that drive CRLM (Fig. 1A). This workflow enabled the isolation of LoVo and HCT116 cell subpopulations with a propensity for liver metastasis, termed LoVo-HM and HCT116-HM, respectively. These cells had a spindle-shaped morphology and higher in vitro migratory and invasive capabilities than the progenitor cells (Supplementary Fig. S1A–S1F). In the in vivo intrasplenic injection model, LoVo-HM and HCT116-HM cells showed a markedly increased number of liver metastatic lesions compared with the progenitor cells (Supplementary Fig. S1G–S1J). Thus, these cells have a pronounced capacity for liver metastasis.
Figure 1.
High SPP1 levels are linked to CRLM and immunotherapy resistance. A, Schematic diagram shows the construction of a CRLM cell line. B, Volcano plot of differentially expressed genes in the RNA-seq analysis of the LoVo and LoVo-HM cells. C, The GSE41568, GSE14297, GSE128213, and LoVo-HM datasets were combined, identifying upregulated genes with a fold change >1.5 and P < 0.05. D, Venn diagram revealing SPP1 as the only gene upregulated in all datasets. E and F, Western blot analysis of SPP1 expression in primary tumors (P) and liver metastases (L) from five patients with colorectal cancer is shown as a heatmap (E) and in PDOs from three patients with colorectal cancer (F). G, ELISA measures SPP1 in the blood of 45 patients with colorectal cancer without metastasis (M0) and 46 with metastasis (M1). H, Imaging assessment before and after immunotherapy in a patient with colorectal cancer #1 with liver metastasis are shown, with tumor diameters analyzed (n = 3 patients). I, Confocal microscopy to evaluate T-cell infiltration into the PDOs from patients with colorectal cancer (n = 3 patients). Scale bar, 50 μm. J, Confocal microscopy shows T-cell cytotoxicity in PDOs of patients with colorectal cancer with propidium iodide (PI) labeling for dead cells, n = 3. Scale bar, 50 μm. K–M, Spatial transcriptomics revealing tumor regression and residual tumor regions (n = 1). Spatial visualization of the cell types (L) and hierarchical clustering of the localized spots using Uniform Manifold Approximation and Projection (UMAP; M). N and O, UMAP plots illustrating SPP1 expression in different cell clusters. P, mIHC examines SPP1 and TME in primary and liver metastases from five patients with colorectal cancer, with representative images. n = 5. Scale bar, 200 μm. Data are presented as mean ± SEM. Statistical analysis: two-tailed unpaired Student t test (F, G, I, and J) and paired-samples Student t test (H and P). *, P < 0.05; **, P < 0.01; ***, P < 0.001. CRC, colorectal cancer; LM, liver metastasis; mIHC, multiplex IHC; PT, primary tumor.
Through RNA-seq, we identified a spectrum of differentially expressed genes in the LoVo-HM and LoVo cells (Fig. 1B). To better screen for candidate genes related to CRLM, we subsequently referred to bulk RNA-seq datasets from three Gene Expression Omnibus (GEO) databases that addressed paired or unpaired primary and metastatic colorectal cancer samples from patient: GSE41568, GSE14297, and GSE128213 (Fig. 1C; refs. 21–23). SPP1 was the only upregulated gene in the liver metastatic colorectal cancer cells/tissues compared with the primary colorectal cancer cells (Fig. 1D). Western blotting confirmed SPP1 upregulation at the protein level in the LoVo-HM and HCT116-HM cells compared with that in the progenitor cells (Supplementary Fig. S1K). We validated these findings by analyzing the tissue proteins from primary and metastatic liver lesions of patients with colorectal cancer. A substantial increase in SPP1 expression was noted in the CRLMs of five patients compared with that in the matched primary tumor tissues (Fig. 1E). This upregulation was further confirmed in three pairs of PDO models derived from these tissues (Fig. 1F). Given that SPP1 is a secreted protein, we measured its concentration in plasma samples using ELISA and found increased plasma SPP1 levels in patients with colorectal cancer with distant metastasis compared with those without distant metastasis (Fig. 1G). The Cancer Genome Atlas (TCGA) data showed that SPP1 was substantially overexpressed in tumors, especially in colorectal cancer (Supplementary Fig. S1L). Additionally, elevated SPP1 levels were linked to advanced pathologic T and N stages in colorectal cancer and correlated with poor overall survival and disease-free survival in the TCGA-colorectal cancer dataset (Supplementary Fig. S1M–S1Q). Considering that patients with CRLM typically exhibit limited survival benefits from immunotherapy, patients in our cohort showed a reduction in the size of primary tumors but a concomitant increase in liver metastases (Fig. 1H). Additionally, we developed a PDO-T-cell cytotoxicity model to assess the differences in immune killing between paired PDOs derived from primary and liver metastatic colorectal cancer tumors in patients. Coculturing of the PDOs with T cells showed that the CRLM-PDOs exhibited increased resistance to T-cell infiltration and cytotoxicity compared with the primary tumor PDOs (Fig. 1I and J). Next, we investigated SPP1’s role in immunotherapy resistance in CRLM using spatial transcriptomics of the primary tumor tissues from patient with colorectal cancer #1 (Fig. 1K). Cluster 1, which was mainly linked to residual tumor regions, demonstrated substantial SPP1 enrichment (Fig. 1L–O). Additionally, multiplexed IHC analysis of matched primary and liver metastatic tumors from five patients with colorectal cancer showed higher SPP1 expression in liver metastases, accompanied by decreased CD8+ T-cell infiltration and increased macrophage and CAF infiltration, creating an immunosuppressive microenvironment (Fig. 1P). Thus, SPP1 may play a crucial role in immunotherapy resistance in CRLM.
SPP1 promotes colorectal cancer metastasis and tumor progression
We investigated SPP1’s biological role in colorectal cancer metastasis. SPP1 overexpression enhanced the in vitro migration, invasion, and wound healing capabilities of the colorectal cancer cell lines LoVo and HCT116 (Fig. 2A–C; Supplementary Fig. S2A–S2C). Contrastingly, the SPP1 knockdown using siRNA led to the suppression of migration, invasion, and wound healing abilities in LoVo-HM and HCT116-HM cells (Fig. 2A–C). OE-SPP1 considerably promoted the expression of genes related to epithelial–mesenchymal transition in LoVo and HCT116 cells, whereas SPP1 knockdown substantially diminished this effect (Supplementary Fig. S2D). Next, we evaluated SPP1’s effect on in vivo metastasis of colorectal cancer using an intrasplenic injection model. Spp1 overexpression considerably promoted liver metastasis in MC38 cells (Fig. 2D–F). Collectively, these results underscore SPP1’s malignant role in driving colorectal cancer metastasis.
Figure 2.
SPP1 promotes the occurrence of colorectal cancer metastasis and immunotherapy resistance. A–C, The impact of SPP1 overexpression or knockdown on LoVo and LoVo-HM cell migration and invasion was assessed using transwell and wound healing assays. Scale bar, 100 μm. n = 3. D–F, Liver metastasis in C57BL/6J mice injected with MC38 cells was evaluated through tumor burden quantification and hematoxylin and eosin (H&E) staining (n = 5 mice/group). Scale bar, 50 μm. G, Schematic representation of the in vivo experiment. CRC, colorectal cancer; LM, liver metastasis; PT, primary tumor. H, Flow cytometry confirmed human CD45+ (hCD45+) cell engraftment at 7 days after implantation (n = 3 mice/group). I and J, Tumor morphology, weight, volume, and IFNγ expression were analyzed (n = 3 mice/group). K–M, ELISA measured IFNγ (K), granzyme B (L), and SPP1 (M) levels in the tumor tissues (n = 3 mice/group). N, Western blot analysis of SPP1 in the tumor tissues (n = 3 mice/group). O, Hematoxylin and eosin analysis of the tumor tissues. P and Q, Masson trichrome staining and IHC were used to evaluate collagen content and αSMA expression. R and S, TUNEL and Ki-67 staining were performed on PDOX tumor tissues. Data are presented as mean ± SEM. P values were determined by two-tailed unpaired Student t test (A, C, E, H–N, and P–S) and one-way ANOVA (B and C). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; n.s., nonsignificant. IOD, integrated optical density.
Next, we investigated SPP1’s role in the immunotherapy resistance of the colorectal cancer cells. Notably, the MC38 model is highly immunogenic, demonstrates an immune-infiltrated phenotype, and responds to PD-1/PD-L1 blockade (24, 25). We developed an immune-resistant MC38 cell line, termed MC38-R, to further investigate the potential of Spp1 knockdown to enhance antitumor immunotherapy efficacy (Supplementary Fig. S2E–S2G). Spp1 expression was increased in the MC38-R cells compared with the parental MC38 cells. This underscores SPP1’s involvement in resistance to anti–PD-1 therapy (Supplementary Fig. S2H and S2I). Given the immune resistance of the MC38-R tumor tissues, we used an intratumoral siRNA injection to knockdown Spp1 (Supplementary Fig. S2J). Subsequently, we combined MC38-R tumors with mu-CAFs to establish a subcutaneous MC38-R C57BL/6J mouse model (Supplementary Fig. S2K). Intratumoral siRNA injections combined with intraperitoneal anti–PD-1 therapy re-sensitized the MC38-R tumors to treatment (Supplementary Fig. S2L and S2M). Spp1 knockdown markedly increased IFNγ and granzyme B expression (Supplementary Fig. S2N and S2O). Furthermore, Spp1 was effectively downregulated in the tumor tissues (Supplementary Fig. S2P). In the si-Spp1 group, αSMA immunofluorescence and Masson trichrome staining for collagen were notably reduced, with a substantial increase in the CD8+ T cells (Supplementary Fig. S2Q–S2S). Furthermore, tumor cell proliferation markedly decreased in the si-SPP1 group (Supplementary Fig. S2T).
Given that the MC38-R model does not completely recapitulate the immune-resistant tumor characteristics of the patients with colorectal cancer, we developed a PDOX model using organoids cultured from the immunotherapy-sensitive primary tumor and immunotherapy-resistant liver metastasis of patient with colorectal cancer #1 to explore SPP1’s role in colorectal cancer immunotherapy resistance. Simultaneously, we successfully cultured CAFs from primary and metastatic tumors of patient with colorectal cancer #1. Subsequently, primary and liver metastatic PDOs were integrated with CAFs and subcutaneously implanted into NOG mice to establish a PDOX model (Fig. 2G). PBMCs were extracted from the peripheral blood of patient with colorectal cancer #1 and were stimulated for proliferation and activation. After 1 week, the T cells were isolated and injected into NOG mice via the tail vein. One week later, human PBMCs were detected in the peripheral blood of NOG mice using flow cytometry with CD45 staining (Fig. 2H). We introduced either empty vectors or OE-SPP1 plasmids into the PDOX tumors originating from the primary tumors and administered siRNA to tumors derived from liver metastases. After 3 weeks, PDOX tumor analysis revealed that SPP1 overexpression enhanced growth, whereas knockdown inhibited it (Fig. 2I). SPP1 overexpression decreased IFNγ+ CD8+ T-cell infiltration, whereas SPP1 knockdown increased their presence (Fig. 2J). ELISA confirmed these findings, showing corresponding IFNγ and granzyme B levels (Fig. 2K and L). SPP1 was successfully expressed after OE-SPP1 plasmid injection and reduced by cholesterol-modified siRNA (Fig. 2M), which was further validated by Western blotting (Fig. 2N). SPP1 overexpression substantially increased α-SMA expression and collagen production, thereby decreasing apoptosis in PDOX tumors. Contrastingly, SPP1 knockdown led to reduced αSMA and collagen levels, promoting apoptosis (Fig. 2O–S). Thus, tumor-derived SPP1 may promote colorectal cancer metastasis and influence tumor immune microenvironment.
