MMP3 overexpression enhances CAR-T cell infiltration and antitumor activity in a CAF-enriched solid tumor model

Abstract

Background

Chimeric antigen receptor (CAR) T cell therapy has shown remarkable success in hematologic malignancies but faces substantial challenges in solid tumors. One of the main obstacles is the extracellular matrix (ECM), which serves as the physical barrier that hinders T cell infiltration into tumor tissues.

Methods

We engineered CAR-T cells targeting mesothelin or B7H3 to co-express matrix metalloproteinase-3 (MMP3). We evaluated the effects of MMP3 overexpression on CAR-T cell proliferation, activation, cytotoxicity, and tumor infiltration using both in vitro Matrigel-based assays and in vivo xenograft and syngeneic models enriched with cancer-associated fibroblasts (CAFs).

Results

MMP3 overexpression did not impair CAR-T cell proliferation, activation, or cytotoxicity. However, it significantly enhanced their capacity to invade through ECM and improved tumor cell killing in vitro. In CAF-enriched xenograft models, MMP3-engineered CAR-T cells demonstrated superior tumor infiltration, expansion, and antitumor activity. Notably, MMP3 overexpression rescued the function of B7H3 CAR-T cells in the stringent CAF-enriched tumor microenvironment, while conventional CAR-T cells showed limited activity. Importantly, MMP3 overexpression also conferred potent antitumor activity in an immunocompetent mouse model, underscoring its therapeutic benefit in a more physiologically and clinically related setting.

Conclusions

These findings suggest that MMP3 engineering is a simple yet effective strategy to overcome stromal barriers and enhance the efficacy of CAR-T cell therapy in solid tumors.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Chimeric antigen receptor (CAR)-T cell therapy has achieved substantial success in hematologic cancers, but its efficacy in solid tumors is limited largely due to extracellular matrix (ECM) and cancer-associated fibroblast (CAF)-mediated stromal barriers. ECM remodeling enzymes such as matrix metalloproteinases (MMPs) have been proposed as potential tools to improve T cell infiltration, yet the therapeutic advantage of engineering CAR-T cells with a single ECM-modifying MMP molecule in physiologically relevant models has remained insufficiently defined.

WHAT THIS STUDY ADDS

• Engineering CAR-T cells to overexpress MMP3 significantly enhances their ability to degrade ECM, thereby enhancing T cell infiltration and antitumor efficacy—especially in CAF-enriched solid tumor models.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our findings propose a straightforward and translatable approach to overcome stromal barriers in CAF-enriched solid tumors by equipping CAR-T cells with ECM-degrading capabilities, thereby providing a theoretical basis for clinical application.

Introduction

Chimeric antigen receptor (CAR)-T cell therapy has shown remarkable success in treating hematological malignancies,¹ but its efficacy in solid tumors is limited by several challenges, notably the dense extracellular matrix (ECM) in the tumor microenvironment (TME), which restricts CAR-T cell infiltration and reduces therapeutic effectiveness.2,4 To address this, various strategies have been explored to degrade ECM components and facilitate CAR-T cell trafficking into tumors.5,7

Matrix metalloproteinases (MMPs) are a family of proteolytic enzymes that can degrade a wide range of ECM components, making them attractive tools for improving CAR-T cell infiltration.^{8 9} Previous studies have demonstrated that the co-expression of MMP9, MMP12, and heparinase enhances CAR-T cell infiltration and antitumor efficacy.¹⁰ However, engineering CAR-T cells to co-express multiple ECM-degrading enzymes presents significant drawbacks, including increased vector complexity, reduced transduction efficiency, and elevated risks of toxicity due to excessive ECM breakdown.¹¹ Therefore, identifying a single ECM-modifying MMP molecule that can enhance both CAR-T infiltration and therapeutic function would offer a more practical and effective approach.

Notably, a prior study showed that overexpression of MMP7 enhanced CAR-T cell infiltration into solid tumors but failed to improve tumor control. This effect may be due to MMP7-mediated cleavage of immune synapse components such as lymphocyte function–associated antigen-1 (LFA-1) and intercellular adhesion molecule (ICAM), which disrupts effective tumor cell engagement and impairs CAR-T function.12,14 These findings highlight the need for alternative ECM-modifying strategies that avoid disrupting immune synapse integrity. MMP3, a broad-spectrum metalloproteinase capable of degrading components such as laminin, fibronectin, and type III, IV and V collagen, has not been associated with damage to key synaptic molecules and thus may offer a better balance between ECM remodeling and functional preservation of CAR-T cells.^{15 16}

Moreover, while prior studies have typically employed simplified tumor models composed of tumor cells embedded in Matrigel to simulate ECM-enriched environments, such models mainly recapitulate basement membrane structure and lack other critical components of the TME.^{10 13 17} In particular, cancer-associated fibroblasts (CAFs) play a central role in ECM deposition and contribute to immunosuppression.18,20 To better mimic the clinical TME, we used a more physiologically relevant tumor model incorporating both Matrigel and CAFs, enabling a more rigorous evaluation of CAR-T cell performance.

