Cutting Edge: Enhanced Antitumor Immunity in ST8Sia6 Knockout Mice

Sialyltransfersases are enzymes that add sialic acids to the termini of either N-linked or O-linked oligosaccharides on glycoproteins or glycolipids. There are six sialyltransferases that generate α2,8-linked sialic acids, that is, ST8Sia1 through ST8Sia6. ST8Sia6 generates disialic acids, specifically on O-linked glycoproteins, by adding a single α2,8-linked sialic acid onto an existing α2,3- or α2,6-linked sialic acid (1). Siglecs are a family of transmembrane proteins that bind and recognize specific sialic acid linkages. Siglecs are generally inhibitory receptors primarily expressed on cells of hematopoietic origin and serve to dampen immune responses (2). Intracellular ITIMs present in the cytoplasmic tails of many Siglecs recruit SHP1 and SHP2 tyrosine phosphatases, leading to immune inactivation (2). Siglec-E recognizes the modification generated by ST8Sia6 (3) and is expressed specifically on macrophages, dendritic cells, neutrophils, and monocytes (4). Although it is known that Siglec-E plays a role in modulating the immune response, it is not known how ST8Sia6 expression in the host influences those immune responses.

ST8Sia6 is expressed by dendritic cells, macrophages, and CD4 T cells. In dendritic cells, type 2 conventional dendritic cells (cDC2s) have substantially higher levels of ST8Sia6 as compared with type 1 conventional dendritic cell (cDC1s) and plasmacytoid dendritic cells (ImmGen). However, unlike T cells, macrophages and dendritic cells also express Siglec-E. Ligands generated by sialyltransferases can interact with Siglecs on opposing cells (defined as a trans interaction) as well as on the same cell (defined as a cis interaction) (5). Previously, we showed that ST8Sia6 overexpression in cancer cells altered macrophage polarization through engagement of ligands generated by ST8Sia6 with Siglec-E (6). In addition, dendritic cells that uptake sialylated Ag, through interaction with Siglec-E, induce immune tolerance and promote the generation of regulatory T cells (Tregs) (7). However, whether endogenous ST8Sia6 helps to set a tolerance threshold in the immune response to tumors is not known.

In this study, we analyzed the immune response in models of cancer in ST8Sia6 knockout (KO) mice. We challenged C57BL/6 and ST8Sia6 KO mice with MC38 and B16 tumor lines. Injection of tumors into ST8Sia6 KO mice led to reduced tumor burden, which increased survival, as compared with injection of tumors into C57BL/6 mice. When examining the tumor microenvironment, we found that intratumoral macrophages and dendritic cells in ST8Sia6 KO mice demonstrated a shift from a suppressive to an activated phenotype as compared with macrophages and dendritic cells in C57BL/6 mice. Additionally, we found that Tregs from tumors in ST8Sia6 KO mice experienced a shift to an inflammatory phenotype. These changes in tumor growth and Treg phenotypes were found to be due to the loss of ST8Sia6 in innate cells. Thus, loss of ST8Sia6 in mice leads to enhanced immune activation after tumor challenge, indicating that ST8Sia6 normally functions to dampen immune responses in vivo.

The MC38 murine colorectal cancer cell line was received from Dr. Yang-Xin Fu (UT Southwestern Medical Center, Dallas, TX) (8). The B16 and B16-OVA murine melanoma cell lines were received from Dr. Richard Vile (Mayo Clinic, Rochester, MN). Cell lines were screened for Mycoplasma contamination and authenticated by short tandem repeat profiling (B16 at American Type Culture Collection and MC38 at IDEXX). Cell lines were cultured as previously described (6).

C57BL/6 (stock no. 000664), OTII (stock no. 004194), and Rag1 KO (stock no. 002216) mice were obtained from The Jackson Laboratory. B6.SJL mice (stock no. 564) were purchased from Charles River. ST8Sia6 KO mice were previously described (3). B6.SJL and OTII mice were interbred to generate OTII CD45.1/CD45.2 mice. Experiments were conducted on both male and female mice between 8 wk and 4 mo of age. Tumor growth and survival experiments were performed with five mice per group, and tumor harvest experiments were performed with four to nine mice per group as denoted in the figures. All mouse work was performed with approval from the Mayo Clinic Institutional Animal Care and Use Committee.

