雌激素重编程STING与ADC的靶点可及性

Figure 1. A paired ‘‘phase II-like’’ PDX preclinical trial reveals sex-dimorphic responses to CLDN18.2-ADC (A) Design of the hCD34+ humanized PDX preclinical trial. (B) Waterfall plot of the best tumor volume change from baseline (best response, defined as the minimum percentage change recorded from day 10 to day 21 after treatment initiation) in CLDN18.2-ADC-treated PDX recipients. Responses were classified according to mRECIST criteria. (C) Objective response rates stratified by recipient sex, donor sex, age, CA19-9 level, differentiation, and CLDN18.2 expression. Response was defined as mCR or mPR; non-response was defined as mSD or mPD. (D) mRECIST response distribution by recipient sex (n = 55 per sex). (E and F) Representative annexin V/PI flow cytometry (E) and quantification of apoptotic tumor cells (F) in treated PDX tumors. (G) CLDN18.2-ADC treatment scheme in male and female spontaneous KPC mice. (H and I) Gross tumor images (H), tumor weight, and tumor inhibition rate (I) in KPC mice (n = 5 per group). (J and K) Representative immunofluorescence images (J) and quantification (K) of Ki67+ tumor cells in KPC mice (n = 5 per group). Scale bars, 50 μm. (L–N) Tumor-cell apoptosis by flow cytometry (L; n = 5 per group), Bax and Bcl-2 immunoblotting (M; n = 5 per group), and Kaplan-Meier survival (N; n = 8 per group) in KPC mice. (O) Treatment scheme for subcutaneous MKN45-CLDN18.2 tumor cell xenografts in male and female mice. (P–S) Tumor growth (P), tumor weight (Q), Ki67+ tumor cells (R), and tumor-cell apoptosis (S) in the gastric cancer model (n = 5 per group). Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (I) (tumor inhibition); paired two-tailed Student’s t test (F); Brown-Forsythe and Welch ANOVA with Dunnett’s T3 correction (I) (tumor weight) and (L); ordinary one-way ANOVA with multiple comparisons (K and Q–S); two-way repeated-measures ANOVA with Geisser-Greenhouse correction and multiple-comparison correction (P); log-rank (Mantel-Cox) test (N). A two-sided exact McNemar test was used for paired binary comparisons of recipient sex, and Rao-Scott chi-square tests with adjustment for PDX-level clustering were used for other categorical variables (C). The Stuart-Maxwell test was used for paired ordinal mRECIST response categories (D). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S1 and Tables S1–S3.

Figure 2. Estrogen, but not androgen, drives sex-dimorphic resistance to CLDN18.2-ADC (A and B) Serum testosterone (A) and estradiol (B) levels in male mice after sham surgery or ORX, and in female mice after sham surgery or OVX, respectively (n = 5 per group) (C) Experimental design of gonadectomy and CLDN18.2-ADC treatment in the KPC cell line orthotopic tumor model. (D–G) Tumor growth (D), tumor inhibition (E), tumor-cell apoptosis (F), and Kaplan-Meier survival (G) in SHAM and ORX male mice treated with the control or CLDN18.2-ADC (n = 5 per group for D–F; n = 8 per group for G). (H–K) Tumor growth (H), tumor inhibition (I), tumor-cell apoptosis (J), and survival (K) in SHAM and OVX female mice treated with the control or CLDN18.2-ADC (n = 5 per group for H–J; n = 8 per group for K). (L) Schematic of 17β-estradiol replacement in OVX female mice bearing orthotopically implanted KPC cell line tumors. (M) Serum estradiol levels in SHAM, OVX, and OVX+E2 mice (n = 5 per group). (N–Q) Tumor growth curves in OVX and OVX+E2 mice treated with the control or CLDN18.2-ADC (n = 5 per group). (R–U) Tumor weight and tumor inhibition (R), Ki67+ tumor cells (S), tumor-cell apoptosis (T), and survival (U) in OVX and OVX+E2 mice treated with the control or CLDN18.2-ADC (n = 5 per group for R–T; n = 8 per group for U). Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (A, B, E, and I) and for tumor inhibition (R); ordinary one-way ANOVA with multiple comparisons (F) and for tumor weight (R); Brown-Forsythe and Welch ANOVA with Dunnett’s T3 correction (J, M, S, and T); a two-way repeated-measures ANOVA with Geisser-Greenhouse correction and multiple-comparison correction (D and H); log-rank (Mantel-Cox) test (G, K, and U). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S1.

