The immune system contains a series of checks and balances that maintain tolerance and prevent autoimmunity. Sialic acid-binding Ig-type lectins (Siglecs) are cell surface receptors found on immune cells and inhibit inflammation by recruiting protein tyrosine phosphatases to ITIMs. Islet-resident macrophages express Siglec-E, and Siglec-E expression decreases on islet-resident macrophages as insulitis progresses in the NOD mouse. The sialyltransferase ST8Sia6 generates α-2,8-disialic acids that are ligands for Siglec-E in vivo. We hypothesized that engaging Siglec-E through ST8Sia6-generated ligands may inhibit the development of immune-mediated diabetes. Constitutive overexpression of ST8Sia6 in pancreatic β cells mitigated hyperglycemia in the multiple low-dose streptozotocin model of diabetes, demonstrating that engagement of this immune receptor facilitates tolerance in the setting of inflammation and autoimmune disease.
Sialyltransferases add sialic acid to oligosaccharides present on glycoproteins and glycolipids in the Golgi. Sialylation has several important functions in cell biology, influencing cell fate, adhesion, and migration. In terms of immunity, cell surface sialic acid-bearing ligands distinguish self from nonself through interaction with receptors of the immune system, sialic acid-binding Ig-type lectins (known as Siglecs). Siglecs are expressed on hematopoietic cells and generally function to inhibit inflammatory processes via recruitment of the protein tyrosine phosphatases SHP-1 and SHP-2 to their ITIM domains (1). For example, bacteria and viruses use sialic acid mimicry to establish infection by modulating the immune system through Siglecs (2). High concentrations of sialic acid promote immune evasion by signaling via Siglecs to inhibit immune cell activation (1). Additionally, hypersialylation is a common feature of tumor cells, and engagement of Siglecs is a mechanism that allows tumors to evade immune surveillance (3).
Immune evasion can occur via the engagement of murine Siglec-E, whose closest human ortholog is Siglec-9 (4). Siglec-E is expressed on tissue-resident macrophages in addition to neutrophils and certain subsets of dendritic cells (5). Its preferential recognition of α-2,8-disialic acid motifs is implicated in a variety of immune suppressive functions. For instance, targeting Siglec-E with nanoparticles equipped with α-2,8-disialic acids abolished inflammation in vivo using LPS-induced models of sepsis and acute respiratory distress syndrome (6). Furthermore, mice globally deficient in Siglec-E demonstrate increased tumor immune surveillance, emphasizing its function as a suppressor of immune activation (3). Thus, Siglec-E is a potent immune modulator, and its engagement has significant implications in the regulation of innate immunity and, by extension, adaptive immunity during inflammation.
ST8Sia6 is a sialyltransferase that catalyzes the addition of α-2,8-disialic acids preferentially onto O-linked glycoproteins (7). In this study, we definitively demonstrate for the first time, to our knowledge, that ST8Sia6 produces ligands for Siglec-E in vivo by using ST8Sia6 knockout (KO) mice. Therefore, we hypothesized that overexpression of ST8Sia6 may engage Siglec-E to reduce inflammation and disease. Within the pancreatic islet microenvironment, the islet-resident macrophage (IRM) expresses high levels of Siglec-E, which decrease as insulitis increases in NOD islets. The NOD mouse spontaneously develops robust insulitis, or lymphocytic infiltration of the islets, followed by overt diabetes as β cell destruction progresses. Although there are many immune cells necessary for disease development, the IRM plays a critical role in Ag presentation and disease initiation (8). IRMs are intimately associated with pancreatic β cells and present β cell Ags to infiltrating CD4+ autoreactive T cells. Depletion of IRMs prevents the onset of type 1 diabetes (T1D) (9). However, IRMs also contribute to β cell health and glucose homeostasis (10). Thus, therapeutically modulating its proinflammatory status may be preferential to overt depletion in T1D. We hypothesized that overexpression of ST8Sia6 in β cells would produce α-2,8-disialic ligands that could engage Siglec-E on IRMs and reduce incidence of diabetes resulting from an inflammatory insult to the β cell. Using multiple low-dose streptozotocin (MLD-STZ), whereby β cell destruction and progression to hyperglycemia is influenced by the innate inflammatory response (11), we found that ST8Sia6 overexpression in β cells appreciably reduced the severity of hyperglycemia and, importantly, preserved β cell mass as well. Thus, to our knowledge, this novel study highlights the potential therapeutic benefit of genetically engineering ST8Sia6, a sialyltransferase that produces preferred ligands for Siglec-E, into cells to protect from deleterious inflammation.