In this study, we investigated the therapeutic potential of MMP3-overexpressing CAR-T cells in a CAF-enriched solid tumor model. We show that MMP3 enhances CAR-T cell infiltration and improves in vivo antitumor efficacy, supporting the feasibility of using a single ECM-degrading molecule to overcome key barriers in solid tumor therapy.

Materials and methods

Cell culture

Human non-small cell lung cancer cell lines, including H1299 and A549, were purchased from American Type Culture Collection (ATCC) and cultivated in Roswell Park Memorial Institute (RPMI)-1640 (Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich). H1299 cells were stably transduced with lentiviral vector encoding mesothelin (MSLN), Green fluorescent protein (GFP) and firefly luciferase; A549 cells were stably transduced with the lentiviral vector encoding GFP and firefly luciferase. Lewis lung cancer (LLC) cells were stably transduced with the lentiviral vector encoding human B7H3 (LLC-hB7H3). Human 293FT cells (ATCC) used for lentiviral or retroviral packaging were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS. Primary human T cells were isolated from the peripheral blood of donors with informed consent and maintained in X-VIVO15 (Lonza) supplemented with 50 ng/mL αCD3 (ACRO Biosystems) and 1 μg/mL αCD28 antibody (TLBiotechnology), 5 ng/mL interleukin (IL)-2 (Peprotech), 5 ng/mL IL-7 (Peprotech) and 5 ng/mL IL-15 (Peprotech) for 3 days. For murine T cells, splenocytes were isolated from 6–8–week–old female C57BL/6J mice (Beijing Vital River Laboratory Animal Technology). Murine T cells were sorted using a mouse CD3+T Cell Isolation Kit (Vazyme) according to the manufacturer’s instructions. Purified murine T cells were resuspended in complete RPMI-1640 medium supplemented with 10% heat-inactivated FBS, 50 µM β-mercaptoethanol, and recombinant mouse IL-2 (10 ng/mL) and activated for 24 hours using plate precoated antimouse CD3ε (5 µg/mL, clone 145–2 C11) and anti-mouse CD28 (2.5 µg/mL, clone 37.51) antibodies. All cells were cultured in a humidified incubator at 37°C with 5% CO₂.

Generation of the control and MMP3 overexpressing CAR

For lentiviral vector construction, MSLN and B7H3-targeting single-chain variable fragment (scFv) domain, CD8α hinge, CD28 transmembrane domain, 4-1BB and CD3ζ intracellular signaling domains were cloned into the pCDH lentiviral vector for the construct of MSLN CAR and B7H3 CAR as reported by our previous study.²¹ Full-length human MMP3 gene was subsequently incorporated into the MSLN CAR or B7H3 CAR lentiviral vector via a Thosea asigna virus 2A peptide (T2A) to generate MMP3-MSLN CAR or MMP3-B7H3 CAR.

For retroviral vector construction, B7H3-targeting scFv domain, murine CD8α hinge, murine CD28 transmembrane domain, murine 4-1BB and CD3ζ intracellular signaling domains were cloned into a retroviral expression vector to generate B7H3 murine CAR (B7H3 mCAR). Full-length murine Mmp3 gene was subsequently incorporated into the B7H3 mCAR retroviral vector to generate a MMP3-overexpressing retroviral vector (MMP3-B7H3 mCAR).

Lentiviral packaging and transduction of human T cells

To produce lentiviral particles, 293FT cells were co-transfected with above lentiviral vectors, the package plasmid and envelope plasmid using PEI (Polysciences) as the transfection reagent, in accordance with the manufacturer’s protocol. The culture media supernatant was harvested and filtered by 0.45 µm filter (Millipore) following 48-hour and 72-hour transfection and concentrated by ultracentrifugation. For lentiviral transduction of human T cells, concentrated virus and 8 μg/mL polybrene were added into the activated T cells. Of note, the transduction efficiency was evaluated by flow cytometry 72 hours post transduction.

Retroviral packaging and generation of murine CAR-T cells

To produce retroviral particles, 293FT cells were co-transfected with the retroviral vectors described above, and the pCL-Eco plasmid using PEI (Polysciences), according to the manufacturer’s protocol. The culture supernatant was harvested and filtered by a 0.45 µm filter (Millipore) following 48-hour and 72-hour transfection and concentrated by ultracentrifugation. The non-tissue culture-treated 48-well plates were precoated with 20 µg/mL Retronectin (Takara) overnight at 4°C and blocked with 2% bovine serum albumin (BSA) for 30 min at room temperature. To generate the B7H3 mCAR-T cells, the concentrated retrovirus encoding B7H3 mCAR was added into the Retronectin-coated plates and centrifuged at 2000×g for 2 hours at 32°C. The activated murine T cells and 10 μg/mL polybrene were then added into the plates and centrifuged at 300×g for 10 min at 32°C. On the next day, the T cells were transduced with the retrovirus encoding murine MMP3 using the same transfection conditions. Transduced murine T cells were cultured for 3 days to allow expansion with fresh murine T cell medium and IL-2 replenished daily.

