Targeting the bicarbonate transporter SLC4A4 overcomes immunosuppression and immunotherapy resistance in pancreatic cancer

Abstract

Solid tumors are generally characterized by an acidic tumor microenvironment (TME) that favors cancer progression, therapy resistance and immune evasion. By single-cell RNA-sequencing analysis in individuals with pancreatic ductal adenocarcinoma (PDAC), we reveal solute carrier family 4 member 4 (SLC4A4) as the most abundant bicarbonate transporter, predominantly expressed by epithelial ductal cells. Functionally, SLC4A4 inhibition in PDAC cancer cells mitigates the acidosis of the TME due to bicarbonate accumulation in the extracellular space and a decrease in lactate production by cancer cells as the result of reduced glycolysis. In PDAC-bearing mice, genetic or pharmacological SLC4A4 targeting improves T cell-mediated immune response and breaches macrophage-mediated immunosuppression, thus inhibiting tumor growth and metastases. In addition, Slc4a4 targeting in combination with immune checkpoint blockade is able to overcome immunotherapy resistance and prolong survival. Overall, our data propose SLC4A4 as a therapeutic target to unleash an antitumor immune response in PDAC.

Fig. 1. SLC4A4 is almost exclusively expressed in the epithelial compartment of PDAC tumors.

a, Uniform manifold approximation and projection (UMAP) map of color-coded cells for the indicated cell types isolated from treatment-naive individuals with PDAC (n = 10 individuals). b, Dot plot of normalized expression of different bicarbonate transporters in the indicated cell types. c, UMAPs showing the expression of EPCAM (marker of pan-epithelial cells), SLC4A4, SPP1 (marker of ductal cells) and SPINK1 (marker of acinar cells). d, Dot plot of normalized expression of SLC4A4 in the epithelial compartment of each individual. e, Violin plot from TCGA data representing SLC4A4 expression in PDAC or adjacent pancreatic tissue (n = 182 individuals). f,g, Representative images of IHC staining for SLC4A4 in PDAC (f) or adjacent pancreatic tissue sections (g); n = 7 individuals; scale bar, 1 mm. P value was assessed by unpaired, two-tailed Student’s t-test (e). Source data

Fig. 2. Slc4a4 targeting reduces tumor growth and metastases.

a, b, Growth (a) and weight (b) of sgNT (n = 15) and sgSlc4a4 (n = 11) subcutaneous Panc02 tumors. Data are representative of two independent experiments. c, Weight of sgNT (n = 6) and sgSlc4a4 (n = 6) orthotopic Panc02 tumors. Data are representative of two independent experiments. d, Quantification of macroscopic metastatic mesenteric lymph nodes of sgNT (n = 6) and sgSlc4a4 (n = 6) orthotopic Panc02 tumors. e, Tumor growth of sgNT (n = 9) and sgSlc4a4 (n = 8) subcutaneous Panc02 tumors using a second gRNA against Slc4a4. f, Weight of sgNT (n = 9) and sgSlc4a4 (n = 9) subcutaneous Panc02 tumors using a second gRNA against Slc4a4. g, Weight of sgNT (n = 9) and sgSlc4a4 (n = 9) orthotopic KPC₁ tumors. Data are representative of two independent experiments. h, Quantification of macroscopic metastatic mesenteric lymph nodes of sgNT (n = 9) and sgSlc4a4 (n = 9) orthotopic KPC₁ tumors. i, j, Representative images of CK19 in livers of sgNT (i; n = 10) and sgSlc4a4 (j; n = 8) orthotopic KPC₁ tumor-bearing mice. The dashed line separates normal liver tissue from metastatic lesions. k, Body weight of sgNT (n = 9) and sgSlc4a4 (n = 9) orthotopic KPC₁ tumor-bearing mice. l, m, Quantification (l) and representative images (m) of macroscopic liver metastatic nodules in mice hydrodynamically injected with sgNT (n = 7) and sgSlc4a4 (n = 6) KPC₁ cells. Arrows indicate liver nodules (m). n, o, Quantification (n) and representative images (o) of macroscopic lung metastatic nodules in mice hydrodynamically injected with sgNT (n = 7) and sgSlc4a4 (n = 6) KPC₁ cells. p, Weight of sgNT (n = 8) and sgSlc4a4 (n = 9) orthotopic KPC₃ tumors. Data are representative of two independent experiments. q, Quantification of macroscopic metastatic mesenteric lymph nodes of sgNT (n = 8) and sgSlc4a4 (n = 9) orthotopic KPC₃ tumors. P values were assessed by unpaired, two-tailed Student’s t-test (b–d, f–h, l, n, p and q) and two-way ANOVA with Sidak’s multiple comparison test (a, e and k). Graphs show mean ± s.e.m.; LM, liver metastasis; scale bars, 50 μm (i and j). Source data