Materials and Methods
C57BL/6J (stock no. 000664), Ins2-cre (stock no. 003573), and LNL-tTA mice (stock no. 008600) were obtained from the Jackson Laboratory. ST8Sia6 KO mice were generated by injecting TALENs into C57BL/6 blastocysts, resulting in a 10-nt deletion in exon 4. This created a frameshift mutation after aa 112 and a truncation that resulted in the loss of the sialyltransferase domain of ST8Sia6. Mice with myc-tagged ST8Sia6 under the control of a tetracycline-responsive element (TRE) were generated as performed previously (12). Male mice between 10 and 14 wk of age were used for MLD-STZ experiments. Prediabetic NOD (<250 mg/dl blood glucose) female mice of various ages were used for assessment of insulitis and Siglec-E expression on IRMs. All animal studies were reviewed and conducted in accordance with the Institutional Animal Care and Use Committee at Mayo Clinic.
Islet isolation, cell culture, and dissociation
Pancreata were isolated by perfusing 3 ml of 1.5 mg/ml cold Collagenase P (MilliporeSigma) dissolved in HBSS (Corning) through the common bile duct, digesting for 10 min at 37°C, and separating islets by centrifugation using a 1.077 g/ml Histopaque gradient (MilliporeSigma). For flow cytometry, islets were dissociated into single cells using enzyme-free Cell Dissociation Buffer (Life Technologies) for 30 min at 37°C. For cell culture, intact islets were rested overnight in RPMI 1640 (Thermo Fisher Scientific) before dissociation and probing with recombinant Siglec-E or before 24-h incubation with 50 ng/ml IFN-γ or 1 mM streptozotocin (STZ).
Single cells (from dissociated islets or thymi) were washed with FACS buffer, blocked with 1:1 mouse/rat serum on ice for 5 min, and stained with Abs or recombinant Siglec-E on ice for 30 min. The directly conjugated fluorescent Abs (BioLegend) used included CD45 (1:1000, clone 30-F11), CD11c (1:100, clone N418), F4/80 (1:200, clone BM8), CD103 (1:100, clone 2E7), CD11b (1:200, clone M1/70), Ly-6C (1:200, clone HK1.4), TCRβ (1:100, clone H57-597), CD19 (1:200, clone 6D5), Siglec-E (1:200, clone M1304A01), Siglec-7 (1:100, clone 6-434), Siglec-9 (1:100, clone K8), CD4 (1:200, clone GK1.5), CD8α (1:200, clone 53-6.7), CD69 (1:200, clone H1.2F3), and H-2Kb (1:200, clone 25-D1.16). Recombinant Siglec-E, Gln20–Phe355 with a C-terminal human IgG1-Fc tag (1:100; BioLegend), was incubated with murine cells alone on ice for 30 min, followed by washing and staining with primary and secondary anti-human IgG (1:100; Jackson ImmunoResearch) Abs on ice for 30 min; for human cells, the same process was performed with both recombinant Siglec-7, Gln19–Gly357 with a C-terminal human IgG1-Fc tag (1:100; R&D Systems), and recombinant Siglec-9, Gln18–Gly348 with a C-terminal human IgG1-Fc tag (1:100; R&D Systems). All experiments included subsequent incubation with Fixable Viability Dye (Tonbo Biosciences) for 10 min at room temperature. Stained cells were analyzed using an Attune NxT Flow Cytometer (Thermo Fisher Scientific), and data were processed using FlowJo (Tree Star) 10.
STZ (MilliporeSigma) was dissolved in 0.1 M citrate buffer (pH 4.5) at a dose of 50 mg/kg and immediately injected i.p. within 5 min. This was repeated daily for 5 d. Nonfasting blood glucose levels were measured with a Contour Next blood glucose meter and test strips (Bayer). For glucose and insulin tolerance tests, blood was analyzed before, 15, 30, 60, and 90 min after i.p. injection of either 1 g/kg glucose or 0.75 U/kg human insulin (Novolin R). Mice were sacrificed at day 32 for histological analysis or at humane end point.