Flow cytometry analysis

Single-cell suspensions were prepared and washed with cold phosphate-buffered saline (PBS) containing 1% BSA. For human T cell surface marker staining, the following antibodies, involving BUV395 anti-CD3, fluorescein isothiocyanate (FITC) anti-FLAG tag (FLAG) for transduction efficacy; phycoerythrin (PE)-CF594 anti-CD4, PE-Cy7 anti-CD8, PE anti-CD197 (CCR7), allophycocyanin (APC) anti-CD45RA, BUV737 anti-CD279(PD-1), APC anti-lymphocyte activation gene 3 (LAG-3), PE anti-T cell immunoglobulin and mucin-domain containing protein 3 (TIM-3) (BD Biosciences) for phenotype evaluation; Brilliant Violet 421 (BV421) anti-CD137 (BD Biosciences) for activation, were used for 30 min at 4°C in the dark. For intracellular staining, T cells were fixed and permeabilized using fixation and permeabilization buffer (BD Biosciences) for 20 min at room temperature in the dark. Next, T cell proliferation was evaluated by staining with Alexa Fluor 647 (AF647) anti-Ki67 antibody (BD Biosciences) for 30 min at 4 °C in the dark. The following antibodies were used to assess CAR surface expression on murine T cells, involving PE anti-CD3, APC anti-FLAG. The stained samples were acquired using fluorescence-activated cell sorting (FACS) Arial II (BD Biosciences) and visualized by FlowJo V.10 software.

Enzyme-linked immunosorbent assay

The MMP3 expression level in the culture supernatant of human and murine T cells was quantified with the human or mouse MMP-3 ELISA kit (Liankebio) following the manufacturer’s instructions. For measurement of the cytokine production level (interferon (IFN)-γ, IL-2, tumor necrosis factor (TNF)-α) of co-culture supernatant, 1×10⁵ effector cells were incubated with 5×10⁴ tumor cells for 24 hours. Subsequently, the co-culture supernatant was collected and analyzed using ELISA kit (R&D system).

Cytotoxicity assay

Carboxyfluorescein succinimidyl ester (CFSE)/propidium iodide (PI)-based cytotoxicity assay was performed as previously described by Zhang et al.²² Briefly, the tumor cells were stained with 5 µM CFSE (BD Biosciences) for 15 min in the 37°C incubator and subsequently cocultured with effector cells at various effector-to-target (E:T) ratios (diverging from 1:1 to 8:1) for 8 hours. Finally, the cells were harvested and stained with PI (BD Biosciences) at room temperature for 15 min and analyzed by an Accuri C6 Flow Cytometer (BD Biosciences,).

Transwell invasion and killing assay

Transwell chambers with 8.0 µm pore (Corning) were used to evaluate T cell invasion and killing capacity, according to the manufacturer’s instructions. For T cell invasion assay, 1×10⁵ T cells (MOCK, which refers to the untransduced control group, CAR-T or MMP3 expressing CAR-T cells) were seeded on the upper chamber precoated with Matrigel (Corning) diluted at 1:8. 500 µL X-VIVO15 medium supplemented with 2% FBS was added into the lower chamber as a chemoattractant. A same experimental design was conducted using the same Transwell chamber without any Matrigel precoating served as the control. Representative images of invaded T cells were captured using a microscope at the indicated time points. The absolute number of invaded T cells was quantified by flow cytometry analysis. The Matrigel invasion percentage was calculated as: (absolute number of cells invading through the Matrigel chamber membrane/ absolute number of migrating through the control insert membrane) × 100.

For the Transwell killing assay, as described above, 2.5×10⁵ T cells (MOCK, CAR-T or MMP3-expressing CAR-T cells) were seeded on the upper chamber precoated with Matrigel diluted at 1:8. 2×10⁵ tumor cells (H1299-MSLN and A549) were seeded on the bottom of the lower chamber of the Transwell. Representative fluorescence images of remaining tumor cells were captured using a microscope at 24 and 48 hours. After 48 hours, all cells in the lower chamber, including invading T cells and residual tumor cells, were harvested for further flow cytometric analysis to determine the proportion of invaded T cells and remaining tumor cells, respectively.

Primary CAF isolation and culture

The primary CAFs were isolated from patients with lung cancer undergoing surgical resection, with informed consent provided by all participants. The surgical specimens were minced into 2 mm fragments, enzymatically digested using tumor dissociation kit (Miltenyi Biotec), and processed using the 37C_h_TDK_2 program. After dissociation, the sample was filtered through a 70 µm cell strainer to remove any undigested tissues and centrifuged at 300 g for 5 min. The primary human-derived CAFs were maintained in DMEM supplemented with 10% FBS. After 48 hours, non-adherent cells were removed by washing, while adherent fibroblast-like cells were expanded and assessed by morphology.

Xenogeneic mouse models

The 6–8–week–old female immunodeficient NXG mice were purchased from HFK Bio-Technology and maintained under pathogen-free conditions. To generate a more physiologically relevant xenograft model, 2×10⁶ H1299-MSLN or A549 cells, and 2×10⁶ CAFs derived from patients with primary lung cancer were resuspended in 200 µL of a 1:1 mixture of PBS and Matrigel and subcutaneously (s.c.) injected into mice. On day 3 following tumor incubation, the mice were grouped according to the average radiance of tumor xenografts. 2×10⁶ MOCK, CAR-T and MMP3 expressing CAR-T cells were infused intravenously into the mice. Tumor burden was routinely monitored using bioluminescence imaging and calipers at indicated time points. No animals were excluded.