Fig. 3. Genetic and pharmacologic Slc4a4 targeting inhibits tumor growth and metastases in immunocompetent mice.

a,b, Weight (a) and macroscopic metastatic mesenteric lymph nodes (b) of sgNT and sgSlc4a4 orthotopic KPC₁ tumors in mice systemically treated with the SLC4A4 inhibitor DIDS (15 mg per kg (body weight) twice daily by i.p. injections) from day 5 to day 15 (sgNT DMSO n = 14, sgNT DIDS n = 16, sgSlc4a4 DMSO n = 8, sgSlc4a4 DIDS n = 9). Data show a pool of two independent experiments; NS, not significant. c,d, Quantification (c; left) and representative images (d) of pHH3 (red) in sgNT (n = 6) and sgSlc4a4 (n = 7) orthotopic Panc02 tumors. Hoechst (in blue) was used to stain the nuclei. c,e, Quantification (c; right) and representative images (e) of pHH3 (red) in sgNT (n = 8) and sgSlc4a4 (n = 7) orthotopic KPC₁ tumors. Hoechst (in blue) was used to stain the nuclei. f,h, Representative images (f) and quantification (h; left) of TUNEL (green) stainings in sgNT (n = 6) and sgSlc4a4 (n = 7) orthotopic Panc02 tumors. Hoechst (in blue) was used to stain the nuclei. g,h, Representative images (g) and quantification (h; right) of TUNEL stainings (green) in sgNT (n = 8) and sgSlc4a4 (n = 8) orthotopic KPC₁ tumors. Hoechst (in blue) was used to stain the nuclei. i,j, Growth (i) and weight (j; left) of sgNT (n = 8) and sgSlc4a4 (n = 8) subcutaneous Panc02 tumors in nude mice. j, Weight of sgNT (n = 7) and sgSlc4a4 (n = 8) KPC₁ orthotopic tumors injected in nude mice (right). P values were assessed by two-way ANOVA with Tukey’s multiple comparison test (a and b), unpaired two-tailed Student’s t-test (c, h and j) and two-way ANOVA with Sidak’s multiple comparison test (i). Graphs show mean ± s.e.m.; scale bars, 20 μm (d–g). Source data

Fig. 4. Slc4a4 targeting decreases extracellular acidification and glycolysis.

a, [¹⁴C]Bicarbonate uptake in sgNT (n = 5) and sgSlc4a4 (n = 5) Panc02 (left) and KPC₁ (right) cells. b, pH_i in sgNT (n = 13 and 13) and sgSlc4a4 (n = 7 and 10) Panc02 (left) or KPC₁ cells (right). c, pH_e in sgNT (n = 17 and 15) and sgSlc4a4 (n = 17 and 15) Panc02 (left) or KPC₁ cells (right). d, Glucose-dependent ECAR in sgNT (n = 11 and 22) and sgSlc4a4 (n = 11 and 24) Panc02 (left) or KPC₁ cells (right). e, ³H₂O release from [³H]glucose in sgNT (n = 5 and 4) and sgSlc4a4 (n = 5 and 4) Panc02 (left) or KPC₁ cells (right). f, Intracellular lactate levels measured by LC–MS in sgNT (n = 9 and 7) and sgSlc4a4 (n = 7 and 7) Panc02 (left) or KPC₁ cells (right). g, Extracellular lactate levels measured by LC–MS in sgNT (n = 9 and 6) and sgSlc4a4 (n = 7 and 7) Panc02 (left) or KPC₁ cells (right). h,i, Lactate measured by microdialysis in culture medium of sgNT (n = 3) and sgSlc4a4 (n = 3) Panc02 (h) or KPC₁ (i) cells at 2, 4, 8 and 16 h. j, [¹³C]Lactate to [¹³C]pyruvate ratio in sgNT (n = 6 and 3) and sgSlc4a4 (n = 6 and 3) Panc02 (left) or KPC₁ cells (right). k–m, pH_i (k), pH_e (l) and pH ratio (m) in sgNT (n = 8) and sgSlc4a4 (n = 8) subcutaneous Panc02 tumors assessed by ³¹P-MRS. Data show a pool of two independent experiments. n, Lactate to pyruvate ratio calculated from area under the curve (AUC) in sgNT (n = 10) and sgSlc4a4 (n = 13) subcutaneous Panc02 tumors assessed by MRS following the administration of hyperpolarized pyruvate. Data show a pool of three independent experiments. o, Lactate concentration measured by LC–MS in tumor interstitial fluid of sgNT (n = 15) and sgSlc4a4 (n = 12) Panc02 subcutaneous tumors (left) or of sgNT (n = 4) and sgSlc4a4 (n = 5) orthotopic KPC₁ tumors (right). Data show a pool of two independent experiments (left). Data were normalized by protein content (a and d–i); n represents independently collected cells (a–j). P values were assessed by unpaired two-tailed Student’s t-test (a, b, d–g and j–o), paired two-tailed Student’s t-test (c) and two-way ANOVA with Sidak’s multiple comparison test (h and i). Graphs show mean ± s.e.m.; AU, arbitrary units. Source data