SPP1 enhances CAF infiltration into the TME and promotes their malignant phenotypes
The effect of tumor-secreted SPP1 on the colorectal cancer microenvironment should be addressed, considering that SPP1 is a secretory protein. Referring to different TCGA datasets for gastrointestinal cancers, we found that SPP1 expression was strongly correlated with the infiltration of various TME cells, especially CAFs (Fig. 3A). Additionally, SPP1 demonstrated a positive correlation with the expression of the key CAF markers, including FAP/α-SMA/S100A4/PDGFRA/PDGFRB (Supplementary Fig. S3A). We further verified SPP1’s impact on different TME cell populations by performing single-cell RNA sequencing of the MC38 xenografts and found an increase in fibroblasts and a concomitant reduction in T cells after Spp1 overexpression (Fig. 3B–E). This suggests that SPP1 facilitates CAFs infiltration into the TME of colorectal cancer. We successfully isolated and cultured CAFs and PDOs from matched primary and liver metastatic tissues to accurately replicate the TME of patients with colorectal cancer (Fig. 3F). We established stable cell lines expressing GFP-tagged SPP1 or GFP-vector via lentivirus, and their culture supernatants stimulated CAFs, causing dose-dependent upregulation of the malignant marker FAP (Supplementary Fig. S3B and S3C). Phenotypically, exposure to either the supernatant from SPP1-overexpressing PDOs or purified SPP1 protein substantially improved CAF migration, invasion, wound healing, and proliferation compared to the control conditions (Fig. 3G–J). Additionally, given that CAFs remodel collagen in the TME, a collagen contraction assay showed that the SPP1-overexpressing supernatant considerably enhanced the CAFs’ collagen contraction capacity (Fig. 3K and L). Thus, SPP1 enhanced CAF infiltration into the TME and promoted malignant phenotypes.
Figure 3.
SPP1 enhances CAF infiltration into the TME and promotes their malignant phenotypes. A, Correlation analysis of SPP1 with key microenvironmental cells in TCGA datasets. DC, dendritic cell; MDSC, myeloid-derived suppressor cell; Treg, regulatory T cell. B–E, Subcutaneous tumor models in the MC38 cells overexpressing SPP1 or a vector control were used to examine cell subpopulations via single-cell RNA-seq. UMAP, Uniform Manifold Approximation and Projection. F, Brightfield images of the CAFs and representative IHC staining images of CDX2, CK20, β-catenin, and Ki-67 in PDOs. H&E, hematoxylin and eosin. G–I, Transwell and wound healing assay of the CAFs with SPP1 overexpression or treatment (n = 3). J, EdU assay for proliferation in CAFs with SPP1 overexpression or treatment. (n = 3). K, Schematic representation of the collagen contraction assays. L, Collagen contraction assay with CAFs treated with or without SPP1 protein (1 µg/mL), n = 3. Data are presented as mean ± SEM. P values were determined using a two-tailed unpaired Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001. CM, conditioned medium.
SPP1 promotes colorectal cancer metastasis through a positive feedback loop mediated by CAF-secreted CXCL12
CAFs modulate the TME via cytokine release, prompting investigation into SPP1’s potential role in modulating CAF-secreted proteins. Mass spectrometry showed that SPP1 stimulation induced the upregulation of several secreted proteins in CAFs, with the most prominent increase observed in CXCL12 secretion (Fig. 4A). Furthermore, reanalysis of previously acquired single-cell sequencing data from MC38 xenografts indicated a substantial increase in the Cxcl12-expressing CAFs lineage elicited by Spp1 overexpression (Fig. 4B–D). Moreover, SPP1 induced CXCL12 secretion in the CAFs, underscoring SPP1's role in modulating TME through CXCL12 (Fig. 4E). Initially, we investigated the potential formation of a positive feedback loop involving proteins secreted by CAFs in response to SPP1 secretion from tumor cells, which may improve the migratory capacity of tumor cells. We treated CAFs with or without SPP1 and applied their conditioned media to LoVo and HCT116 cells (Fig. 4F). Conditioned media from the SPP1-treated CAFs improved tumor cell migration and invasion (Fig. 4G and H). Thus, factors secreted by CAFs as a result of SPP1 induction facilitate tumor cell migration. Based on this premise, we hypothesized that CXCL12 secreted by CAFs in response to SPP1 could enhance colorectal cancer cell migration. We confirmed this hypothesis by introducing a CXCL12-neutralizing antibody into the conditioned media derived from SPP1-stimulated CAFs, which led to a substantial decrease in the observed effects (Fig. 4G and H). This result suggests that SPP1 secretion by the tumor cells induces CXCL12 release by the CAFs, thereby creating a positive feedback loop that augments tumor cell migration. Additionally, we investigated the direct effect of the purified CXCL12 protein on the colorectal cancer cells (Fig. 4I). Treatment with the purified CXCL12 protein substantially increased the migration, invasion, and wound healing capabilities of the colorectal cancer cells. Contrastingly, applying a CXCL12-neutralizing antibody markedly attenuated these effects (Fig. 4J and K). Western blot analysis showed that CXCL12 stimulation induced a molecular phenotype of the epithelial–mesenchymal transition genes, whereas the CXCL12 neutralizing antibody reversed these changes (Fig. 4L). SPP1, which is secreted by tumors, enhances CXCL12 secretion from the CAFs, thereby facilitating the migration and invasion of tumor cells. This observation raises the question of whether a positive feedback loop exists between SPP1 and CXCL12. Furthermore, TGFβ is a crucial regulatory factor in CAF activation, and existing literature indicates that CXCL12 secreted by the CAFs can improve TGFβ expression in the colorectal cancer cells (26). An analysis of the TCGA-colorectal cancer datasets showed that CXCL12 expression is markedly positively correlated with SPP1 and TGFβ (Fig. 4M). In LoVo and HCT116 cells, we found that CXCL12 facilitates SPP1 and TGFβ expression, whereas CXCL12-neutralizing antibodies reversed these upregulations (Fig. 4N). Additionally, SPP1/TGFβ secretion by the colorectal cancer cells was stimulated by CXCL12 but reversed by CXCL12-neutralizing antibodies (Fig. 4O). These results suggest a positive feedback loop through which cancer-secreted SPP1 stimulates the generation of a CXCL12-expressing CAFs lineage, whereas CXCL12 secreted by CAFs promotes colorectal cancer metastasis.
Figure 4.
SPP1 promotes colorectal cancer metastasis through a positive feedback loop mediated by CAF-secreted CXCL12. A, Mass spectrometry analyzed supernatants from SPP1-stimulated and unstimulated CAFs, showing fold changes in secreted proteins (SPP1/control). B, A bubble chart (Fig. 3C) displays commonly secreted protein levels in fibroblasts. C and D, Uniform Manifold Approximation and Projection (UMAP) plots and quantitative analysis reveal CXCL12 expression in fibroblasts within OE-SPP1 and vector groups. E, ELISA measured CXCL12 in CAF supernatants with/without SPP1 (1 µg/mL), n = 3. F, A flowchart shows CAF-conditioned medium’s (CM) impact on colorectal cancer (CRC) cell migration and invasion. G and H, Transwell and wound healing assays evaluated the effects of CAF-conditioned media or CXCL12-neutralizing antibody (100 ng/mL) on colorectal cancer cell migration and invasion (n = 3). I–K, Flowchart illustrating the effects of CXCL12 or neutralizing antibody treatment on the colorectal cancer cell migration and invasion, assessed via transwell and wound healing assays (n = 3). L, The effect of CXCL12 (100 ng/mL) or a neutralizing antibody (100 ng/mL) on the epithelial–mesenchymal transition markers expression in the colorectal cancer cells was analyzed using Western blotting (n = 3). M, Correlation analysis of CXCL12 with SPP1 and TGFB1 in the TCGA dataset. N and O, The effect of CXCL12 (100 ng/mL) or neutralizing antibody (100 ng/mL) on the SPP1 and TGFβ expression in the colorectal cancer cells was evaluated using Western blotting (N) or ELISA (O), n = 3. Results are presented as mean ± SEM. P values were calculated using a two-tailed unpaired Student t test (E), whereas one-way ANOVA was used for the other comparisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
SPP1 inhibits T-cell infiltration and cytotoxicity via CXCL12 secretion from the CAFs
CAFs promote the progression of CRLM via different mechanisms (27–29). Notably, SPP1 inhibits T-cell cytotoxicity (30). However, whether the inhibitory effect of the tumor-derived SPP1 on T-cell cytotoxicity is dependent on CAFs remains unclear. Using our established coculture model of the tumor cells and T cells, we found that in the absence of the CAFs, SPP1 overexpression in tumor cells partially inhibited T-cell–mediated cytotoxicity against the colorectal cancer cells (Supplementary Fig. S4A–S4C). Contrastingly, when CAFs were added to the upper chamber of the tumor–T-cell coculture system, T-cell cytotoxicity against the tumor cells was considerably reduced (Supplementary Fig. S4A–S4C). Furthermore, in a coculture system without CAFs, recombinant human SPP1 (rhSPP1) partially suppressed T-cell cytotoxicity in colorectal cancer cells (Supplementary Fig. S4D–S4F). However, when the CAFs were added to the upper chamber, CAFs exposed to purified SPP1 for 24 hours showed markedly impaired T-cell cytotoxicity against the tumor cells compared with the untreated CAFs (Supplementary Fig. S4D–S4F). Thus, CAFs may play a key role in amplifying SPP1’s effect in suppressing T-cell–mediated tumor killing. We used a PDO-CAF-T-cell coculture model to investigate SPP1’s role in inhibiting T-cell cytotoxicity (Fig. 5A). In the absence of CAFs, SPP1 overexpression in PDOs partially inhibited T-cell infiltration and cytotoxicity (Fig. 5B and C). Contrastingly, when CAFs were added to the upper chamber, T-cell infiltration and cytotoxicity against the SPP1-overexpressing organoids were substantially reduced (Fig. 5B and C). These experiments show that SPP1 suppresses T-cell infiltration and cytotoxicity against PDOs, with the CAFs further amplifying SPP1’s inhibitory effect on the T cells.
Figure 5.
SPP1 inhibits T-cell infiltration and cytotoxicity via CXCL12 secretion from CAFs. A, Schematic of the coculture system with PDOs, T cells, and CAFs. CRC, colorectal cancer; E:T, effector to target. B and C, Confocal microscopy assessing the effect of SPP1 overexpression on T-cell infiltration and cytotoxicity in PDOs with or without CAFs (n = 3). D and E, Impact of rhSPP1 (1 µg/mL) or CXCL12-neutralizing antibody (100 ng/mL) on T-cell infiltration and cytotoxicity in PDOs (n = 3). Results are presented as mean ± SEM. P values were determined using one-way ANOVA. *, P < 0.05; **, P < 0.01; ***, P < 0.001; n.s., nonsignificant. PI, propidium iodide.