To evaluate the distribution of T cells in tumor tissues, a similar xenograft model was constructed as described above. Until tumor size reached approximately 100 mm³, the mice were randomly grouped and administrated intravenously with MOCK, CAR-T and MMP3 expressing CAR-T cells. The peripheral blood was collected, subjected to erythrocyte lysis, and conducted flow cytometry analysis to assess the number and antitumor effect of CAR-T cells. The tumor xenograft tissues were dissected, enzymatically dissociated into single-cell suspensions for further flow cytometry analysis on day 8 post infusion.

Syngeneic mouse models

The 6–8–week–old female immunocompetent C57BL/6J mice were purchased from Beijing Vital River Laboratory Animal Technology, and maintained under pathogen-free conditions. LLC-hB7H3 cells and mouse CAFs were resuspended in 200 µL of a 1:1 mixture of PBS and Matrigel and s.c. injected into mice. 3 days after tumor inoculation, the mice were intraperitoneally administered with 200 mg/kg cyclophosphamide (CTX) for lymphodepletion. The next day, MOCK, B7H3 mCAR-T and MMP3-B7H3 mCAR-T cells were infused intravenously. The tumor size and body weight were measured every 7 days after infusion.

H&E and immunohistochemical staining

Mouse tumors were dissected and fixed in 10% neutral buffered formalin for 72 hours at room temperature, dehydrated through a graded series of ethanol, cleared in xylene, embedded in paraffin, and sectioned at 5 µm thickness. The sections were deparaffinized in xylene and rehydrated through descending ethanol concentrations. H&E (Servicebio) and IHC (ZSGB-BIO) staining was performed according to the standard protocol. Representative images were captured using a light microscope (Olympus).

Statistical analysis

The statistical analysis and data visualization were performed using GraphPad Prism V.9.5.0, FlowJo V.10, and Adobe Illustrator 2020 software. Data were represented as mean±SEM (standard error of the mean) from triplicate independent experiments. Statistical significance was determined using unpaired two-tailed Student’s t-test, one-way or two-way analysis of variance with Tukey’s multiple comparisons test, or Mann-Whitney U test as appropriate. The statistical details of the individual experiment were described in the figure legends. Statistical significance was denoted as follows: not significant (ns), *p<0.05, **p<0.01, ***p<0.001.

Results

MMP3 overexpression does not compromise In vitro functionality of CAR-T

To evaluate the impact of MMP3 overexpression on CAR-T cell functionality, we constructed second-generation CARs targeting MSLN, with or without MMP3 co-expression (figure 1A). Although MMP3-MSLN CAR-T cells exhibited a slightly lower transduction efficiency compared with conventional MSLN CAR-T cells (45.6% vs 61.4%), the difference was not statistically significant (p>0.05) (figure 1B). ELISA assays confirmed robust MMP3 secretion by MMP3-MSLN CAR-T cells, whereas MOCK and MSLN CAR-T cells produced negligible levels (figure 1C).

