Targeting cancer glycosylation repolarizes tumor-associated macrophages allowing effective immune checkpoint blockade

Abstract

Immune checkpoint blockade (ICB) has substantially improved the prognosis of patients with cancer, but the majority experiences limited benefit, supporting the need for new therapeutic approaches. Up-regulation of sialic acid-containing glycans, termed hypersialylation, is a common feature of cancer-associated glycosylation, driving disease progression and immune escape through the engagement of Siglec receptors on tumor-infiltrating immune cells. Here, we show that tumor sialylation correlates with distinct immune states and reduced survival in human cancers. The targeted removal of Siglec ligands in the tumor microenvironment, using an antibody-sialidase conjugate, enhanced antitumor immunity and halted tumor progression in several murine models. Using single-cell RNA sequencing, we revealed that desialylation repolarized tumor-associated macrophages (TAMs). We also identified Siglec-E as the main receptor for hypersialylation on TAMs. Last, we found that genetic and therapeutic desialylation, as well as loss of Siglec-E, enhanced the efficacy of ICB. Thus, therapeutic desialylation represents an immunotherapeutic approach to reshape macrophage phenotypes and augment the adaptive antitumor immune response.

Fig. 1.. Tumor sialylation is associated with immune suppression and reduced survival in patients with cancer.

(A) Clustering of correlations between the expression of sialic acid–modifying enzymes and the expression of immune genes in all solid cancers from The Cancer Genome Atlas (TCGA) database. (B) Kaplan-Meier survival curve of patients with clear cell renal carcinoma (KIRC) or (C) squamous cell carcinoma of the lung (LUSC) is shown, divided into quartiles on the basis of their low, intermediate-low, intermediate-high, and high expression of gene set 1, using TCGA data. (D) Correlations between gene set 1 expression and gene signatures of tumor-infiltrating immune cell types are shown for all patients with LUSC from TCGA data. r values are shown on a color scale, from blue to red. (E) Dot plots displaying correlations between gene set 1 expression and signatures of T_regs and M2 macrophages (P < 0.01). (F) Dot plots displaying correlations between gene set 1 expression and signatures of CD8⁺ T cells and T cell dysfunction in cancer (P < 0.01). (G) Expression of CD137 (4-1BB) was measured on CD8⁺ T cells from primary LUSC samples from individual patients after a 48-hour incubation with or without V. cholerae sialidase (n = 13). (H) Experimental design: Mice bearing subcutaneous or intramammary wild-type or GNE-KO tumors were treated intraperitoneally with four doses of anti–PD-1 or anti–CTLA-4 antibodies (10 mg/kg) individually or in combination, beginning at a tumor size of about 80 mm³. (I) Effect of PD-1 blockade on the survival of mice bearing subcutaneous wild-type or GNE-KO MC38 tumors (n = 5 to 6 mice per group). (J) Effect of combined PD-1 and CTLA-4 blockade on the survival of mice bearing subcutaneous wild-type or GNE-KO MC38 tumors (n = 14 to 17 mice per group). (K) Experimental design: Mice bearing established (about 500 mm³) subcutaneous wild-type or GNE-KO MC38 tumors were treated intraperitoneally with two doses of anti–PD-1 and anti–CTLA-4 antibodies (10 mg/kg). Seven days after the first treatment, tumors were resected and immunophenotyped. (L) Frequency of IFN-γ⁺TNF⁺⁻IL2⁺ CD8⁺ T cells after restimulation in single-cell suspensions of MC38 wild-type or GNE-KO tumors treated with PD-1 and CTLA-4 blockade (n = 4 to 6 mice per group). n indicates the number of biological replicates. Error bars represent means ± SEM. Statistical analyses were performed using the log-rank (Mantel-Cox) test for the TCGA survival data or the Gehan-Wilcoxon test for the mouse survival data, followed by Bonferroni’s correction for multiple comparisons. A paired two-tailed Student’s t test was used in (G) and a one-way ANOVA followed by a post hoc Šidák correction for multiple comparisons in (L). *P ≤ 0.05, **P ≤ 0.01, and ****P ≤ 0.0001.

