Design of a bispecific peptide-nanozyme conjugate for cancer immunotherapy

Summary

Despite advances in cancer immunotherapy, clinical efficacy remains constrained by immunosuppressive tumor microenvironment (TME), including PD-L1-mediated T cell dysfunction and CXCL8-driven myeloid cell recruitment. To address this, a bispecific peptide-nanozyme conjugate (BsPNEC) is engineered. Leveraging iterative structure-guided optimization, we first develop q6w, a proteolysis-resistant D-peptide targeting CXCR1/2, and conjugate it to a PD-L1-blocking peptide to generate a bispecific peptide qGA. To augment the therapeutic efficacy, qGA is conjugated to Fe₃O₄ nanozymes with peroxidase-mimetic activity. The Fe₃O₄ nanozymes catalytically decompose H₂O₂ into reactive oxygen species (ROS), thus activating the cGAS-STING pathway to potentiate CD8⁺ T cell infiltration and activation in anti-PD-1-resistant tumor model. The BsPNEC platform integrates tumor-targeted delivery, magnetic resonance imaging (MRI) contrast capabilities, and robust inhibition of tumor growth. Our findings present a synergistic immunotherapeutic strategy that simultaneously skews immunosuppressive TME and amplifies T cell immune response.

Graphical abstract

Highlights

• •A bispecific D-peptide targets PD-L1 and CXCR1/2 to combat tumor immunosuppression
• •Fe₃O₄ nanozyme activates cGAS-STING pathway via ROS to enhance CD8⁺ T cell tumor infiltration
• •Bispecific peptide-nanozyme conjugate (BsPNEC) integrates tumor-targeted MRI and immunotherapy
• •BsPNEC exhibits therapeutic efficacy in an anti-PD-L1-resistant pancreatic cancer model

Gao et al. develop a PD-L1 and CXCR1/2-targeting bispecific peptide-nanozyme conjugate (BsPNEC) that reduces immunosuppressive myeloid cell infiltration, activates cGAS-STING via ROS to enhance CD8⁺ T cell tumor infiltration, and validates efficacy in an anti-PD-L1-resistant pancreatic cancer model.

Introduction

Cancer immunotherapy, which harnesses the immune system to eliminate tumors, has demonstrated transformative clinical promise. Yet durable responses remain limited to about 20% of patients, constrained by tumor-intrinsic adaptive resistance and immunosuppressive cellular networks within the tumor microenvironment (TME).¹ First, tumor cell-intrinsic PD-L1 upregulation drives T cell exhaustion by sustained PD-1/PD-L1 axis activation. Concurrently, there is infiltration of immunosuppressive myeloid cells, which actively suppress cytotoxic T cell infiltration, notably through cytokine cascades and metabolic competition.^2,3 Thus, effective therapeutic strategies must simultaneously enhance T cell function while attenuating immunosuppressive cell infiltration to synchronize coordinated antitumor immunity.

Accumulating evidence identifies chemokine signaling as a pivotal regulatory mechanism of the immunosuppressive TME.⁴ CXCL8, a chemokine overexpressed across tumors, drives tumor progression by recruiting neutrophils and myeloid-derived suppressor cells (MDSCs), polarizing TAMs toward an M2 phenotype,^5,6 and directly stimulating tumor cell invasion via its seven transmembrane G protein-coupled receptors CXCR1/2.^7,8 Clinically, elevated plasma CXCL8 correlates with poor outcomes of immune checkpoint blockade (ICB), while preclinical studies reveal that PD-1/PD-L1 inhibition paradoxically amplifies CXCL8 production, which establishes a self-reinforcing immunosuppressive loop.^9,10,11 These findings position dual CXCR1/2 and PD-1/PD-L1 blockade as a rational strategy to potentiate ICB efficacy.

This therapeutic rationale has spurred clinical evaluation of combination regimens, including anti-PD-1 with CXCL8/CXCR1/2-targeting agents. Early-phase trials reported enhanced responses in patients with hepatocellular carcinoma receiving PD-1/CXCL8 dual blockade¹² and demonstrated safety when combining CXCR1/2 small-molecule inhibitors with pembrolizumab (NCT03161431). Inhibition of CXCR1/2 signaling attenuates the infiltration of immunosuppressive cell populations, thereby alleviating their suppressive constraints on antitumor immune responses. This strategic intervention synergistically potentiates the therapeutic efficacy of PD-L1 blockade, ultimately culminating in enhanced T cell-mediated tumoricidal activity.⁹ However, monoclonal antibodies exhibit limited penetration into solid tumors due to their large molecular size, while small-molecule drugs, despite their inherent cell permeability, risk off-target toxicity due to nonspecific intracellular accumulation.¹³ In contrast, the inherent cell membrane impermeability of peptide therapeutics limits systemic exposure while enabling high-affinity engagement of cell-surface targets.¹⁴ This strategic advantage positions peptides as precision tools to disrupt immunosuppressive chemokine gradients without compromising safety.

In addition to immunosuppressive axis blockade, effective immunotherapy requires promoting tumor infiltration of T cells. Activation of the cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway primes dendritic cells (DCs) for antigen cross-presentation and recruits CD8⁺ T cells into immunologically “cold” tumors.¹⁵ However, conventional STING agonists often provoke systemic inflammation necessitating tumor-specific activation.¹⁶ Previous work circumvented this limitation using Fe₃O₄ nanozymes, which have been approved by the Food and Drug Administration (FDA) as magnetic resonance imaging (MRI) contrast agents.^17,18 As magnetic nanoparticles with intrinsic peroxidase (POD)-like activity, these nanozymes catalyze localized reactive oxygen species (ROS) generation in the acidic and H₂O₂-rich tumor cells and TME. Localized ROS production induces mitochondrial damage and mitochondrial DNA (mtDNA) release that activates tumor-specific cGAS-STING signaling with limited cytotoxicity to healthy tissues.¹⁹ The ease of synthesis and modification of Fe₃O₄ nanozymes further positions them as a versatile platform for peptide drug delivery.

To address the dual bottlenecks of immunosuppressive cell infiltration and impaired T cell function in the TME, we pursued a structure-guided rationale by first mapping the CXCL8-CXCR1/2 interaction interface. Leveraging these insights, we designed a CXCR1/2-targeting lead peptide and iteratively optimized its stability and binding affinity by sequentially employing peptide truncation, retro-inversion, and site-directed mutagenesis. This optimization yielded a proteolysis-resistant D-peptide q6w. To further potentiate T cell-mediated immunity, we engineered a chimeric peptide, qGA, by conjugating q6w with a PD-L1-blocking peptide previously screened by our team.²⁰ This bispecific peptide achieves coordinated TME reprogramming through synchronized inhibition of immunosuppressive cell recruitment (via CXCL8-CXCR1/2 axis disruption) and reactivation of T cell function (via PD-1/PD-L1 checkpoint blockade). To further enhance CD8⁺ T cell infiltration, we conjugated qGA to Fe₃O₄ nanozymes, generating a tumor-targeted bispecific peptide-nanozyme conjugate (BsPNEC). Critically, BsPNEC integrates CXCR1/2/PD-L1 blockade with nanozyme-driven cGAS-STING activation, where Fe₃O₄ catalyzes H₂O₂-to-ROS conversion in the TME to induce mtDNA release, thereby promoting DC priming and T cell trafficking. This multimodal BsPNEC platform with theranostic profiles thus offers a targeted and multifaceted approach to overcome current immunotherapeutic limitations.

Results

High expression of CXCL8 and its receptor in cancer tissues and myeloid cells

To delineate the role of CXCL8-CXCR1/2 axis in TME, we analyzed patient samples across multiple cancer types. Pan-cancer transcriptomics revealed pronounced CXCL8 overexpression in tumor tissues such as cholangiocarcinoma, esophageal carcinoma, and colon adenocarcinoma relative to normal tissues (Figure 1A). Moreover, elevated CXCL8 expression correlated with poorer overall survival (Figure 1B). Re-analysis of single-cell RNA sequencing of patients with colorectal cancer (http://crc-icb.cancer-pku.cn/) uncovered a myeloid-centric expression pattern that CXCL8 was widely expressed in macrophages, neutrophils, and DCs, whereas its receptors CXCR1 and CXCR2 showed preferential enrichment in neutrophils (Figures 1C, 1D, S1A, and S1B). These findings suggest that the CXCL8-CXCR1/2 axis may orchestrate the behavior and function of myeloid cells. Moreover, CXCL8 expression is elevated in patients with low response to anti-PD-1 therapy (Figure 1E).

Figure 1.

Pan-cancer analysis of the CXCL8-CXCR1/2 axis in tumor microenvironment and immunotherapy response

(A) Differential expression of CXCL8 in tumor (red) versus normal (blue) samples across multiple cancer types from The Cancer Genome Atlas (TCGA) database. Cancer types include cervical squamous carcinoma (CESC), cholangiocarcinoma (CHOL), colorectal carcinoma (COAD), glioblastoma (GBM), head and neck squamous cell carcinoma (HNSC), bladder carcinoma (BLCA), renal papillary cell carcinoma (KIRP), pancreatic adenocarcinoma (PAAD), rectal carcinoma (READ), and gastric adenocarcinoma (STAD).

(B and C) (B) Kaplan-Meier survival analysis of patients with colorectal cancer from the TCGA database, stratified by CXCL8 expression levels (low, n = 346; high, n = 346), showing significant survival differences (p < 0.05) (C) UMAP visualization of single-cell RNA sequencing data from patients with colorectal cancer (GEO: GSE236581), highlighting CXCL8 and CXCR1/2 expression across immune and tumor cell populations. UMAP, Uniform Manifold Approximation and Projection

(D) Expression levels of CXCL8 in various cell types from patients with colorectal cancer (GEO: GSE236581), with violin plots showing significant upregulation in macrophages (Mph), neutrophils (Neu), and conventional dendritic cells (cDC).