Flow cytometry of isolated tumors was performed as described in Friedman et al. (6). Fluorescently conjugated flow cytometry Abs CD45-BV785 (catalog no. 103149, BioLegend), CD45.1-BV785 (catalog no. 110743, BioLegend), CD11b-FITC (catalog no. 101206, BioLegend), CD11c-PE (catalog no. 117308, BioLegend), CD11c-BV421 (catalog no. 117330, BioLegend), F4/80-PE-Cy7 (catalog no. 123114, BioLegend), Ly6C-PerCP (catalog no. 128028, BioLegend), Ly6G-BV510 (catalog no. 127633, BioLegend), Ly6G-allophycocyanin (catalog no. 127614, BioLegend), Gr1-PerCP (catalog no. 108426, BioLegend), IA/IE (MHC class II)-BV605 (catalog no. 107639, BioLegend), TNFRII-biotin (catalog no. 113403, BioLegend), CD40-PE-Dazzle (catalog no. 124630, BioLegend), CD80-BV510 (catalog no. 104741, BioLegend), CD86-allophycocyanin (catalog no. 105012, BioLegend), streptavidin-BV605 (catalog no. 405229, BioLegend), TCRβ-PE-Cy7 (catalog no. 109222, BioLegend), CD4-PerCP (catalog no. 100538, BioLegend), CD8α-FITC (catalog no. 100706, BioLegend), CD25-BV421 (catalog no. 101923, BioLegend), PD-1-allophycocyanin (catalog no. 135210, BioLegend), CD62L-BV510 (catalog no. 104441, BioLegend), CD69-BV605 (catalog no. 104530, BioLegend), and recombinant Siglec-E-Fc (catalog no. 551506, BioLegend) were purchased from BioLegend. Retinoic acid–related orphan receptor γt (RORγt)-PE-CF594 (catalog no. 562684, BD Horizon) was purchased from BD Horizon. Fixable viability dye Ghost Red 780 (catalog no. 13-0865-T100, Tonbo Biosciences) and Foxp3-PE (catalog no. 50-5773-U100, Tonbo Biosciences) were purchased from Tonbo Biosciences. Arg-1-allophycocyanin (catalog no. 12-3697-82, Invitrogen) and inducible NO synthase-PE (catalog no. 17-5920-82, Invitrogen) were purchased from Invitrogen. Anti-human IgG-allophycocyanin (catalog no. 109-135-098, Jackson ImmunoResearch Laboratories) was purchased from Jackson ImmunoResearch Laboratories. Magnetic enrichment and negative selection Abs CD8-biotin (catalog no. 100704, BioLegend), NK1.1-biotin (catalog no. 108704, BioLegend), B220-biotin (catalog no. 103204, BioLegend), CD11b-biotin (catalog no. 101207, BioLegend), CD19-biotin (catalog no. 115503, BioLegend), CD11c-biotin (catalog no. 117303, BioLegend), Gr-1-biotin (catalog no. 108404, BioLegend), and Ter-119-biotin (Cat. 116203, BioLegend) were purchased from BioLegend. TCRγδ-biotin (catalog no. 13-5811085, eBioscience) was purchased from eBioscience.

MC38, B16 and B16-OVA were injected into the flanks of mice as described in Friedman et al. (6). Tumors were isolated and dissociated as described in Friedman et al. (6).