CXCL12 can influence the formation of an immunosuppressive microenvironment, thereby inhibiting antitumor immune responses (31, 32). We used a coculture system comprising PDOs, T cells, and CAFs to explore whether SPP1 secreted by tumor cells inhibits T-cell infiltration and cytotoxicity by enhancing CXCL12 secretion from the CAFs. The presence of rhSPP1-treated CAFs markedly decreased T-cell infiltration and cytotoxicity against PDOs, an effect that was reversed using CXCL12-neutralizing antibodies (Fig. 5D and E). Moreover, the exogenous addition of CXCL12 notably decreased T-cell cytotoxicity against the tumor cells, an effect that was reversed by CXCL12-neutralizing antibodies (Supplementary Fig. S4G–S4I). Thus, the tumor cell–derived SPP1 inhibits T-cell infiltration and cytotoxicity by enhancing CXCL12 secretion from the CAFs.
SPP1 activates the β-catenin/HIF1α axis in the CAFs to drive CXCL12 secretion
We further elucidated the molecular mechanisms by which SPP1 regulates CXCL12 expression in the CAFs by analyzing the RNA-seq data from the CAFs derived from primary tumors and liver metastases of patients with colorectal cancer. Enrichment of the Wnt signaling pathway suggested that it plays a key role in CAFs derived from liver metastases (Supplementary Fig. S5A). Western blot analysis of the CAFs from rhSPP1-treated CRLM revealed substantial upregulation of β-catenin and HIF1α expression, with increased expression observed over time (Fig. 6A and B). Moreover, conditional medium from high liver metastatic cell lines upregulated β-catenin and HIF1α in the CAFs compared with the parental lines (Fig. 6C). Similarly, the medium from the parental cell lines overexpressing SPP1 plasmids also increased β-catenin and HIF1α levels (Fig. 6D). Conversely, the medium from the SPP1 knockdown high liver metastatic cell lines showed decreased expression (Fig. 6E). Thus, SPP1 may promote the upregulation of β-catenin and HIF1α protein levels in the CAFs.
Figure 6.
SPP1 activates the β-catenin/HIF1α axis in the CAFs to drive CXCL12 secretion. A, Western blotting assessed key signaling pathway in CAFs after 24 hours of SPP1 protein stimulation. B–E, β-catenin and HIF1α expressions were analyzed following SPP1 or conditioned medium treatments, including from SPP1-overexpressing or -knockdown cells. F–H, HIF1α degradation was evaluated with MSAB or si-CTNNB1 transfection after cycloheximide (CHX) treatment, and HIF1α levels were measured after MSAB (1 µmol/L) or MG132 (20 µmol/L) pretreatment. I and J, Coimmunoprecipitation examined the HIF1α and β-catenin interaction. K and L, Immunofluorescence and nuclear–cytoplasmic fractionation assays assessed HIF1α and β-catenin localization (n = 3). Scale bar, 25 μm. M–O, CXCL12 levels in conditioned media were measured after SPP1 (1 µg/mL) or MSAB treatments (24 hours). P, Correlation analysis of HIF1α and CXCL12 expression in 50 CAF samples using transcriptome data. Q, Dual-luciferase assays evaluated CXCL12 promoter activity (n = 3). R and S, T-cell migration and infiltration were analyzed with or without SPP1 protein or MSAB treatment, n = 3. Scale bar, 50 μm. Western blotting (A–J and L) and ELISA (M–O) were repeated three times, with data representative of three independent experiments. Results are presented as mean ± SEM. P values were determined by one-way ANOVA (M–O, R, and S) and two-tailed unpaired Student t test (F, G, and Q). *, P < 0.05; **, P < 0.01; ***, P < 0.001. R and S, Created with Figdraw.com.
β-Catenin, a key component of the Wnt signaling pathway, interacts with HIF1α to promote its stabilization and facilitate its translocation to the nucleus (33). Thus, we hypothesize that SPP1 improves HIF1α stability by upregulating β-catenin in the CAFs. Treatment with the β-catenin–specific inhibitor MSAB or β-catenin siRNA markedly reduced HIF1α protein levels without affecting its mRNA levels (Fig. 6F and G; Supplementary Fig. S5B–S5D). Furthermore, HIF1α can be degraded through the ubiquitin–proteasome pathway (34). Western blotting revealed a substantial reduction in the HIF1α levels after MSAB treatment, which was rescued by the proteasome inhibitor MG132 (Fig. 6H). Coimmunoprecipitation assays confirmed the interaction between β-catenin and HIF1α within CAFs (Fig. 6I and J). Taken together, these data suggest that β-catenin enhances the protein stability of HIF1α by inhibiting its ubiquitin–proteasome degradation pathway. HIF1α accumulation in the cytoplasm results in its translocation into the nucleus, in which it binds to hypoxia response elements in the promoters of the target genes, thereby regulating their transcriptional expression (35). Immunofluorescence and nuclear–cytoplasmic fractionation assays revealed that SPP1 treatment promoted β-catenin and HIF1α colocalization and nuclear translocation, which were inhibited by MSAB treatment (Fig. 6K and L). Building on previous research suggesting that HIF1α enhances CXCL12 expression in endothelial cells (36) and that SPP1 stimulation promotes CXCL12 secretion from CAFs (37), we investigated whether this effect is mediated by β-catenin and HIF1α upregulation. MSAB treatment effectively inhibited the SPP1-induced secretion of CXCL12 from the CAFs (Fig. 6M and N). Furthermore, silencing HIF1α decreased CXCL12 levels in the CAFs, further supporting its role in SPP1-induced CXCL12 upregulation (Fig. 6O; Supplementary Fig. S5E). Correlation analysis of the RNA-seq data obtained from 50 CAF samples derived from colorectal cancer showed a considerable positive correlation between the HIF1α and CXCL12 expression (Fig. 6P). This suggests that HIF1α regulates CXCL12 expression. We reviewed the existing literature that identified two potential HIF1α binding sites (HBS1 and HBS2) in the CXCL12 promoter (36) to confirm whether HIF1α directly interacts with the CXCL12 promoter to enhance transcription. Dual-luciferase assays revealed that overexpression of HIF1α-ΔODD activated the full-length CXCL12 promoter and the HBS1-containing promoter but not HBS2. This indicated that HIF1α specifically improves CXCL12 transcription via HBS1 (Fig. 6Q).
The SPP1–CD44 receptor complex promotes tumor cell stemness and improves tumor proliferation and metastasis (38, 39). We isolated fibroblasts from adjacent colorectal cancer tissues, primary tumors, liver metastases, and peripheral blood T cells to evaluate CD44 expression in various cell types using qRT-PCR. The transcription levels of CD44 were markedly higher in the fibroblasts than in the T cells, with the highest expression observed in the CAFs derived from liver metastases (Supplementary Fig. S5F). Thus, we hypothesized that the CAFs are the primary target cells that receive SPP1 signals from the colorectal cancer cells. To test this hypothesis, CD44 knockdown in liver metastasis–derived CAFs impaired the SPP1-induced upregulation of β-catenin and HIF1α (Supplementary Fig. S5G–S5J) and attenuated the rhSPP1-induced upregulation of β-catenin and HIF1α (Supplementary Fig. S5K). Additionally, CD44 knockdown reversed rhSPP1-induced HIF1α nuclear translocation (Supplementary Fig. S5L). This indicated that SPP1 activates the β-catenin/HIF1α axis in the CAFs through CD44 binding.
The CAF-restructured extracellular matrix can act as a physical barrier that obstructs immune cell infiltration, whereas CAFs are also known to create an immunosuppressive microenvironment by secreting cytokines, thereby impeding T-cell infiltration into the tumor tissues (40). In vitro barrier assays revealed that the SPP1-stimulated CAFs improved barrier function to migratory T cells, an effect attenuated by β-catenin inhibition using MSAB (Fig. 6R). In vitro infiltration assays further confirmed that SPP1-stimulated CAFs strongly inhibited T-cell migration, which was reversed by MSAB treatment (Fig. 6S). Collectively, our data suggest that SPP1 activates the β-catenin/HIF1α signaling pathway in the CAFs, resulting in increased CXCL12 secretion, which subsequently impedes T-cell infiltration and cytotoxic activity.
Talabostat mesylate reverses CRLM progression and synergizes with PD-1 blockade therapy
SPP1 secretion by the tumor cells promotes metastatic progression and activates CAF remodeling in the immunosuppressive TME by producing CXCL12. Phase IIa trials demonstrated that oral BXCL701 (talabostat, a CAF inhibitor), in combination with pembrolizumab, showed promising antitumor activity in late-line, refractory metastatic castration-resistant prostate cancer (41). Thus, we used an MC38 subcutaneous tumor model in C57BL/6L mice to investigate the roles of SPP1 and talabostat in mediating immunotherapy resistance (Fig. 7A). MC38 cells overexpressing SPP1 showed enhanced antitumor immune responses compared with the controls (Fig. 7B–D). Additionally, combining talabostat mesylate with anti–PD-1 therapy considerably attenuated resistance induced by SPP1 (Fig. 7B–D). Masson trichrome staining indicated increased collagen synthesis in the SPP1 overexpression group, which was effectively inhibited by talabostat mesylate (Supplementary Fig. S6A). Multiplexed IHC analysis showed a substantial upregulation of the CAFs activation marker αSMA in the SPP1 overexpression group, which was attenuated by talabostat mesylate (Supplementary Fig. S6B). Immunofluorescence showed reduced CD8+ T-cell infiltration in the SPP1 overexpression group; however, the combination of talabostat mesylate and anti–PD-1 therapy enhanced infiltration (Supplementary Fig. S6C). Further analysis revealed increased tumor cell proliferation in the SPP1 overexpression group, whereas talabostat mesylate combined with anti–PD-1 significantly promoted apoptosis (Supplementary Fig. S6D–S6F). ELISA results indicated that SPP1 overexpression suppressed IFNγ and granzyme B expression while promoting TGFβ expression, effects that were reversed by the combination therapy (Fig. 7E).
Figure 7.
Talabostat mesylate reverses CRLM progression and synergizes with PD-1/PD-L1 blockade therapy. A, Schematic of the subcutaneous mouse model established with MC38 cells. B–D, Representative tumor morphology, weight, and volume (n = 5 mice/group). E, ELISA measured the IFNγ, granzyme B, and TGFβ levels in the tumor tissues. F, Schematic representation of the colorectal cancer (CRC) PDX model in PBMC-reconstituted NOG mice. G–I, Tumor morphology, weight, and volume data (n = 5 mice/group). J, ELISA measured the SPP1, IFNγ, granzyme B, and TGFβ levels in the tumor tissues. K and L, Hematoxylin and eosin staining of metastatic livers from C57BL/6J mice implanted with MC38-GFP or MC38-SPP1 cells, treated with talabostat mesylate, showed liver weight and tumor burden (n = 5 mice/group). Scale bar, 1 mm. M and N, Flow cytometric analysis of the IFNγ+ CD8+ and GZMB+ CD8+ T-cell populations in liver metastases (n = 5 mice/group). Results are presented as mean ± SEM. P values were determined via one-way ANOVA (D, E, H, J, and L–N). *, P < 0.05; **, P < 0.01; ***, P < 0.001. aP, anti–PD-1 antibody; aP + i, anti–PD-1 antibody + talabostat mesylate.