(E) CXCL8 expression in immune cell subsets from patients with colorectal cancer (GEO: GSE205506), comparing responders (tumor_PCR) and non-responders (tumor_non-PCR) to anti-PD-1 therapy, indicating higher CXCL8 levels in non-responders.

Under physiological conditions, CXCL8 is barely detectable but can be rapidly induced by pro-inflammatory stimuli such as lipopolysaccharide (LPS). After treating the mouse macrophage cell line RAW264.7 with LPS, we observed an increase in the expression of CXCL1/2 (the murine homologs of human CXCL8),^21,22 and a certain degree of upregulation in the expression of CXCR1/2 (Figure S2A). Based on these findings, we selected RAW264.7 macrophages as a tractable model for in vitro functional validation of CXCL8-CXCR1/2 biology.

The D-enantiomeric peptide D10 blocks CXCL8-CXCR1/2 axis

The interaction between CXCL8 and CXCR1 involves two distinct binding steps. During the initial binding phase, the N-loop-terminal region of CXCL8 (residues 14–20) interacts electrostatically with the N-terminal notched region of CXCR1 (residues 21–27). Subsequently, the hydrophobic interaction causes a change in CXCL8 steering, and the ELR motif binds to the EC ring of CXCR1 in an electrostatic interaction. This secondary engagement induces conformational change in CXCR1 to activate downstream signaling.²³ A similar phenomenon was also observed in the interaction between CXCL8 and CXCR2.²⁴

Based on the stepwise binding framework, the lead peptide 18pep was designed by incorporating the N-loop-anchoring domain and ELR signaling motif, with cysteine residues replaced to prevent cyclization (Figure 2A). To investigate the necessity of the ELR motif for peptide-binding targets, two additional peptides (14pep and 16pep) were generated through sequential truncation of two N-terminal amino acids. These peptides were synthesized and expected to inhibit the migration of tumor-associated immune cells by blocking the CXCL8-CXCR1/2 axis. Transwell migration experiments of RAW264.7 showed that ELR-intact 18pep exhibited the most potent inhibitory effect on cell migration, significantly outperforming ELR-deficient control peptides (Figure S2B). To delineate the minimal functional fragment of 18pep, we systematically truncated the peptide in two-amino acid increments from its C terminus, generating a series of peptides ranging from 16- to 8-mers (Table S1). Through transwell migration assays, the 10-mer peptide (T10) was identified as the minimal functional unit retaining comparable inhibitory activity (Figure S2C).

Figure 2.

Screening of CXCL8-CXCR1/2-blocking peptides

(A) Schematic representation of the modification process. Upper- and lower-case letters indicate L- and D-amino acids, respectively.

(B) Transwell migration assay showing D10 peptide’s effects on LPS-stimulated RAW264.7 macrophage. Scale bars, 100 μm.

(C) ZDOCK predicted docking models of D10 peptide with CXCR1/2. Green highlights indicate interaction regions; blue dashed lines represent residue interactions.

(D) Transwell migration assay of RAW264.7 macrophages toward top 8 affinity-ranked mutant peptides. Scale bars, 100 μm.

(E) Quantitative analysis of peptides effects on RAW264.7 migration.

(F) Quantitative reverse-transcription PCR (RT-qPCR) analysis of M1 (TNFA) and M2 (IL-10, ARG1) markers in BMDMs cultured for 3 days in CT26-conditioned medium with PBS or 50 μM peptides.

Data are presented as means ± SEM from three independent experiments. ∗∗∗p < 0.001; #p < 0.05; ###p < 0.001; ordinary one-way ANOVA with Tukey’s multiple comparisons test.

Notably, the T10 is composed of the natural L-type amino acids, which confer poor proteolytic stability.²⁵ To address this, we replaced it with a reverse-ordered D peptide to obtain the D10 peptide and performed transwell experiments to investigate whether its migration inhibition effect was reduced. Strikingly, D10 peptide not only preserved target engagement but also exhibited enhanced migration inhibition relative to T10 (Figure 2B), which may be attributed to its resistance to enzymatic degradation and prolonged bioavailability. Through these optimizations, we identified a metabolically stable peptide capable of potently disrupting CXCL8-CXCR1/2 signaling.

Iterative optimization identified peptide q6w as a stabilized CXCR1/2 antagonist with dual immunomodulatory function

Structural modeling localized the D10 peptide within the CXCR1/2 N-terminal groove, a critical interface for CXCL8 binding (Figure 2C). To further improve its CXCR1/2 affinity, we conducted molecular docking, dynamics simulations, and single amino acid mutations of D10. This screening process revealed 8 mutant peptides with the highest theoretical affinity enhancement (Table S2). Compared to the D10 peptide, the four mutant peptides (s7y, s7k, q6p, and q6w) exhibited comparable or even stronger inhibition of cell migration (Figures 2D and 2E).

Since CXCL8-CXCR1/2 promotes the differentiation of tumor-associated macrophages toward a pro-tumor M2 phenotype in the TME,²² we further evaluated the capacity of these four mutant peptides to counteract the polarization of mouse bone marrow-derived macrophages (BMDMs). Results showed that exposure to CT26 conditioned medium (CM) significantly downregulated the M1 phenotypic marker TNFA but upregulated the M2 phenotypic markers IL-10 and ARG1. q6w reversed this effect by partially restoring TNFA expression, while suppressing IL-10 and ARG1 (Figure 2F). Notably, combined treatment with CXCL2 neutralization and the CXCR1/2-blocking peptide q6w did not produce a synergistic effect, suggesting partial functional overlap between these interventions (Figure S3). These findings support a model wherein q6w promotes pro-inflammatory BMDM polarization by disrupting CXCLs-CXCR1/2 signaling within the CT26 CM microenvironment. Pharmacokinetic profiling further revealed the superiority of q6w as its D-amino acid backbone conferred strong resistance to proteolytic hydrolysis in 10% human serum (Figure S4A), while binding assays confirmed preferential affinity for CXCR2 over CXCR1 (Figure S4B), aligning with the dominant role of CXCR2 in myeloid recruitment.

Through sequential optimization—truncation to refine pharmacophore topology, D-enantiomer substitution to circumvent proteolysis, and in silico-directed mutagenesis to maximize receptor engagement, we engineered q6w as a metabolically stable, high-affinity CXCR1/2 antagonist. This systematic engineering strategy positions q6w as a tailored therapeutic agent capable of disrupting both myeloid-driven immunosuppression and M2 polarization within the TME.

The bispecific peptide qGA simultaneously targets CXCR1/2 and PD-L1

Emerging evidence implicates adaptive feedback between CXCL8-CXCR1/2 activation and PD-L1 upregulation on tumor cells as a key immune evasion mechanism during immune checkpoint blockade therapy.¹¹ A positive correlation between PD-L1 and CXCL8 was found in patients with colorectal cancer by using Spearman’s correlation analysis (r = 0.54, p < 0.001; Figure S5A). This indicates that simultaneous blockade of the CXCL8-CXCR1/2 and PD-L1-PD-1 axes is beneficial to synergistically enhance the anti-tumor effect. Building upon this foundation, we conjugated our high-affinity CXCR1/2-blocking peptide q6w with OPBP-1(8-12),²⁰ a PD-L1-blocking peptide identified previously, via flexible GS linkers. Two configurations, AGq (N-terminal fusion) and qGA (C-terminal fusion), were synthesized (Figure 3A; Table S3).

Figure 3.

Design and validation of bispecific peptides

(A) Schematic of bispecific peptide design and mechanism.

(B and C) MST assay of bispecific peptides’ binding affinity to CXCR1/2.

(D) Transwell migration assay was performed to assess LPS-stimulated RAW264.7 macrophage migration toward bispecific peptides. Scale bars, 100 μm. ΔFnorm represents the normalized fluorescence difference.

(E) Quantification of bispecific peptide effects on RAW264.7 migration.

(F) RT-qPCR analysis of M1 (TNFA) and M2 (IL-10, ARG1) markers in BMDMs cultured with CT26-conditioned medium with PBS or 50 μM peptides.

(G and H) Blocking effect of bispecific peptide on hPD-1/hPD-L1 or mPD-1/mPD-L1 interaction at 100 μM concentration.

(I) Flow cytometry analysis of CD8⁺IFN-γ⁺ T cell percentage in PBMCs stimulated with anti-CD3/CD28 antibodies and 100 μM peptide.

All data were calculated from three independent experiments. ∗∗p < 0.01; ∗∗∗p < 0.001; ##p < 0.01; ###p < 0.001; ns, not significant; ordinary one-way ANOVA with Tukey’s multiple comparisons test.

Microscale thermophoresis (MST) revealed that both AGq and qGA showed a significantly enhanced binding affinity for CXCR1 compared to q6w and that qGA exhibited a 2-fold increase in CXCR2 binding affinity (Figures 3B and 3C). Consistent with this, the ability of these bispecific peptides to inhibit cell migration was also greatly enhanced (Figures 3D and 3E). Notably, qGA also significantly restored TNFA expression in BMDMs and down-regulated IL-10 and ARG1, thereby inhibiting their differentiation toward the pro-tumor M2 polarization (Figure 3F).