Fig. 5. Slc4a4 targeting unleashes a CD8⁺ T cell-mediated immune response.

a, Percentage of CD8⁺ cells in sgNT (n = 6, 6 and 7) or sgSlc4a4 (n = 6, 5 and 8) subcutaneous Panc02 (left), orthotopic KPC₁ (middle) and KPC₃ tumors (right). b, CD8⁺:T_reg cell ratio in sgNT (n = 5, 6 and 7) and sgSlc4a4 (n = 6, 5 and 8) subcutaneous Panc02 (left), orthotopic KPC₁ (middle) and KPC₃ tumors (right). c,d, Quantification (c) and representative histograms (d) of the MFI of IFNγ (left) and CD69 (right) in CD8⁺ cells in sgNT (n = 5) and sgSlc4a4 (n = 5) subcutaneous Panc02 tumors. FMO: fluorescence minus one. e, MFI of IFNγ (left) and CD69 (right) in CD8⁺ cells in sgNT (n = 6 and 6) and sgSlc4a4 (n = 5 and 7) orthotopic KPC₁ tumors. f, CD8 staining in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic Panc02 tumors. g,h, Representative images (g) and quantification (h) of CD8 (red) staining in sgNT (n = 8) and sgSlc4a4 (n = 8) orthotopic KPC₁ tumors. Nuclei are stained with Hoechst (blue). i, MFI of IFNγ (left) and CD69 (right) in CD8⁺ cells in sgNT (n = 6 and 5) and sgSlc4a4 (n = 7 and 5) orthotopic KPC₃ tumors. j, CD8 staining in orthotopic sgNT (n = 8) and sgSlc4a4 (n = 8) KPC₃ tumors. k, Viable (%) Panc02-ovalbumin (Panc02-OVA) cells cocultured with OT-1 T cells in T cell medium (control (Ctrl), n = 3) with lactic acid (Lac; n = 3), HCl (n = 3) or sodium lactate (NaLac; n = 3). l, CD8⁺ T cell proliferation cultured in conditioned medium from sgNT (n = 5) or sgSlc4a4 (n = 5) Panc02 cells supplemented with Lac (n = 6), HCl (n = 6) or NaLac (n = 6); +act, maximal; –act, basal. m, Weight of subcutaneous Panc02 tumors in mice treated with anti-CD8 or control IgG (IgG; sgNT-IgG n = 12, sgSlc4a4-IgG n = 11, sgNT-anti-CD8 n = 5 and sgSlc4a4-anti-CD8 n = 5). n, Weight of orthotopic KPC₁ tumors in mice treated with anti-CD8 or IgG (sgNT-IgG n = 13, sgSlc4a4-IgG n = 13, sgNT-anti-CD8 n = 8 and sgSlc4a4-anti-CD8 n = 7). o, Weight of orthotopic KPC₃ tumors in mice treated with anti-CD8 or IgG (sgNT-IgG n = 15, sgSlc4a4-IgG n = 14, sgNT-anti-CD8 n = 7 and sgSlc4a4-anti-CD8 n = 5). Data in m–o are representative of pools of two independent experiments. n represents independently collected cells (k and l). P value was assessed by unpaired, two-tailed Student’s t-test (a–c, e, f and h–j), two-way ANOVA with Sidak’s multiple comparison test (k), one-way ANOVA with Tukey’s multiple comparison test (l) and two-way ANOVA with Tukey’s multiple comparison test (m–o). Graphs show mean ± s.e.m.; scale bar, 20 μm (g). Source data