PDX models are valuable because of their clinical relevance in maintaining tumor and stromal microenvironmental heterogeneity. Herein, we used a PDX model derived from the primary colorectal cancer tumors at F3 passage (Fig. 7F). Flow cytometry confirmed the successful reconstitution of immune profiles in the peripheral blood (Supplementary Fig. S6G). After 3 weeks of treatment, the tumors were excised for analysis. SPP1 overexpression exhibited a marked decrease in the antitumor immune response, whereas the combination of talabostat mesylate and anti–PD-1 substantially reduced SPP1-induced immune resistance (Fig. 7G–I). Immunofluorescence analysis revealed increased α-SMA expression and collagen deposition in the SPP1 overexpression group, which was suppressed by talabostat mesylate and anti–PD-1 treatment (Supplementary Fig. S6H). Furthermore, CD8+ T-cell infiltration was diminished in the SPP1 overexpression group but was substantially enhanced by the combination treatment (Supplementary Fig. S6H). Tumor cell proliferation was higher in the SPP1 overexpression group, whereas combination treatment promoted apoptosis (Supplementary Fig. S6H). ELISA confirmed successful SPP1 expression after intratumoral plasmid administration (Fig. 7J). Moreover, SPP1 overexpression decreased IFNγ and granzyme B levels while increasing TGFβ; however, talabostat mesylate and PD-1 inhibitors mitigated these effects (Fig. 7J). Furthermore, we established an MC38 spleen injection liver metastasis model to evaluate whether targeting CAFs could reverse SPP1-mediated CRLM. Treatment with talabostat mesylate markedly decreased the liver metastatic tumor burden induced by SPP1 overexpression (Fig. 7K and L). Talabostat mesylate reversed the immunosuppressive microenvironment induced by SPP1, increasing IFNγ+ CD8+ T cells and GZMB+ CD8+ T cells (Fig. 7M and N). Thus, targeting CAFs can counteract SPP1-mediated liver metastasis and the immunosuppressive liver microenvironment. Additionally, talabostat mesylate decreased the immunofluorescence intensity of αSMA and collagen and enhanced tumor cell apoptosis (Supplementary Fig. S6I–S6K). Thus, the CAFs-targeting inhibitor, talabostat mesylate, can effectively suppress SPP1-mediated CRLM and immunotherapy resistance.
Blocking the SPP1/CXCL12 axis alleviates immunosuppression in the liver microenvironment and augments the benefits of immunotherapy
Previous studies have highlighted the crucial role of SPP1 overexpression in tumor cells and macrophages in promoting immune resistance during primary tumor progression and liver metastasis in the colorectal cancer (11, 42). Our study mainly focused on the SPP1 secreted by highly metastatic liver tumor cells. We used clodronate liposomes to deplete macrophages in vivo to exclude potential interference from macrophage-derived SPP1 (43, 44). First, we established an intrasplenic injection model using MC38 cells that stably overexpressed SPP1 or vector (Supplementary Fig. S7A). In the Mac+ group, SPP1 overexpression promoted liver metastasis and resistance to immunotherapy. Even after macrophage depletion, SPP1 facilitated liver metastasis and immune resistance in the MC38 cells (Supplementary Fig. S7B and S7C). Flow cytometry confirmed a substantial decrease in macrophage numbers in the liver metastases of the Mac△ group, indicating effective macrophage depletion (Supplementary Fig. S7D). Overexpression of SPP1 markedly reduced the infiltration of IFNγ+ CD8+ T cells and GZMB+ CD8+ T cells within liver metastases (Supplementary Fig. S7E and S7F). Furthermore, the immunofluorescence intensity of αSMA and collagen was substantially elevated in the SPP1 overexpression group, whereas tumor cell apoptosis was markedly reduced (Supplementary Fig. S7G). These findings highlight the crucial role of tumor-derived SPP1 in promoting CRLM and immunotherapy resistance, independent of macrophage influence.
SPP1 secreted by highly metastatic liver colorectal cancer cells forms a positive feedback loop with CXCL12 secreted by the CAFs. Targeting this signaling axis has crucial therapeutic potential for reversing liver metastatic immune resistance. Current research indicates that the receptors for CXCL12 include CXCR4 and CXCR7 (45). CXCR4 is highly expressed in colorectal cancer tumors and immune cells, whereas CXCR7 is less prevalent in the tumor cells and absent in the CD8+ T cells (45, 46). Thus, we selected the CXCR4 antagonist, plerixafor, to inhibit CXCL12’s immunosuppressive function. The results of an intrasplenic injection model using MC38 cells revealed that Spp1 overexpression significantly promoted liver metastasis and decreased tumor sensitivity to αPD-1 (anti–PD-1 antibody) therapy. Contrastingly, combination treatment with αSPP1 (anti-SPP1 antibody) or plerixafor and αPD-1 considerably inhibited liver metastasis progression (Fig. 8A–D). The combination therapy enhanced infiltration of IFNγ+ CD8+ T cells and GZMB+ CD8+ T cells in liver metastases compared with anti–PD-1 monotherapy (Fig. 8E and F). Furthermore, the immunofluorescence intensity of αSMA and collagen was notably reduced in the combination group, whereas tumor cell apoptosis was markedly increased (Supplementary Fig. S7H). We developed a humanized immune system mouse model of cecal orthotopic liver metastasis to simulate the pathologic process of liver metastasis in patients with colorectal cancer more accurately (Fig. 8G). Flow cytometry confirmed the successful reconstitution of the immune profiles in the peripheral blood (Supplementary Fig. S7I). After 2 weeks of treatment, in vivo imaging showed markedly reduced liver tumor signals and decreased tumor burden in the combination therapy group (Fig. 8H and I). Histologic analysis and ELISA results revealed substantially fewer liver metastases and higher IFNγ levels in the dual-drug combination group compared with αPD-1 alone (Fig. 8J and K). Thus, targeting SPP1 and the CXCL12 receptor CXCR4 can effectively reverse immunotherapy resistance in CRLM.
Figure 8.
Blocking the SPP1/CXCL12 axis alleviates immunosuppression in the liver microenvironment and augments the benefits of immunotherapy. A, Flowchart of the intrasplenic injection model of liver metastasis using OE-SPP1 MC38 cells (i.s.v., intrasplenic injection; i.p., intraperitoneal injection). B–D, Representative tumor morphology, hematoxylin and eosin staining, liver weight, and tumor burden (n = 5 mice/group). Scale bar, 1 mm. E and F, Flow cytometric analysis of IFNγ+ CD8+ and GZMB+ CD8+ T cells in liver metastases (n = 5 mice/group). G, Flowchart of the cecal orthotopic injection model of liver metastasis in the NOG mice using HCT116-HM cells. H and I, Luciferase images and bioluminescence quantification of metastatic livers. J, Hematoxylin and eosin staining and the number of liver metastases (n = 5 mice/group). K, ELISA analysis of IFNγ levels in liver metastases (n = 5 mice/group). L–N, ELISA of SPP1 and CXCL12 in peripheral blood of responders (n = 25) and nonresponders (n = 12) in immunotherapy-treated colorectal cancer cohorts. O, Diagram of tumor-derived SPP1 activation of CAFs to promote immunotherapy resistance in CRLM. Data are presented as mean ± SEM. P values were determined using one-way ANOVA (C–F, and I–K) and two-tailed unpaired Student t test (L and M). *, P < 0.05; **, P < 0.01; ***, P < 0.001. O, Created in BioRender. Liu, F. (2025) https://BioRender.com/k7tx8am.
Additionally, we investigated the potential of plasma SPP1 levels as a biomarker for evaluating the response of patients with CRLM to immunotherapy. Baseline plasma samples were collected from 37 patients with CRLM undergoing immunotherapy and stratified into two groups: responders (25 patients with complete response or partial response) and nonresponders (12 patients with progressive disease or stable disease). ELISA showed markedly higher plasma SPP1 levels in the nonresponder group, suggesting an inverse correlation between SPP1 levels and immunotherapy effectiveness (Fig. 8L). Moreover, plasma CXCL12 concentrations were substantially higher in nonresponders than in responders. Additionally, a positive correlation was found between SPP1 and CXCL12 expression in the peripheral blood (Fig. 8M and N). Furthermore, in a cohort of 156 patients with gastric cancer receiving immunotherapy, increased SPP1 expression was linked to poor immune-related overall survival and immune-related progression-free survival (Supplementary Fig. S8A and S8B; Supplementary Table S3). Increased SPP1 expression was also noted in patients who were resistant to immunotherapy (Supplementary Fig. S8C–S8F). Thus, targeting the SPP1/CXCL12 axis alleviates immunosuppression in the liver microenvironment and improves the therapeutic efficacy of immunotherapy. Additionally, SPP1 may serve as a promising biomarker for predicting CRLM treatment outcomes (Fig. 8O).
Discussion
Colorectal cancer is a major global health burden, ranking as the third most commonly diagnosed cancer and second leading cause of cancer-related deaths globally (47, 48). The liver is the most common colorectal cancer metastatic site, and the emergence of liver metastases in patients with colorectal cancer poses a substantial challenge to clinical management. Immunotherapy targeting PD-1/PD-L1 is a revolutionary treatment option for colorectal cancer by harnessing the body’s immune system to target and destroy malignant cells. However, its efficacy remains limited for patients with liver metastases. Thus, identifying the factors that shape the immunosuppressive TME is crucial for driving liver metastasis for therapeutic sensitization. SPP1, a secretory protein, was upregulated in CRLM and contributed to liver metastases progression and immune evasion in these lesions. Herein, we identified that after interacting with CD44, SPP1 expression facilitates CAFs’ immunosuppressive function by activating a β-catenin/HIF1α/CXCL12 axis, thereby producing a TME that favors tumor growth and shields the cancer from immune destruction. Further analysis of clinical cohorts confirmed that plasma SPP1 levels were markedly increased in patients with distant metastasis and a poor response to immunotherapy was observed. This result is consistent with previous foundational research and underscores the SPP1’s critical role in regulating the progression of CRLM and resistance to immunotherapy.
Herein, highly metastatic colorectal cancer cells promoted SPP1 secretion, which in turn induced CXCL12 secretion by CAFs within the liver metastatic niche. CXCL12 facilitates further liver metastasis of tumor cells and upregulates TGFβ and SPP1 expression in the colorectal cancer cells, thereby promoting CXCL12+ CAF formation. This creates a positive feedback loop between the tumor cells and CAFs. Moreover, CXCL12 inhibits CD8+ T-cell infiltration and activation, contributing to the establishment of an immunosuppressive TME. SPP1 is a secreted glycoprotein that is closely related to various biological functions. SPP1 is overexpressed in several cancers and correlates with poor prognosis (49, 50). By integrating Hi-C, assay for transposase-accessible chromatin using sequencing, and RNA-seq technologies, SPP1’s enhancer–promoter loop gradually strengthened during colorectal cancer progression and liver metastasis, with expression levels being lower in normal tissues, increased in primary tumors, and further upregulated in liver metastatic lesions (51). In addition to tumor cells, aberrantly high SPP1 expression in immune cell subpopulations is markedly related to poor prognosis. Single-cell transcriptomics showed enrichment of SPP1+ macrophages in CRLM. These macrophages likely promote tumor progression by mediating angiogenesis and remodeling the metastatic microenvironment, with the degree of enrichment correlating with worse patient prognosis (11). Thus, SPP1, whether secreted by tumor cells or derived from macrophages, participates in tumor malignancy, underscoring its potential as a therapeutic target and prognostic biomarker.