Crucially, blockade assays demonstrated that the bispecific peptide could significantly inhibit PD-1/PD-L1 interaction (Figures 3G and 3H). Moreover, conjugation of OPBP1(8-12) to either the N- or C terminus of q6w demonstrated enhanced blocking activity compared to the original OPBP1(8-12) peptide, with the C-terminally modified bispecific peptide qGA showing the strongest inhibitory effect. To investigate whether qGA could activate T cell function, we examined the proportion of interferon (IFN)-γ-producing CD8⁺ T cells in human peripheral blood mononuclear cells (PBMCs). Notably, qGA treatment markedly potentiated the ability of human primary CD8⁺ T cells to secrete IFN-γ compared with the single PD-L1-blocking peptide OPBP-1(8-12) (Figure 3I), which was comparable to the effect of anti-PD-L1 and q6w combined (Figure S5B), underscoring its capacity to counteract T cell exhaustion.

Through rational design, we successfully synthesized bispecific peptide qGA that exhibits affinity for both CXCR1/2 and PD-L1, which not only inhibits the polarization of macrophages toward the M2 phenotype in vitro but also reinstates cytotoxic CD8⁺ T cell function.

Bispecific peptide qGA exhibits favorable biocompatibility and proteolytic resistance

To assess the safety of the bispecific peptide qGA, we examined the erythrocyte hemolysis rate and the proliferation of mouse colon cancer cells CT26 and human normal hepatocytes HL7702 after in vitro treatment with the peptide. The results demonstrated that qGA exerted dose- and time-dependent inhibitory effects on CT26 cell proliferation, while showing minimal impact on the proliferation of HL7702 cells (Figure S6A). Additionally, qGA exhibited no erythrocyte lysis (Figure S6B). These results, combined with sustained stability in 10% human serum (Figure S6C), validate biocompatibility and translational suitability of qGA.

Bispecific peptide qGA significantly enriched in tumor tissue and inhibits CT26 tumor growth

In vivo biodistribution analysis of the bispecific peptide qGA was performed using fluorescein isothiocyanate (FITC)-conjugated qGA in CT26-bearing BALB/c mice. At 6 h post-injection, qGA exhibited significant tumor accumulation, followed by the kidneys and liver, compared to the control peptide GA. Notably, qGA signals could still be detected at the tumor site even at 24 h post-injection, suggesting prolonged target engagement (Figure 4A). These findings confirm that qGA, due to its high affinity for CXCR1/2 and PD-L1, and especially for CXCR1/2 (Figure S7), possesses favorable tumor tissue targeting and prolonged duration of action and is primarily excreted through the kidneys.

Figure 4.

Tissue distribution and antitumor efficacy of bispecific peptide in CT26 tumor-bearing mice

(A) Representative fluorescence images and quantitative distribution of major organs and tumors at 6, 12, and 24 h post intraperitoneal injection.

(B) Treatment regimen. CT26 tumor-bearing mice received intraperitoneal (i.p.) injections of 0.5 mg/kg OPBP-1(8-12), 1 mg/kg q6w, 1.55 mg/kg qGA, or 2.5 mg/kg reparixin for 2 weeks.

(C) Tumor growth curves (n = 5).

(D) Tumor weights of each group were measured at final sampling (n = 5 mice per group, no exclusions).

(E) Individual tumor growth curves for each treatment group (n = 5 mice per group, no exclusions).

(F) Body weight changes in tumor-bearing mice during treatment (n = 5 mice per group, no exclusions).

Data are represented as means ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ordinary one-way ANOVA with Tukey’s multiple comparisons test.

We established a CT26 tumor model with a treatment regimen as shown in Figure 4B, and the drugs were administered in equimolar amounts. Results demonstrated that the bispecific peptide qGA significantly suppressed the growth of CT26 tumors, exhibiting superior inhibitory effects compared to both reparixin (a small-molecule CXCR1/2 inhibitor) and single-targeting peptides (Figures 4C–4E), confirming synergistic benefits of combined PD-L1/CXCR1/2 blockade. Additionally, it had no impact on the body weight of the mice bearing the tumors (Figure 4F), indicating a favorable safety profile for the peptide. We also investigated whether the bispecific peptide qGA could also inhibit the infiltration of immunosuppressive cells and enhance CD8⁺ T cell infiltration into tumor tissues in vivo. Results showed that qGA treatment significantly increased the M1/M2 macrophage ratio and the infiltration of CD8⁺ T cells into tumor tissues, while efficiently decreasing tumor-associated neutrophil (TAN) infiltration (Figure S8). In summary, these results demonstrate that qGA exhibits synergistic antitumor activity in vivo.

Construction and characterization of BsPNECs

While the bispecific peptide qGA demonstrated robust antitumor efficacy, its capacity to enhance CD8⁺ T cell infiltration remained suboptimal. To increase T cell infiltration, we leveraged the cGAS-STING pathway, which is a key axis promoting DC-mediated antigen cross-presentation. Our previous work established that Fe₃O₄ nanozymes with peroxidase-mimetic (POD) activity catalyze localized H₂O₂ decomposition into ROS, inducing mtDNA release to activate tumor-specific cGAS-STING signaling.¹⁹ Given the biocompatibility of Fe₃O₄ as an FDA-approved agent^17,26 and the dual-targeting capability of qGA, we engineered a peptide-nanozyme conjugate by coupling qGA peptide with Fe₃O₄, utilizing a reactive peptide substrate GPA of fibroblast activation protein as a linker.^27,28 The functionalized nanozyme termed BsPNEC was constructed through an EDC/NHS coupling reaction, as illustrated in Figure 5A.

Figure 5.

Design, characterization, and in vitro activity of the BsPNEC

(A) Schematic illustration of BsPNEC preparation via EDC/NHS coupling chemistry to conjugate peptides onto Fe₃O₄ nanozyme.

(B) FTIR spectra of Fe₃O₄, GPA-qGA, and BsPNEC.

(C) Zeta potential measurements before and after conjugation.

(D–G) (D and F) Hydrodynamic diameter distribution and (E and G) representative TEM images of Fe₃O₄ and BsPNEC.

(H) Cell viability assessed by MTT assay in RAW264.7 and CT26 cells treated with various nanozyme concentrations at 24, 48, and 72 h.

(I) Peroxidase-like activity kinetics showing Vmax and Km values before and after conjugation.

(J) Mean fluorescent intensity (MFI) of ROS production in CT26 cells following nanozyme stimulation.

(K) RT-qPCR analysis of IFNβ and CXCL10 mRNA expression levels.

(L) RT-qPCR analysis of IFNβ and CXCL10 mRNA expression levels following pretreatment with 3 mM NAC (ROS scavenger).

(M and N) Assessment of cGAS-STING activation by BsPNEC following STING knockout in CT26 cells, as determined by RT-qPCR and western blot analysis.

(O) Schematic illustration of BsPNEC-mediated STING pathway activation.

Data are represented as means ± SEM. All data were calculated from three independent experiments. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; #p < 0.05; ###p < 0.001; ns, not significant; ordinary one-way ANOVA with Tukey’s multiple comparisons test and unpaired two-tailed Student’s t tests.

When the ratio of GPA-qGA to Fe₃O₄ was 1:1, the peptide loading capacity was 53.28% ± 0.30%. Characterization was performed using Fourier transform infrared (FTIR) spectroscopy. For BsPNEC, the characteristic peaks at 3,500–3,050 cm⁻¹ and 1,650–1,500 cm⁻¹ indicated the presence of amide bonds (Figure 5B), suggesting successful peptide conjugation. The change in zeta potential from −36.03 to −16.42 mV on the surface of the nanoparticles after conjugation further supported this result (Figure 5C). Dynamic light scattering (DLS) was used to detect the average hydrodynamic sizes before and after conjugation, with mean hydration diameters of 153.2 ± 9.45 nm and 177.8 ± 26.34 nm, respectively (Figures 5D and 5F). Transmission electron microscopy (TEM) analysis revealed that both possessed spherical morphologies, with mean particle diameters of 10.76 ± 2.82 nm and 13.71 ± 1.20 nm, respectively (Figures 5E and 5G). Moreover, the significant discrepancy between TEM and DLS particle size measurements stems from TEM revealing the actual, dry-state diameter, while DLS measuring the larger hydrodynamic diameter in the hydrated state.

Functional characterization revealed tumor-selective cytotoxicity. BsPNEC had no notable effect on the proliferation of RAW264.7, but exhibited inhibitory effects on the proliferation of CT26 cells at high concentrations and with prolonged exposure (Figure 5H). This may be attributed to higher levels of H₂O₂ in tumor cells and lower levels in macrophages and the heightened sensitivity of tumor cells to high levels of ROS. These reactive species directly impair tumor cells through oxidative stress, ultimately inducing apoptosis.