Fig. 6. Slc4a4 targeting affects TAMs only in presence of T cell-derived factors.

a, Percentage of F4/80⁺ TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 5) subcutaneous Panc02 tumors. b,c, Representative images (b) and quantification (c) of immunofluorescence stainings for CD206 (red) and F4/80 (green) in sgNT (n = 5) and sgSlc4a4 (n = 4) subcutaneous Panc02 tumors. d, Percentage of MHC class II⁺ TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 4) subcutaneous Panc02 tumors. e, MFI of MHC class II in TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 4) subcutaneous Panc02 tumors. f, Percentage of F4/80⁺ TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic KPC₁ tumors. g, Percentage of MHC class II⁺ (left) and CD206⁺ (right) TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic KPC₁ tumors. h, MFI of MHC class II (left) and CD206 (right) in TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic KPC₁ tumors. i,j, Quantification (i) and representative images (j) of immunofluorescence stainings for CD206 (red) and F4/80 (green) in sgNT (n = 5) and sgSlc4a4 (n = 5) subcutaneous Panc02 tumors injected in nude mice. k, Percentage of F4/80⁺ TAMs (FACS) in sgNT (n = 7) and sgSlc4a4 (n = 8) orthotopic KPC₁ tumors injected in nude mice. l, MFI of CD206 in TAMs (FACS) in sgNT (n = 7) and sgSlc4a4 (n = 8) orthotopic KPC₁ tumors injected in nude mice. m, MFI of MHC class II in BMDMs (FACS) cocultured with sgNT (n = 3) and sgSlc4a4 (n = 3) Panc02 cells in the absence or presence of different concentrations of IFNγ (25, 50 and 100 ng ml^–1). n represents independently collected cells (m). P value was assessed by unpaired, two-tailed Student’s t-test (a, c–i, k and l) and two-way ANOVA with Sidak’s multiple comparison test (m). Graphs show mean ± s.e.m.; scale bars, 20 μm (b and j). Source data

Fig. 7. Overexpression of Ldha counteracts the effects of Slc4a4 targeting.

a, LDHA protein levels assessed by western blotting analysis in sgNT and sgSlc4a4 KPC₁ cells overexpressing an empty vector (EV) or Ldha (Ldha-OE). b, [¹³C]Lactate:[¹³C]pyruvate ratio in sgNT empty vector (n = 3), sgSlc4a4 empty vector (n = 3), sgNT Ldha-OE (n = 3) and sgSlc4a4 Ldha-OE (n = 3) KPC₁ cells. c,d, Intracellular (c) and extracellular (d) lactate concentration measured by LC–MS in sgNT empty vector (n = 4 and 4), sgSlc4a4 empty vector (n = 4 and 4), sgNT Ldha-OE (n = 7 and 4) and sgSlc4a4 Ldha-OE (n = 5 and 4) KPC₁ cells. e, Lactate concentration measured by LC–MS in tumor interstitial fluid of sgNT empty vector (n = 3), sgSlc4a4 empty vector (n = 3), sgNT Ldha-OE (n = 3) and sgSlc4a4 Ldha-OE (n = 3) KPC₁ tumors. f,g, Intracellular (f) and extracellular (g) pH of sgNT empty vector (n = 8 and 30), sgSlc4a4 empty vector (n = 12 and 30), sgNT Ldha-OE (n = 17 and 30) and sgSlc4a4 Ldha-OE (n = 15 and 30) KPC₁ cells. h, Glucose-dependent ECAR of sgNT empty vector (n = 23), sgSlc4a4 empty vector (n = 24), sgNT Ldha-OE (n = 24) and sgSlc4a4 Ldha-OE (n = 23) KPC₁ cells. i, ³H₂O release from [³H]glucose in sgNT empty vector (n = 4), sgSlc4a4 empty vector (n = 4), sgNT Ldha-OE (n = 4) and sgSlc4a4 Ldha-OE KPC₁ (n = 4) cells. j, Weight of sgNT empty vector (n = 13), sgSlc4a4 empty vector (n = 13), sgNT Ldha-OE (n = 14) and sgSlc4a4 Ldha-OE (n = 14) orthotopic KPC₁ tumors. Data show a pool of two independent experiments. k, Macroscopic metastatic mesenteric lymph nodes of sgNT empty vector (n = 7), sgSlc4a4 empty vector (n = 8), sgNT Ldha-OE (n = 5) and sgSlc4a4 Ldha-OE (n = 6) orthotopic KPC₁ tumors. l, Percentage of CD8⁺ T cells (FACS) in sgNT empty vector (n = 4), sgSlc4a4 empty vector (n = 3), sgNT Ldha-OE (n = 4) and sgSlc4a4 Ldha-OE (n = 3) orthotopic KPC₁ tumors. m, MFI of IFNγ in CD8⁺ T cells (FACS) in sgNT empty vector (n = 4), sgSlc4a4 empty vector (n = 3), sgNT Ldha-OE (n = 4) and sgSlc4a4 Ldha-OE (n = 3) orthotopic KPC₁ tumors. n, Lamp2 expression assessed by RT–qPCR analysis in sgNT empty vector (n = 5), sgSlc4a4 empty vector (n = 5), sgNT Ldha-OE (n = 5) and sgSlc4a4 Ldha-OE (n = 5) orthotopic KPC₁ tumors. Data were normalized by protein content (c, d, h and i). The experiment in a was repeated three times with similar results. n represents independently collected cells (b–i). P value was assessed by two-way ANOVA with Tukey’s multiple comparison test (b–n). Graphs show mean ± s.e.m. Source data