Spleens and lymph nodes from either C57BL/6, ST8Sia6 KO, or OTII CD45.1/CD45.2 mice were harvested and dissociated as described above. Samples were pooled and washed with PBS. Samples were resuspended in RoboSep buffer (catalog no. 20104, STEMCELL Technologies) at a concentration of 2 × 10⁸ cells per ml for sorting. Cells were magnetically enriched for CD4⁺ T cells using the EasySep mouse streptavidin RapidSpheres isolation kit (STEMCELL Technologies, catalog no. 19860). Biotin-conjugated Abs against CD8, TCRγδ, NK1.1, B220, CD11b, CD19, CD11c, Gr-1, and Ter-119 were used to remove all non-CD4⁺ T cell populations. The enriched CD4⁺ T cells were washed three times with PBS and resuspended in 5 ml of PBS. Then, 5 × 10⁶ cells, in a volume of 100 μl, were retro-orbitally injected into B16-OVA tumor-bearing C57BL/6 and ST8Sia6 KO mice, after which 8 × 10⁶ cells, in a volume of 100 μl, were retro-orbitally injected into MC38 tumor-bearing Rag1 KO mice.

Data were graphed and analyzed using Prism (GraphPad Software). Tumor growth kinetics were compared using a two-way ANOVA, and survival data were assessed using a Mantel–Cox test. Individual parameter datasets across two groups were compared using an unpaired t test. Error bars represent SEM.

Previously, we demonstrated that overexpression of ST8Sia6 in murine cancer cells causes increased tumor growth and reduced survival (6). This change in growth and survival was dependent on Siglec-E expression in the host (6). However, the importance of ST8Sia6 in the host immune response was not addressed. Thus, we sought to determine how ST8Sia6 in the mouse influences the immune system in the context of cancer. C57BL/6 and ST8Sia6 KO mice were injected s.c. with either MC38, B16, or B16-OVA tumor cells. ST8Sia6 KO mice injected with MC38 exhibited reduced tumor growth (p < 0.0006) and increased survival (p = 0.0019) as compared with C57BL/6 mice (Fig. 1A). Similarly, ST8Sia6 KO mice injected with B16 exhibited reduced tumor growth (p < 0.0001) and increased survival (p = 0.0023) as compared with C57BL/6 mice (Fig. 1B). In addition, ST8Sia6 KO mice injected with B16-OVA exhibited reduced tumor growth (p = 0.0063) and increased survival (p = 0.008) as compared with C57BL/6 mice (Fig. 1C). Therefore, tumor growth is reduced in ST8Sia6 KO mice.

To understand how the absence of ST8Sia6 on immune cells alters the immune response, tumor-associated macrophages (TAMs) and dendritic cells (TADCs) were analyzed for expression of proteins associated with immune suppression (TNFRII, Arg-1) and activation (CD40, CD80, CD86). TNFRII has been associated with a tolerogenic or suppressive phenotype in dendritic cells and Tregs, although the exact mechanism leading to this suppression is currently unknown (9). Arg-1 expression in M2-like macrophages suppresses immune responses (10). CD40, CD80, and CD86 are costimulatory molecules that are upregulated on APCs during immune activation (11). These proteins are expressed on APCs and serve to activate T cells (11). TAMs were further categorized into either classical (CD11c⁻F4/80⁺) or hybrid (CD11c⁺F4/80⁺) macrophages, whereas TADCs were categorized into either classical (CD11b⁻) dendritic cells or cDC2s (CD11b⁺F4/80⁻). Classical (CD11b⁻) dendritic cells are comprised of cDC1s and plasmacytoid DCs. Gating trees for quantification of all flow cytometry populations are shown in Supplemental Figs. 1 and 2. No differences in the types of immune cells infiltrating the tumors were observed (Supplemental Fig. 3).

In MC38 and B16, both hybrid macrophages and cDC2s from ST8Sia6 KO mice demonstrated a significant decrease in TNFRII expression and a significant increase in CD40 as compared with tumors from C57BL/6 mice (Fig. 2). Also, classical macrophages from ST8Sia6 KO mice also had increased CD40 expression, and classical dendritic cells also had decreased TNFRII expression in both the MC38 and B16 models (Supplemental Fig. 4). In addition, hybrid macrophages from ST8Sia6 KO mice exhibited a decrease in Arg-1 in the MC38 tumor model. cDC2s also demonstrated an increase in CD80 expression in ST8Sia6 KO mice in both the MC38 and B16 tumors. Other changes in expression consistent with increased activation and decreased suppression were also observed; however, these were not present in both the MC38 and B16 tumors (Fig. 2, Supplemental Fig. 4). Thus, the absence of ST8Sia6 on immune cells altered the phenotype of macrophages and dendritic cells toward promoting immune activation.