BsPNEC induces ROS production and subsequent activation of the STING pathway

To validate functional integrity post-conjugation, we confirmed that BsPNEC retained potent POD-like activity equivalent to unconjugated Fe₃O₄ nanozymes (Figure 5I). When co-cultured with CT26 cells for 6, 12, and 24 h, BsPNEC induced substantial ROS accumulation (Figure 5J), inducing oxidative stress commensurate with mtDNA release. IFNβ and CXCL10 represent core downstream effector molecules of the cGAS-STING signaling pathway, with their expression levels demonstrating a definitive positive correlation with pathway activation.^29,30,31 This ROS burst drove robust cGAS-STING activation, evidenced by rapid upregulation of IFNβ and CXCL10 mRNA in just 6 h of co-incubation with CT26 cells, consistent with the effect of the STING agonist cGAMP. Crucially, while cGAMP-induced activation was only transiently detectable at 6 h, BsPNEC-mediated cGAS-STING pathway activation persisted for up to 12 h (Figure 5K), underscoring its prolonged immunostimulatory capacity. Moreover, the elevation of IFNβ and CXCL10 was attenuated after treatment with NAC, an ROS scavenger, suggesting that nanozyme-mediated cGAS-STING activation is partially dependent on ROS generation (Figure 5L). After further knockout of STING in CT26, the role of BsPNEC in activating cGAS-STING was also attenuated, which was manifested by the decreased elevation of IFNβ and CXCL10, as well as the decreased expression of pTBK1 and pIRF3, indicating that nanozyme-induced cGAS-STING activation requires functional STING signaling (Figures 5M and 5N). The activation of cGAS-STING by BsPNEC also endows it with the ability to promote DC maturation, which is manifested by the increased expression of maturation-related markers CD80, CD86, and CD40 by BMDCs, and can promote the proliferation of T cells and the secretion of IFN-γ in the coculture system (Figure S9). Collectively, BsPNEC synergizes dual-axis blockade of CXCR1/2 and PD-L1 to dismantle immunosuppression, leverages TME-targeted ROS generation to ignite cGAS-STING signaling, and sustains innate immune activation to enhance CD8⁺ T cell infiltration (Figure 5O). This orchestrated mechanism positions BsPNEC as a tumor-centric immunotherapeutic platform engineered to override adaptive resistance.

BsPNEC inhibits CT26 tumor growth and reprograms the tumor microenvironment

Subsequently, we verified whether BsPNEC has better anti-tumor immune effects in vivo using the CT26 tumor model. As illustrated in Figure 6A, mice received treatment every other day. Notably, BsPNEC outperformed both qGA and Fe₃O₄ monotherapies (Figures 6B–6D) with no systemic toxicity, as evidenced by stable body weight (Figure 6E). We further analyzed the immune cell populations in the tumor, draining lymph node (DLN), and spleen following BsPNEC treatment. Results showed that BsPNEC not only elevated the ratio of M1/M2 macrophages in tumor tissues but also substantially reduced the TAN infiltration (Figures 6F and 6G). Most importantly, we observed robust recruitment of CD8⁺ T cells into tumors (Figure 6H), with a marked increase in IFN-γ-secreting CD8⁺ T cells, which are critical for effective anti-tumor immunity (Figure 6I). Beyond local effects, mice treated with BsPNEC showed a significant increase in the proportion of IFN-γ-secreting CD8⁺ T cells in DLN and spleen compared with the equivalent dose of Fe₃O₄ and qGA alone (Figure S10). These findings suggest that BsPNEC can effectively inhibit tumor growth by reprogramming the TME and triggering a strong systemic immune response.

Figure 6.

Therapeutic efficacy of systemic BsPNEC administration in CT26 tumor-bearing mice

(A) Treatment regimen. CT26 tumor-bearing mice received intravenous (i.v.) injections of 1.55 mg/kg qGA, 5 mg/kg Fe₃O₄, and 5 mg/kg BsPNEC every other day for 2 weeks.

(B) Tumor growth curves (n = 5 mice per group, no exclusions).

(C) Tumor weights of each group were measured at final sampling (n = 5 mice per group, no exclusions).

(D) Individual tumor growth curves for each treatment group (n = 5 mice per group, no exclusions).

(E) Body weight variation during treatment (n = 5 mice per group, no exclusions).

(F) Ratio of macrophage M1 (CD45⁺CD11b⁺F4/80⁺CD11c⁺) and M2 (CD45⁺CD11b⁺F4/80⁺CD206⁺) in tumors measured by flow cytometry.

(G) Flow cytometry measurement of the percentage of tumor-infiltrating neutrophils (CD45⁺CD11b⁺Ly6G⁺Ly6C^int).

(H) Flow cytometry measurement of the percentage of tumor-infiltrating CD8⁺ T cells (CD45⁺CD3⁺CD8⁺).

(I) Percentage of IFN-γ-secreting CD8⁺ T cells (CD3⁺CD8⁺IFN-γ⁺) in tumors detected by flow cytometry.

(J) MRI schematic diagram.

(K) MRI of BsPNEC in vivo and statistical chart (n = 3 mice per group, no exclusions). In vivo T2-weighted MRI of Fe₃O₄ nanoparticles and BsPNEC. PBS was used as a negative control.

Data are represented as means ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; #p < 0.05; ##p < 0.01; ordinary one-way ANOVA with Tukey’s multiple comparisons test.

BsPNEC enables tumor-targeted magnetic resonance imaging

As magnetic nanoparticles, especially the stable magnetite Fe₃O₄, can be used to enhance the signal in T1- and T2-weighted images for imaging tumors,³² we evaluated the imaging properties of BsPNEC in vitro and in vivo using T2-weighted MRI. Both Fe₃O₄ and BsPNEC significantly reduced the T2 relaxation time, demonstrating their capacity to enhance T2-weighted contrast (Figure S11). However, only BsPNEC conjugated with a dual-targeting peptide was able to augment the T2 signal intensity specifically at the tumor site in vivo (Figures 6J and 6K), reflecting its preferential accumulation via dual-receptor targeting. This tumor-specific contrast enhancement positions BsPNEC as a precision theranostic platform capable of simultaneously delivering immunotherapy and enabling non-invasive monitoring of drug biodistribution.

Simultaneously inhibiting CXCR1/2 can enhance the anti-tumor effect of PD-L1 blockade

Given the established role of CXCR1/2 inhibition in augmenting PD-1/PD-L1 blockade efficacy and the notorious resistance of pancreatic tumors to PD-1 blockade, we assessed the efficacy of the dual-targeting agent BsPNEC in this immunologically “cold” model (Figure 7A). Results demonstrated superior antitumor efficacy in treatment groups receiving qGA or BsPNEC, which mediate concurrent blockade of CXCR1/2-CXCL8 and PD-1/PD-L1 pathways, in contrast to the limited efficacy observed with anti-PD-L1 monotherapy (Figures 7B–7E). Mechanistic profiling revealed that although the PD-L1 monotherapy expanded tumor-infiltrating T cells, it failed to reduce immunosuppressive neutrophils. This persistent myeloid suppression directly undermined therapeutic efficacy. In contrast, both qGA and BsPNEC simultaneously amplified cytotoxic CD8⁺ T cell infiltration and reduced the neutrophil proportion by > 50% (Figures 7F–7H). Moreover, systemic immune activation was evidenced by expanded populations of IFN-γ-secreting CD8⁺ T cells in DLN and spleen, respectively (Figures 7I and 7J). Besides, qGA and BsPNEC have high safety in mice (Figure S12). In order to better reflect the synergistic effect of these two targets, we further explored the antitumor effect of anti-PD-L1 and CXCR1/2-blocking peptide q6w combined. The results showed that the combination of anti-PD-L1 and q6w could also inhibit tumor growth, suggesting that CXCR1/2 blockade has a positive effect on overcoming the drug resistance to anti-PD-L1 (Figure S13). These findings establish concurrent CXCR1/2 inhibition as an effective therapeutic strategy to overcome myeloid-driven resistance, effectively converting immunologically inert tumors into T cell-inflamed microenvironments. This dual-targeting paradigm thus addresses a fundamental limitation of PD-L1 checkpoint blockade.

Figure 7.

Therapeutic efficacy of systemic BsPNEC administration in pancreatic tumor-bearing mice

(A) Treatment regimen. Pancreatic tumor-bearing mice received i.v. or i.p. injections of 1.55 mg/kg qGA, 5 mg/kg Fe₃O₄, 5 mg/kg BsPNEC every day and 10 mg/kg IgG, 10 mg/kg anti-PD-L1 every 3 days for 2 weeks.

(B) Tumor growth curves (n = 5 mice per group, no exclusions).

(C) Tumor weights of each group were measured at final sampling (n = 5 mice per group, no exclusions).

(D) Body weight variation during treatment (n = 5 mice per group, no exclusions).

(E) Individual tumor growth curves for each treatment group (n = 5 mice per group, no exclusions).

(F) Flow cytometry measurement of the percentage of tumor-infiltrating CD8⁺ T cells (CD45⁺CD3⁺CD8⁺).

(G) Flow cytometry measurement of the percentage of tumor-infiltrating neutrophils (CD45⁺CD11b⁺Ly6G⁺Ly6C^int).

(H) Percentage of IFN-γ-secreting CD8⁺ T cells (CD3⁺CD8⁺IFN-γ⁺) in tumors detected by flow cytometry.

(I) Percentage of IFN-γ-secreting CD8⁺ T cells (CD3⁺CD8⁺IFN-γ⁺) in draining lymph nodes detected by flow cytometry.

(J) Percentage of IFN-γ-secreting CD8⁺ T cells (CD3⁺CD8⁺IFN-γ⁺) in spleen detected by flow cytometry.

Data are represented as means ± SEM. ∗∗p < 0.01; #p < 0.05; ##p < 0.01; ###p < 0.001; Ordinary one-way ANOVA with Tukey’s multiple comparisons test.

Discussion

The CXCL8-CXCR1/2 axis represents a pivotal orchestrator of myeloid-driven immunosuppression and emerges as a critical target for overcoming ICB resistance. As the prototype chemokine for polymorphonuclear leukocyte recruitment, CXCL8 drives tumorigenesis by concurrently recruiting neutrophils, immunosuppressive myeloid cells, and monocytes, while reprogramming the TME toward pro-tumorigenic states.^33,34 Clinically, elevated serum CXCL8 correlates with increased TAN infiltration, poor survival, and PD-L1 upregulation on tumor and immune cells, thereby dampening ICB efficacy.^11,35,36,37 Our observations demonstrate pronounced neutrophil infiltration within the pancreatic TME, a well-documented feature of this model.^38,39 This significant accumulation of TANs appears to constitute a critical immunosuppressive axis that undermines the efficacy of immune checkpoint inhibition in this model. Given the abundant expression of CXCR1 and CXCR2 on neutrophil surfaces, pharmacological inhibition of CXCL8-CXCR1/2 signaling emerges as a compelling therapeutic strategy. Katoh et al. demonstrated that CXCR2 inhibition reduced neutrophil recruitment, alleviated T cell suppression, and curtailed both chronic inflammation and tumor progression in a colitis-associated colorectal cancer model,⁴⁰ underscoring the therapeutic potential of CXCL8 pathway blockade. Our study further extends this concept by demonstrating that dual targeting of CXCR1/2 and PD-L1 receptors co-opted by both myeloid and tumor cells enables coordinated disruption of immunosuppressive networks.

