Lactate Oxidase Nanocapsules Boost T Cell Immunity and Efficacy of Cancer Immunotherapy

Subhead 1: n(LOx) enhances the stability of lactate oxidase while retaining enzymatic activity.

Recombinant LOx from microorganisms possesses a short half-life and potential immunogenicity. To circumvent this limitation, we encapsulated LOx within a thin shell of polymer using our single-enzyme encapsulating platform (25–27). The encapsulation of LOx involves the enrichment of poly(ethylene glycol) methyl ether acrylate (PEGMA, monomer) and bisacrylamide (BIS, crosslinker) around LOx molecules via noncovalent interactions, and subsequent growth of a thin polymer shell around each LOx molecule by in situ polymerization (fig. S1). The encapsulation process leads to the formation of LOx nanocapsules, denoted hereinafter as n(LOx).

Compared with native LOx with an average size of about 10 nm, n(LOx) showed a spherical morphology with an average size of roughly 25 nm, as measured by transmission electron microscopy (Fig. 1A) and dynamic light scattering (fig. S2A). We further validated the encapsulation of LOx using protein gel electrophoresis, where n(LOx) remained in the stacking gel while native LOx migrated to the separating gel (Fig. 1B). Furthermore, the encapsulation changed the negative charge of native LOx (zeta potential, −7 mV) to a near-neutral zeta potential for n(LOx) (+2 mV) (fig. S2B). The n(LOx) retained more than 90% activity of the native LOx (fig. S2C); the Michaelis constant (K_m) was slightly increased from 0.16 mM to 0.55 mM (Fig. 1C), which is not expected to substantially affect the enzyme kinetics given the high concentrations of lactate in tumors (10 to 30 mM) (28). Moreover, tumors secrete various proteases, such as matrix metalloproteinases, degrading free proteins within the TME (29). To assess the proteolytic stability of n(LOx), we measured the activity of native LOx and n(LOx) in the presence of a model protease, trypsin (10 μg/mL). Unlike native LOx, which completely lost the activity within 20 minutes, n(LOx) retained more than 80% activity even after a 2-hour incubation (fig. S2D). The enhanced enzyme stability is essential to assure its continuous function in tumors.

The ability of n(LOx) to reduce lactate concentrations was first examined using various cancer cell lines, including MDA-MB-231, 4T1, and B16F10, which abundantly produced lactate in the culture media (fig. S2, E and F). Specifically, adding native LOx or n(LOx) (10 μg/mL) to MDA-MB-231 cell cultures reduced the lactate concentration in the culture media from 12 mM to 6 mM in 1 hour (Fig. 1D). Meanwhile, in the presence of H₂DCFDA (a fluorescent probe for H₂O₂), the cancer cells treated with native LOx or n(LOx) exhibited substantially higher fluorescent signals than phosphate-buffered saline (PBS)-treated controls, indicating increased concentrations of H₂O₂ (Fig. 1E and fig. S2, G and H). Consistent results were also obtained in tumor spheroids of MDA-MB-231 (Fig. 1F and fig. S3), confirming the ability of n(LOx) to reduce lactate and generate H₂O₂ in cancer cells.

The in vivo stability of n(LOx) was examined in a murine breast cancer model based on 4T1 cells. One week after engrafting luciferase-expressing 4T1 (4T1-Luc) cells into the mammary glands of BALB/c mice, native LOx or n(LOx) (0.5 mg/kg) labeled with Dylight 755 was injected into the tumors. Consistent with the enhanced stability of n(LOx) (fig. S2D), the fluorescent signals of n(LOx) were well retained within the tumor region 5 days after injection (Fig. 1G and fig. S4, A to C), whereas the fluorescent signals of native LOx rapidly became nearly undetectable at 24 hours (Fig. 1G). Further study suggested that n(LOx) exhibited a local half-life of 35 hours, which is longer than that of native LOx (about 6 hours), suggesting the favorable in vivo stability of n(LOx) (fig. S4D).

The ability of n(LOx) to reduce lactate concentrations and generate H₂O₂ was also examined in this breast cancer model. After intratumoral injection of native LOx or n(LOx), the average lactate concentration was reduced from 16 mM to 14 mM or 12 mM, respectively (Fig. 1H). Using Amplex Red Reagent as an H₂O₂ probe, the fluorescence intensity for the tumors treated with native LOx or n(LOx) was elevated by 2.5- and 4.6-fold, respectively, compared to the control group (Fig. 1I). It is also worth noting that, despite the reduced lactate concentrations, there was no change in tumor pyruvate values (fig. S4E). Overall, these results confirm the successful synthesis of n(LOx) that reduces lactate and generates H₂O₂ in tumors. n(LOx) also exhibited stability against the proteolysis in the TME, supporting its potential as a candidate cancer immunotherapy.

Subhead 2: n(LOx) suppresses the proliferation of T_reg cells, activates cytokine-producing T cells, and promotes tumor infiltration of T cells.