Pancreatic sections were stained with Abs against insulin (1:400, polyclonal/A0564; Agilent Technologies) or myc-tag (1:200, clone 9B11; Cell Signaling Technology) overnight at 4°C. After washing, slides were incubated with secondary Ab (5 μg/ml; Invitrogen) for 1 h at room temperature and imaged under a fluorescent microscope. For analysis of total pancreas sections, blocks were cut at two different depths to obtain unique sections of tissue and were serially imaged using a Zeiss Axio Observer Z1 microscope and stitched together using ZEN pro software (Carl Zeiss Microscopy). Regions of interest were drawn around the entire pancreas based on DAPI staining and β cells based on insulin staining. The percentage of insulin+ area was calculated using ImageJ.
Statistical tests used are denoted in each figure legend and were calculated using GraphPad Prism.
IRMs express Siglec-E, and cell surface expression is reduced with inflammation
The IRM is an immune cell that occupies the pancreatic islets under homeostatic conditions and has a crucial role in autoimmune diabetes development (13). Murine IRMs were examined for expression of Siglec-E, Siglec-H, and Siglec-F, which could potentially counter local inflammation in the islet microenvironment. Only Siglec-E was highly expressed relative to Fluorescence Minus One controls (Fig. 1A; gating strategy in Supplemental Fig. 1). Furthermore, expression of Siglec-7 and Siglec-9 (human orthologs of murine Siglec-E) was assessed on IRMs obtained from human nondiabetic cadaveric islets. Interestingly, both Siglec-7 and Siglec-9 were expressed on human IRMs (Supplemental Fig. 2A, 2B).
To examine whether Siglec-E expression is modulated by inflammation, its expression on IRMs during the course of insulitis in prediabetic NOD mice was examined. As insulitis progressed, the frequency of Siglec-E+ IRMs was significantly reduced (Fig. 1B). This reduction in Siglec-E was not due to the influx of CD11b+ Ly-6C+ inflammatory monocytes, as gating on Ly-6C− cells revealed similar results (Fig. 1C). Additionally, Siglec-E is not expressed on infiltrating T and B cells (Fig. 1D). CD4+ T cells are one of the first immune cells to infiltrate the islet in the NOD model of diabetes, and the diabetogenic T cell response in NOD mice is Th1-skewed, characterized by IFN-γ production (14). Interestingly, IFN-γ was recently demonstrated to reduce the expression of Siglec-15 in human monocytes (15). Therefore, IFN-γ produced by local T cells in the islet microenvironment may downregulate Siglec-E. To address this, intact islets isolated from C57BL/6J mice were cocultured with 50 ng/ml IFN-γ for 24 h. IRM Siglec-E expression was significantly reduced upon IFN-γ treatment compared with untreated islets (Fig. 1E).
The sialyltransferase ST8Sia6 generates ligands for Siglec-E
During T cell development, gene expression of ST8Sia6 increases with maturation and correlates with generation of ligands for Siglec-E (16). To demonstrate that ST8Sia6 generates ligands for Siglec-E in vivo, ST8Sia6 KO mice were generated. Following positive selection at the double-positive stage, CD4 single-positive thymocytes mature and transition from CD69+ H2-Kb− semimature to CD69+ H2-Kb+ mature-1 (M1) and finally CD69− H2-Kb+ mature-2 (M2) cells. By probing with recombinant Siglec-E, Siglec-E ligands were significantly reduced in maturing thymocytes at the M2 stage in ST8Sia6 KO mice, definitively demonstrating that ST8Sia6 generates ligands for Siglec-E in vivo (Fig. 2A, 2B, Supplemental Fig. 3A). As a control, treatment of wild-type thymocytes with 0.1 U/ml sialidase completely abolished binding of recombinant Siglec-E to M2 cells (Supplemental Fig. 3B). This suggests that ST8Sia6 may have implications in immune processes by generating ligands for relevant Siglecs, such as Siglec-E. Therefore, we examined whether ST8Sia6 overexpression could dampen the immune response in a disease model.