Moreover, emerging evidence underscores the necessity of considering the combined contributions and interchangeable roles of macrophages and neutrophils within the TME when developing neutrophil-targeted therapies.³⁵ Targeting the receptor-ligand pairs co-expressed by these two cell types can yield enhanced antitumor effects, and PD-L1 has been proved to be a promising target.^36,41 PD-L1⁺ TANs and PD-L1⁺ TAMs typically exert negative regulatory effects on T cells, thereby suppressing their immune functions. Consequently, targeting PD-L1 in a combinatorial approach may be advantageous for synergistically augmenting the efficacy of tumor immunotherapy.

Peptides offer distinct advantages over conventional biologics for such combinatorial targeting. When it comes to inhibiting protein interactions, affinity peptides tailored to the local structural features of protein macromolecule interactions exhibit specificity on par with antibodies. Moreover, these peptides offer numerous advantages in terms of synthesis technology, cost-effectiveness, penetration into solid tumors, immunogenicity, and safety.^42,43 The CXCR1/2-targeting peptide q6w developed in this study demonstrated robust efficacy in reducing neutrophil infiltration at tumor sites in vivo. However, q6w exhibited minimal impact on macrophage infiltration despite its capacity to inhibit migration of the murine macrophage cell line RAW264.7 in vitro. This disparity likely stems from significantly higher CXCR1/2 surface density on neutrophils compared to macrophages in the TME. Consequently, following systemic administration, the peptide preferentially engages CXCR1/2 on neutrophils, thereby attenuating their tumor-homing capacity while sparing macrophages. However, monotherapy failed to counteract PD-L1-mediated adaptive resistance, necessitating the bispecific agent qGA. The chimeric peptide qGA targeting both CXCR1/2 and PD-L1 exhibited superior activity both in vitro and in vivo, significantly reducing neutrophil infiltration into tumors while markedly increasing the proportion of M1-like TAMs and CD8⁺ T cells. Notably, qGA also demonstrated excellent tumor tissue-targeting properties, suggesting its potential utility in aiding the design of various tumor-targeting therapy systems.

The integration of qGA with Fe₃O₄ nanozymes further amplifies therapeutic efficacy through spatiotemporal immunomodulation. In this study, we coupled qGA to Fe₃O₄ nanoparticles to construct a BsPNEC and demonstrated that it can also stimulate tumor cells to produce ROS, directly impair tumor cells through oxidative stress, and activate the cGAS-STING signaling pathway. This activation significantly enhanced the recruitment and activity of CD8⁺ T cells at the tumor site, resulting in the inhibition of tumor growth in murine models. Meanwhile, considering that Fe₃O₄ can be used for MRI imaging, we also studied the tumor imaging function of BsPNEC. Fe₃O₄ nanoparticles lack intrinsic tumor-targeting capabilities, representing a significant limitation to their clinical utility. While in vitro experiments demonstrated that Fe₃O₄ could upregulate IFNβ and CXCL10, in vivo results clearly indicated its poor tumor-targeting capability and limited antitumor efficacy. In contrast, BsPNEC achieves active targeting to tumor tissues, enhancing T2-weighted MRI signals at the tumor site and exhibiting excellent tumor-targeting properties, which holds promise for the integrated diagnosis and treatment of tumors.

Despite these advances, challenges remain. The limited surface modifications of BsPNECs result in poor dispersibility and aggregation, leading to rapid uptake by macrophages and accumulation in the liver upon systemic administration. To address this, further optimization of its surface properties, such as glucan or polyethylene glycol modification, is essential to maintain a small particle size in solution, thereby enhancing its effective concentration at the tumor site.⁴⁴ Additionally, the covalent modification of peptides with groups like aryl sulfonyl fluoride group can significantly improve target affinity, offering potential applications in LYTAC-based protein degradation systems.⁴⁵ Beyond nanozyme conjugation, peptides also show promise in tumor immunotherapy and diagnostic imaging, such as peptide-radionuclide conjugates that deliver radioactive isotopes to tumor cells for dual diagnostic and therapeutic effects.^46,47 Future efforts could explore conjugating the dual-targeted peptide qGA with nuclear or other small-molecule drugs to develop more clinically valuable antitumor agents. Also, further preclinical safety evaluations are required to advance its clinical translational potential.

In conclusion, we have proposed and designed a multifunctional and biocompatible BsPNEC that can simultaneously target CXCR1/2 and PD-L1 on the surfaces of myeloid and tumor cells, which exhibits tumor targeting, MRI, and therapeutic potentials. This BsPNEC platform paves another avenue for drug design and cancer immunotherapy.

Limitations of the study

First, our findings are primarily generated from murine cell lines and animal models, while the clinically translational potential needs to be further investigated. Notably, the lack of a direct murine ortholog for human CXCL8 necessitates further validation in human settings to confirm the efficacy and safety of our therapeutic strategy. Future studies could incorporate patient-derived organoids or human CXCL8 transgenic mouse models to enhance the translational relevance.

Second, the current BsPNEC platform lacks optimized surface modification, resulting in rapid hepatic accumulation. This reduces the effective drug exposure at tumor sites and compromises the dual-targeting efficiency of the bispecific peptide, leading to suboptimal MRI contrast enhancement for tumor diagnosis (Figure 6K). To address this, future iterations will focus on surface engineering to reduce non-specific uptake.

Finally, the universality of this platform needs to be further elucidated by introducing alternative tumor-targeting ligands and functional materials. Moreover, the long-term biosafety and pharmacokinetics of BsPNEC require validation through rigorous preclinical studies for advancing its translational potential.

Resource availability

Lead contact

Requests for further information should be directed to and will be fulfilled by the lead contact, Yanfeng Gao (gaoyf29@mail.sysu.edu.cn).

Materials availability

This study did not generate new materials or unique reagents.

Data and code availability

• •Data: This paper analyzes existing, publicly available data, which are available as noted in the key resources table. All data reported in this paper will be shared by the lead contact upon request.
• •Code: This study does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported by grants from the Shenzhen Science and Technology Program (JCYJ20241202130045054), Shenzhen Medical Research Fund (D2403007), the Guangdong Basic and Applied Basic Research Foundation (2022B1515120085), and Fundamental Research Funds for the Central Universities (24xkjc021).