The ability of n(LOx) to suppress the proliferation of T_reg cells associated with elevated concentrations of lactate was examined using human peripheral blood mononuclear cells (PBMC). In the presence of 20 mM lactate, a lactate concentration similar to those in tumors, the relative proliferation rate of T_reg (FOXP3⁺CD25^hi) markedly increased by 22% (Fig. 2A), and the percentage of T_reg cells in the CD4⁺ T cell compartment increased from 7.2% to 8% (Fig. 2B). In order to dissect the roles of lactate reduction and H₂O₂ generation both induced by n(LOx), catalase (CAT), an enzyme that effectively decomposes H₂O₂ into H₂O and O₂, was employed alongside n(LOx) as n(LOx)/CAT to only retain lactate reduction by removing H₂O₂ produced by n(LOx). Addition of n(LOx) and n(LOx)/CAT to the culture media restored T_reg proliferation rates and the percentages of T_reg cells of total CD4⁺ T cells to that of control values, suggesting that the reduced T_reg population was mainly attributed to the reduced lactate concentrations, rather than H₂O₂ generation (Fig. 2, A and B, and fig. S5, A to C).

The ability of n(LOx) to increase the population of cytokine-producing T cells was examined in PBMC/MDA-MB-231 co-cultures, where MDA-MB-231 cells were first treated with n(LOx) or n(LOx)/CAT for 24 hours followed by co-culturing with PBMC for 6 hours. The culture media with n(LOx) showed higher populations of both IFN-γ and interleukin-2 (IL-2) producing CD4⁺ and CD8⁺ T cells, whereas their populations in the n(LOx)/CAT group were similar to those of the controls (Fig. 2, C to F). Consistently, IFN-γ-producing total T cells (about 8.2%) and IL-2-producing total T cells (about 3.7%) were comparable in the n(LOx)/CAT and control groups, whereas the n(LOx) group showed markedly higher populations of IFN-γ-producing T cells (31.2%) and IL-2-producing T cells (13%) (fig. S5, D to G). This study indicates a critical role for H₂O₂ in activation of T cells for cytokine production. For confirmation, H₂O₂ was directly added to the co-cultures; as expected, we observed increased populations of the cytokine-producing T cells, although the efficacy was lower than n(LOx), which generates H₂O₂ continuously and maintains relatively high concentrations of H₂O₂ in the cultures (fig. S5, D to I).

The ability of n(LOx) to promote tumor infiltration and activation of T cells for cytokine production was further examined in the murine 4T1 breast cancer model. One week after 4T1 engraftment, native LOx or n(LOx) (0.5 mg/kg) was injected into the tumors at day 7, 10, 13, and 16. At day 25 after tumor engraftment, tumor tissues were collected. Both native LOx and n(LOx) increased IFN-γ secretion, as well as the population of CD4⁺ and CD8⁺ T cells in tumors (Fig. 2G and fig. S6). Improved tumor infiltration of CD11b⁺ pro-inflammatory macrophages, as well as natural killer cells, were also observed in tumors with the treatment of n(LOx) (fig. S6). Overall, these results confirm that n(LOx) effectively suppressed the proliferation of T_reg cells under elevated concentrations of lactate, while promoting the tumor infiltration and activation of IFN-γ- and IL-2-producing T cells. In addition, the hepatic and renal functions (fig. S7, A to G) and hematoxylin and eosin (H&E) staining of the major organs (fig. S7H) of the mice treated with native LOx or n(LOx) showed no changes compared with controls, indicating favorable biocompatibility of n(LOx).

Subhead 3: n(LOx) boosts the efficacy of PD-L1 checkpoint blockades in a murine melanoma model.

The ability of n(LOx) to boost anti-tumor immunity and the efficacy of immune checkpoint blockade therapy was first studied in a murine melanoma model (B16F10). Mice were treated with native LOx or n(LOx) (0.5 mg/kg) by intratumoral injection after tumor engraftment (day 7, 9, and 11); anti-programmed cell death ligand 1 (aPD-L1, 100 μg/mouse) was injected intraperitoneally twice per week (Fig. 3A). Animals were carefully monitored and euthanized when the tumor volume reached 2000 mm³. Although aPD-L1 failed to control the tumor growth, n(LOx) markedly inhibited the tumor progression, and a combination of n(LOx) and aPD-L1 (n(LOx)/aPD-L1) further suppressed the tumor growth, resulting in dramatically reduced tumor volumes (< 100 mm³) after 30 days (Fig. 3B). Although native LOx also slowed down the tumor growth, its efficacy was lower than that of n(LOx), consistent with its short half-life and poor stability (fig. S8, A and B). Compared with aPD-L1, which slightly improved the median survival time (from 15 days to 17 days), n(LOx) extended the median survival time to 27 days, whereas mice treated with n(LOx)/aPD-L1 were alive after 30 days, suggesting n(LOx) boosted the effectiveness of aPD-L1 (Fig. 3C). The effectiveness of n(LOx) was further compared with CpG oligodeoxynucleotide, a commercial immunostimulator widely used in preclinical cancer models. We found that n(LOx) also outperformed CpG in suppressing tumor growth (fig. S9, A and B). To exclude the potential impact from the polymer shell of n(LOx), we prepared nanocapsules using a benign protein, bovine serum albumin (BSA). The treatment of BSA nanocapsules (n(BSA)) failed to control tumor growth (fig. S9C), confirming that the polymer shell did not contribute to the effectiveness of n(LOx). In addition, flow cytometry analysis indicate that n(LOx) mainly retained and functioned in the extracellular space of tumors (fig. S9D).