Fig. 8. Slc4a4 targeting overcomes immunotherapy resistance.

a, MFI of PD-1 in CD8⁺ T cells in sgNT (n = 5, 6 and 12) and sgSlc4a4 (n = 5, 7 and 12) subcutaneous Panc02 (left) and orthotopic KPC₁ (middle) and KPC₃ tumors (right). b, MFI of CTLA-4 in Foxp3⁺ T cells in sgNT (n = 6, 6 and 8) and sgSlc4a4 (n = 6, 7 and 8) subcutaneous Panc02 (left) and orthotopic KPC₁ (middle) and KPC₃ tumors (right). c, MFI of PD-L1 in CD45^– cells in sgNT (n = 6, 6 and 8) and sgSlc4a4 (n = 6, 7 and 8) subcutaneous Panc02 (left) and orthotopic KPC₁ (middle) and KPC₃ tumors (right). d, Growth of sgNT and sgSlc4a4 subcutaneous Panc02 tumors treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 6, sgNT-anti-PD-1/anti-CTLA-4 n = 6, sgSlc4a4-IgG n = 6, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 5). e, Weight of sgNT and sgSlc4a4 subcutaneous Panc02 tumors treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 11, sgNT-anti-PD-1/anti-CTLA-4 n = 9, sgSlc4a4-IgG n = 9, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 9). f, Survival curve of sgNT and sgSlc4a4 subcutaneous Panc02 tumor-bearing mice treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 8, sgNT-anti-PD-1/anti-CTLA-4 n = 7, sgSlc4a4-IgG n = 7, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 8). g, Weight of sgNT and sgSlc4a4 orthotopic KPC₁ tumors treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 9, sgNT-anti-PD-1/anti-CTLA-4 n = 8, sgSlc4a4-IgG n = 9, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 8). h, Survival curve of sgNT and sgSlc4a4 orthotopic KPC₁ tumor-bearing mice treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 8, sgNT-anti-PD-1/anti-CTLA-4 n = 8, sgSlc4a4-IgG n = 8, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 8). i, Survival curve of sgNT and sgSlc4a4 orthotopic KPC₁ tumor-bearing mice treated with anti-PD-1 (sgNT-IgG n = 9, sgNT-anti-PD-1 n = 9, sgSlc4a4-IgG n = 9, sgSlc4a4-anti-PD-1 n = 9). j, Survival curve of sgNT and sgSlc4a4 orthotopic KPC₁ tumor-bearing mice treated with anti-CTLA-4 (sgNT-IgG n = 8, sgNT-anti-CTLA-4 n = 8, sgSlc4a4-IgG n = 8, sgSlc4a4-anti-CTLA-4 n = 8). k, Weight of sgNT and sgSlc4a4 orthotopic KPC₃ tumors treated with anti-PD-1 and anti-CTLA-4 (sgNT-IgG n = 16, sgNT-anti-PD-1/anti-CTLA-4 n = 14, sgSlc4a4-IgG n = 14, sgSlc4a4-anti-PD-1/anti-CTLA-4 n = 14). Data in e and k are representative of a pool of two independent experiments. Treatment regimen is indicated by the arrows (mice were treated three times per week with up to six injections; d, f and h–j). P value was assessed by unpaired, two-tailed Student’s t-test (a–c), two-way ANOVA with Sidak’s multiple comparison test (d), two-way ANOVA with Tukey’s multiple comparison test (e, g and k) and log-rank (Mantel–Cox) test (f and h–j). Graphs show mean ± s.e.m. Source data