Fig. 2.. Tumor-targeted sialidase effectively desialylates the tumor microenvironment.

(A) Schematic representation of the tumor-targeted sialidase constructs: trastuzumab, the trastuzumab-sialidase conjugate E-301, and the loss-of-function (LOF) mutated, enzymatically inactive version E-301 LOF. (B) In vitro titration of trastuzumab, E-301, and E-301 LOF on EMT6-HER2 cells. Desialylation was assessed by PNA staining 24 hours after treatment relative to that after maximal desialylation (n = 3). (C) Experimental setup to test in vivo desialylation: Mice bearing established (500-mm³) intramammary EMT6-HER2 tumors were treated with a single dose of PBS, trastuzumab, or E-301 [10 mg/kg, intraperitoneally (i.p.)], and desialylation was assessed at 24 and 72 hours after treatment. (D) Representative immunofluorescence images of untreated, trastuzumab-treated, and E-301–treated EMT6-HER2 tumors at 24 hours after treatment, stained with anti-human Fc secondary, PNA, and MAL II. Scale bars, 2 mm. (E) Quantification of immunofluorescence staining with PNA and MAL II. The sum of the staining intensity was normalized to the respective 4′,6-diamindino-2-phenylindole–stained area. (F) Flow cytometric analysis of intramammary EMT6-HER2 tumor sialylation [same tumors as in (E)] by lectin staining of tumor cell suspensions at 24 and 72 hours after treatment. Geometric mean fluorescence intensities (MFIs) of PNA, MAL II, and secondary anti-human Fc staining are shown (D to F) (n = 2 mice per group). (G) Experimental setup comparing desialylation of intramammary wild-type and GNE-KO EMT6-HER2 tumors: Mice bearing established (500-mm³) intramammary wild-type or GNE-KO EMT6-HER2 tumors were treated with a single dose of E-301 LOF or E-301 (10 mg/kg, i.p.), and desialylation was assessed at 48 hours after treatment. (H) Geometric MFIs of PNA and MAL II staining of tumor cells. (I) Geometric MFIs of PNA and MAL II staining of tumor-infiltrating CD45⁺ cells (H and I) (n = 3 mice per group). (J) Desialylation of different intratumoral immune cell populations from EMT6-HER2 tumors by E-301 was measured by PNA staining 48 hours after treatment with trastuzumab or E-301 (n = 4 to 5 mice per group). Conv, conventional. (K) Liquid chromatography–mass spectrometry–based N-glycan analysis of EMT6-HER2 cells after overnight in vitro treatment with E-301 LOF or E-301 (n = 3 samples per group). (L) Experimental setup comparing desialylation of subcutaneous B16D5 and B16D5-HER2 tumors at 48 hours after intraperitoneal treatment with a single dose of trastuzumab, E-301, or E-301 LOF (10 mg/kg). (M) Geometric MFIs of PNA and MAL II staining of tumor cells. (N) Fold change in the geometric MFI of MAL II staining after E-301 treatment relative to E-301 LOF treatment (M and N) (n = 4 mice per group). N indicates the number of biological replicates. Error bars represent means ± SEM. Statistical analyses were performed using two-way ANOVAs followed by post hoc Šidák corrections for multiple comparisons, and an unpaired two-tailed Student’s t test was used to assess fold change differences in (N). ns, not significant; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

Fig. 3.. Tumor-targeted sialidase inhibits tumor growth by activating the adaptive immune system.