Compared with PBS and aPD-L1, n(LOx) and n(LOx)/aPD-L1 increased the populations of CD45⁺ tumor-infiltrating lymphocytes (TILs) from about 8% to 24.6% and 24.9% at day 14, respectively, and maintained the high populations until the end of the experiment at day 30 (Fig. 3D and fig. S10, A to E). Populations of dendritic cells, macrophages, and T cells, which are subtypes of TILs that collectively contribute to anti-tumor immunity, all increased after n(LOx) or n(LOx)/aPD-L1 treatment (Fig. 3, E to G). Furthermore, n(LOx) and n(LOx)/aPD-L1 treatment also boosted the populations of activated CD4⁺ T cells (CD69⁺/CD4⁺) from 37% (PBS) to 49% and 53% (Fig. 3H), as well as activated CD8⁺ T cells (CD69⁺/CD8⁺) from 44% (PBS) to 59% and 61%, respectively (Fig. 3I), accompanied by the increases in IFN-γ-producing CD4⁺ and CD8⁺ T cells (Fig. 3, J and K). The increased TILs and IFN-γ-producing T cells likely boosted the effectiveness of aPD-L1. In addition, depletion of CD4⁺ or CD8⁺ T cells compromised the efficacy of n(LOx) and resulted in faster tumor progression, confirming that the efficacy of n(LOx) mainly focuses on T cells (fig. S11, A to F).

Beyond boosting anti-tumor immunity and therapeutic efficacy of immune checkpoint blockade, n(LOx) may also induce abscopal effects. After intratumoral injection of n(LOx) in the right flank (0.5 mg/kg at day 7 and 9), a secondary tumor was engrafted in the left flank at day 10 (Fig. 3L). Compared to the PBS-treated control, the growth of the secondary tumors was suppressed in mice treated with n(LOx) (Fig. 3M and fig. S12A). Populations of TILs in the secondary tumors, including CD45⁺ TILs, CD3⁺ T cells, and CD11c⁺ dendritic cells, were increased (Fig. 3N and fig. S12, B to H). In addition, an increased population of M1 macrophages, but decreased populations of M2 macrophages and T_reg cells, were observed in the secondary tumors (Fig. 3N and fig. S12, E to K). These results suggest that n(LOx) treatment in the primary tumors induced anti-tumor immunity in the secondary tumors.

To further understand how n(LOx) boosted anti-tumor immunity, single-cell RNA sequencing (scRNA-seq) analysis of the tumors was performed, including the cancer cells, fibroblasts, and T cells (fig. S13A). Gene ontology (GO) analysis indicates that n(LOx) markedly upregulated the genes related to T cell activation and leukocyte migration in the tumors (Fig. 4A). To gain insight into the gene networks in responding to the n(LOx) treatment, we aligned our data to the main genes associated with the crosstalk to T cells. As shown in Fig. 4B, n(LOx) upregulated gene pathways that positively regulate the activation of T cells and alpha-beta T cells specifically. In addition, pathways related to cell chemotaxis and cell killing were upregulated after n(LOx) treatment (fig. S13B). The genes associated with cell killing and activation of immune cells were also upregulated in the T cells (Fig. 4C). Further analysis of T cells reveals that n(LOx) also induced increases in expression of genes related to the cytokine response, such as Ifng and Irf1 (Fig. 4D), as well as a skewed T-helper 1 (Th1) type immune response (Fig. 4E). Such upregulation boosted the cytotoxic responses of Th1 cells and stimulated the proliferation of cytotoxic T cells for tumor suppression. Multiple chemokine genes, such as Cxcl1, Cxcl3, and Cxcl10, were also stimulated in different cell types, including the activated fibroblasts and monocytes (Fig. 4F). Upregulation of chemokine-production genes likely promoted the recruitment of T cells and their migration to the tumors. These findings suggest that n(LOx) not only promoted T cell activation and anti-tumor responses (fig. S12C) at the transcriptional level, but also promoted the recruitment of circulating T cells to tumors.

Subhead 4: n(LOx) promotes T cell immunity and boosts the efficacy of PD-L1 checkpoint blockade in a humanized mice model.