Extended Data Fig. 1. RNA-seq profiling and protein analysis for SLC4A4.

a, Uniform Manifold Approximation and Projection (UMAP) representation of Leiden clusters identifying 19 cell types/states. b,c, UMAP (b) and proportionality bar plot (c) color-coded for individual patients. d, Dot plot showing the expression of marker genes used for cell type annotation. e, UMAP map of color-coded cells for the indicated cell types isolated from treatment-naïve PDAC patients (n = 24 patients). f, Dot plot of normalized expression of different bicarbonate transporters in the indicated cell types. g, Representative images of immunohistochemistry (IHC) staining for SLC4A4 in each patient from the RNA-seq cohort (n = 10 patients). h, Dot plot of normalized expression of SLC4A4 in the epithelial compartment of each patient. i,j, Representative images of SLC4A4 IHC stainings in orthotopic KPC tumor sections (i) or normal pancreas (j) (n = 3 mice). Scale-bar: 0,5 mm (g,i,j). Source data

Extended Data Fig. 2. Slc4a4 targeting does not affect in vitro proliferation, cell cycle and apoptosis of PDAC cells.

a, SLC4A4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in Panc02 cells. b, SLC4A4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#1 cells. c, SLC4A4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#2 cells originating from a different clone than in b. d, CAS9 protein levels assessed by WB analysis before (day 0) or after 7 day of doxycycline treatment (day 7) followed by 7 day of doxycycline removal (day 14) in Panc02 cells. e,f, Proliferation curves assessed by Incucyte analysis of sgNT (n = 12) and sgSlc4a4 (n = 12) Panc02 (e) or sgNT (n = 12) and sgSlc4a4 (n = 12) KPC_#1 (f) cells. g,h, Cell cycle distribution analysis (FACS) assessed by PI staining in sgNT (n = 4) and sgSlc4a4 (n = 4) Panc02 (g) or sgNT (n = 12) and sgSlc4a4 (n = 11) KPC_#1 (h) cells. i,j, Percentage of PI⁺ Annexin V⁺ apoptotic cells (FACS) in sgNT (n = 4) and sgSlc4a4 (n = 4) Panc02 (i) or in sgNT (n = 8) and sgSlc4a4 (n = 8) KPC_#1 (j) cells in culture for 24, 48 and 72 h. k, SLC4A4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 (2^nd gRNA) n = 3) and WB (right) analysis in Panc02 cells. l,m, Weight (l) and quantification of macroscopic metastatic mesenteric lymph nodes (m) of sgNT (n = 10) and sgSlc4a4 (n = 9) orthotopic KPC_#2 tumors. n, SLC4A4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#3 cells o, Body weight of nude mice with sgNT (n = 8) and sgSlc4a4 (n = 8) orthotopic KPC_#1 tumors. The experiments in a,b,c,d,k,n have been repeated three times with similar results. n represents independently harvested cells (a-c,e-k,n). P value was assessed by unpaired Student’s t-test (a,b,c,k,l,m,n), and two-way ANOVA with Sidak’s multiple comparison test (e-j,o). Graphs show mean ± SEM. PI = propidium iodide. Source data

Extended Data Fig. 3. Slc4a4 targeting does not affect LDHA, MCT1, MCT4 expression.

a, Basal OCR in sgNT (n = 20) and sgSlc4a4 (n = 11) Panc02 and sgNT (n = 10) and sgSlc4a4 (n = 20) KPC_#1 cells. b, Glucose measured by microdialysis in culture medium of sgNT (n = 3) and sgSlc4a4 (n = 3) Panc02 cells at 2, 4, 8 and 16 h. Data are normalized by protein content. c, Glucose measured by microdialysis in culture medium of sgNT (n = 3) and sgSlc4a4 (n = 3) KPC_#1 cells at 2, 4, 8 and 16 h. Data are normalized by protein content. d, Extracellular to intracellular lactate ratio measured by LC/MS in sgNT (n = 8) and sgSlc4a4 (n = 6) Panc02 cells (bottom) and in sgNT (n = 6) and sgSlc4a4 (n = 7) KPC#1 cells (top). e, LDHA expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in Panc02 cells. f, LDHA expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#1 cells. g, MCT1 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in Panc02 cells. h, MCT1 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#1 cells. i, MCT4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in Panc02 cells. j, MCT4 expression assessed by qRT-PCR (left, sgNT n = 3 and sgSlc4a4 n = 3) and WB (right) analysis in KPC_#1 cells. The experiments in e-j have been repeated three times with similar results. n represents independently harvested cells. P value was assessed by unpaired, two-tailed Student’s t-test (a,d-j) and two-way ANOVA with Sidak’s multiple comparison test (b-c).Graphs show mean ± SEM. Source data