(A) Experimental setup: Mice bearing intramammary EMT6-HER2 tumors were treated intraperitoneally with four doses of trastuzumab, E-301 LOF, or E-301 (10 mg/kg), beginning at a tumor size of about 100 mm³. (B) Growth of individual intramammary EMT6-HER2 tumors treated with trastuzumab, E-301 LOF, or E-301 (n = 6 to 8 mice per group). (C) Survival of mice bearing intramammary EMT6-HER2 tumors treated with trastuzumab, E-301 LOF, or E-301 (pooled from two experiments, n = 12 mice per group). (D) Rechallenge of the tumor-free mouse from (C) and tumor-naïve control mice with subcutaneous EMT6 or EMT6-HER2 tumor cells in each flank, respectively (n = 1 to 4 mice per group). (E) Experimental setup: Mice bearing subcutaneous B16D5-HER2 tumors were treated with four doses of trastuzumab, E-301 LOF, or E-301 (10 mg/kg, i.p.), with the first dose administered once the tumor size reached about 80 mm³. (F) Growth of individual subcutaneous B16D5-HER2 tumors treated with trastuzumab, E-301 LOF, or E-301. (G) Survival of mice bearing B16D5-HER2 tumors treated with trastuzumab, E-301 LOF, or E-301 (F and G) (n = 6 to 8 mice per group). (H) Growth of individual B16D5-HER2 tumors treated with E-301 LOF or E-301 after CD8⁺ T cell depletion. (I) Impact of CD8⁺ T cell depletion on the survival of mice bearing B16D5-HER2 tumors treated with E-301 LOF or E-301 (H and I) (n = 6 to 8 mice per group). n indicates the number of biological replicates. Statistical analyses were performed using the log-rank (Mantel-Cox) test or the Gehan-Wilcoxon test, followed by Bonferroni’s correction for multiple comparisons for all survival analyses. ***P ≤ 0.001.

Fig. 4.. Therapeutic desialylation repolarizes TAMs.

(A) Experimental setup for single-cell RNA sequencing (scRNA-seq) of immune infiltrates after E-301 treatment: Mice bearing palpable (100-mm³) subcutaneous B16D5-HER2 tumors were treated with two doses of E-301 LOF or E-301 (10 mg/kg), alone or in combination with anti–PD-1/CTLA-4 antibodies. CD45⁺ tumor-infiltrating immune cells were isolated and sorted 7 days after the first injection for scRNA-seq. (B) scRNA-seq gene expression data were processed, sorted into clusters, and are presented in a dimensional reduction projection (UMAP), showing the different identified immune cell populations. Labels have been added on the basis of the expression of marker genes. (C) UMAP projections are shown separated by condition. Clusters 3 and 6 are highlighted in blue, and cluster 14 is highlighted in red. (D) Contribution of each condition to each cluster of CD45⁺ cells. (E) Subclustering of all macrophages. (F) Contribution of each condition to each macrophage cluster. (G) UMAP projections of macrophages are shown separated by condition. Clusters 2 and 13 are highlighted in red. (H and I) Dot plot representation of differentially expressed genes between the macrophage clusters showing genes characteristic for M2 polarization (H) or reflecting more general macrophage function (I). Size reflects the percentage of each cluster expressing a given gene, and average-scaled expression is indicated on the color gradient. Clusters 2 and 13 are boxed in and highlighted with arrows (A to I) (n = 5 pooled mice per condition).

Fig. 5.. Tumor desialylation repolarizes TAMs in murine and human tumors.