The efficacy of n(LOx)/aPD-L1 in human cancers was examined using a humanized mouse model of triple-negative breast cancer (TNBC) (table S1). The humanized mouse model was engrafted with allogeneic human PBMC in NOD scid gamma (NSG) mice to reconstitute a functional human T cell system with reduced rejection of xenografted human tumors. PD-L1⁺ human breast cancer cells (MDA-MB-231) were injected into the subcutaneous space of the mammary glands in NSG mice. PBMC were intravenously injected 10 days after tumor engraftment. Two weeks after PBMC injection (day 0), n(LOx) was injected intratumorally at day 0 and 2 at 0.5 mg/kg; starting at day 2, aPD-L1 (100 μg/mice) was intraperitoneally administered twice per week (Fig. 5A). Similar to the treatment of melanoma (30), intratumoral injection of drugs has been widely applied to patients with TNBC to avoid off-target side effects (31, 32).

To dissect the role of lactate reduction and H₂O₂ generation, treatments of n(LOx)/CAT (lactate-concentration reduction only) and n(LOx) (lactate reduction and H₂O₂ generation) were examined. The abilities of n(LOx) to generate H₂O₂ and CAT to decompose H₂O₂ produced by n(LOx) in tumors were confirmed with in vivo imaging using a H₂O₂ probe (L-012). Tumors treated with n(LOx) showed intense luminescence signals, verifying that n(LOx) produced H₂O₂ in tumors. In comparison, no signal in the tumor site was detected in mice treated with n(LOx)/CAT due to the elimination of H₂O₂ by the CAT (Fig. 5B).

In terms of tumor progression, aPD-L1 alone failed to slow down the tumor growth; in contrast, n(LOx)/CAT, n(LOx), and n(LOx)/aPD-L1 suppressed tumor growth with increasing effectiveness (Fig. 5C). In addition, n(LOx)/CAT/aPD-L1 inhibited the tumor growth comparably with n(LOx) (Fig. 5C). This observation suggests that lactate reduction improved the efficacy of aPD-L1, and the improvement was further enhanced by H₂O₂ generation. As a result, mice treated with n(LOx)/aPD-L1 showed the best-controlled tumor volume (< 280 mm³) throughout the study (Fig. 5C and fig. S14A).

In terms of tumor infiltration of T cells, n(LOx) increased the population of CD4⁺ T cells from 3% to 7%, and the population of CD8⁺ T cells from 1% to 6%, respectively, in comparison with the PBS-treated control (Fig. 5D). In contrast, n(LOx)/CAT did not increase these populations of T cells. This suggests that H₂O₂ generation is the main driving force for T cell infiltration. Compared with PBS, aPD-L1 failed to increase the percentage of CD4⁺ and CD8⁺ T cells (Fig. 5E). In contrast, n(LOx)/aPD-L1 increased the percentage of CD4⁺ and CD8⁺ T cells to 11% and 9%, respectively, in comparison with n(LOx) (Fig. 5, D and E). Compared with n(LOx)/CAT, n(LOx)/CAT/aPD-L1 also improved T cell infiltration, demonstrating that lactate reduction also enhances aPD-L1 therapy, yet less effectively than the simultaneous lactate reduction and H₂O₂ generation.

Consistent with the enhanced tumor infiltration of T cells, n(LOx) and aPD-L1 also alleviated tumor immunosuppression associated with abundance of T_reg cells and exhausted cytotoxic T cells with PD-1 overexpression (33). Compared with PBS, n(LOx)/CAT and n(LOx) reduced the population of FOXP3⁺ T_reg cells from 15% to 8% and 5% (Fig. 5F), and the PD-1⁺ population in CD8⁺ T cells from 93% to 66% and 35% (Fig. 5G), respectively. Reduction in T_reg cell population was also observed after the treatment of n(LOx)/aPD-1 (fig. S14B). Compared with PBS, aPD-L1 reduced the PD1⁺ population in CD8⁺ T cells from 93% to 64%; n(LOx)/CAT/aPD-L1 and n(LOx)/aPD-L1 further reduced the population to 52% and 17%, respectively (Fig. 5G).

Beyond the alleviated immunosuppression and enhanced tumor infiltration of T cells, n(LOx) and aPD-L1 also improved the effector functions of T cells in humanized mice. aPD-L1 alone helped to recruit T cells in tumors but failed to increase IFN-γ production (Fig. 5H and fig. S14, C and D). In comparison, n(LOx) effectively increased the production of IFN-γ (Fig. 5H), IL-2 (fig. S15A) and tumor necrosis factor-alpha (TNF-α, Fig. 5I), which were further improved by aPD-L1. In addition, n(LOx)/CAT and n(LOx), as well as their combination with aPD-L1, reduced the production of an immunosuppressive cytokine, interleukin 4 (IL-4) (Fig. 5J and fig. S15B). n(LOx) also increased the population of CD44^highCD62L^low effector memory T cells (Fig. 5K), as well as CD44^highCD62L^high central memory T cells (fig. S15C), corresponding to a decrease in the percentage of CD44^lowCD62L^high naïve T cells (fig. S15D). Thus, these data show that combining aPD-L1 therapy with n(LOx) potently boosts T cell immunity and promotes control of tumors in humanized mice.