Extended Data Fig. 4. Slc4a4 targeting does not alter tumor perfusion and hypoxia.

a, Size of sgNT (n = 8) and sgSlc4a4 (n = 8) subcutaneous Panc02 tumors assessed by MRI. Data show a pool of two independent experiments. b, Representative images of lectin⁺ (in green) perfused vessels (in red, CD34⁺) in sgNT (n = 6) and sgSlc4a4 (n = 6) subcutaneous Panc02 tumors. c, Quantification of vessel perfusion in sgNT (n = 6) and sgSlc4a4 (n = 6) subcutaneous Panc02 tumors. d, Quantification of vessel density in sgNT (n = 6) and sgSlc4a4 (n = 6) subcutaneous Panc02 tumors. e,f, Quantification (e) and representative images (f) of hypoxic areas (in green, PIMO⁺) in sgNT (n = 6) and sgSlc4a4 (n = 4) subcutaneous Panc02 tumors. Hoechst (in blue) was used to stain the nuclei. g, Lamp2 expression assessed by qRT-PCR analysis in sgNT (n = 6) and sgSlc4a4 (n = 5) subcutaneous Panc02 tumors (left) or in sgNT (n = 10) and sgSlc4a4 (n = 9) orthotopic KPC_#1 tumors (right). h, Dynamics of lactate production in sgNT (n = 10) and sgSlc4a4 (n = 13) subcutaneous Panc02 tumors assessed by MRS following the administration of hyperpolarized pyruvate. Data show a pool of three independent experiments. P value was assessed by unpaired, two-tailed Student’s t-test (a,c-e,g,h). Graphs show mean ± SEM. Scale bars: 50 μm (b,f). a.u., arbitrary unit. Source data

Extended Data Fig. 5. Central role of CD8⁺ T cells in the anti-tumor phenotype following the targeting of Slc4a4 in cancer cells.

a, Gating strategy for T cell compartment (numbers represent percentage of cells out of parental population). b, Representative flow cytometry plots of CD8⁺ T cell within sgNT and sgSlc4a4 tumors. c, Representative histograms of the MFI of IFNγ (left) and CD69 (right) in CD8⁺ T cells (FACS) in sgNT and sgSlc4a4 orthotopic KPC_#1 tumors. d, Percentage of CD8⁺ T cells (FACS) in KPC_#1 orthotopic tumors in DMSO- (n = 5) or DIDS- (n = 6) treated mice. e, CD8⁺ to T_reg cell ratio obtained by dividing the percentage of CD8⁺ T cells (out of CD45⁺ cells) by the percentage of Foxp3⁺CD4⁺ T cells (out of CD45⁺ cells) in orthotopic KPC_#1 tumors from DMSO- (n = 5) or DIDS- (n = 6) treated mice. f, Percentage of IFNγ in CD8⁺ T cells (FACS) in orthotopic KPC_#1 tumors from DMSO- (n = 5) or DIDS- (n = 6) treated mice. g, Representative pictures of CD8 (in red) immunofluorescence stainings in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic Panc02 tumors. Hoechst (in blue) was used to stain the nuclei. h, Representative histograms of the MFI of IFNγ (left) and CD69 (right) in CD8⁺ T cells (FACS) in orthotopic KPC_#3 tumors. i, Representative pictures of CD8 (in red) immunofluorescence stainings in sgNT (n = 8) and sgSlc4a4 (n = 8) orthotopic KPC_#3 tumors. Hoechst (in blue) was used to stain the nuclei. j, MFI of IFNγ (left) and GZMB (right) in CD8⁺ OT-I T cells (FACS) after 24-hour co-culture with sgNT (n = 6) and sgSlc4a4 (n = 6) Panc02-OVA cells. n represents independently harvested cells (j). P value was assessed by unpaired, two-tailed Student’s t-test (d,e,f,j). Graphs show mean ± SEM. Scale bars: 20 μm (g,i). Source data

Extended Data Fig. 6. Effect of CD8⁺ T cell depletion following targeting of Slc4a4 in cancer cells.