(A) Experimental setup to phenotype changes in immune infiltrates after E-301 treatment: Mice bearing established (500-mm³) subcutaneous B16D5-HER2 tumors were treated with two doses of trastuzumab, E-301 LOF, or E-301 (10 mg/kg, i.p.) and immune infiltrates analyzed after 7 days by flow cytometry. (B) Frequencies of MHC-II⁺CD206⁻ (M1) and MHC-II⁻CD206⁺ (M2) cells among CD11b⁺F4/80⁺ tumor-associated macrophages (TAMs) are shown, as well as the ratio of M1 to M2 TAMs (n = 7). (C) Experimental setup for in vitro coculture of primary mouse CD11b⁺ TAMs and CD8 T cells. Mice carrying established subcutaneous B16D5-HER2 tumors were treated intraperitoneally with two doses of trastuzumab or E-301 (10 mg/kg), and CD11b⁺ TAMs were bead-isolated from single-cell suspensions 7 days after treatment. (D) TAMs were cultured for 48 hours before their polarization was analyzed. Frequencies of MHC-II⁺ M1 and CD206⁺ M2 TAMs were quantified by flow cytometry. (E) Trastuzumab- or E-301–treated TAMs were cocultured with naïve CD8⁺ T cells in the presence of agonistic anti-CD3/28 antibodies, and T cell activation and proliferation were quantified after 48 hours (D and E) (n = 9 to 10 replicates). (F) Experimental setup for in vitro coculture of primary human CD14⁺ TAMs from LUSC tumors and matched autologous CD8 TILs. TAMs were bead-isolated from single-cell suspensions of primary LUSC tumors and treated with E-301 LOF or E-301. (G) Flow cytometric analysis of TAM polarization in T cell cocultures after in vitro E-301 LOF or E-301 treatment. MHC-II⁺⁻ CD206⁻ M1 and MHC-II⁻CD206⁺ M2 macrophages are shown. (H) Proliferation and activation of CD8 TILs after coculture with autologous E-301 LOF− or E-301–treated primary TAMs (n = 3). n indicates the number of biological replicates. Error bars represent means ± SEM. Statistical analyses were performed using one-way (B) or two-way (D to H) ANOVAs followed by post hoc Šidák corrections for multiple comparisons. *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Fig. 6.. Efficacy of tumor-targeted sialidase is dependent on Siglec-E on TAMs.

(A) Expression of Siglece, Siglecf, Siglec1, and Siglecg in all macrophage clusters from scRNA-seq data (Fig. 4E). Sample distribution shown as violin plots. (B) UMAP projections of all CD45⁺ cells and all macrophages from scRNA-seq data. Expression of Siglece is shown as a color gradient from blue (low) to green (high). (C) t-distributed stochastic neighbor embedding (t-SNE) projection of multicolor flow cytometric immunophenotyping of pooled B16D5-HER2 and EMT6-HER2 tumors. Cell populations have been assigned on the basis of marker expression as shown in fig. S6 (A and B). (D) Staining intensity for Siglec-E is shown as a color gradient from blue (low) to red (high). (E) Representative histogram of Siglec-E staining on CD11b⁺F4/80⁺ TAMs from B16D5-HER2 tumors. Isotype control staining is shown as an empty histogram and anti–Siglec-E staining in red. (F) Experimental setup: Mice deficient for Siglec-E (EKO) bearing subcutaneous B16D5-HER2 tumors were treated intraperitoneally with four doses of trastuzumab, E-301 LOF, or E-301 (10 mg/kg) beginning at a tumor size of about 80 mm³. (G) Survival of EKO mice bearing subcutaneous B16D5-HER2 tumors after trastuzumab, E-301 LOF, or E-301 treatment (n = 6 mice per group). (H) Experimental setup: C57BL/6 mice and mice lacking Siglec-E (EKO) were subcutaneously injected with wild-type or GNE-KO MC38 tumor cells, and tumor growth was monitored over time. (I) Survival of C57BL/6 and EKO mice after subcutaneous injection of MC38 wild-type or GNE-KO tumor cells (n = 13 to 17 mice per group). (J) Experimental setup: Siglece^flox/flox mice were crossed to CD11c^cre mice. Cre-expressing mice (Siglece^ΔCD11c), lacking Siglec-E on all CD11c-expressing cells, were compared to their WT (Siglece^WT) littermate controls. Mice were subcutaneously injected with either MC38 tumor cells and left untreated or with B16D5-HER2 tumor cells in combination with intraperitoneal treatment with four doses of E-301 LOF or E-301 (10 mg/kg) beginning at a tumor size of about 80 mm³. (K) Average tumor growth of subcutaneously injected MC38 cells in Siglece^WT and Siglece^ΔCD11c mice (n = 10 to 11 per group). (L) Siglec-E expression was measured on different tumor-infiltrating myeloid immune cell types in Siglece^ΔCD11c mice compared to WT littermate controls by flow cytometry. Siglec-E expression shown as fold change over fluorescence minus one (FMO) control staining. (M) Frequencies of CD206⁻MHC-II⁺ (M1), CD206⁺MHC-II⁺, and CD206⁺MHC-II⁻ (M2) macrophages among CD11b⁺F4/80⁺ TAMs. Siglece^ΔCD11c mice were compared to WT littermate controls (L and M) (n = 6 to 7 mice per group). (N) Average tumor growth of B16D5-HER2 tumors in Siglece^WT and Siglece^ΔCD11c mice treated intraperitoneally with four doses of E-301 LOF or E-301 (n = 7 to 9 mice per group). (O) Experimental setup for myeloid cell depletion in mice bearing GNE-KO or wild-type MC38 tumors. C57BL/6 mice were treated with either anti-Ly6G antibodies to deplete PMN-MDSCs or anti-CSF1R to deplete TAMs; mice were then injected subcutaneously with wild-type or GNE-KO MC38 tumors cells, and tumor growth was monitored over time. (P) Average tumor growth of subcutaneous wild-type or GNE-KO MC38 tumors in mice treated with anti-Ly6G antibodies or anti-CSF1R antibodies. (Q) Tumor volumes of wild-type or GNE-KO MC38 tumors after anti-Ly6G or anti-CSF1R treatment on day 12 after tumor cell injection (P and Q) (n = 7 to 8 mice per group). (R) Experimental setup for the isolation of TAMs from subcutaneous B16D5-HER2 tumors from Siglece^ΔCD11c or Siglece^WT mice and coculture with CD8 T cells. (S) Macrophages were isolated from B16D5-HER2 tumors from Siglece^ΔCD11c or Siglece^WT mice after treatment with E-301 LOF or E-301. Frequencies of macrophages with the indicated phenotypes were quantified by flow cytometry. MHC-II⁺CD206⁻ M1 and MHC-II⁻CD206⁺ M2 polarization was measured. (T) CD8 T cell activation was measured after coculture with TAMs from Siglece^ΔCD11c or Siglece^WT mice and E-301 treatment (n = 3). n indicates the number of biological replicates. Error bars represent means ± SEM. Statistical analyses were performed using the log-rank (Mantel-Cox) test or the Gehan-Wilcoxon test, followed by Bonferroni’s correction for multiple comparisons for all survival analyses. Differences between groups were tested using two-way ANOVAs followed by post hoc Šidák corrections for multiple comparisons. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