Study design

The aim of this study was to use nanocapsules of lactate oxidase (n(LOx) as a multifunctional activator to boost T cell immunity and the efficacy of cancer immunotherapy. To accomplish this, we evaluated the effectiveness of n(LOx) in removing lactate produced by cancer cells and activating effector T cells. The efficacy of n(LOx) in recruiting circulating T cells and activating T cells in tumors was examined in a murine breast cancer model (4T1). The therapeutic efficacy of n(LOx) in combination with aPD-L1 therapy was evaluated in a murine melanoma model and a humanized model of triple-negative breast cancer. Tumor growth measurement, survival curve analysis, RNA sequencing, and immune cell population analysis were performed to examine the effect of n(LOx). All animal experiments were performed in accordance with the protocols approved by the UCLA Chancellor’s Animal Research Committee (ARC). For each experiment, age-matched animals were randomly assigned to different groups. The determination of sample sizes was empirically based on published work. Investigators were blinded throughout the animal experiments, including treatment administration, tumor size measurement, flow cytometry staining analysis, and histological analysis. All details on sample size, statistical analysis, and experimental design are provided in data files S1 and S2, as well as in the figures and legends.

Materials, cells, and animals

Lactate oxidase (LOx) was purchased from Toyobo Inc. Poly (ethylene glycol) methyl ether acrylate (PEGMA, average Mn = 480), N,N’-methylenebisacrylamide (BIS), N-(3-aminopropyl) methacrylamide (APm), ammonium persulfate (APS), N,N,N’,N’-tetramethylethylenediamine (TEMED), sodium L-lactate, catalase (CAT) from bovine liver, L-lactate dehydrogenase (LDH), horseradish peroxidase (HRP), 3,3’,5,5’-tetramethylbenzidine (TMB), and agarose were purchased from Sigma Aldrich. Q-Sepharose Fast Flow resin was purchased from GE Healthcare. Phosphate buffered saline (PBS/10x), 2’,7’-dichlorodihydrofluorescein diacetate (H₂DCFDA), 6-chloromethyl (CM)-H₂DCFDA, and Dylight 755 NHS ester were purchased from Fisher Scientific. Ultra-LEAF anti-mouse PD-L1 (B7-H1) antibody was purchased from BioLegend. Durvalumab (anti-human PD-L1 antibody) was purchased from UCLA Health. CellTiter-Blue cell viability assay kit was purchased from Promega Corporation. Amplex Red reagent (10-acetyl-3,7-dihydroxyphenoxazine) was purchased from Cayman Chemical. Dulbecco’s Modified Eagle Medium (DMEM), Eagle’s Minimum Essential Medium (EMEM), Roswell Park Memorial Institute Medium 1640 (RPMI-1640), fetal bovine serum (FBS), penicillin/streptomycin (P/S), and all the other cell culture reagents were obtained from Fisher Scientific unless otherwise specified. Matrigel membrane matrix was purchased from Corning. All the antibodies are listed in the table S2. 4T1, B16F10, MDA-MB-231, HeLa, and U87 cells were purchased from American Type Culture Collection (ATCC). HOK-16B cells were a gift from Prof. Cun-yu Wang’s lab at UCLA. Luciferase-expressing 4T1 (4T1-Luc) cells were prepared through lentiviral transfection at UCLA core facility. PBMC were purchased from the UCLA/CFAR Virology Core Laboratory. All animals were treated in accordance with the Guide for Care and Use of Laboratory Animals under the protocol approved by the UCLA Chancellor’s ARC. Female BALB/c mice and male C56BL/6J mice were purchased from Jackson Laboratory. Female NOD/SCID/IL-2Rγnull (NSG) mice were purchased from Radiation Oncology Animal Core at UCLA. All animals were maintained at a pathogen-free facility at UCLA.

Instruments

Transmission electron microscopy (TEM) images were acquired with a 120 kV T12 cryo-electron microscope (FEI). Dynamic light scattering data were performed on a Zetasizer Nano ZS instrument (Malvern). Fluorescence intensity of different reagents was measured with an Infinite M200 Pro microplate reader (Tecan). UV-VIS absorption was measured with a One^C Microvolume Nanodrop (Thermo Scientific). Fluorescence imaging of cells and tissues was imaged with a Leica SP8 fluorescent microscope or a Leica TCS-SP8-SMD confocal microscope. Tissue sections with immunohistochemistry (IHC) staining were imaged with an Aperio ScanScope AT microscope. Optical imaging of live animals or organ samples was acquired with an IVIS Lumina II Imaging System (PerkinElmer) and analyzed with Living Image (PerkinElmer). The flow cytometry analysis was performed on the Fortessa (BD Biosciences) or Attune NxT flow cytometer (Thermo Fisher Scientific).