Overexpression of ST8Sia6 in pancreatic β cells increases ligands for Siglec-E
A system was created to spatiotemporally drive overexpression of ST8Sia6 in β cells. In brief, myc-tagged ST8Sia6 under the control of TRE was knocked into the COL1A1 locus and interbred with RIP-cre (17) and LNL-tTA (18) mice to generate RIP-cre LNL-tTA ST8Sia6 mice (Fig. 3A). When all three alleles are present, ST8Sia6 is overexpressed in pancreatic β cells in the absence of doxycycline. For simplicity, RIP-cre LNL-tTA ST8Sia6 mice will be referred to as βST, and mice harboring any other combination of two or fewer alleles will be referred to as littermates. To confirm ST8Sia6 expression in β cells, immunofluorescence staining against insulin and myc-tag on pancreatic sections from βST and littermate mice was performed. As expected, ST8Sia6 overexpression was observed in βST islets. In addition, there were no obvious abnormalities in islet morphology, and insulitis was absent (Fig. 3B). To examine whether overexpression of ST8Sia6 produced ligands that could engage Siglec-E, we probed β cells from βST and littermate mice with recombinant Siglec-E. β cells were identified by flow cytometry as large, autofluorescent, and highly granular. Recombinant Siglec-E bound at significantly greater levels to βST β cells compared with littermate controls, demonstrating that overexpression of ST8Sia6 led to increased production of ligands for Siglec-E on the β cell surface (Fig. 3C). Human Siglec-7 and Siglec-9 are the closest human orthologs of murine Siglec-E. Therefore, we were interested in whether there was comparable baseline binding to the human β cell surface, as observed with murine littermate β cells when probed with recombinant Siglec-E. Human nondiabetic cadaveric β cells were probed with recombinant Siglec-7 and Siglec-9, and we found that only ligands for Siglec-9 were expressed on human β cells (Supplemental Fig. 2A, 2B).
β cells from littermate and βST mice were treated with 0.1 U/ml sialidase. Unexpectedly and unlike thymocytes, this only marginally reduced the ability of recombinant Siglec-E to bind to the β cell surface. Therefore, the targets of ST8Sia6 may differ, depending on cell type. As a detergent must be present for sialidase to hydrolyze glycolipids, sialylated glycolipids were maintained during our live cell analysis. Intriguingly, recombinant Siglec-E bound to sialidase-treated βST β cells significantly more than sialidase-treated littermate β cells (Supplemental Fig. 3C). This suggests ST8Sia6 may target a glycolipid substrate in murine β cells.
β cell ST8Sia6 overexpression mitigates severe hyperglycemia induced by MLD-STZ–induced diabetes
Because of the β cell’s close contact with Siglec-E+ IRMs in the islet microenvironment, the effects of both ST8Sia6 deficiency and β cell overexpression of ST8Sia6 in the context of diabetes induced by MLD-STZ were examined. Administration of STZ in low doses induces a DNA damage response, leading to elevated proinflammatory molecules IL-1β and Cxcl10 and subsequent insulitis (11). In contrast to administration of high-dose STZ, which causes rapid necrosis of β cells and severe hyperglycemia, MLD-STZ–induced diabetes occurs progressively in male C57BL/6J mice because of limited apoptosis of β cells (19). Blood glucose levels in ST8Sia6 KO mice were similar to C57BL/6J controls after MLD-STZ (Fig. 4A). In stark contrast, βST mice exhibited significantly lower blood glucose levels compared with littermate controls after administration of MLD-STZ, indicating a protective effect of ST8Sia6 overexpression in β cells during diabetes development (Fig. 4B). The presence of cre did not have a confounding effect on this observation, as comparing cre+ littermates to βST yielded similar results (Supplemental Fig. 4). To define whether β cell overexpression of ST8Sia6 impacts glucose homeostasis, an i.p. glucose tolerance test (IPGTT) and an i.p. insulin tolerance test (IPITT) comparing βST and littermate mice was performed. There was no statistically significant difference between βST and littermate controls in either IPGTT (Fig. 4C) or IPITT (Fig. 4D), demonstrating that ectopic expression of ST8Sia6 does not alter glucose homeostasis in βST mice. As βST resistance to hyperglycemia after MLD-STZ challenge could be due to preservation of β cells, insulin+ cells of total pancreatic sections were examined at experimental end point (day 32). There was a significantly increased percentage of insulin+ area in βST pancreatic sections (Fig. 4E) compared with littermate controls. Although significantly reduced compared with littermates, an increase in blood glucose after MLD-STZ in βST mice over baseline was observed. As STZ is a β cell toxin that elicits apoptosis even in low doses, an increase in βST blood glucose after MLD-STZ was not unexpected. To determine whether overexpression of ST8Sia6 altered the intrinsic effects of STZ on β cell survival, islets from βST and littermate mice were exposed to 1 mM STZ for 24 h. Similar disruption in islet morphology and equivalent β cell death was noted, demonstrating that βST β cells are not intrinsically resistant to STZ-induced damage and death (Fig. 4F). Notably, the difference in blood glucose between βST and littermates diverged at approximately day 16, when an immune response and insulitis would be expected to occur. As overexpression of ST8Sia6 occurs locally within the islets, protection from hyperglycemia may stem from local immune modulation, although the precise mechanism is unclear.