Figure 1. Construction, transduction efficiency, proliferation, phenotype, in vitro antitumor activity and invasion ability of MMP3-overexpressing MSLN CAR-T cells. (A) Schematic diagram of second-generation MSLN CAR constructs. Left: control MSLN CAR lentiviral vector encoding an MSLN scFv, a CD8α hinge and CD28 transmembrane domain, and 4-1BB and CD3ζ intracellular signaling domains. Right: MMP3-overexpressing MSLN CAR construct incorporating MMP3 via a T2A peptide. (B) Representative flow cytometry histograms and quantification of transduction efficiency of MSLN CAR-T and MMP3-MSLN CAR-T cells 3 days after lentiviral transduction (n=3). (C) MMP3 secretion levels in 24-hour culture supernatants of MSLN CAR-T and MMP3-MSLN CAR-T cells, measured by ELISA (n=3). (D) Proliferation capacity assessed by absolute viable cell counts and viability percentages 3 days after transduction (n=3). (E) Ki67 expression levels determined by flow cytometry at 3 days post transduction (n=3). (F–I) Phenotypic analysis of MMP3-overexpressing MSLN CAR-T cells. (F) CD4/CD8 subset distribution (n=3); (G) expression of exhaustion markers (PD-1, LAG-3, TIM-3) (n=3); (H–I) T cell differentiation status (n=3). (J) Representative histograms and quantification of CD137 (4-1BB) expression following co-culture with H1299-MSLN cells, assessed by flow cytometry (n=3). (K) IFN-γ, IL-2, and TNF-α levels in culture supernatants after 24-hour co-culture with H1299-MSLN cells at an E:T ratio of 2:1, measured by ELISA (n=3). (L) Cytotoxicity of CAR-T cells against H1299-MSLN cells at indicated E:T ratios, evaluated by CFSE/PI assay (n=4). (M) Schematic diagram of the Matrigel invasion assay. CAR-T cells were seeded in Matrigel-coated upper chambers, with X-VIVO15 medium containing 2% FBS as a chemoattractant in the lower chambers. After 48 hours, invaded T cells in the lower chamber were collected and quantified by flow cytometry. (N) Representative brightfield images and quantification of Matrigel invasion by MMP3-overexpressing MSLN CAR-T cells at 24, 36, and 48 hours. GM6001, a broad-spectrum MMP inhibitor, was used to block enzymatic activity (n=3). Scale bar, 100 µm. (O) Schematic diagram of the Matrigel-based cytotoxicity assay. CAR-T cells seeded in the upper chamber invaded through Matrigel to target GFP-expressing tumor cells in the lower chamber. After 48 hours, invaded T cells and residual tumor cells were analyzed by confocal microscopy and flow cytometry. (P) Representative fluorescence images of GFP+tumor cells at 24 and 48 hours. Scale bar, 100 µm. (Q) Flow cytometry quantification of infiltrated T cells (CD3+) and remaining tumor cells (GFP+) at 48 hours. Percentages indicated the proportion of each population (n=3). All data are presented as mean±SEM from three independent experiments. Statistical significance was determined using unpaired two-tailed Student’s t-test (B), one-way ANOVA with Tukey’s multiple comparisons test (C–E, G, J, K, N and Q), or two-way ANOVA with Tukey’s multiple comparisons test (L). *p<0.05, **p<0.01, ***p<0.001. ANOVA, analysis of variance; CAR, chimeric antigen receptor; CFSE, carboxyfluorescein succinimidyl ester; E:T, effector-to-target; FBS, fetal bovine serum; GFP, green fluorescent protein; IFN, interferon; IL, interleukin; Ki67, Ki-67 (proliferation marker); LAG-3, lymphocyte activation gene 3; MOCK, untransduced control; MMP, matrix metalloproteinase; MSLN, mesothelin; ns, not significant; PD-1, programmed cell death protein-1; PI, propidium iodide; scFv, single-chain variable fragment; T2A, Thosea asigna virus 2A peptide; Tcm, central memory T cells; Teff, effector T cells; Tem, effector memory T cells; TIM-3, T cell immunoglobulin and mucin-domain containing protein 3; Tn, naive T cells; TNF, tumor necrosis factor; X-VIVO 15, X-VIVO 15 serum-free hematopoietic cell medium (Lonza).

Importantly, MMP3 overexpression did not alter CAR-T cell proliferation (figure 1D–E), CD4/CD8 ratios, exhaustion marker expression, or memory phenotype distribution (figure 1F–I). On stimulation with MSLN-positive tumor cells (H1299-MSLN), both MMP3-MSLN CAR-T cells and conventional CAR-T cells demonstrated comparable upregulation of activation markers (CD137) (figure 1J), cytokine production (IFN-γ, IL-2, TNF-α) (figure 1K), and antigen-specific cytotoxicity across various E:T ratios (figure 1L).

Similar results were obtained using B7H3-targeting CAR-T cells. Both CAR-T cells exhibited comparable CAR expression levels (43.0% vs 40.1%) (online supplemental Figure S1A and B), maintained proliferation, phenotype, activation, cytokine secretion, and cytolytic function, while producing abundant MMP3 (online supplemental Figure S1C-L).

These data collectively indicate that MMP3 overexpression does not impair the fundamental biological functions of CAR-T cells in vitro.

MMP3 overexpression enhances CAR-T invasion and cytotoxicity through ECM barriers

We next investigated whether MMP3 enhanced CAR-T cell infiltration through ECM components using Matrigel invasion assays (figure 1M). MMP3-MSLN CAR-T cells demonstrated significantly greater infiltration compared with MOCK and conventional MSLN CAR-T cells (figure 1N). This enhanced invasion was abrogated by GM6001, a broad-spectrum MMP inhibitor, confirming that the effect was MMP-dependent (figure 1N). Similar findings were observed for MMP3-B7H3 CAR-T cells (online supplemental Figure S2A).

To determine whether improved invasion led to enhanced tumor cytotoxicity, CAR-T cells were allowed to invade through Matrigel toward H1299-MSLN tumor cells seeded in the lower chamber (figure 1O). MMP3-MSLN CAR-T cells significantly reduced tumor cell viability, as evidenced by decreased GFP fluorescence and flow cytometric quantification (figure 1P,Q). This enhanced cytotoxicity was also abrogated by GM6001, further supporting that the effect was MMP-dependent (figure 1P,Q). MMP3-B7H3 CAR-T cells similarly exhibited superior tumor clearance in Matrigel invasion assay (online supplemental Figure S2B and C).

Thus, while MMP3 overexpression did not intrinsically enhance CAR-T cytotoxicity, it facilitated more efficient elimination of tumor cells by promoting infiltration through physical barriers.

MMP3-modified MSLN CAR-T cells exhibit superior antitumor efficacy in vivo

Recognizing the limitations of conventional Matrigel-based tumor models, we developed a more physiologically relevant xenograft model by co-injecting H1299-MSLN tumor cells with both Matrigel and CAFs¹⁷ (figure 2A). The inclusion of CAFs enriched the stromal components of the TME, enhancing ECM deposition and creating a more stringent barrier to immune cell infiltration.¹⁸ This CAF-enriched model enabled more accurate assessment of CAR-T cell performance under clinically relevant TME conditions.