The activated phenotype present in macrophages and dendritic cells within the tumors from ST8Sia6 KO mice could influence T cell frequencies. Our previous findings demonstrated that when ST8Sia6 was overexpressed in MC38, Treg proportions increased in a Siglec-E–dependent manner (6). However, this change was indirect, as neither CD4 nor CD8 T cells express Siglec-E (6). The alterations in innate cell activation in ST8Sia6 KO mice could lead to changes in Treg frequencies or phenotypes. To investigate this, C57BL/6 and ST8Sia6 KO mice were either injected with MC38 or B16 tumor cells. Tumors were harvested after 12 d, at a point when there was substantial lymphocyte infiltration. The frequencies of CD4 and CD8 T cells were unaltered within MC38 or B16 tumors in C57BL/6 and ST8Sia6 KO mice (Fig. 3). Frequencies of Tregs also remained unchanged, as was the case for the CD8/CD4 ratios (Fig. 3). CD8/Treg ratios were significantly different only in MC38 tumors (Fig. 3A). However, the phenotype of Tregs did significantly change with regard to expression of TNFRII and RORγt in tumors from ST8Sia6 KO mice compared with C57BL/6 mice. TNFRII expression is associated with functionally suppressive Tregs (12). RORγt-positive Tregs have been associated with an inflammatory phenotype (13). Tregs from MC38 and B16 tumors in ST8Sia6 KO mice exhibited a reduction in TNFRII as compared C57BL/6 mice (Fig. 3). Additionally, Tregs from MC38 and B16 tumors in ST8Sia6 KO mice exhibited an increase in RORγt as compared with C57BL/6 mice (Fig. 3). Therefore, these changes in expression could indicate a shift to a less suppressive Treg in the absence of ST8Sia6.

ST8Sia6 is expressed in macrophages and dendritic cells with particularly high expression in cDC2s. T cells also express ST8Sia6 with high expression in CD4 T cells (ImmGen). Macrophages, dendritic cells, and Tregs were all phenotypically altered with the loss of ST8Sia6 in tumor-bearing mice. We sought to determine whether the loss of ST8Sia6 expression in myeloid cells was altering T cell phenotypes or whether the loss of ST8Sia6 expression in T cells was altering myeloid cell phenotypes. To address this question, C57BL/6 and ST8Sia6 KO mice were both injected with B16-OVA tumor cells. ST8Sia6-sufficient OTII CD45.1/CD45.2 T cells were isolated and transferred into day 8 tumor-bearing mice. Tumors were harvested at day 11 and Treg phenotypes were analyzed. Similar to our previous results, OTII Treg frequencies remained unaffected from tumors present in either C57BL/6 or ST8Sia6 KO mice (Fig. 4A). However, OTII Tregs from tumors present in ST8Sia6 KO mice exhibited a reduction in TNFRII and an increase in RORγt as compared with tumors present in C57BL/6 mice (Fig. 4A). Additionally, tumors weighed significantly less in ST8Sia6 KO mice as compared with C57BL/6 mice, replicating the tumor growth pattern observed in (Fig. 1 (Fig. 4A). We then examined whether the effect on Treg phenotype was cell intrinsic. Rag1 KO mice were injected with MC38 tumors cells along with CD4 and CD8 T cells isolated from either C57BL/6 or ST8Sia6 KO mice. Tumors were harvested at day 11 and Treg phenotypes were analyzed. Treg frequencies, CD8/CD4 ratios, and CD8/Treg ratios between C57BL/6 and ST8Sia6 KO mice remained unchanged (Fig. 4B). However, TNFRII and RORγt frequencies in Tregs were not significantly different between C57BL/6 and ST8Sia6 KO mice (Fig. 4B). Tumor weights were also not significantly different between C57BL/6 and ST8Sia6 KO mice (Fig. 4B). Within the tumor microenvironment, hybrid macrophages, cDC2s, classical macrophages, classical dendritic cells, and Tregs were examined in WT and ST8Sia6 KO mice for changes in the generation of ligands for Siglec-E. In ST8Sia6 KO mice, all of these immune cells exhibited reduced binding to recombinant Siglec-E, indicating decreased generation of ST8Sia6-generated ligands as compared with immune cells from tumors in C57BL/6 mice (Fig. 4C, 4D). These results indicated that the loss of ST8Sia6 in APCs was responsible for the reduction in tumor growth and changes in Treg phenotypes, and that loss of ST8Sia6 limited to T cells did not alter tumor growth.