At this time, we cannot rule out that β cell ST8Sia6 overexpression has protective effects outside of producing ligands and immunomodulating IRMs via Siglec-E engagement. All autoantigens in T1D are proteins that are linked to the secretory pathway (20). As ST8Sia6 is expressed in the Golgi, sialic acid incorporation into autoantigens may prevent recognition by altering incorporation into MHC and preventing T cell recognition or by allowing polarization of islet-infiltrating T cells to a regulatory phenotype. However, few autoantigens have been identified in the context of MLD-STZ, and the role of autoreactive T cells remains controversial in this model, so an influence on innate immunity is more likely.
β cell destruction following MLD-STZ was diminished in βST mice, indicating a profound protective effect of ST8Sia6 on cells targeted by an inflammatory insult. Because expression of Siglec-E is reduced as insulitis progresses, the opportunity to engage this receptor with sialylated ligands may be a therapeutic option for people with T1D early in the disease process. Alternatively, altering islets or β cells before transplantation using gene therapy or physical incorporation of sialic acids into supporting biomaterial or associated nanoparticles presents a unique therapeutic opportunity not currently available in islet transplantation.
Since the first procedure using the Edmonton protocol in 1999, islet transplantation has established itself as a promising therapy for patients with longstanding T1D, and β cell replacement using replenishable sources, such as human embryonic stem cell–derived β cells, has the potential to become curative (21). Still, current drug regimens in islet transplantation use suppressors of the T cell response and IL-2 production, exposing patients to risk of infection and malignancies (22). However, innate immunity plays a critical role in graft rejection and is required to prime an adaptive response. The incorporation of sialylation locally within the graft may synergize with Ag release and presentation by Siglec-bearing APCs to induce a tolerogenic response to a broad array of native graft-derived Ags. Therefore, engineering sialylated ligands to target Siglecs on graft-infiltrating innate immune cells may offer an innovative local immune therapy option in the context of β cell replacement.
Future research will determine whether inducing ST8Sia6 overexpression only after diabetes onset leads to remission. Finally, backcrossing βST mice to the NOD background, which develops immunity directed against pancreatic β cells, will determine whether protection is still observed using this robust model of spontaneous autoimmunity. In conclusion, Siglec-E is expressed on IRMs and is downregulated in the presence of insulitis, and IFN-γ is sufficient to mediate downregulation. Using two novel mouse models, we have shown that ST8Sia6 produces ligands for Siglec-E, and that overexpression of ST8Sia6 in pancreatic β cells significantly increases Siglec-E binding on the β cell surface. Finally, ST8Sia6 overexpression in β cells protects against severe hyperglycemia when diabetes is induced using MLD-STZ, initiating a new avenue of research into the role of the Siglec/sialic acid axis as a potential modulator of immune-mediated diabetes.
We thank Barsha Dash and David Friedman for thoughtful discussions, Matthew Brown for assistance in quantifying percentage insulin+ area, Aleksey Matveyenko for his assistance with the human cadaveric islet studies, Karl Clark and Stephen Ekker for providing TALENs to generate ST8Sia6 KO mice, and the Mayo Clinic Transgenic and Knockout Core for generation of ST8Sia6 KO mice and TRE-ST8Sia6-myc-tag knock-in mice.
This work was supported by National Institute of Allergy and Infectious Diseases, National Institutes of Health Grant R21AI138858-01A1 (to V.S.S.). P.J.B. was supported by the Mayo Clinic Wilbur T. and Grace C. Pobanz Predoctoral Fellowship.
The online version of this article contains supplemental material.
Abbreviations used in this article:
i.p. glucose tolerance test
i.p. insulin tolerance test
CD69− H2-Kb+ mature-2
multiple low-dose streptozotocin
sialic acid-binding Ig-type lectin
RIP-cre LNL-tTA ST8Sia6 mice
type 1 diabetes
The authors have no financial conflicts of interest.