CAFs’ pro-tumorigenic roles in tumor progression make them potential therapeutic targets, although there are several challenges in targeting these cells. However, with a growing understanding of CAF biology, interest in CAF-targeted therapies is increasing, as evidenced by multiple preclinical studies (52). FAP has emerged as a promising target, and preclinical data suggest that talabostat mesylate, a FAP-targeting inhibitor, improves the efficacy of immunotherapy in pancreatic ductal adenocarcinoma, Lewis lung carcinoma, and gastric cancer models (19, 53–55). A phase I trial of FAP-expressing stromal cells with the humanized monoclonal antibody F19 (sibrotuzumab or BIBH 1) demonstrated clinical safety and potential inhibition of tumor progression (56, 57). However, phase II trials on metastatic colorectal cancer failed to show sufficient efficacy, leading to limited responses (58). Additionally, targeting α-SMA or depleting Col1 in pancreatic cancer models could paradoxically promote tumor proliferation and metastasis (59, 60). These findings underscore the complexity of CAF-targeted strategies, highlighting the requirement for a deeper understanding of their molecular mechanisms and careful selection of target molecules. Furthermore, CAFs demonstrate pro- and antitumor activities depending on the subset (61, 62). Thus, the main challenge lies in identifying and selectively targeting CAF subsets that promote tumorigenesis. Herein, CXCL12, a key molecule secreted by CAFs in liver metastases, suppressed CD8+ T-cell infiltration, contributing to an immune-desert state within the tumor. Treatment with CXCL12 receptor antagonists markedly enhanced CD8+ T-cell infiltration, and when combined with αPD-1 therapy, it synergistically enhanced antitumor effects. Furthermore, SPP1 markedly upregulated CXCL12 secretion in CAFs, suggesting that dynamic monitoring of peripheral blood markers could help identify patients who may benefit from this therapeutic approach. However, the optimal clinical threshold for using SPP1 as a predictive biomarker of efficacy must be validated through multicenter prospective cohort studies to establish standardized clinical application guidelines.
Thus, our study underscores SPP1’s crucial role in driving liver metastasis and immunosuppressive TME in colorectal cancer. SPP1 contributes to immunotherapy resistance observed in patients with CRLM by manipulating the CAF-driven CXCL12 axis; therefore, targeting the SPP1–CAF–CXCL12 signaling pathway offers a promising strategy to overcome this resistance and improve outcomes in patients with colorectal cancer. As a biomarker and a therapeutic target, SPP1 holds great potential for developing precise treatments for colorectal cancer, paving the way for more effective management of liver metastases and enhanced responses to immunotherapy.
Supplementary Material
Figure S1 Verification of high-metastasis mouse model and the relationship between SPP1 and clinical features. A-F Results of bright field imaging, migration, and invasion assays for LoVo, LoVo-HM, HCT116 and HCT116-HM cells are presented, n = 3 biologically independent experiments. Scale bar, 50 μm. G-J Representative metastatic livers gross morphology from NOD/SCID mice intrasplenically implanted with LoVo, LoVo-HM, HCT116 or HCT116-HM cells (n = 5 per group). K Western blot analysis of SPP1 protein levels in LoVo, LoVo-HM, HCT116, and HCT116-HM cells, n = 3 biologically independent experiments. L Expression of SPP1 gene in different cancer and normal tissues of TIMER2.0 database. M, N Association of SPP1 expression with OS (M) and DFS (N) in CRC based on TCGA data. O-Q Correlation of SPP1 expression with pathological stage, N stage and T stage in CRC patients. All results are presented as mean ± SEM. The data in M, N were determined by Kaplan–Meier analysis with the log-rank test. P values were determined by two-tailed unpaired Student’s t test (C-F, H, J, P) and one-way ANOVA (O, Q). **P < 0.01, ***P < 0.001.
Figure S2 SPP1 promotes the occurrence of CRC metastasis and immunotherapy resistance. A The effect of overexpression of SPP1 on the migration and invasion of HCT116 cells was assessed using transwell assays, n = 3 biologically independent experiments. Scale bar, 200 μm. B The effect of SPP1 knockdown on the migration and invasion of HCT116-HM cells was assessed using transwell, n = 3 biologically independent experiments. Scale bar, 200 μm. C Wound healing assays were used to evaluate the effect of SPP1 overexpression (left) or knockdown (right) on the migration of HCT116 and HCT116-HM cells, n = 3 biologically independent experiments. D Western blot analysis was used to evaluate the effect of SPP1 overexpression or knockdown on the expression of EMT markers in LoVo and HCT116 cells. E A flowchart illustrating the establishment of an MC38-R string with acquired resistance to anti-PD-1 antibody. F, G Representative tumor gross morphology with tumor volume and tumor weight data (n = 5/group). H H&E analysis in tumor tissues from each group, with expression levels of α-SMA and CD8 evaluated by mIHC analysis. Scale bar, 100 μm. IOD: Integrated Optical Density. I The expression levels of SPP1 in MC38-WT and MC38-R were evaluated using western blot analysis. J qRT-PCR and western blot were used to assess the knockdown efficiency of si-Spp1 in MC38 cells. K C57BL/6 mice were subjected to subcutaneous co-injection of CAFs and MC38-R tumors. Mice were randomized to receive either si-NC or si-SPP1 via intratumoral injection, followed by intraperitoneal administration of anti-PD-1. L-M Representative tumor gross morphology with tumor weight and tumor volume data (n = 5/group). N, O ELISA was performed to assess the expression levels of granzyme B and IFN-γ in tumor tissues from each group. The expression of the control group was taken as 1, and the expression in the remaining groups was expressed as a multiple of the control group. P Western blot analysis of SPP1 protein levels in tumor tissues from each group. Q H&E analysis in tumor tissues from each group. Scale bar, 100 μm. R, S Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA and CD8 evaluated by IHC analysis. Scale bar, 100 μm. T TUNEL and Ki-67 staining in MC38 allografts to assess cell apoptosis and proliferation. Scale bar, 100 μm. Western blot experiments in (D, I, J) were repeated three times, the data are representative of three biologically independent experiments. All results are presented as mean ± SEM. P values were determined by two-tailed unpaired Student’s t test (A, C, F-H, L-P, R-T) and one-way ANOVA (B, C, J). *P < 0.05, **P < 0.01, ***P < 0.001.
Figure S3 SPP1 positively correlates with CAF marker expression and promotes FAP expression. A Correlation analysis of SPP1 with CAF marker genes from TCGA data. B Brightfield and fluorescence images of PT-PDOs #1 and LM-PDO #1 stably expressing GFP-SPP1. Green fluorescence indicates SPP1 expression. Scale bar, 500 μm. C Supernatants from PDOs overexpressing SPP1 or Vector were collected at various time points to stimulate CAFs, followed by western blot analysis of FAP level, n = 3 biologically independent experiments. CM, Conditional Medium.
Figure S4 CAF enhances SPP1-induced suppression of T-cell cytotoxicity. A Schematic representation of the tumor cell and T cell coculture system with or without CAFs. B, C After 24 hours of coculture with or without CAFs, propidium iodide staining was used to identify LoVo (B) or HCT116 (C) cells killed by T cells. Fluorescence microscopy assessed the effects of SPP1 overexpression on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. D Schematic representation of the tumor cell and T cell coculture system, with or without CAFs and SPP1 protein (1 µg/mL) treatment. E, F After 24 hours of coculture with or without CAFs, propidium iodide staining was used to identify LoVo (E) or HCT116 (F) cells killed by T cells. Fluorescence microscopy assessed the effects of SPP1 protein (1 µg/mL) treatment on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. G Schematic representation of the tumor cell and T cell coculture system, with or without CXCL12 protein (100 ng/mL) or CXCL12-neutralizing antibody (100 ng/mL) treatment. H, I After 24 hours of coculture, propidium iodide labeling identified LoVo and HCT116 cells killed by T cells. Fluorescence microscopy assessed the effects of CXCL12 protein (100 ng/mL) or CXCL12-neutralizing antibody (100 ng/mL) treatment on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. All results are presented as mean ± SEM. P values were determined by one-way ANOVA. ns, no significant difference, *P < 0.05, ***P < 0.001. PI, propidium iodide.
Figure S5 SPP1-CD44 interaction activates β-catenin/HIF-1α signaling in CAFs. A GSEA analysis of RNA-seq data from CAFs derived from primary tumors (PT-CAFs) and liver metastases (LM-CAFs) of CRC patients. B qRT-PCR and western blot assessed the knockdown efficiency of si-CTNNB1 in CAFs. C qRT-PCR was used to evaluate the mRNA expression of HIF1A in CAFs with CTNNB1 knockdown (left) or MSAB treatment (right), n = 3 biologically independent experiments. D Western blot analysis was performed to assess the protein expression of HIF-1α in CAFs with CTNNB1 knockdown or MSAB treatment. E qRT-PCR and western blot assessed the knockdown efficiency of si-HIF1A in CAFs. F qRT-PCR was used to measure CD44 mRNA levels in T cells and different kind of CAFs. NAF: normal-associated fibroblasts; PT: primary tumor; LM: liver metastasis, n = 3 biologically independent experiments. G qRT-PCR and western blot assessed the knockdown efficiency of si-CD44 in CAFs. H, I Western blot analysis was conducted to evaluate HIF-1α and β-catenin protein expression in control (si-NC) and CD44 knockdown CAFs, with or without conditional medium from HCT116-HM (H) or LOVO-HM (I). J CAF cells were transfected with si-NC or si-CD44 and subsequently treated with conditioned medium from SPP1-overexpressing HCT116 or LoVo cells. The expression levels of β-catenin and HIF-1α were then assessed by western blot. K Western blot analysis was conducted to evaluate HIF-1α and β-catenin protein expression in control (si-NC) and CD44 knockdown CAFs, with or without rhSPP1 protein (1 µg/mL) treatment. L Nuclear-cytoplasmic fractionation was performed to assess the nuclear translocation of HIF-1α in control (si-NC) and CD44 knockdown CAFs, with or without treatment with rhSPP1 protein (1 µg/mL). Western blot experiments in (A, C, F-J) were repeated three times, the data are representative of three biologically independent experiments. All results are presented as mean ± SEM. P values were determined by two-tailed unpaired Student’s t test (C) and one-way ANOVA (B, E-G). ns, no significant difference, **P < 0.01, ***P < 0.001.
Figure S6 Talabostat mesylate reverses the progression of CRLM and synergizes with PD-1 blockade therapy A-C Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA and CD8 evaluated by mIHC analysis. D, E Ki-67 and TUNEL staining were performed on PDOX tumor tissues to evaluate cell apoptosis and proliferation. F H&E analysis in tumor tissues from each group. G Flow cytometry validated the engraftment of hCD45+ cells 14 days post implantation in each group of mice (n = 3/group). H Collagen content was assessed using Masson’s trichrome staining, while expression levels of α-SMA and CD8 were evaluated through mIHC analysis. Ki-67 and TUNEL staining were performed on PDX tumor tissues to assess cell apoptosis and proliferation. H&E analysis was conducted on tumor tissues from each group. I, J Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA evaluated by IHC analysis. IOD: Integrated Optical Density. K TUNEL staining was performed to evaluate cell apoptosis. All results are presented as mean ± SEM. P values were determined by one-way ANOVA. **P < 0.01, ***P < 0.001.