Figure 3. Estrogen induces neutrophil-dependent resistance to CLDN18.2-ADC (A) Experimental design for scRNA-seq of tumor-infiltrating CD45+ cells from SHAM and OVX female mice bearing KPC organoidluc tumors (n = 3 per group). (B–D) Single-cell RNA sequencing analysis of tumor-infiltrating CD45+ cells, including UMAP visualization of 41,260 cells (B), canonical marker expression (C), and the proportions of major cell populations (D). (E and F) Uniform manifold approximation and projection (UMAP) of neutrophils (E) and Gene Ontology analysis of differentially expressed genes (F) between SHAM and OVX groups. (G and H) Representative flow cytometry (G) and quantification (H) of tumor-infiltrating neutrophils (n = 5 per group). (I) Flow-cytometric analysis of tumor-infiltrating neutrophils after vehicle or E2 treatment in OVX female mice (n = 5 per group). (J and K) Representative multiplex immunofluorescence staining (J) and Ly6G+ neutrophil quantification (K) after vehicle or E2 treatment (n = 5 per group). Scale bars, 100 μm. (L) Experimental design for anti-Ly6G-mediated neutrophil depletion in female mice bearing KPC organoidluc tumors. (M) Peripheral-blood neutrophils after isotype or anti-Ly6G treatment (n = 10 per group). (N–P) Tumor weight (N), tumor-cell apoptosis (O), and Kaplan-Meier survival (P) after neutrophil depletion and CLDN18.2-ADC treatment (n = 5 per group for N and O; n = 8 per group for P). (Q) Experimental design for the adoptive transfer of bone-marrow-derived neutrophils into S100a8-DTA female mice bearing KPC organoidluc tumors. (R and S) Representative bioluminescence images (R) and tumor total flux (S) after vehicle or neutrophil transfer (n = 5 per group). (T–X) Effects of neutrophil transfer on tumor growth and survival, including tumor weight (T), apoptosis (U), Ki67 staining (V) and quantification (W), and Kaplan- Meier analysis (X) (n = 5 per group for T, U, and W; n = 8 per group for X). Scale bars, 50 μm. Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (H, I, K, and M); ordinary one-way ANOVA with multiple comparisons (N, T, and W); Brown- Forsythe and Welch ANOVA with multiple comparisons (O, S, and U); log-rank (Mantel-Cox) test (P and X). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S2.

Figure 4. Estrogen-induced NET formation restricts tumor-antigen accessibility and impairs ADC cytotoxicity (A) Heatmap of gene expression based on RNA sequencing of tumor-infiltrating neutrophils isolated from SHAM and OVX female mice (n = 3 per group). (B and C) Representative flow-cytometry histograms (B) and quantification (C) of neutrophil ROS levels detected by H2DCFDA staining (n = 5 per group). (D) Flow-cytometric analysis of Ly6G+MPO+ neutrophils in SHAM and OVX tumors (n = 5 per group). (E and F) CitH3+MPO+ cell quantification (E) and representative multiplex immunofluorescence staining (F) in SHAM and OVX tumors (n = 5 per group). Scale bars, 100 μm. (G) Experimental design of the tumor cell-neutrophil co-culture assay with estrogen and CLDN18.2-ADC treatment. (H) Scanning electron microscopy images of NETs with or without E2 treatment. Scale bars, 10 μm. (I) Representative fluorescence images of NETs and tumor-cell apoptosis with or without E2 treatment. Scale bars, 15 μm. (J) CLDN18.2-ADC dose-response curves with or without E2 treatment. (K) Schematic of ATTO647 labeling of CLDN18.2-ADC. (L) Representative fluorescence images and quantification of tumor-cell-associated ATTO647-ADC in tumor cell-neutrophil co-cultures (n = 15 wells per group). ATTO647-labeled ADC is shown in green pseudocolor, tumor cells in red pseudocolor, and NETs in blue pseudocolor. Scale bars, 10 μm. (M) Experimental design of in vivo ATTO647-ADC tracing in SHAM and OVX mice bearing KPC organoidluc tumors. (N and O) Representative immunofluorescence images (N) and quantification (O) of intratumoral ATTO647-ADC localization (n = 5 per group). Scale bars, 50 μm. (P) Model of NET-mediated functional antigen shielding. (Q) Generation of neutrophil-lineage Padi4-deficient (S100a8-Padi4Δ) mice. (R–T) Representative immunofluorescence images (R), CitH3-occupied tumor area (S), and tumor-cell apoptosis (T) in wild-type (WT) and S100a8-Padi4Δ female mice treated with the control or CLDN18.2-ADC (n = 5 per group). Scale bars, 50 μm. (U–W) Representative bioluminescence images (U), tumor total flux (V), and Kaplan-Meier survival (W) in WT and S100a8-Padi4Δ female mice (n = 5 per group for U and V; n = 8 per group for W). Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (D, E, and O); Mann-Whitney U test (L); ordinary one-way ANOVA with multiple com­ parisons (S); Brown-Forsythe and Welch ANOVA with multiple comparisons (C, T, and V); four-parameter nonlinear regression model (J); log-rank (Mantel-Cox) test (W). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figures S3–S5 and Tables S4–S9.