Figure 2. In vivo antitumor efficacy and infiltration of MMP3-overexpressing MSLN CAR-T cells in a CAF-enriched xenograft model. (A) Schematic overview of CAF-enriched xenograft model establishment and CAR-T cell infusion. NXG mice were subcutaneously injected with H1299-MSLN cells mixed with CAFs (2×10⁶ cells each). Mice were infused with MOCK cells, MSLN CAR-T cells, or MMP3-MSLN CAR-T cells (2×10⁶ cells per mouse) on day 0. Tumor growth was monitored by BLI and caliper measurements. (B–C) Representative BLI images (B) and kinetics of BLI signal intensity (C) at indicated time points (n=5 per group). (D) Tumor volume curves calculated from caliper measurements (length×width²/2) (n=5 per group). (E) Body weight monitoring throughout the study (n=5 per group). (F) Schematic overview of immune infiltration analysis. Peripheral blood and tumor tissues were collected on days 4 and 8 post-CAR-T infusion, respectively, for flow cytometric analysis. (G–I) Flow cytometry analysis of CD3+T cells (G), CAR+T cells (H), and CD137+CAR T cells (I) in peripheral blood (n=3 per group). (J–L) Flow cytometry analysis of CD3+T cells (J), CAR+T cells (K), and CD137+CAR T cells (L) in tumor tissues (n=3 per group). (M) Representative H&E staining images of tumor tissues collected 8 days post CAR-T cell infusion. (N) Representative immunohistochemical staining images of CD3+T cell infiltration in tumor tissues collected 8 days post CAR-T cell infusion. All data are presented as mean±SEM. Statistical significance was determined using Mann-Whitney U test (C), two-way ANOVA with Tukey’s multiple comparisons test (D, E), and unpaired two-tailed Student’s t-test (G–L). *p<0.05, **p<0.01, ***p<0.001. ANOVA, analysis of variance; BLI, bioluminescence imaging; CAF, cancer-associated fibroblast; CAR, chimeric antigen receptor; FACS, fluorescence-activated cell sorting; IHC, immunohistochemistry; i.v., intravenous; MOCK, untransduced control; MMP, matrix metalloproteinase; MSLN, mesothelin; ns, not significant; s.c., subcutaneous.

In the CAF-enriched H1299-MSLN xenograft model, conventional MSLN CAR-T cells induced significant tumor regression compared with MOCK T cells, validating their antigen-specific antitumor activity in vivo (figure 2B–D). Notably, MMP3-MSLN CAR-T cells achieved more rapid and sustained tumor regression than conventional MSLN CAR-T cells (figure 2B–D), with no associated body weight loss (figure 2E).

To further examine the infiltration of CAR-T cells, additional mice were infused 8-day post-tumor inoculation (figure 2F). Flow cytometric analysis of peripheral blood revealed that treatment with MMP3-MSLN CAR-T cells resulted in significantly elevated levels of circulating CD3+and CAR+T cells, as well as a higher proportion of activated (CD137+) T cells (figure 2G–I). Consistently, analysis of tumor tissues demonstrated markedly increased infiltration of total and activated CAR-T cells in the MMP3-MSLN group compared with controls (figure 2J–L). Notably, immunohistochemical (IHC) staining further confirmed enhanced intratumoral T cell infiltration in mice receiving MMP3-MSLN CAR-T cells (figure 2M,N).

These results confirmed that MMP3 overexpression promotes MSLN CAR-T cell expansion, infiltration, and antitumor efficacy in a stroma-rich tumor model.

MMP3 overexpression rescues antitumor activity Of B7H3 CAR-T cells

To further assess the generalizability of MMP3 overexpression, we applied the same CAF-enriched mouse model approach using A549 cells expressing B7H3 (figure 3A). Notably, the proportion of tumor cells, CAFs, and Matrigel used to establish the B7H3 and MSLN xenografts was identical, ensuring comparable stromal barrier conditions across models.

Figure 3. In vivo antitumor efficacy and infiltration of MMP3-overexpressing B7H3 CAR-T cells in a CAF-enriched xenograft model. (A) Schematic overview of CAF-enriched xenograft model establishment and CAR-T cell infusion. NXG mice were subcutaneously injected with A549 cells mixed with CAFs (2×10⁶ cells each). Mice were infused with MOCK cells, B7H3 CAR-T cells, or MMP3-B7H3 CAR-T cells (2×10⁶ cells per mouse) on day 0. Tumor growth was monitored by BLI and caliper measurements. (B–C) Representative BLI images (B) and kinetics of BLI signal intensity (C) at indicated time points (n=5 per group). (D) Tumor volume curves calculated from caliper measurements (length×width²/2) (n=5 per group). (E) Body weight monitoring throughout the study (n=5 per group). (F) Schematic overview of immune infiltration analysis. Peripheral blood and tumor tissues were collected on days 4 and 8 post-CAR-T infusion, respectively, for flow cytometric analysis. (G–I) Flow cytometry analysis of CD3+T cells (G), CAR+T cells (H), and CD137+CAR T cells (I) in peripheral blood (n=3 per group). (J–L) Flow cytometry analysis of CD3+T cells (J), CAR+T cells (K), and CD137+CAR T cells (L) in tumor tissues (n=3 per group). (M) Representative H&E staining images of tumor tissues collected 8 days post CAR-T cell infusion. (N) Representative immunohistochemical staining images of CD3⁺ T cell infiltration in tumor tissues collected 8 days post CAR-T cell infusion. All data are presented as mean±SEM. Statistical significance was determined using Mann-Whitney U test (C), two-way ANOVA with Tukey’s multiple comparisons test (D, E), and unpaired two-tailed Student’s t-test (G–L). *p<0.05, **p<0.01, ***p<0.001. ANOVA, analysis of variance; BLI, bioluminescence imaging; CAF, cancer-associated fibroblast; CAR, chimeric antigen receptor; FACS, fluorescence-activated cell sorting; IHC, immunohistochemistry; i.v., intravenous; MOCK, untransduced control; MMP, matrix metalloproteinase; s.c., subcutaneous.