Figure S7 Tumor-derived SPP1 promotes the progression of CRLM and immunotherapy resistance independent of macrophages. A Flowchart of the intrasplenic injection model of liver metastasis in mice using OE-SPP1 or OE-vector MC38 cells. i.s.v., intrasplenic injection; i.p., intraperitoneal injection. Clo: Clodronate-liposomes. B Representative H&E staining of metastatic livers from C57BL/6J mice intrasplenically implanted with SPP1-overexpressing or vector-overexpressing MC38 cells, followed by treatment with anti-PD-1 antibody. Scale bar, 1 mm. C Quantification of liver weight and tumor burden in liver metastases from the indicated mouse groups (n = 5 per group), with tumor burden assessed by the tumor-to-liver area ratio in tissue sections. D Flow cytometric analysis of macrophage populations in liver metastases from the indicated mouse groups (n = 5 per group). Mac+ represents the control group, MacΔ represents the macrophage depletion group. E, F Flow cytometric analysis of IFN-γ+ CD8+ and GZMB+ CD8+ T cell populations in liver metastases from the indicated mouse groups (n = 5 per group). G, H Collagen content was evaluated by Masson’s trichrome staining, α-SMA expression levels were assessed via IHC analysis, and cell apoptosis was measured by TUNEL staining. Integrated Optical Density (IOD) was used for quantification in all assays (n = 5 per group). Scale bar = 40 μm. I Flow cytometry validated the engraftment of hCD45+ cell 7 days post implantation in each group of mice (n = 3/group). All results are presented as mean ± SEM. P values were determined by one-way ANOVA. ns, no significant difference, *P < 0.05, **P < 0.01, ***P < 0.001.
Figure S8 High plasma SPP1 levels are associated with poor response to immunotherapy in GC patients. A, B Statistical analysis of irOS (A) and irPFS (B) in the SPP1 lower and SPP1 higher groups. C-F Statistical analysis of the best treatment efficacy results in the SPP1 lower and SPP1 higher groups. Data in scatter plots indicate mean ± SEM. ***P < 0.001. Log-rank test (A, B), Student’s t test (C-E), one-way ANOVA (F). CR, Complete Response; PR, Partial Response; PD, Progressive Disease; SD, Stable Disease; irOS, immune-related Overall Survival; irPFS, immune-related Progression-Free Survival; ORR, Objective Response Rate; DCR, Disease Control Rate.
Primers for RT-qPCR
Sequences of siRNAs
Plasma SPP1-stratified demographic features of 156 GC patients treated with ICBs.
Acknowledgments
We thank FigDraw (www.figdraw.com) and BioRender (www.biorender.com) for creating the flow charts and working model images and granting us the copyright ownership. We would like to thank Editage (www.editage.cn) for English language editing. The National Natural Science Foundation of China (no. 82403443), Beijing Natural Science Foundation (no. 7254308), National Natural Science Foundation of China (no. 82373252), the Science Foundation of Peking University Cancer Hospital (no. BJCH2024GG03), the China Postdoctoral Science Foundation (no. 2024T170034), the Natural Science Foundation of Inner Mongolia (2024MS08010), and the State Key Laboratory of Natural and Biomimetic Drugs (K202412) funded this research.
Footnotes
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Data Availability
Publicly available bulk RNA-seq data generated by others that were used by the authors in this study were obtained from the GEO under accession numbers GSE41568, GSE14297, and GSE128213. Single-cell RNA-seq and bulk RNA-seq data generated in this study have been deposited in GEO under accession numbers GSE298348 and GSE284019. The spatial transcriptome dataset reported in this study has been deposited in the Genome Sequence Archive-Human (https://ngdc.cncb.ac.cn/gsa-human/) with accession number HRA013308. Access to the raw data can be obtained by requesting the data following the Genome Sequence Archive guidelines. All requests will be reviewed and approved by the Data Access Committee. The expression data of SPP1 analyzed in this study were obtained from the TCGA dataset at https://portal.gdc.cancer.gov/, which includes data from multiple cancer types. All the other raw data generated in this study are available upon request from the corresponding author.
Authors’ Disclosures
S. Liu reports grants from Beijing Natural Science Foundation and China Postdoctoral Science Foundation during the conduct of the study. No disclosures were reported by the other authors.
Authors’ Contributions
S. Liu: Writing–original draft. Z. Zhang: Writing–original draft. Z. Wang: Writing–review and editing, revised and commented the manuscript. C. Liu: Data curation. G. Liang: Data curation. T. Xu: Data curation. Z. Li: Resources, clinical data collection. X. Duan: Resources. G. Xu: Resources. X. Feng: Resources. Q. Feng: Resources. Q. Wang: Resources. D. Han: Validation. C. Zhang: Validation. J. Li: Funding acquisition, writing–review and editing. L. Shen: Supervision, writing–review and editing.
References
- 1. Ciardiello F, Ciardiello D, Martini G, Napolitano S, Tabernero J, Cervantes A. Clinical management of metastatic colorectal cancer in the era of precision medicine. CA Cancer J Clin 2022;72:372–401. [DOI] [PubMed] [Google Scholar]
- 2. Amin MB, Greene FL, Edge SB, Compton CC, Gershenwald JE, Brookland RK, et al. The eighth edition AJCC cancer staging manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging. CA Cancer J Clin 2017;67:93–9. [DOI] [PubMed] [Google Scholar]
- 3. Van Gestel YR, de Hingh IHJT, Van Herk-Sukel MP, van Erning FN, Beerepoot LV, Wijsman JH, et al. Patterns of metachronous metastases after curative treatment of colorectal cancer. Cancer Epidemiol 2014;38:448–54. [DOI] [PubMed] [Google Scholar]
- 4. Väyrynen V, Wirta EV, Seppälä T, Sihvo E, Mecklin JP, Vasala K, et al. Incidence and management of patients with colorectal cancer and synchronous and metachronous colorectal metastases: a population-based study. BJS Open 2020;4:685–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Jooste V, Lepage C, Manfredi S, Bouvier AM. Trends in incidence of infrequent and frequent synchronous metastases from colorectal cancer. Dig Liver Dis 2024;57:83–8. [DOI] [PubMed] [Google Scholar]
- 6. Liu S, Zhang Z, Wang Z, Li J, Shen L. Genome-wide CRISPR screening identifies the pivotal role of ANKRD42 in colorectal cancer metastasis through EMT regulation. IUBMB Life 2024;76:803–19. [DOI] [PubMed] [Google Scholar]
- 7. Zhang Z, Zheng S, Lin Y, Sun J, Ding N, Chen J, et al. Genomics and prognosis analysis of epithelial-mesenchymal transition in colorectal cancer patients. BMC Cancer 2020;20:1135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Jin Z, Sinicrope FA. Mismatch repair-deficient colorectal cancer: building on checkpoint blockade. J Clin Oncol 2022;40:2735–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zhou H, Liu Z, Wang Y, Wen X, Amador EH, Yuan L, et al. Colorectal liver metastasis: molecular mechanism and interventional therapy. Signal Transduct Target Ther 2022;7:70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Wu Y, Yang S, Ma J, Chen Z, Song G, Rao D, et al. Spatiotemporal immune landscape of colorectal cancer liver metastasis at single-cell level. Cancer Discov 2022;12:134–53. [DOI] [PubMed] [Google Scholar]
- 11. Liu Y, Zhang Q, Xing B, Luo N, Gao R, Yu K, et al. Immune phenotypic linkage between colorectal cancer and liver metastasis. Cancer Cell 2022;40:424–37.e425. [DOI] [PubMed] [Google Scholar]
- 12. Yang S, Qian L, Li Z, Li Y, Bai J, Zheng B, et al. Integrated multi-omics landscape of liver metastases. Gastroenterology 2023;164:407–23.e417. [DOI] [PubMed] [Google Scholar]
- 13. Zhu Y, Yang J, Xu D, Gao XM, Zhang Z, Hsu JL, et al. Disruption of tumour-associated macrophage trafficking by the osteopontin-induced colony-stimulating factor-1 signalling sensitises hepatocellular carcinoma to anti-PD-L1 blockade. Gut 2019;68:1653–66. [DOI] [PubMed] [Google Scholar]
- 14. Chhabra Y, Weeraratna AT. Fibroblasts in cancer: unity in heterogeneity. Cell 2023;186:1580–609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Liu S, Zhang Z, Jiang L, Zhang M, Zhang C, Shen L. Claudin-18.2 mediated interaction of gastric cancer cells and cancer-associated fibroblasts drives tumor progression. Cell Commun Signal 2024;22:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Giguelay A, Turtoi E, Khelaf L, Tosato G, Dadi I, Chastel T, et al. The landscape of cancer-associated fibroblasts in colorectal cancer liver metastases. Theranostics 2022;12:7624–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Tsoumakidou M. The advent of immune stimulating CAFs in cancer. Nat Rev Cancer 2023;23:258–69. [DOI] [PubMed] [Google Scholar]
- 18. Arpinati L, Scherz-Shouval R. From gatekeepers to providers: regulation of immune functions by cancer-associated fibroblasts. Trends Cancer 2023;9:421–43. [DOI] [PubMed] [Google Scholar]
- 19. Jiang F, Jia K, Chen Y, Ji C, Chong X, Li Z, et al. ANO1-mediated inhibition of cancer ferroptosis confers immunotherapeutic resistance through recruiting cancer-associated fibroblasts. Adv Sci (Weinh) 2023;10:e2300881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Zhang Z, Liu S, Wang Z, Wang S, Jiang L, Wang X, et al. Influential upregulation of KCNE4: propelling cancer associated fibroblasts-driven colorectal cancer progression. Cancer Cell Int 2024;24:103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Lu M, Zessin AS, Glover W, Hsu DS. Activation of the mTOR pathway by oxaliplatin in the treatment of colorectal cancer liver metastasis. PLoS One 2017;12:e0169439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Stange DE, Engel F, Longerich T, Koo BK, Koch M, Delhomme N, et al. Expression of an ASCL2 related stem cell signature and IGF2 in colorectal cancer liver metastases with 11p15.5 gain. Gut 2010;59:1236–44. [DOI] [PubMed] [Google Scholar]