Author contributions

Y.G., D.C., and X. Zhu conceived the research and designed the experiments; X.S., D.C., R.X., X. Ye, Y.X., M.W., F.L., X.N., X. Yang, Y.S., W.Z., Y. Liu, W.S., W.L., Y. Li, X. Zhao, J.H., and X.C. conducted the experiments and acquired and analyzed the data; D.C., X.S., X. Zhu, and Y.G. analyzed and interpreted the results. All authors contribute to writing the manuscript. All authors revised the manuscript, discussed the results and approved the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-mouse CD45 FITC (30-F11)	eBioscience	MCD4501; RRID:AB_2621689
Anti-mouse CD3 PerCP-efluor710 (17A2)	eBioscience	46-0032-82; RRID:AB_3064875
Anti-mouse CD8α APC (53–6.7)	eBioscience	MA1-10302; RRID:AB_3696272
Anti-mouse CD8α PE (53–6.7)	eBioscience	12-0081-82; RRID:AB_2904149
Anti-mouse IFN-γ APC (XMG1.2)	eBioscience	17-7311-82; RRID:AB_2830458
Anti-mouse CD11c PE-Cy7 (N418)	eBioscience	25-0114-82; RRID:AB_2621837
Anti-mouse CD206 APC-efluor780 (MR6F3)	eBioscience	47-2061-82; RRID:AB_2802285
Anti-mouse CD11b PerCP-efluor450 (M1/70)	eBioscience	48-0112-80; RRID:AB_3713129
Anti-mouse Ly6G Percp eflour710 (1A8)	eBioscience	46-9668-82; RRID:AB_3713133
Anti-mouse Ly6C APC (Monts1)	elabscience	P0CW03; RRID:AB_3696317
Anti-mouse CD80 Percp eFluor 710 (16-10A1)	eBioscience	46-0801-82; RRID:AB_2621699
Anti-mouse CD86 FITC (GL-1)	eBioscience	11-0862-82; RRID:AB_3665944
Anti-mouse CD40 FITC (HM40-3)	eBioscience	11-0402-82; RRID:AB_312948
Human Fcγ specific PE	eBioscience	12-4998-82; RRID:AB_465926
Anti-human CD45 FITC (HI30)	eBioscience	MHCD4501; RRID:AB_2621695
Anti-human CD8 PE-Cy5 (RPA-T8)	eBioscience	15-0088-42; RRID:AB_1149051
Anti-human IFN-γ PE (4S.B3)	eBioscience	12-7319-42; RRID:AB_315233
Anti-human CD3 (OKT3)	Dakewe Biotech	317326; RRID:AB_2621540
Anti-human CD28 (CD28.2)	Dakewe Biotech	302934; RRID:AB_627002
GAPDH Rabbit mAb	ABclonal	A19056; RRID:AB_2862549
Phospho-TBK1 Rabbit mAb	ABclonal	AP1026; RRID:AB_2798526
Phospho-IRF3-S396 Rabbit mAb	ABclonal	AP1412; RRID:AB_3067078
HRP-conjugated Goat anti-Rabbit IgG	ABclonal	AS014; RRID:AB_2940954
mCXCL2/MIP-2 mAb	R&D	MAB452-SP; RRID:AB_2230058
Anti-human PD-L1 (B7-H1) (29E.2A3)	BioxCell	BE0285; RRID:AB_2228868
Bacterial and virus strains
pLVX-puro-EGFP	Laboratory preservation	N/A
pMD2.G	Laboratory preservation	N/A
psPAX2	Laboratory preservation	N/A
Chemicals, peptides, and recombinant proteins
Phorbol 12-myristate 13-acetate (PMA)	Sigma-Aldrich	P8139
Ionomycin	Sigma-Aldrich	407951
CFSE Cell Division Tracker Kit	BioLegend	B394587
N-acetylcysteine (NAC)	Selleck	S1623
Carboxyl modified Fe3O4 nanoparticles	Meilunbio	MB9864
Cy5.5 NHS ester	APExBIO	A8103
FITC NHS ester	Ruixibio	R-FF-005
2-Morpholinoethanesulfonic acid (MES)	Sigma-Aldrich	Cas: 4432-31-9
N-Hydroxysuccinimide (NHS)	Aladdin	H109330
1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide (EDC)	Aladdin	E106172
OVA antigen peptide (SIINFEKL)	Laboratory synthesis	N/A
Human PD-L1-Fc fusion protein	SinoBiological	10084-H02H
Murine PD-L1-Fc fusion protein	SinoBiological	50010-M02H
Lipopolysaccharide (LPS)	Sigma-Aldrich	L2880
murine granulocyte macrophage colony stimulating factor (mGM-CSF)	PeproTech	315–03
Critical commercial assays
Cellular RNA extraction kit	Vazyme	RC112
Reactive oxygen species detection kit	Beyotime	S0033S
Mem Preparation Kit	Beijing Priority	P1203
Deposited data
colorectal cancer (CRC) transcriptomes	GEO: GSE236581	N/A
anti-PD-1 treatment response-associated CXCL8 dynamics	GEO: GSE205506	N/A
Experimental models: Cell lines
CHOK1-hPD-1	Laboratory preservation	N/A
CHOK1-mPD-1	Laboratory preservation	N/A
Mouse-derived macrophages RAW264.7	Laboratory preservation	N/A
CT26	Laboratory preservation	N/A
CT26-sgNT	Laboratory preservation	pLK O .1
CT26-sgSTING	Laboratory preservation	N/A
KPC (Pdx1-Cre; LSL-KrasG12D; LSL-Trp53R172H/+)	From Liu Yunhua laboratory at Zhejiang University	N/A
HEK-293T	Laboratory preservation	N/A
Experimental models: Organisms/strains
C57BL/6	BesTest	Female
BALB/c	BesTest	Female
OT-1 transgenic mice	Laboratory cultivation	N/A
Oligonucleotides
mIFNβ primers: Forward - GATGAACT CCACCAGCAGACA, Reverse - CATCCAGGCGTAGCTGTTGTA	Ruibo Biotechnology	N/A
mCXCL10 primer: Forward -AAGTG CTGCCGTCATTTTCTG, Reverse - TAGGCTCGCAGGGATGATTTC	Ruibo Biotechnology	N/A
mTNFA primers: Forward - ATGGC CTCCCTCTCATCAGT, Reverse - ATAGCAAATCGGCTGACGGT	Ruibo Biotechnology	N/A
mIL-10 primers: Forward - TGCTG CCTGCTCTTACTGAC, Reverse - CTAGGAGCATGTGGCTCTGG	Ruibo Biotechnology	N/A
mARG1 primer: Forward - AGCAC TGAGGAAAGCTGGTC, Reverse - CAGACCGTGGGGGTTCTTCACA	Ruibo Biotechnology	N/A
mCXCL1 primer: Forward - CACTG CACCCAAACCGAAGT, Reverse - TGGGGACACCTTTTAGCATCTT	Ruibo Biotechnology	N/A
mCXCL2 primer: Forward - ATCCAG AGCTTGAGTGTGACG, Reverse - GGATGATTTTCTGAACCAGGGG	Ruibo Biotechnology	N/A
mCXCR1 primer: Forward - CTGGGA AACTCGCTGGTGAT, Reverse - AGGGGTGTGCCAAAAATCCA	Ruibo Biotechnology	N/A
mCXCR2 primer: Forward - GGGTCG TACTGCGTATCCTG, Reverse –AGACAAGGACGACAGCGAAG	Ruibo Biotechnology	N/A
β-actin primer: Forward-GGTGGG AATGGGGTCAGAAGG. Reverse - GTACATGGGCTGGGGGTGTTGA	Ruibo Biotechnology	N/A
Software and algorithms
ITK-SNAP	University of North Carolina	https://www.itksnap.org/pmwiki/pmwiki.php
GraphPad Prism 8	N/A	https://www.graphpad.com/
PyMol	DeLano Scientific LLC	https://pymol.org/2/
MO Affinity Analysis	Nano Temper Technologies	N/A
Other
Magnetic Resonance Imaging (MRI)	United Imaging	uPMR 9.4T
Flow cytometry	Beckman and BD/LSRForte ssa	N/A
Fluorescence imaging of small animals	Revvity	IVIS Lumina K Series III
Quantitative Real-time PCR (qRT-PCR)	ABI Quantstudio5	N/A
Monolith NT.115 system	Nano Temper Technologies	N/A
MOE	Yukangsheng	N/A
Transmission Electron Microscope (TEM)	Tecnai G2 spirit 120 kV	N/A
Malvern particle size analyzer	Zetasizer Pro	N/A
Fourier transform infrared spectroscopy (FTIR)	PE/Spectrum Two	N/A
microplate spectrophotometer	BioTek	N/A

Experimental model and study participant details

Cell lines

CHOK1-hPD-1 and CHOK1-mPD-1 were constructed by our laboratory and cultured in Roswell Park Memorial Institute (RPMI) 1640 medium. Mouse-derived macrophages RAW264.7, mouse-derived colon cancer cell line CT26, mouse-derived pancreatic cancer cell line KPC (Pdx1-Cre; LSL-KrasG12D; LSL-Trp53R172H/+) (gifted by Prof. Yunhua Liu from Zhejiang University), HEK293T-mCXCR1 and HEK293T-mCXCR2 were cultured in Dulbecco’s Modified Eagle Medium (DMEM). All the cells were cultured at 37°C under 5% CO₂ in an incubator with the above media (Gibco, Grand Island, USA) supplemented with 10% fetal bovine serum (FBS) (BI, ISR), 100 μg/mL streptomycin (Solarbio, China) and 100 U/mL penicillin (Solarbio, China). Finally, we confirmed that all cell lines used in this study were free from mycoplasma contamination by PCR.

Mice

BALB/c and C57BL/6 female mice at six weeks old were obtained from BesTest (Zhuhai, China) and housed in a specific pathogen-free facility under a controlled temperature of 24 ± 1°C. All animal experiments were approved by the Ethics Committee of Sun Yat-sen University and carried out following national guidelines.

Study approval

All animal experiments were approved by the Ethics Committee of Sun Yat-sen University (approval number: SYSU-IACUC-2024-001469) and carried out following national guidelines.

Method details

Bioinformatics analysis

To investigate the role of CXCL8 across cancer types, transcriptomic and clinical data were acquired from the GDC TCGA dataset through the UCSC Xena browser (https://xenabrowser.net/datapages/). RNA-seq profiles and corresponding clinical metadata were analyzed to compare CXCL8 expression between tumor and matched normal tissues, followed by Kaplan-Meier survival analysis stratified by median CXCL8 expression levels. For single-cell resolution, colorectal cancer (CRC) transcriptomes were examined using the Single Cell RNA-seq Data Visualization and Analysis platform (http://crc-icb.cancer-pku.cn/), with GEO: GSE236581 interrogated to evaluate CXCL8 expression patterns in tumor microenvironments. Additionally, anti-PD-1 treatment response-associated CXCL8 dynamics were assessed using GEO: GSE205506. Finally, potential immune regulatory mechanisms were explored through correlation analysis on GEPIA (http://gepia.cancer-pku.cn/), where Spearman’s rank test was applied to examine associations between CXCL8 and immune checkpoint gene CD274 in TCGA-COAD/READ cohorts.

Peptide synthesis

All peptides used in this study were synthesized in our laboratory by Fmoc solid-phase synthesis according to the standard protocol.⁴⁸ Reversed-phase high-performance liquid chromatography (RP-HPLC) was applied for peptide purification, with all compounds demonstrating >95% purity by HPLC analysis. The molecular weights were confirmed by electrospray ionization-mass spectrometry (ESI-MS).

Quantitative RT-PCR analysis of CXCL1/2 and CXCR1/2

RAW264.7 cells (1×10⁶ cells/well) were seeded in 6-well plates and allowed to adhere. After removing the culture medium, cells were treated with 250 ng/mL LPS for 48 h, with PBS-treated cells serving as negative controls. Total RNA was extracted, and reversely transcribed to cDNA. Quantitative RT-PCR was carried out to determine the expression levels of CXCL1, CXCL2, CXCR1, and CXCR2.