Cell culture

4T1 and 4T1-Luc cells were cultured in RPMI-1640, supplemented with 10% FBS and 1% P/S. B16F10 and MDA-MB-231 cells were cultured in DMEM, supplemented with 10% FBS and 1% P/S. HeLa and U87 cells were cultured in IMDM, supplemented with 10% FBS and 1% P/S. PBMC were cultured in RPMI-1640, supplemented with 10% FBS and 1% P/S. HOK-16B cells were cultured in Gibco Defined Keratinocyte-SFM media supplemented with defined keratinocyte-SFM growth supplement (1 mL per 500 mL media).

Synthesis and characterization of LOx nanocapsules (n(LOx))

LOx was dissolved in 1x PBS at a concentration of about 5 mg/mL and dialyzed overnight at 4 °C to remove the residual inorganic salts. The concentration of LOx was measured by its absorption at 450 nm. Subsequently, PEGMA (16000:1, n/n, PEGMA:LOx), APm (800:1, n/n, APm:LOx), and BIS (1600:1, n/n, BIS:LOx) were added to the LOx solution. The in situ polymerization was initiated with APS (800:1, n/n, APS:LOx) and TEMED (2:1, w/w, TEMED:APS). The reaction mixture was kept at 4 °C for 2 hours and dialyzed in HEPES buffer (25 mM, pH 7.5) overnight to remove the unreacted molecules. n(LOx) was purified with an anion exchange column (Q-Sepharose) using HEPES buffer (25 mM, pH 7.5) as the elution buffer. The purified n(LOx) samples were concentrated with Amicon Centrifugal Filter Units (MWCO, 30kDa) and the final concentration of n(LOx) was determined by its absorption at 450 nm. TEM, SDS page, and DLS were used to confirm the synthesis of n(LOx).

Enzyme kinetics assay

The activity of LOx samples was measured by the generation of hydrogen peroxide (H₂O₂) during the oxidation of lactate. The concentration of H₂O₂ was assessed by the color change of TMB in the presence of H₂O₂ and HRP. Briefly, sodium L-lactate (4.5 mg/mL), TMB (1.0 mg/mL), and HRP (0.005 mg/mL) were mixed in PBS. LOx samples (0.001 mg/mL) were then added to the reaction mixture to initiate the reaction. The absorbance at 650 nm was recorded continuously with a UV-VIS spectrophotometer (Beckman), and the linear portion of the absorption curve was used to calculate the reaction rate.

To measure the K_m and V_max of LOx samples, various concentrations of sodium L-lactate (0, 0.04882813, 0.09765625, 0.1953125, 0.390625, 0.78125, 1.5625, 3.125, 6.25, 12.5, and 25 mM) were added to the reaction mixture of TMB (1.0 mg/mL) and HRP (0.005 mg/mL). Subsequently, LOx samples (0.001 mg/ml) were added, and the absorbance of the reaction mixture at 650 nm was monitored for 2 minutes to derive the kinetics with the Michaelis-Menten equation.

Measurement of lactate removal by LOx samples in the cell culture media

4T1, B16F10, MDA-MB-231, HeLa, and U87 cells were seeded in 96-well plates at 5,000 cells/well. The culture media of these cells were collected daily for 5 days. The concentration of lactate was determined using a standard assay with LDH, as described in the Supplementary Materials. To measure the lactate removal by LOx samples, MDA-MB-231 cells were seeded in 96-well plates at 5,000 cells/well. Three days after seeding the cells, LOx samples (10 μg/mL) were added to the media. Samples from the mixture were collected every 20 minutes for 1 hour and the concentration of lactate was determined.

Detection of H₂O₂

H₂O₂ generation in the cell culture media was detected by H₂DCFDA, which can be converted to the highly fluorescent 2’,7’-dichlorofluorescein (DCF) upon oxidation by H₂O₂. 4T1, B16F10, MDA-MB-231, HeLa, and U87 cells were seeded at 5,000 cells/well and cultured for 3 days. H₂DCFDA (5 μM) was added to the culture media and incubated with the cells for 30 minutes before adding LOx samples at different concentrations (0, 0.25, 0.5, 1, and 2 μg/mL). The fluorescent signals of DCF (Ex. = 485 nm, Em. = 535 nm) at 30 minutes were measured with a plate reader.

To investigate the intracellular H₂O₂ generation in 3D tumor spheroids (Supplementary Materials), MDA-MB-231 tumor spheroids were prepared as indicated above. LOx samples were added to the culture media at 0.5 μg/mL. Four hours after the incubation, CM-H₂DCFDA (5 μM) was added to the culture media and incubated with the spheroids for 30 minutes. The spheroids were then washed with PBS and counterstained with DAPI. The 3D spheroids were observed under a fluorescent microscope.