Figure 5. Estrogen redirects tumor-cell STING signaling toward NF-κB to promote NET formation (A–C) Experimental scheme (A), representative immunofluorescence images (B), and quantification (C) of NET formation in neutrophils cultured alone or co- cultured with tumor cells with or without E2 treatment (n = 3). Scale bars, 50 μm. (D–F) Experimental scheme (D), representative images of neutrophil migration and NET immunofluorescence staining (E), and quantification (F) of NET formation in neutrophils treated with conditioned media from CLDN18.2- or CLDN18.2+ tumor cells with or without E2 treatment (n = 3). Scale bars, 50 μm. (G–I) Design of STING inhibition with SN-011 (G) and quantification of CitH3+ neutrophils (H) and neutrophil migration (I) after exposure to conditioned media from CLDN18.2+ tumor cells (n = 3). (J–L) Quantification of MPO+CitH3+ cells (J), the experimental design (K), and representative immunofluorescence images (L) in SHAM and OVX mice bearing WT or Sting1-knockout (KO) KPC organoidluc tumors (n = 5 per group). Scale bars, 50 μm. (M) Schematic of STING signaling through the TBK1-IRF3/type I interferon and IKK-NF-κB axes. (N) Heatmap of RT-qPCR analysis of NF-κB-related and interferon-response genes in CLDN18.2+ tumor cells treated with vehicle or E2 (n = 3). (O) Immunoblot analysis of STING pathway signaling in CLDN18.2+ tumor cells with or without E2 treatment. (P–R) Representative immunofluorescence images of NET staining and neutrophil migration (P), quantification of CitH3+ neutrophils (Q), and neutrophil migration (R) after exposure to conditioned media from CLDN18.2+ tumor cells treated with estrogen and/or TPCA-1 (n = 3). Scale bars, 50 μm. (S and T) CitH3+ neutrophils (S) and neutrophil migration (T) after exposure to conditioned media from CLDN18.2+ tumor cells treated with E2 and/or GSK8612 (n = 3). (U and V) The TPCA-1 treatment scheme (U) and representative immunofluorescence images with the quantification of intratumoral NET formation (V) in female mice bearing KPC organoidluc tumors (n = 5 per group). Scale bars, 100 μm. Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (V); ordinary one-way ANOVA with multiple comparisons (C, F, H, I, and Q–T); Brown- Forsythe and Welch ANOVA with multiple comparisons (J). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figures S6–S10.