- 23. Okamoto T, Duverle D, Yaginuma K, Natsume Y, Yamanaka H, Kusama D, et al. Comparative analysis of patient-matched PDOs revealed a reduction in OLFM4-associated clusters in metastatic lesions in colorectal cancer. Stem Cell Rep 2021;16:954–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Ho W, Gomes-Santos I L, Aoki S, Datta M, Kawaguchi K, Talele NP, et al. Dendritic cell paucity in mismatch repair-proficient colorectal cancer liver metastases limits immune checkpoint blockade efficacy. Proc Natl Acad Sci U S A 2021;118:e2105323118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Efremova M, Rieder D, Klepsch V, Charoentong P, Finotello F, Hackl H, et al. Targeting immune checkpoints potentiates immunoediting and changes the dynamics of tumor evolution. Nat Commun 2018;9:32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Tan HX, Gong WZ, Zhou K, Xiao ZG, Hou FT, Huang T, et al. CXCR4/TGF-β1 mediated hepatic stellate cells differentiation into carcinoma-associated fibroblasts and promoted liver metastasis of Colon cancer. Cancer Biol Ther 2020;21:258–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Liu L, Ba Y, Yang S, Zuo A, Liu S, Zhang Y, et al. FOS-driven inflammatory CAFs promote colorectal cancer liver metastasis via the SFRP1-FGFR2-HIF1 axis. Theranostics 2025;15:4593–613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Li C, Chen T, Liu J, Wang Y, Zhang C, Guo L, et al. FGF19-induced inflammatory CAF promoted neutrophil extracellular trap formation in the liver metastasis of colorectal cancer. Adv Sci (Weinh) 2023;10:e2302613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Wang F, Long J, Li L, Wu ZX, Da TT, Wang XQ, et al. Single-cell and spatial transcriptome analysis reveals the cellular heterogeneity of liver metastatic colorectal cancer. Sci Adv 2023;9:eadf5464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Klement JD, Paschall AV, Redd PS, Ibrahim ML, Lu C, Yang D, et al. An osteopontin/CD44 immune checkpoint controls CD8+ T cell activation and tumor immune evasion. J Clin Invest 2018;128:5549–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Lu Y, Liu Y, Zuo X, Li G, Wang J, Liu J, et al. CXCL12(+) tumor-associated endothelial cells promote immune resistance in hepatocellular carcinoma [J]. J Hepatol, 2025, 82: 634–48. [DOI] [PubMed] [Google Scholar]
- 32. Garg B, Giri B, Modi S, Sethi, V, Castro, I, Umland, O, et al. , NFκB in pancreatic stellate cells reduces infiltration of tumors by cytotoxic T cells and killing of cancer cells, via up-regulation of CXCL12. Gastroenterology 2018;155:880–91.e888. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Dong S, Liang S, Cheng Z, Zhang X, Luo L, Li L, et al. ROS/PI3K/Akt and Wnt/β-catenin signalings activate HIF-1α-induced metabolic reprogramming to impart 5-fluorouracil resistance in colorectal cancer. J Exp Clin Cancer Res 2022;41:15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Yang L, Shi J, Zhong M, Sun P, Zhang X, Lian Z, et al. NXPH4 mediated by m(5)C contributes to the malignant characteristics of colorectal cancer via inhibiting HIF1A degradation. Cell Mol Biol Lett 2024;29:111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Lee JH, Son S, Ko Y, Lim H, Lee M, Kang MG, et al. Nidogen-1 suppresses cell proliferation, migration, and glycolysis via integrin β1-mediated HIF-1α downregulation in triple-negative breast cancer. Sci Rep 2025;15:10633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Ceradini DJ, Kulkarni AR, Callaghan MJ, Tepper OM, Bastidas N, Kleinman ME, et al. Progenitor cell trafficking is regulated by hypoxic gradients through HIF-1 induction of SDF-1. Nat Med 2004;10:858–64. [DOI] [PubMed] [Google Scholar]
- 37. Butti R, Nimma R, Kundu G, Bulbule A, Kumar TVS, Gunasekaran VP, et al. Tumor-derived osteopontin drives the resident fibroblast to myofibroblast differentiation through Twist1 to promote breast cancer progression. Oncogene 2021;40:2002–17. [DOI] [PubMed] [Google Scholar]
- 38. Cheng M, Liang G, Yin Z, Lin X, Sun Q, Liu Y. Immunosuppressive role of SPP1-CD44 in the tumor microenvironment of intrahepatic cholangiocarcinoma assessed by single-cell RNA sequencing. J Cancer Res Clin Oncol 2023;149:5497–512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Nallasamy P, Nimmakayala RK, Karmakar S, Leon F, Seshacharyulu P, Lakshmanan I, et al. Pancreatic tumor microenvironment factor promotes cancer stemness via SPP1-CD44 axis. Gastroenterology 2021;161:1998–2013.e1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Menche C, Schuhwerk H, Armstark I, Gupta P, Fuchs K, van Roey R, et al. ZEB1-mediated fibroblast polarization controls inflammation and sensitivity to immunotherapy in colorectal cancer. EMBO Rep 2024;25:3406–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Zhang J, Aggarwal R, Tagawa ST, Linch MD, Petrylak DP, Costin D, et al. BXCL701: first-in-class oral activator of systemic innate immunity combined with pembrolizumab, in patients with metastatic castration-resistant prostate cancer (mCRPC) of adenocarcinoma phenotype—phase 2a results. J Clin Oncol 2022;40(suppl):125. [Google Scholar]
- 42. Qi J, Sun H, Zhang Y, Wang Z, Xun Z, Li Z, et al. Single-cell and spatial analysis reveal interaction of FAP(+) fibroblasts and SPP1(+) macrophages in colorectal cancer. Nat Commun 2022;13:1742. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Zhang Z, Chen C, Yang F, Zeng YX, Sun P, Liu P, et al. Itaconate is a lysosomal inducer that promotes antibacterial innate immunity. Mol Cell 2022;82:2844–57.e2810. [DOI] [PubMed] [Google Scholar]
- 44. Cai J, Peng J, Zang X, Feng J, Li R, Ren P, et al. Mammary leukocyte-assisted nanoparticle transport enhances targeted milk trace mineral delivery. Adv Sci (Weinh) 2022;9:e2200841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Yang Y, Li J, Lei W, Wang H, Ni Y, Liu Y, et al. CXCL12-CXCR4/CXCR7 axis in cancer: from mechanisms to clinical applications. Int J Biol Sci 2023;19:3341–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Mezzapelle R, Leo M, Caprioglio F, Colley LS, Lamarca A, Sabatino L, et al. CXCR4/CXCL12 activities in the tumor microenvironment and implications for tumor immunotherapy. Cancers (Basel) 2022;14:2314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209–49. [DOI] [PubMed] [Google Scholar]
- 48. Zhang Z, Ju F, Chen F, Wu H, Chen J, Zhong J, et al. GDC-0326 enhances the effects of 5-Fu in colorectal cancer cells by inducing necroptotic death. Onco Targets Ther 2021;14:2519–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Kumari A, Kashyap D, Garg VK. Osteopontin in cancer. Adv Clin Chem 2024;118:87–110. [DOI] [PubMed] [Google Scholar]
- 50. Panda VK, Mishra B, Nath AN, Butti R, Yadav AS, Malhotra D, et al. Osteopontin: a key multifaceted regulator in tumor progression and immunomodulation. Biomedicines 2024;12:1527. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Xu X, Gan J, Gao Z, Li R, Huang D, Lin L, et al. 3D genome landscape of primary and metastatic colorectal carcinoma reveals the regulatory mechanism of tumorigenic and metastatic gene expression. Commun Biol 2025;8:365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Chen X, Song E. Turning foes to friends: targeting cancer-associated fibroblasts. Nat Rev Drug Discov 2019;18:99–115. [DOI] [PubMed] [Google Scholar]
- 53. Huang RSP, Decker B, Murugesan K, Hiemenz M, Mata DA, Li G, et al. Pan-cancer analysis of CD274 (PD-L1) mutations in 314,631 patient samples and subset correlation with PD-L1 protein expression. J Immunother Cancer 2021;9:e002558. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Lei M, Liu J, Gao Y, Dai W, Huang H, Jiang Q, et al. DPP inhibition enhances the efficacy of PD-1 blockade by remodeling the tumor microenvironment in lewis lung carcinoma model. Biomolecules 2024;14:391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Fitzgerald AA, Wang S, Agarwal V, Marcisak EF, Zuo A, Jablonski SA, et al. DPP inhibition alters the CXCR3 axis and enhances NK and CD8+ T cell infiltration to improve anti-PD1 efficacy in murine models of pancreatic ductal adenocarcinoma. J Immunother Cancer 2021;9:e002837. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Welt S, Divgi CR, Scott AM, Garin-Chesa P, Finn RD, Graham M, et al. Antibody targeting in metastatic colon cancer: a phase I study of monoclonal antibody F19 against a cell-surface protein of reactive tumor stromal fibroblasts. J Clin Oncol 1994;12:1193–203. [DOI] [PubMed] [Google Scholar]
- 57. Scott AM, Wiseman G, Welt S, Adjei A, Lee FT, Hopkins W, et al. A Phase I dose-escalation study of sibrotuzumab in patients with advanced or metastatic fibroblast activation protein-positive cancer. Clin Cancer Res 2003;9:1639–47. [PubMed] [Google Scholar]
- 58. Hofheinz RD, Al-Batran SE, Hartmann F, Hartung G, Jäger D, Renner C, et al. Stromal antigen targeting by a humanised monoclonal antibody: an early phase II trial of sibrotuzumab in patients with metastatic colorectal cancer. Onkologie 2003;26:44–8. [DOI] [PubMed] [Google Scholar]
- 59. Chen Y, Kim J, Yang S, Wang H, Wu CJ, Sugimoto H, et al. Type I collagen deletion in αSMA(+) myofibroblasts augments immune suppression and accelerates progression of pancreatic cancer. Cancer Cell 2021;39:548–65.e546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60. Özdemir BC, Pentcheva-Hoang T, Carstens J L, Zheng, X, Wu, CC, Simpson, TR, et al. Depletion of carcinoma-associated fibroblasts and fibrosis induces immunosuppression and accelerates pancreas cancer with reduced survival. Cancer Cell 2014;25:719–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61. Zhang Y, Liu G, Zeng Q, Wu W, Lei K, Zhang C, et al. CCL19-producing fibroblasts promote tertiary lymphoid structure formation enhancing anti-tumor IgG response in colorectal cancer liver metastasis. Cancer Cell 2024;42:1370–85.e1379. [DOI] [PubMed] [Google Scholar]
- 62. Chen S, Fan L, Lin Y, Qi Y, Xu C, Ge Q, et al. Bifidobacterium adolescentis orchestrates CD143(+) cancer-associated fibroblasts to suppress colorectal tumorigenesis by Wnt signaling-regulated GAS1. Cancer Commun (Lond) 2023;43:1027–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1 Verification of high-metastasis mouse model and the relationship between SPP1 and clinical features. A-F Results of bright field imaging, migration, and invasion assays for LoVo, LoVo-HM, HCT116 and HCT116-HM cells are presented, n = 3 biologically independent experiments. Scale bar, 50 μm. G-J Representative metastatic livers gross morphology from NOD/SCID mice intrasplenically implanted with LoVo, LoVo-HM, HCT116 or HCT116-HM cells (n = 5 per group). K Western blot analysis of SPP1 protein levels in LoVo, LoVo-HM, HCT116, and HCT116-HM cells, n = 3 biologically independent experiments. L Expression of SPP1 gene in different cancer and normal tissues of TIMER2.0 database. M, N Association of SPP1 expression with OS (M) and DFS (N) in CRC based on TCGA data. O-Q Correlation of SPP1 expression with pathological stage, N stage and T stage in CRC patients. All results are presented as mean ± SEM. The data in M, N were determined by Kaplan–Meier analysis with the log-rank test. P values were determined by two-tailed unpaired Student’s t test (C-F, H, J, P) and one-way ANOVA (O, Q). **P < 0.01, ***P < 0.001.