PBMC studies

To analyze the population of T_reg cells, PBMC were first stimulated using T cell activation/expansion kit (BioLegend). PBMC were then labeled with CFSE cell division tracker (BioLegend) and cultured in low glucose RPMI-1640 (300 mg/L) media. PBMC were seeded in a 6-well plate at 2×10⁶ cells/well, and IL-2 (100 IU/mL) was added into the cell culture media. Lactic acid (20 mM) was added to the culture media before the addition of LOx samples at 0.05 μg/mL. To remove lactate without producing H₂O₂, CAT (5 μg/mL) was added simultaneously with LOx samples. The cells were incubated overnight the population of T_reg cells was analyzed with flow cytometry. The flow cytometry gating strategies are summarized in fig. S16.

To analyze the IFN-γ and IL-2 production by T cells, MDA-MB-231 cells were seeded in a 6-well plate at 1×10⁵ cells/well and cultured for 3 days before adding LOx samples (2 μg/mL). For the H₂O₂ group, H₂O₂ was added to MDA-MB-231 cells at a final concentration of 250 μM. The cells were incubated with LOx samples for 24 hours, followed by adding 2×10⁶ PBMC in each well. PBMC were then incubated with the LOx-treated MDA-MB-231 cells for 6 hours. The T cell population and IFN-γ/IL-2 production were then analyzed with flow cytometry.

4T1 mouse model

4T1 cells (5×10⁵ cells) in RPMI-1640/Matrigel (1:1, v/v) were subcutaneously injected into the second mammary gland of female BALB/c mice. Seven days after tumor injection, 18 mice were randomly divided into 3 groups (n=6) and treated with PBS, native LOx (0.5 mg/kg), or n(LOx) (0.5 mg/kg) intratumorally every 3 days at day 7, 10, 13, and 16 after tumor engraftment. At day 25 after tumor engraftment, tumor tissues were collected for histological analysis. Through the study, the body weight of all animals was monitored. For imaging purposes, 4T1-Luc cells (5×10⁵ cells) in RPMI-1640/Matrigel (1:1, v/v) were used, and the details are described in the Supplementary Materials.

To detect the concentrations of lactate, pyruvate, and H₂O₂, tumor samples were immersed in cold PBS (100 mg tissue per 100 μL PBS) and homogenized with a Bead Mill 24 homogenizer. The homogenate was centrifuged at 20,000 g for 15 minutes, and the supernatant was collected. To measure the abundance of H₂O₂ in tumors, Amplex Red reagent was used as an indicator, and the supernatant was diluted by 10-fold. The reaction mixture was prepared in a 96-well plate with 20 μL supernatant and 80 μL Amplex Red solution (50 μM Amplex Red + 0.1 U/mL HRP). After 30 minute incubation, the fluorescent signals (Ex. = 530 nm, Em. = 590 nm) were analyzed with a plate reader. The concentrations of pyruvate in the supernatant were measured with a pyruvate assay kit (Cayman Chemical). To measure the liver and renal functions after the treatment of native LOx or n(LOx), about 200 μL serum samples were collected from each animal and sent to IDEXX Bioresearch for screening (test case # 50627).

B16F10 mouse model

B16F10 cells (5×10⁵ cells) in DMEM/Matrigel (1:1, v/v) were subcutaneously injected into the right flank of male C57BL/6J mice. Seven days after tumor injection, mice were randomly divided into 5 groups (n=6 to 8 mice per group). Mice in the native LOx or n(LOx) group were intratumorally injected with native LOx (0.5 mg/kg) or n(LOx) (0.5 mg/kg), respectively, at days 7, 9, and 11. Mice in the aPD-L1 or n(LOx)/aPD-L1 group received intraperitoneal injections of aPD-L1 (100 μg/mouse) twice per week starting on day 11. Tumor growth was measured with a digital caliper, and the tumor volume (mm³) was calculated as (length x width²)/2. The animals were euthanized and tumor samples were collected when the tumor volume reached about 2000 mm³ or at day 30. Single-cell suspensions were then obtained from the tumor tissues and the tumor-infiltrating lymphocytes (TILs) were analyzed with flow cytometry. To analyze the TILs at an earlier time point (day 14), mice were randomly divided into 4 groups (n=6). Mice in the n(LOx) group were intratumorally injected with native LOx (0.5 mg/kg) or n(LOx) (0.5 mg/kg), respectively, at day 7, 9, and 11. Mice in the aPD-L1 or n(LOx)/aPD-L1 group received the intraperitoneal injection of aPD-L1 (100 μg/mouse) at day 11. All the mice were euthanized at day 14 and the TILs were analyzed with flow cytometry. To engraft the secondary tumors, B16F10 cells (5×10⁵ cells) in DMEM/Matrigel (1:1, v/v) were subcutaneously injected into the right flank of male C57BL/6J mice. Mice in the n(LOx) group (n=6) were intratumorally injected with n(LOx) (0.5 mg/kg) at day 7 and 9. At day 10, B16F10 cells (5×10⁵ cells) in DMEM/Matrigel (1:1, v/v) were subcutaneously injected into the left flank of male C57BL/6J mice. The animals were euthanized at day 23 and the TILs were analyzed with flow cytometry.