Transwell migration assay

Transwell chambers with an 8 μm pore size were placed into 24-well plates. Each chamber was seeded with 2×10⁵ cells and cultured for 24 h until cells reached 90–95% confluency. The original medium was discarded, and fresh medium containing 50 μM peptide was added to the upper chamber. The lower chamber was filled with fresh medium containing 250 ng/mL LPS. After 48 h of coculturing, the medium was aspirated, and the chambers were washed with PBS. Cells were fixed with 4% paraformaldehyde for 30 min, stained with 0.2% crystal violet for 25 min, and photographed under a microscope (10× objective lens).

Molecular docking and molecular dynamics mutations

The three-dimensional (3D) structure of the D10 peptide was predicted using the PEP-FOLD4 online server (RPBS Web Portal, https://mobyle.rpbs.univ-paris-diderot.fr/) by submitting its amino acid sequence. Meanwhile, the model of CXCR1/2 was predicted using the AlphaFold protein structure database (UniProt accession), followed by structural refinement using MOE molecular modeling software. Molecular docking was performed via the ZDOCK server (https://zdock.umassmed.edu/) to predict the binding modes of D10 with CXCR1/2, and the interaction patterns were analyzed in MOE. Subsequently, the peptide-protein complexes were protonated, energy-minimized (with protein constraints), and subjected to molecular dynamics simulation (NPA algorithm, OPLS-AA force field, 100 ps equilibration at 300 K, 50 ns production run at 300 K) for further mutation studies.

Construction of HEK-293T^mCXCR1 and HEK-293T^mCXCR2

The CXCR1 or CXCR2 overexpression plasmid was generated by inserting CXCR1 or CXCR2 CDS sequence into the pLVX-puro-EGFP plasmid vector. HEK-293T cells were co-transfected with either pLVX-puro-mCXCR1-EGFP or pLVX-puro-mCXCR2-EGFP lentiviral transfer vectors along with the pMD2.G and psPAX2 helper plasmids at a 4:3:1 ratio using polyethylenimine (PEI). Following 48 h of incubation, the viral supernatant was harvested and filtered through a 0.45 μm membrane for target cell infection. When HEK-293T cells reached approximately 30% confluence, lentiviral supernatant with 8 μg/mL of polybrene was added for infection. After 24 h, the medium was replaced with fresh complete medium, followed by selection with puromycin at 48 h post-infection to establish stably transduced cell populations.

Microscale thermophoresis (MST)

The membrane proteins of HEK-293T^mCXCR1 and HEK-293T^mCXCR2 were extracted using the Membrane Preparation Kit. The peptides were solubilized with PBST at a maximum concentration of 200 μM, followed by serial 2-fold dilutions to generate 16 concentrations. Equal volumes of peptide solutions and the above membrane protein samples were then mixed and incubated at RT for 10 min. Samples were then loaded into capillary tubes, and analyses were performed using the Monolith NT.115 system (Nano Temper Technologies, Germany). The dissociation constant (K_D) values were analyzed by MO Affinity Analysis software.

Mouse bone marrow-derived macrophages (BMDM) polarization experiment

BMDMs were generated from the hind limbs of C57BL/6 mice. Briefly, the hind limbs were removed aseptically, and the bone marrow was flushed out with PBS. The cell suspension was filtered through a 200-mesh screen and centrifuged at 3000 rpm for 5 min to collect the cells. Erythrocytes were lysed using ACK Lysis Buffer, followed by two washes with PBS. Cells were then cultured in DMEM medium supplemented with 20 ng/mL murine granulocyte macrophage colony stimulating factor (mGM-CSF). Starting on day 3, half-medium changes with supplementation of half-dose mGM-CSF were performed daily until day 5 of differentiation, at which point the adherent cells were considered BMDMs.

For polarization experiments, 1×10⁶ BMDMs were seeded in a 6-well plate and treated with CT26-conditioned medium containing 50 μM peptide. At 48 h post-stimulation, half of the medium was replaced with fresh CT26-conditioned medium supplemented with 50 μM peptide. Total RNA was extracted at 72 h using a cellular RNA extraction kit (RC112, Vazyme). Gene expression levels of mTNFA, mIL-10 and mARG1 were quantified by quantitative real-time PCR (qRT-PCR) on a qRT-PCR instrument (ABI Quantstudio5, USA). The internal reference was β-actin. To assess the impact of mCXCL2 derived from CT26-conditioned medium on phenotypic alterations, 2 μg/mL mCXCL2 was added to the induction system, and the aforementioned process was repeated.

Human serum enzyme degradation assay

The peptide was dissolved in PBS (pH = 7.2) and incubated with 10% human serum solution at a final concentration of 200 μM to evaluate its stability. An initial aliquot was collected immediately (0 h timepoint), while the remaining mixture was maintained at 37°C for continuous monitoring. Samples were subsequently withdrawn at predetermined intervals (0.5, 1, 2, 4, 6, 24, and 48 h), with the proteolytic reaction immediately terminated by adding an equal volume of 10% perchloric acid followed by vigorous vortex mixing. After centrifugation at 10,000 × g for 20 min at 4°C, the resulting supernatants were filtered through 0.22 μm membranes and analyzed by RP-HPLC. Stationary phase: C18 analytical column (150 × 4.6 mm × 5 μm, 100 Å, Waters), mobile phase: H₂O (1‰ TFA) and acetonitrile, flow rate: 1 mL/min, absorption: 220 nm (UV-Vis detector).

PD-1/PD-L1 blocking assay

Peptides were dissolved in PBS (pH = 7.2) to a final concentration of 100 μM. Human PD-L1-Fc or murine PD-L1-Fc fusion proteins (50 ng) were incubated with 50 μL of peptide solution for 30 min on ice after brief vortex mixing. Tube containing proteins but without peptide served as positive control. Subsequently, CHOK1-mPD-1 or CHOK1-hPD-1 cells were incubated with the above mixture for 30 min at 4°C. After incubation, cells were washed with PBS and stained with anti-human Fc-PE antibody in the dark for 30 min. Cells were collected and the mean fluorescent intensity (MFI) was detected by flow cytometry (Beckman Coulter, USA) with the PE channel. The blocking rates were calculated as the formula: blocking rate (%) = (MFI value of the positive control – MFI of the tested peptides)/MFI value of the positive control × 100%.

Human peripheral blood mononuclear cell (PBMC) stimulation experiment

PBMCs were isolated from fresh whole blood of healthy donors using density gradient centrifugation. Briefly, whole blood was mixed 1:1 with PBS and the mixture was then carefully layered onto the lymphocyte separation solution at a 2:1 ratio (mixture: lymphocyte separation solution). Centrifugation was performed at 2000 rpm for 20 min at room temperature with low acceleration. After completion, the PBMC layer was retrieved, and resuspended in 1640 medium.

For T cell stimulation, the obtained PBMCs were activated with anti-human CD3 (1 μg/mL) and anti-human CD28 (0.5 μg/mL) in the presence of 100 μM peptides or 10 μg/mL anti-human PD-L1. Protein transport inhibitor (GolgiPlug) was added at the final 4 h of 72 h co-culture. Cells were then harvested, washed with PBS, and stained with anti-human CD45-FITC and anti-human CD8-PE-Cy5 for 30 min at 4°C. Subsequently, cells were fixed by adding 200 μL of 4% paraformaldehyde and permeabilized using a membrane disruptor. Intracellular IFN-γ was detected with anti-human IFN-γ-PE. Flow cytometry (Beckman Coulter, USA) was performed to detect the proportion of CD8⁺IFN-γ⁺ T cells.

In vivo distribution of peptides

CT26 cells were collected and subcutaneously inoculated into the right flank of 6-week-old female BALB/c mice at 2 × 10⁵ cells per mouse. When the tumor volume reached about 40–70 mm³, mice received intraperitoneal injection of either 5 mg/kg FITC- conjugated qGA peptide or an equimolar dose of control peptide FITC-conjugated GA peptide (GAGAAGGAGGGGG), 3.5 mg/kg FITC-q6w and 2.5 mg/kg Cy5.5-OPBP-1(8–12). At 6, 12, and 24 h post-injection, the mice were sacrificed, and tissues, including the brain, heart, liver, spleen, lung, kidneys, lymph nodes, and tumors were collected for ex vivo fluorescence imaging. Fluorescence intensity was measured using an imaging system (IVIS Lumina III, PerkinElmer, USA) and normalized to PBS-injected background values.

Validation of in vivo anti-tumor effect of bispecific peptides

For CT26 subcutaneous xenograft model, CT26 cells were implanted as above-mentioned. When the tumor volume reached approximately 50 mm³, the mice were randomly divided into 6 groups and administered daily intraperitoneal injections (200 μL/mouse) for two weeks. The dosages were as follows: 0.5 mg/kg OPBP-1 (8–12), 1 mg/kg q6w, 1.55 mg/kg qGA, and 2.5 mg/kg Reparixin was administered every other day.⁴⁹ Body weights were monitored using an electronic balance, and tumor volumes were measured with a digital vernier caliper every other day. The tumor volumes were calculated using the formula: V = 1/2 × a (length) × b (width) × c (height).

Preparation and characterization of bispecific peptide-nanozyme conjugates (BsPNEC)

The coupling process was initiated by dispersing Fe₃O₄ nanoparticles (1 mg/mL) in MES buffer via sonication, followed by carboxyl group activation through 15 min incubation at 37°C with NHS/EDC. Peptides dissolved in DMSO were subsequently introduced to the activated nanoparticles and allowed to react for 2 h at 37°C. After the reaction, the BsPNEC was isolated by centrifugation (12,000 rpm, 20 min) and washed three times with ultrapure water. Coupling efficiency was quantified by RP-HPLC analysis of unconjugated peptides in the 0.22 μm-filtered supernatant. Successful conjugation was confirmed through Fourier transform infrared spectroscopy (FTIR) (PE/Spectrum Two, USA) by analyzing characteristic functional groups in potassium bromide-pressed pellets. The particle size and zeta potential of BsPNEC were analyzed by Malvern particle size analyzer (Zetasizer Pro, UK). For morphological assessment, samples were dispersed in anhydrous ethanol, deposited onto carbon-coated copper grids, and air-dried before imaging using TEM (Tecnai G2 spirit 120 kV, the Netherlands).