a, Cytotoxic activity of CD8⁺ OT-1 T cells co-cultured with sgNT (n = 5) or sgSlc4a4 (n = 5) Panc02 OVA spheroids. n represents independently harvested cells. b,c, Schematic overview of in vivo CD8⁺ T cell depletion in Panc02 (b) and KPC_#1/3 (c) tumor models. Mice were treated with anti-CD8 (αCD8) depleting antibodies three days before tumor injection (−3). After subcutaneous Panc02 (b) or orthotopic KPC_#1/3 (c) cell injection, αCD8 depleting antibodies were administered once per week (day 7, 14 and 21 after tumor injection). d, Representative flow cytometry plots of CD8⁺ T cell depletion within sgNT and sgSlc4a4 tumors. e, Efficiency of depletion (FACS) of CD8⁺ T cells within sgNT (n = 5) and sgSlc4a4 (n = 5) subcutaneous Panc02 (left), in sgNT (n = 5) and sgSlc4a4 (n = 4) orthotopic KPC_#1 (middle) and in sgNT (n = 5) and sgSlc4a4 (n = 4) orthotopic KPC_#3 tumors (right) upon αCD8 antibody treatment. Data are shown as percentage of depletion in αCD8-treated tumors, relative to the control IgG-treated groups. f, Growth of sgNT and sgSlc4a4 Panc02 tumor subcutaneously injected in mice treated with an anti-CD8 depleting antibody (αCD8) or control IgG (IgG) (sgNT IgG n = 12, sgSlc4a4 IgG n = 12, sgNT αCD8 n = 5, sgSlc4a4 αCD8 n = 5). All statistics are referred to the sgNT-IgG condition. P value was assessed by unpaired, two-tailed Student’s t-test (a,e) and two-way ANOVA with Tukey’s multiple comparison test (f). Graphs show mean ± SEM. Source data

Extended Data Fig. 7. Slc4a4 targeting does not affect per se macrophage polarization in vitro.

a, Gating strategy for TAM compartment (numbers represent percentage of cells out of parental population). b, Representative histograms of the MFI of MHCII in F4/80+ cells (FACS) in sgNT and sgSlc4a4 subcutaneous Panc02 tumors. c, Representative histograms of the MFI of MHCII (left) and CD206 (right) in F4/80+ cells (FACS) in sgNT and sgSlc4a4 orthotopic KPC#1 tumors. d,e, MFI of CD11c (d) and CD204 (e) in TAMs (FACS) in sgNT (n = 5) and sgSlc4a4 (n = 6) orthotopic KPC#1 tumors. f, Percentage of MHC-II + (left) and CD206 + (right) TAMs (FACS) in sgNT (n = 7) and sgSlc4a4 (n = 8) orthotopic KPC#1 tumors injected in nude mice. g, MFI of MHC-II in TAMs (FACS) in sgNT (n = 7) and sgSlc4a4 (n = 8) orthotopic KPC#1 tumors injected in nude mice. h,i, MFI of MHC-II (h) and CD206 (i) in BMDMs (FACS) co-cultured with sgNT (n = 3) and sgSlc4a4 (n = 3) Panc02 cells. n represents independently harvested cells (h,i). P value was assessed by unpaired, two-tailed Student’s t-test (d-i). Graphs show mean ± SEM. Source data

Extended Data Fig. 8. Immune check-point molecules.

a, Gating strategy for immune checkpoint molecules (numbers represent percentage of cells out of parental population).

Extended Data Fig. 9. Scheme illustrating the role of SLC4A4 in PDAC progression.

SLC4A4, the most abundant bicarbonate transporter in PDAC and almost exclusively expressed in the cancer cell compartment, favors the accumulation of intracellular bicarbonate. Its activity maintains an optimal intracellular pH (pH_i), thus preventing acidosis-dependent glycolysis inhibition, and, in a feed forward loop, further decreasing the extracellular pH (pH_e) because of the excretion of lactate in the extracellular space. Thus, SLC4A4 proficient tumors are characterized by an acidic TME and high levels of lactate that dampen CD8⁺ T cell activation and favor M2-like skewing of TAMs. On the other hand, genetic or pharmacological inhibition of SLC4A4 decreases bicarbonate uptake by the cancer cells, thus preventing TME acidosis while reducing the pH_i, which inhibits glycolysis in a pH-dependent manner, and consequently impairs lactate production and export. Overall, SLC4A4 targeting decreases acidosis and extracellular lactate boosting anti-tumoral CD8⁺ T cell response and M1-like skewing, ultimately harnessing tumor progression and overcoming resistance to ICB.