Despite these equivalent conditions, conventional B7H3 CAR-T cells exhibited limited antitumor activity, with tumor growth curves closely resembling those of MOCK T cell-treated mice (figure 3B–D). This suggests that B7H3 CAR-T cells possess intrinsically weaker antitumor activity in the context of a dense, CAF-enriched TME. Consequently, the B7H3 model represents a more stringent system for evaluating therapeutic enhancements compared with the MSLN model.

Importantly, MMP3-B7H3 CAR-T cells demonstrated significantly enhanced tumor control compared with both conventional B7H3 CAR-T cells and MOCK T cells (figure 3B–D), without inducing systemic toxicity (figure 3E). Flow cytometric analyses of peripheral blood and tumor tissues revealed a trend toward enhanced accumulation and activation of CAR-T cells in the MMP3-B7H3 group compared with controls (figure 3F–L). Importantly, immunohistochemical staining further corroborated that MMP3 overexpression substantially increased intratumoral T cell infiltration (figure 3M,N). These findings suggest that MMP3 overexpression can overcome stromal barriers and rescue B7H3 CAR-T cell efficacy even when baseline antitumor activity is insufficient in the CAF-enriched TME.

MMP3 overexpression enhances the invasive ability and antitumor efficacy of murine B7H3 CAR-T cells

To determine whether MMP3 overexpression improves murine CAR-T cell infiltration and tumor control, we generated murine B7H3 CAR-T cells (B7H3 mCAR) and murine MMP3 overexpressing B7H3 CAR-T (MMP3-B7H3 mCAR). Both CAR-T cells exhibited comparable CAR expression levels (figure 4A). MMP3-B7H3 mCAR-T cells produced substantially higher levels of mouse MMP3 as measured by ELISA, while MOCK and B7H3 mCAR-T cells showed minimal secretion (figure 4B). On LLC-hB7H3 cell stimulation, both CAR-T cells exhibited similar cytokine (IFN-γ) production (figure 4C), and antigen-specific cytotoxicity across multiple E:T ratios (figure 4D).

Figure 4. MMP3 overexpression enhances the invasive ability and antitumor efficacy of murine B7H3 CAR-T cells. (A) Representative flow cytometry histograms and quantification of surface CAR expression of B7H3 mCAR-T and MMP3-B7H3 mCAR-T cells 3 days after retrovirus transduction (n=3). (B) MMP3 secretion levels in 24-hour culture supernatants of B7H3 mCAR-T and MMP3-B7H3 mCAR-T cells, measured by ELISA (n=3). (C) IFN-γ levels in culture supernatants after 24-hour co-culture with LLC-hB7H3 cells at an E:T ratio of 2:1, measured by ELISA (n=3). (D) Cytotoxicity of both CAR-T cells against LLC-hB7H3 cells at indicated E:T ratios, evaluated by CFSE/PI assay (n=3). (E) Representative brightfield images and quantification of Matrigel invasion by MMP3-overexpressing B7H3 mCAR-T cells at 12 and 24 hours. Scale bar, 100 µm. (F) Schematic overview of syngeneic model establishment and CAR-T cell infusion. C57BL/6J mice were subcutaneously injected with LLC-hB7H3 (2×10⁶ cells per mouse). Mice were infused with MOCK cells, B7H3 mCAR-T cells, or MMP3-B7H3 mCAR-T cells (2×10⁶ cells per mouse) on day 0. (G) Tumor growth was monitored by caliper measurements (length×width²/2) (n=5 per group). (I) Tumor weights of individual mice (n=5 per group). (J) Body weight monitoring throughout the study (n=5 per group). All data are presented as mean±SEM. Statistical significance was determined using unpaired two-tailed Student’s t-test (A), one-way ANOVA with Tukey’s multiple comparisons test (B, C, E, I), or two-way ANOVA with Tukey’s multiple comparisons test (D, G, J). *p<0.05, **p<0.01, ***p<0.001. CAR, chimeric antigen receptor; CFSE, carboxyfluorescein succinimidyl ester; CTX, cyclophosphamide; E:T, effector-to-target; IFN, interferon; i.p., intraperitoneal; i.v., intravenous; LLC, Lewis lung cancer; mCAR, murine CAR; MOCK, untransduced control; MMP, matrix metalloproteinase; ns, not significant; PI, propidium iodide; s.c., subcutaneous.