Cell proliferation experiment

Cells were seeded in 96-well plate at a density of 4,000 cells/well and adhered overnight. After discarding the original culture medium, cells were incubated for an additional 8 h in serum-free medium. Subsequently, peptides were added to cell culture medium at a final concentration of 100 μM, 50 μM, 25 μM, 12.5 μM, 6.25 μM, and 0 μM, followed by incubation for 24, 48, or 72 h. Following treatment, 5 mg/mL MTT solution was added to each well. After incubation for 4 h, the original culture medium was carefully discarded, and DMSO was added to dissolve formazan crystals. Absorbance was measured at 490 nm using a microplate spectrophotometer (BioTek, USA). For the cytotoxicity assessment of BsPNEC, the same procedure was followed, with drug concentrations of 100, 30, 10, 3, 1, and 0.3 μg/mL.

Red blood cell (RBC) hemolysis test

The hemolytic potential of peptides was evaluated using freshly prepared human RBCs. Peptides were dissolved in PBS at a stock concentration of 100 μM, followed by serial dilution to achieve final concentrations ranging from 6.25 to 100 μM. Equal volumes of peptide solutions were mixed with 2% (v/v) RBC suspension (prepared in PBS) and incubated at 37°C for 1 h. After incubation, samples were centrifuged at 1,000 × g for 5 min at 25°C. The supernatant absorbance was measured at 540 nm using a microplate reader. The ultrapure water and PBS served as positive and negative controls, respectively.

Peroxidase-like activity assay

Peroxidase-like activity was conducted using a standard colorimetric assay in 0.1 M HAc-NaAc buffer (pH = 3.5). The reaction system contained 30% H₂O₂ as substrate and tetramethylbenzidine (TMB) as chromogenic agent, with Fe₃O₄ nanozyme and BsPNEC serving as peroxidase. All components were mixed in 1.5 mL microcentrifuge tubes and incubated at 40°C for 15 min. Following the reaction, aliquots were transferred to a 96-well flat-bottom microplate, and absorbance was measured at 652 nm using a microplate reader. Kinetic parameters (K_M and V_MAX) were determined by nonlinear regression analysis of the Michaelis-Menten curves using GraphPad Prism 8 software.

Reactive oxygen species (ROS) detection

CT26 cells were seeded in a 24 well plate at a density of 2×10⁵ cells/well and allowed to adhere. Subsequently, Fe₃O₄ nanozyme and BsPNEC were added at a final concentration of 200 μg/mL. After incubation for 6, 12, and 24 h, the cells were collected and stained with 10 μM 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) as an ROS-sensitive fluorescent probe. The cells were then incubated at 37°C for 20 min, with gentle inversion every 5 min to ensure uniform probe distribution. Following incubation, unbound DCFH-DA was removed by washing three times with PBS. Finally, intracellular ROS levels were quantified by flow cytometry (Beckman Coulter, USA).

cGAS-STING activating effect of BsPNEC

CT26 cells were seeded in a 6-well plate at a density of 5×10⁵ cells/well and allowed to adhere overnight. Cells were then treated with Fe₃O₄ nanozyme and BsPNEC (200 μg/mL final concentration) or cGAMP (1.5 μg/mL) as a positive control. After incubation for 6, 12, and 24 h, total RNA was extracted, and reversely transcribed to cDNA. Gene expression levels of mIFNβ and mCXCL10 were detected by qRT-PCR, with β-actin serving as the internal reference. Following a 4-h pretreatment with the antioxidant N-acetylcysteine (NAC) to scavenge ROS, cells were incubated with 200 μg/mL BsPNEC for 12 h before sample collection. CT26 cells with STING knockout were treated with 200 μg/mL BsPNEC for 12 h, followed by quantification of IFNβ and CXCL10 expression levels. Phosphorylated TBK1 (pTBK1) and phosphorylated IRF3 (pIRF3) expression was additionally assessed via Western blotting.

Effects of BsPNEC on BMDC maturation and antigen cross-presentation

Bone marrow cells were isolated from 6-week-old female C57BL/6 mice and induced for 5 days with 20 ng/mL mGM-CSF and mIL-4. Subsequently, the cells were treated for 24 h with the following stimuli: 10 μg/mL OVA_257-264 (SIINFEKL), 100 μg/mL Fe₃O₄-OVA, 100 μg/mL BsPNEC, or 5 μg/mL LPS. Surface expression of CD80, CD86, and CD40 was analyzed by flow cytometry.

Concurrently, CD8⁺ T cells from OT-I transgenic mice, which possess T cell receptors specific for the OVA_257-264 peptide (SIINFEKL), were purified from the spleens and lymph nodes, labeled with 5 μM carboxyfluorescein succinimidyl ester (CFSE) to track cell proliferation, and co-cultured at a 10:1 ratio (OT-Ⅰ: BMDC) with BMDCs from each treatment group for 72 h. T cell proliferation and IFNγ secretion were assessed by flow cytometry.

In vivo anti-tumor validation of BsPNEC

CT26 subcutaneous tumor xenograft models were established as above described. When the tumor volume reached approximately 50 mm³, mice were randomly divided into 4 groups. Treatments were administered via tail vein injection every two days for 2 weeks. The dosages were as follows: 1.55 mg/kg qGA, 5 mg/kg Fe₃O₄ or BsPNEC.

For pancreatic tumor model, KPC cells were collected and subcutaneously inoculated into the right flank of 6-week-old female C57BL/6 mice at 1 × 10⁶ cells per mouse. When the tumor volume reached approximately 50 mm³, the mice were randomly divided into 6 groups and treated for two weeks. The dosages were as follows: 1.55 mg/kg qGA, 5 mg/kg Fe₃O₄ or BsPNEC, 1 mg/kg q6w, 10 mg/kg IgG or anti-PD-L1. Among them, qGA was administered intraperitoneally every day. Fe₃O₄ and BsPNEC were administered via tail vein every day. IgG and anti-PD-L1 were administered intraperitoneally every three days. Body weights were monitored using an electronic balance, and tumor volumes were measured using a digital vernier caliper every other day. The tumor volume of the mice was calculated using the following formula: V = 1/2 × a (length) × b (width) × c (height).

Ex vivo testing and analysis of tumor infiltrating immune cells

After treatment completion, tumor tissues were removed and digested with 100 U/mL collagenase IV and DNase I into single-cell suspensions. Cells were then washed twice with PBS and equally distributed into three aliquots for parallel immune cell subset analysis. For CD8⁺ T cells infiltration, cells were stained with anti-mouse CD45-FITC, anti-mouse CD3-PerCP-eFluor710, and anti-mouse CD8α-APC for 30 min at 4°C. After staining, cells were washed and resuspended in flow cytometry buffer and were further analyzed by flow cytometry. For macrophage polarization analysis, samples were stained with anti-mouse CD45-FITC, anti-mouse CD11b-PerCP-Cy5.5, anti-mouse F4/80-APC, anti-mouse CD11c-PE-Cy7, and anti-mouse CD206-APC-efluor780 to distinguish M1 and M2 subsets. Neutrophil populations were identified using anti-mouse CD45-FITC, anti-mouse CD11b-PerCP-efluor450, anti-mouse Ly6G-PerCP-efluor710, and anti-mouse Ly6C-APC, with all staining procedures performed under identical conditions.

For functional assessment of tumor-infiltrating T cells, intracellular factor staining for IFN-γ was performed. First, cells were plated in 24-well plates and treated with PMA (20 ng/mL), ionomycin (1 μM), and the protein transport inhibitor. After 4 h, cells were collected and surface-stained with anti-mouse CD3-PerCP-eFlur710 and anti-mouse CD8-PE at 4°C for 30 min. After fixation and permeabilization, intracellular staining was performed with anti-mouse IFN-γ-APC to identify IFN-γ-producing CD8⁺ T cells using a flow cytometer.

The histopathological analysis of major organs and serum biochemical parameter assays were outsourced to Servicebio Biotechnology.

Magnetic resonance imaging (MRI)

A rodent receiver coil and 9.4T MRI scanner were used. When CT26 tumor-bearing BALB/c mice reached 100–200 mm³ in volume, they were intravenously injected via the tail vein with either 10 mg/kg BsPNEC or an equivalent dose of Fe₃O₄ nanozyme. Six hours after the injection, T2 weighted MRI imaging was performed. The specific parameters are as follows: TR 3000 ms; TE 48 ms; FOV read 35 × 35 mm; matrix 256 × 256; slice thickness 1 mm; averages 2.

Quantification and statistical analysis

Data are presented as means ± SEM. GraphPad Prism8 and Microsoft Excel were used for data analysis. Statistical analyses were performed using either unpaired two-tailed Student’s t-tests or one-way ANOVA to determine group differences, as appropriate for experimental design and data structure. All of the statistical details of experiments can be found in the figure legends. Statistical significance was determined as follows.

• •∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 for comparisons between negative control/positive control groups and drug-treated groups.
• •#P < 0.05, ##P < 0.01, and ###P < 0.001 for comparisons among drug-treated groups.

Published: January 21, 2026