Fig. 7.. Targeting tumor sialylation or Siglec-E enhances the efficacy of ICB.

(A) Experimental setup: Mice bearing subcutaneous B16D5-HER2 tumors were treated with four doses of E-301 LOF or E-301, in combination with anti–PD-1/CTLA-4 antibodies (10 mg/kg, i.p.), with the first dose administered once the tumor size reached about 80 mm³. (B) Growth of individual B16D5-HER2 tumors treated with E-301 LOF or E-301 in combination with anti–PD-1/CTLA-4 antibodies. (C) Survival of mice bearing B16D5-HER2 tumors treated with E-301 LOF or E-301 in combination with anti–PD-1/CTLA-4 antibodies (B and C) (n = 6 to 8 mice per group). (D) Experimental setup for the treatment of wild-type C57BL/6 and EKO mice bearing subcutaneous MC38 tumors with four doses of anti–PD-1 alone or in combination with anti–CTLA-4 intraperitoneally, beginning once the tumor size reached about 80 mm³. (E) Effect of anti–PD-1 ICB on the growth of individual MC38 tumors in wild-type C57BL/6 and EKO mice. (F) Survival of wild-type C57BL/6 and EKO mice bearing MC38 tumors after treatment with PD-1 blockade (E and F) (n = 4 to 6 mice per group). (G) Effect of combined anti–PD-1 and anti–CTLA-4 ICB on the growth of individual MC38 tumors in wild-type C57BL/6 and EKO mice. (H) Survival of wild-type C57BL/6 and EKO mice bearing MC38 tumors after treatment with PD-1 and CTLA-4 blockade (G and H) (n = 13 to 18 mice per group). n indicates the number of biological replicates. Statistical analyses were performed using the Gehan-Wilcoxon test followed by Bonferroni’s correction for multiple comparisons for all survival analyses. *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.