Our previous findings demonstrated the role of Siglec-E in modulating the immune system when ST8Sia6 was overexpressed in cancer cells (6). However, the role of ST8Sia6 within the host immune system remained unknown. We demonstrate in the present study that the loss of ST8Sia6 in tumor-bearing mice reduced tumor growth, increased survival, and altered the immune response as compared with ST8Sia6-sufficient tumor-bearing mice. Using MC38, B16, and B16-OVA flank tumor mouse models, tumor growth was decreased in ST8Sia6 KO mice as compared with C57BL/6 mice. Within the tumor microenvironment of tumors present in ST8Sia6 KO mice compared with tumors present in C57BL/6 mice, macrophages and dendritic cells altered their phenotype toward enhanced immune activation. Tregs from tumors in ST8Sia6 KO mice also displayed an increase in RORγt and a reduction in TNFRII, suggesting a possible shift to a less suppressive Treg. These alterations in the immune response to tumors were due to the loss of ST8Sia6 in macrophages and dendritic cells and not T cells. Therefore, the loss of ST8Sia6 in the host induces a shift in myeloid cells to exhibit an inflammatory phenotype, which leads to a reduction in tumor growth and the development of less suppressive Tregs.

Inflammatory responses in APCs, such as macrophages and dendritic cells, play a major role in stimulating T cells to promote antitumor responses (14). We observed increased CD40, which is crucial for antitumor responses (15). In contexts outside of cancer, overstimulation of the immune system can lead to autoimmune disorders, such as type 1 diabetes and inflammatory bowel disease (16, 17). One consequence of cancer immunotherapies designed to remove barriers of immune suppression is the development of autoimmune disorders. Many studies have demonstrated that subsets of patients receiving cancer immunotherapies develop diabetes or colitis (16, 17). ST8Sia6-mediated sialylation may serve as a protective barrier by dampening immune responses leading to inflammatory diseases. Previously, we showed that overexpression of ST8Sia6 delayed the onset and frequency of hyperglycemia in mice in the streptozotocin-induced model of diabetes (3). If mice were to lose expression of ST8Sia6 in the NOD model of type 1 diabetes, one could hypothesize that disease progression would increase.

ST8Sia6 produces ligands that bind Siglec-E (3, 6). Tumor-associated immune cells in ST8Sia6 KO mice have reduced binding to recombinant Siglec-E. It has been shown that Siglec-E plays a critical role in regulating immune responses. Loss of Siglec-E in a mouse model of atherosclerosis led to increased disease progression (18). Additionally, loss of Siglec-E in a model of pneumonia resulted in increased survival and bacterial clearance along with an increase in inflammatory cytokine production in bronchoalveolar lavage fluid (19). Removal of Siglec-E would prevent interaction with ST8Sia6-generated ligands. Thus, the recognition of sialylated ligands by Siglec-E may dictate how ST8Sia6 regulates the immune system. By extension, loss of ST8Sia6 in models of atherosclerosis and pneumonia could replicate the results seen with the loss of Siglec-E. Sialic acids generated by sialyltransferases, such as ST8Sia6, can participate in trans and cis interactions (5). Because macrophages and dendritic cells contain both ST8Sia6 and Siglec-E, cis interactions between ST8Sia6-generated ligands and Siglec-E on myeloid cells could serve as a barrier for subsequent immune responses. Therefore, removal of ST8Sia6 would sever this interplay and promote inflammatory responses.

We thank Dr. Richard Vile for the B16 and B16-OVA tumor cell lines. We also thank members of the Shapiro and Dong laboratories for thoughtful discussions.

The authors have no financial conflicts of interest.