Figure 6. CXCL1 from estrogen-reprogrammed tumor cells promotes CXCR2-dependent NET formation (A–C) Size fractionation of conditioned medium from estrogen-treated CLDN18.2+ tumor cells using a 3-kDa molecular-weight cutoff filter (A), followed by quantification of neutrophil migration (B) and CitH3+ neutrophils (C) in response to the <3-kDa and >3-kDa fractions (n = 3). (D and E) Quantification of neutrophil migration (D) and CitH3+ neutrophils (E) in response to untreated or heat-inactivated conditioned medium (n = 3). (F and G) Secretome proteomics of vehicle- and estrogen-treated CLDN18.2+ tumor cells (F) and the overlap of differentially expressed proteins with NF-κB target genes defined by the HALLMARK_TNFA_SIGNALING_VIA_NFKB gene set (G). (H) CXCL1 concentrations in the conditioned medium measured by ELISA (n = 3). (I–K) Neutrophil migration (I), CitH3+ neutrophils (J), and ROS levels (K) after exposure to the conditioned medium with IgG or CXCL1-neutralizing antibody (n = 3). (L) CXCL1 secretion after E2 and/or TPCA-1 treatment measured by ELISA (n = 3). (M and N) Serum CXCL1 levels in male and female KPC organoidluc-bearing mice (M; n = 6 per group) and in patients with CLDN18.2−or CLDN18.2+ PDAC (N; n = 88). (O) Association between tumor CXCL1 levels and neutrophil infiltration in patients with PDAC (n = 69). (P) Representative multiplex immunofluorescence images of CXCL1 and NET formation in CLDN18.2−and CLDN18.2+ PDAC. Scale bars, 50 μm. (Q and R) Experimental design and representative images (Q), with quantification of PDAC organoid apoptosis (R), after neutrophil co-culture and CLDN18.2-ADC treatment with or without a CXCL1-neutralizing antibody (n = 3). Scale bars, 100 μm. (S–W) Treatment scheme (S), intratumoral MPO+CitH3+ neutrophils (T), tumor weight (U), tumor-cell apoptosis (V), and Kaplan-Meier survival (W) after CXCL1 neutralization and CLDN18.2-ADC treatment (n = 5 per group for T–V; n = 8 per group for W). (X and Y) Generation of neutrophil-lineage Cxcr2-deficient (S100a8-Cxcr2Δ) mice (X) and quantification of intratumoral MPO+CitH3+ neutrophils in WT and S100a8-Cxcr2Δ female mice (Y; n = 5 per group). Data are presented as mean ± SD. Unpaired two-tailed Student’s t test (H and M); Welch’s unpaired t test (Y); Mann-Whitney U test (N); ordinary one-way ANOVA with multiple comparisons (B–E, I–L, R, and T–V); linear regression analysis (O); log-rank (Mantel-Cox) test (W). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 7. Sivelestat-mediated NET inhibition enhances CLDN18.2-ADC efficacy in pancreatic and gastric cancer models (A and B) Representative immunofluorescence images (A) and CitH3+ neutrophil quantification (B) in neutrophils treated with conditioned media from E2-treated CLDN18.2+ tumor cells with or without sivelestat (n = 3). Scale bars, 50 μm. (C) Tumor-cell apoptosis in tumor cell-neutrophil co-cultures treated with CLDN18.2-ADC with or without sivelestat under E2 stimulation (n = 3). (D) CLDN18.2-ADC dose-response curves in tumor cell-neutrophil co-cultures treated with E2 and/or sivelestat (n = 3). (E and F) The treatment scheme (E) and body weight (F) of female KPC mice receiving CLDN18.2-ADC and/or sivelestat (n = 5 per group). (G–I) Representative immunofluorescence images (G), CitH3-occupied tumor area (H), and tumor-cell apoptosis (I) in KPC tumors (n = 5 per group). Scale bars, 100 μm. (J) Representative images of the intratumoral localization of ATTO647-labeled CLDN18.2-ADC. Scale bars, 100 μm. (K–N) Representative CK19/Ki67 immunofluorescence (K), Ki67+ tumor cells (L), tumor weight (M), and Kaplan-Meier survival (N) in female KPC mice (n = 5 per group for L and M; n = 8 per group for N). Scale bars, 50 μm. (O–S) The treatment scheme of subcutaneous MKN45-CLDN18.2 tumor-cell xenografts established in humanized mice (O), CitH3-occupied tumor area (P), Ki67+ tumor cells (Q), tumor-cell apoptosis (R), and tumor weight (S) after CLDN18.2-ADC and/or sivelestat treatment (n = 5 per group). Data are presented as mean ± SD. Ordinary one-way ANOVA with multiple comparisons (B, C, L, M, P, Q, and S); Brown-Forsythe and Welch ANOVA with multiple comparisons (H, I, and R); two-way repeated-measures ANOVA (F); four-parameter nonlinear regression (D); log-rank (Mantel-Cox) test (N). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figures S11–S13.