In Matrigel invasion assays, MMP3-B7H3 mCAR-T cells displayed significantly enhanced invasive activity compared with the control B7H3 mCAR-T cells, which was blocked by the MMP inhibitor GM6001 (figure 4E).

To evaluate the in vivo antitumor activity of MMP3-overexpressing murine CAR-T cells, C57BL/6J mice were s.c. inoculated with LLC-hB7H3 cells and treated with CTX, followed by intravenous infusion of CAR-T cells (figure 4F). Compared with the control B7H3 mCAR-T cells, MMP3-B7H3 mCAR-T cells significantly inhibited tumor growth (figure 4G–I). No significant differences in body weight were observed among the groups, indicating good tolerability (figure 4J). These findings demonstrate that MMP3 overexpression facilitates ECM degradation and enhances the infiltration and antitumor activity of CAR-T cells in the immunocompetent setting.

In summary, our results highlight the critical role of stromal components in limiting CAR-T cell efficacy within solid tumors. The CAF-enriched models established here more accurately recapitulate the physical and immunosuppressive barriers of the clinical TME. Importantly, MMP3 overexpression substantially enhanced CAR-T cell expansion, infiltration, and tumor eradication ability across different CAR targets. Consistent results were observed in both immunocompetent and immunodeficient mouse models, underscoring the broad applicability of this strategy to overcome stromal resistance and improve therapeutic outcomes against solid tumors.

Discussion

In this study, we demonstrated that MMP3 overexpression significantly enhances the functionality of CAR-T cells against solid tumors. Specifically, MMP3 modification improved CAR-T cell infiltration and cytotoxicity in vitro and promoted superior tumor control in vivo within a CAF-enriched xenograft and syngeneic models. These findings establish MMP3 as an effective strategy to overcome the ECM-mediated barriers that limit CAR-T cell efficacy in solid tumors.

Previous studies have explored enhancing CAR-T cell infiltration by remodeling the tumor ECM.^{10 23} One prominent approach involved co-expression of multiple MMPs (eg, MMP9, MMP12) and heparanase, which effectively augmented CAR-T cell infiltration and antitumor activity.¹⁰ However, this method requires the simultaneous overexpression of several enzymes, complicating genetic engineering, potentially impairing T cell viability, and posing translational challenges. To simplify this strategy, prior research attempted single MMP overexpression, such as MMP7; while MMP7 improved CAR-T cell infiltration, it failed to enhance in vivo antitumor efficacy, possibly due to impaired immune synapse formation or insufficient ECM degradation breadth.^{12 13}

In contrast, our study identified MMP3 as a single modification that both promotes CAR-T cell infiltration and significantly enhances antitumor activity. Mechanistically, MMP3 possesses broad substrate specificity, degrading various ECM components including collagen, fibronectin, and laminin,^{24 25} which may facilitate deeper tumor penetration without compromising CAR-T cell functionality. This dual benefit highlights MMP3 as a more practical and efficient option compared with multi-enzyme strategies or MMP7 alone.

Importantly, we observed distinct responses between the MSLN and B7H3 CAR-T models. Although both xenograft models were established using the same proportions of tumor cells, CAFs, and Matrigel, conventional B7H3 CAR-T cells exhibited negligible antitumor efficacy, unlike their MSLN-targeting counterparts. This suggests that B7H3 CAR-T cells possess intrinsically weaker antitumor activity under stromal-rich conditions, rendering the B7H3 model more stringent for evaluating therapeutic interventions. Remarkably, even within this more challenging environment, MMP3-B7H3 CAR-T cells achieved significant tumor control, underscoring the robustness and broad applicability of MMP3 engineering.

Additionally, our study employed a CAF-enriched tumor model, which better recapitulates the complexity of the solid TME compared with conventional models using only Matrigel and tumor cells.^{10 26} CAFs are major producers of ECM proteins and immunosuppressive factors; thus, their inclusion creates a more physiologically relevant setting for evaluating CAR-T cell performance.^{26 27} The success of MMP3-modified CAR-T cells in this stringent model further supports the translational potential of this approach.

To determine whether MMP3 overexpression enhances CAR-T cell function under physiological immune conditions, we employed an immunocompetent syngeneic tumor model. Lymphodepletion prior to CAR-T cell infusion can facilitate engraftment and improve the expansion, persistence and performance of transferred CAR-T cells, which has been widely established as the standard practice in both preclinical and clinical settings.^{28 29} Therefore, we performed lymphodepletion before CAR-T cell infusion in the syngeneic mouse model. Importantly, MMP3 overexpressing B7H3 mCAR-T cells also exhibited superior tumor control ability in the syngeneic tumor model, which more faithfully recapitulates a clinically related TME, further reinforcing the therapeutic advantages of MMP3 engineering.

In conclusion, our findings establish MMP3 overexpression as a simple yet effective strategy to enhance CAR-T cell infiltration and antitumor activity in solid tumors. By simplifying the genetic modification process and maintaining T cell functionality, MMP3 engineering offers a promising direction for the development of next-generation CAR-T cell therapies targeting solid malignancies.

Data availability statement

No data are available.