To deplete CD4⁺ and CD8⁺ T cells, mice (n=6) received intraperitoneal injections of 150 μg anti-mouse CD4 antibody (BioLegend, GK1.5) or anti-mouse CD8a antibody (BioLegend, 53–6.7) twice per week starting from one week before tumor engraftment. The depletion of T cells was verified with flow cytometry analysis of splenocytes. B16F10 cells (5×10⁵ cells) in DMEM/Matrigel (1:1, v/v) were subcutaneously injected into the right flank of male C57BL/6J mice. <ice received intratumoral injections of n(LOx) (0.5 mg/kg) at day 7, 9, and 11 after tumor injection. The tumor volume was measured throughout the experiment.

Single-cell RNA sequencing and data analysis

C57BL/6J mice bearing B16F10 tumors received intratumoral injections of n(LOx) (0.5 mg/kg) at day 7 and 9 after tumor engraftment. At day 13, the mice were euthanized and tumor tissues were harvested. Single cell suspensions were obtained by passing the tumor tissues through a MACS SmartStrainer (30 μm). The red blood cells (RBC) were lysed with RBC lysis buffer (BioLegend) and dead cells were removed with the Dead Cell Removal Kit (Miltenyi Biotec). The cells were then counted and processed according to the protocol of the Chromium Next GEM Single Cell 5’ kit (10x Genomics, product 1000265). The library preparation were completed using the 10x library construction kit (product 1000215). After library preparation, the samples were pooled together and sequenced in one batch on a NovaSeq (Illumina) at UCLA Technology Center for Genomics & Bioinformatics.

The raw data were first preprocessed by 10x’s Cellranger pipeline. After the preprocessing, the single-cell 10x count tables were imported into Seurat in R for the following processing. The gene counts were log-transformed and principal component analysis with 30 dimensions was used to dimensionally reduce the gene count data. Graphically based clustering and manual annotation were applied to identify the different cell types. To identify the genes that were significantly differentially expressed between control and n(LOx) treatment, we used Wilcoxon rank-sum tests with the Benjamini–Hochberg adjusted P-values below 0.05 and absolute values of the log2 of the fold change between the average expression of the raw molecular counts above 0.25. The biological processes GO terms of the significantly differentially expressed genes were calculated through the R package ClusterProfiler with a false discovery rate (FDR) < 0.05.

MDA-MB-231 humanized mouse model

Currently available humanized mouse models include PBMC-humanized, SRC (scid-repopulating cell)-humanized, and BLT (bone marrow, liver, and thymus)-humanized mice, summarized in table S1. Among different humanized mouse models, the PBMC humanized model stands out as the most widely adopted choice for investigating immune cell activation and anti-tumor responses in the context of immunotherapy, primarily due to its ability to effectively reconstitute high frequencies of mature human T cells. Thus, the PBMC humanized model was selected to evaluate the therapeutic efficacy of n(LOx) and the combination treatment with human-specific anti-PD-L1 antibody. MDA-MB-231 cells (1.5×10⁶ cells) in DMEM/Matrigel (1:1, v/v) were subcutaneously injected into the second mammary gland of female NSG mice. Ten days after tumor injection, PBMC (2×10⁶ cells) were intravenously injected into the mice. Two weeks after PBMC injections (day 0), mice were randomly divided into 6 groups (n=7 to 8 mice per group). Mice in the n(LOx) group received intratumoral injections of n(LOx) (0.5 mg/kg) at day 0 and 2. Mice in the n(LOx)/CAT group were injected with n(LOx) (0.5 mg/kg) and CAT (1 mg/kg) to remove the H₂O₂. Mice receiving aPD-L1 were given durvalumab (100 μg/mouse) intraperitoneally twice per week starting on day 2. Tumor growth was measured with a digital caliper, and the tumor volume (mm³) was calculated as (length x width²)/2. The animals were euthanized, and tumor samples were collected when the tumor volume reached about 2000 mm³ or at day 24. Single cell suspensions were then obtained from the tumor tissues and T cells were analyzed by flow cytometry.

Statistical Analysis

All raw, individual-level data are presented in data file S1. Measurements in this study were obtained from distinct samples. GraphPad Prism 8 was used to perform statistical analysis. Graphs include means and error bars, with the latter representing s.d. or s.e.m. as indicated. Statistical significance was calculated by one-way ANOVA with Tukey’s multiple comparison test or unpaired t-test. The survival curve was analyzed by Mantel-Cox test.