Figure S2 SPP1 promotes the occurrence of CRC metastasis and immunotherapy resistance. A The effect of overexpression of SPP1 on the migration and invasion of HCT116 cells was assessed using transwell assays, n = 3 biologically independent experiments. Scale bar, 200 μm. B The effect of SPP1 knockdown on the migration and invasion of HCT116-HM cells was assessed using transwell, n = 3 biologically independent experiments. Scale bar, 200 μm. C Wound healing assays were used to evaluate the effect of SPP1 overexpression (left) or knockdown (right) on the migration of HCT116 and HCT116-HM cells, n = 3 biologically independent experiments. D Western blot analysis was used to evaluate the effect of SPP1 overexpression or knockdown on the expression of EMT markers in LoVo and HCT116 cells. E A flowchart illustrating the establishment of an MC38-R string with acquired resistance to anti-PD-1 antibody. F, G Representative tumor gross morphology with tumor volume and tumor weight data (n = 5/group). H H&E analysis in tumor tissues from each group, with expression levels of α-SMA and CD8 evaluated by mIHC analysis. Scale bar, 100 μm. IOD: Integrated Optical Density. I The expression levels of SPP1 in MC38-WT and MC38-R were evaluated using western blot analysis. J qRT-PCR and western blot were used to assess the knockdown efficiency of si-Spp1 in MC38 cells. K C57BL/6 mice were subjected to subcutaneous co-injection of CAFs and MC38-R tumors. Mice were randomized to receive either si-NC or si-SPP1 via intratumoral injection, followed by intraperitoneal administration of anti-PD-1. L-M Representative tumor gross morphology with tumor weight and tumor volume data (n = 5/group). N, O ELISA was performed to assess the expression levels of granzyme B and IFN-γ in tumor tissues from each group. The expression of the control group was taken as 1, and the expression in the remaining groups was expressed as a multiple of the control group. P Western blot analysis of SPP1 protein levels in tumor tissues from each group. Q H&E analysis in tumor tissues from each group. Scale bar, 100 μm. R, S Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA and CD8 evaluated by IHC analysis. Scale bar, 100 μm. T TUNEL and Ki-67 staining in MC38 allografts to assess cell apoptosis and proliferation. Scale bar, 100 μm. Western blot experiments in (D, I, J) were repeated three times, the data are representative of three biologically independent experiments. All results are presented as mean ± SEM. P values were determined by two-tailed unpaired Student’s t test (A, C, F-H, L-P, R-T) and one-way ANOVA (B, C, J). *P < 0.05, **P < 0.01, ***P < 0.001.
Figure S3 SPP1 positively correlates with CAF marker expression and promotes FAP expression. A Correlation analysis of SPP1 with CAF marker genes from TCGA data. B Brightfield and fluorescence images of PT-PDOs #1 and LM-PDO #1 stably expressing GFP-SPP1. Green fluorescence indicates SPP1 expression. Scale bar, 500 μm. C Supernatants from PDOs overexpressing SPP1 or Vector were collected at various time points to stimulate CAFs, followed by western blot analysis of FAP level, n = 3 biologically independent experiments. CM, Conditional Medium.
Figure S4 CAF enhances SPP1-induced suppression of T-cell cytotoxicity. A Schematic representation of the tumor cell and T cell coculture system with or without CAFs. B, C After 24 hours of coculture with or without CAFs, propidium iodide staining was used to identify LoVo (B) or HCT116 (C) cells killed by T cells. Fluorescence microscopy assessed the effects of SPP1 overexpression on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. D Schematic representation of the tumor cell and T cell coculture system, with or without CAFs and SPP1 protein (1 µg/mL) treatment. E, F After 24 hours of coculture with or without CAFs, propidium iodide staining was used to identify LoVo (E) or HCT116 (F) cells killed by T cells. Fluorescence microscopy assessed the effects of SPP1 protein (1 µg/mL) treatment on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. G Schematic representation of the tumor cell and T cell coculture system, with or without CXCL12 protein (100 ng/mL) or CXCL12-neutralizing antibody (100 ng/mL) treatment. H, I After 24 hours of coculture, propidium iodide labeling identified LoVo and HCT116 cells killed by T cells. Fluorescence microscopy assessed the effects of CXCL12 protein (100 ng/mL) or CXCL12-neutralizing antibody (100 ng/mL) treatment on T-cell cytotoxicity, n = 3 biologically independent experiments. Scale bar, 100 μm. All results are presented as mean ± SEM. P values were determined by one-way ANOVA. ns, no significant difference, *P < 0.05, ***P < 0.001. PI, propidium iodide.
Figure S5 SPP1-CD44 interaction activates β-catenin/HIF-1α signaling in CAFs. A GSEA analysis of RNA-seq data from CAFs derived from primary tumors (PT-CAFs) and liver metastases (LM-CAFs) of CRC patients. B qRT-PCR and western blot assessed the knockdown efficiency of si-CTNNB1 in CAFs. C qRT-PCR was used to evaluate the mRNA expression of HIF1A in CAFs with CTNNB1 knockdown (left) or MSAB treatment (right), n = 3 biologically independent experiments. D Western blot analysis was performed to assess the protein expression of HIF-1α in CAFs with CTNNB1 knockdown or MSAB treatment. E qRT-PCR and western blot assessed the knockdown efficiency of si-HIF1A in CAFs. F qRT-PCR was used to measure CD44 mRNA levels in T cells and different kind of CAFs. NAF: normal-associated fibroblasts; PT: primary tumor; LM: liver metastasis, n = 3 biologically independent experiments. G qRT-PCR and western blot assessed the knockdown efficiency of si-CD44 in CAFs. H, I Western blot analysis was conducted to evaluate HIF-1α and β-catenin protein expression in control (si-NC) and CD44 knockdown CAFs, with or without conditional medium from HCT116-HM (H) or LOVO-HM (I). J CAF cells were transfected with si-NC or si-CD44 and subsequently treated with conditioned medium from SPP1-overexpressing HCT116 or LoVo cells. The expression levels of β-catenin and HIF-1α were then assessed by western blot. K Western blot analysis was conducted to evaluate HIF-1α and β-catenin protein expression in control (si-NC) and CD44 knockdown CAFs, with or without rhSPP1 protein (1 µg/mL) treatment. L Nuclear-cytoplasmic fractionation was performed to assess the nuclear translocation of HIF-1α in control (si-NC) and CD44 knockdown CAFs, with or without treatment with rhSPP1 protein (1 µg/mL). Western blot experiments in (A, C, F-J) were repeated three times, the data are representative of three biologically independent experiments. All results are presented as mean ± SEM. P values were determined by two-tailed unpaired Student’s t test (C) and one-way ANOVA (B, E-G). ns, no significant difference, **P < 0.01, ***P < 0.001.
Figure S6 Talabostat mesylate reverses the progression of CRLM and synergizes with PD-1 blockade therapy A-C Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA and CD8 evaluated by mIHC analysis. D, E Ki-67 and TUNEL staining were performed on PDOX tumor tissues to evaluate cell apoptosis and proliferation. F H&E analysis in tumor tissues from each group. G Flow cytometry validated the engraftment of hCD45+ cells 14 days post implantation in each group of mice (n = 3/group). H Collagen content was assessed using Masson’s trichrome staining, while expression levels of α-SMA and CD8 were evaluated through mIHC analysis. Ki-67 and TUNEL staining were performed on PDX tumor tissues to assess cell apoptosis and proliferation. H&E analysis was conducted on tumor tissues from each group. I, J Collagen content assessed by Masson’s trichrome staining, and expression levels of α-SMA evaluated by IHC analysis. IOD: Integrated Optical Density. K TUNEL staining was performed to evaluate cell apoptosis. All results are presented as mean ± SEM. P values were determined by one-way ANOVA. **P < 0.01, ***P < 0.001.
Figure S7 Tumor-derived SPP1 promotes the progression of CRLM and immunotherapy resistance independent of macrophages. A Flowchart of the intrasplenic injection model of liver metastasis in mice using OE-SPP1 or OE-vector MC38 cells. i.s.v., intrasplenic injection; i.p., intraperitoneal injection. Clo: Clodronate-liposomes. B Representative H&E staining of metastatic livers from C57BL/6J mice intrasplenically implanted with SPP1-overexpressing or vector-overexpressing MC38 cells, followed by treatment with anti-PD-1 antibody. Scale bar, 1 mm. C Quantification of liver weight and tumor burden in liver metastases from the indicated mouse groups (n = 5 per group), with tumor burden assessed by the tumor-to-liver area ratio in tissue sections. D Flow cytometric analysis of macrophage populations in liver metastases from the indicated mouse groups (n = 5 per group). Mac+ represents the control group, MacΔ represents the macrophage depletion group. E, F Flow cytometric analysis of IFN-γ+ CD8+ and GZMB+ CD8+ T cell populations in liver metastases from the indicated mouse groups (n = 5 per group). G, H Collagen content was evaluated by Masson’s trichrome staining, α-SMA expression levels were assessed via IHC analysis, and cell apoptosis was measured by TUNEL staining. Integrated Optical Density (IOD) was used for quantification in all assays (n = 5 per group). Scale bar = 40 μm. I Flow cytometry validated the engraftment of hCD45+ cell 7 days post implantation in each group of mice (n = 3/group). All results are presented as mean ± SEM. P values were determined by one-way ANOVA. ns, no significant difference, *P < 0.05, **P < 0.01, ***P < 0.001.
Figure S8 High plasma SPP1 levels are associated with poor response to immunotherapy in GC patients. A, B Statistical analysis of irOS (A) and irPFS (B) in the SPP1 lower and SPP1 higher groups. C-F Statistical analysis of the best treatment efficacy results in the SPP1 lower and SPP1 higher groups. Data in scatter plots indicate mean ± SEM. ***P < 0.001. Log-rank test (A, B), Student’s t test (C-E), one-way ANOVA (F). CR, Complete Response; PR, Partial Response; PD, Progressive Disease; SD, Stable Disease; irOS, immune-related Overall Survival; irPFS, immune-related Progression-Free Survival; ORR, Objective Response Rate; DCR, Disease Control Rate.
Primers for RT-qPCR
Sequences of siRNAs
Plasma SPP1-stratified demographic features of 156 GC patients treated with ICBs.
Data Availability Statement
Publicly available bulk RNA-seq data generated by others that were used by the authors in this study were obtained from the GEO under accession numbers GSE41568, GSE14297, and GSE128213. Single-cell RNA-seq and bulk RNA-seq data generated in this study have been deposited in GEO under accession numbers GSE298348 and GSE284019. The spatial transcriptome dataset reported in this study has been deposited in the Genome Sequence Archive-Human (https://ngdc.cncb.ac.cn/gsa-human/) with accession number HRA013308. Access to the raw data can be obtained by requesting the data following the Genome Sequence Archive guidelines. All requests will be reviewed and approved by the Data Access Committee. The expression data of SPP1 analyzed in this study were obtained from the TCGA dataset at https://portal.gdc.cancer.gov/, which includes data from multiple cancer types. All the other raw data generated in this study are available upon request from the corresponding author.