A sulfonamide sialoside analog for targeting Siglec-8 and -F on immune cells

Abstract

The Siglec family of cell surface receptors have emerged as attractive targets for cell directed therapies due to their restricted expression on immune cells, endocytic properties and ability to modulate receptor signaling. Human Siglec-8, for instance has been identified as a therapeutic target for the treatment of eosinophil and mast cell disorders. A promising strategy to target Siglecs involves the use of liposomal nanoparticles with multivalent display of Siglec ligands. A key challenge for this approach is the identification of a high affinity ligand for the target Siglec. Here we report the development of a ligand of Siglec-8 and its closest murine functional ortholog Siglec-F that is capable of targeting liposomes to cells expressing Siglec-8 or -F. A glycan microarray library of synthetic 9-N-sulfonyl sialoside analogs was screened to identify potential lead compounds. The best ligand, 9-N-(2-naphthyl-sulfonyl)-Neu5Acα2–3-[6-O-sulfo]-Galβ1–4GlcNAc (6′-O-sulfo ^NSANeu5Ac) combined the lead 2-naphthyl sulfonyl C-9 substituent with the preferred sulfated scaffold. The ligand 6′-O-sulfo ^NSANeu5Ac was conjugated to lipids for display on liposomes to evaluate targeted delivery to cells. Targeted liposomes showed strong in vitro binding/uptake and selectivity to cells expressing Siglec-8 or -F, and when administered to mice, exhibit in vivo targeting to Siglec-F⁺ eosinophils.

Graphical Abstract

The Sialic acid-binding immunoglobulin-type lectins (Siglecs) are a family of cell surface receptors with restricted expression on one or a few immune cell types.^1–2 Due to their expression patterns, endocytic properties, and ability to modulate receptor signaling, Siglecs have emerged as attractive targets for cell directed therapies.^3–16 Examples include targeting Siglec-1 on macrophages,^6–7 Siglec-2 on B cells,^8–10 Siglec-3 on mast cells,^11–13 Siglec-9 on T cells,¹⁴ and Siglec-15 on osteoclasts.^{15, 17} Furthermore, Siglec-8 has been identified as a therapeutic target for the treatment of eosinophil and mast cell disorders, with antibodies targeting Siglec-8 in clinical trials.^18–22 A promising alternative to antibodies to target Siglecs involves use of a multivalent display of Siglec ligands presented on nanoparticles or polymers.³ Indeed, ligand mediated targeting of Siglecs has been used to deliver diagnostic or therapeutic agents to a variety of immune cells.^{6, 23–28} A key challenge for this strategy is the identification of selective, high affinity ligands for the target Siglec.

Natural sialylated glycan ligands of Siglecs have variable selectivity and generally low monovalent affinity (0.1–3 mM). Glycan ligands gain avidity through multivalent interactions.^29–31 These glycan scaffolds, however, can serve as starting points to develop more optimal synthetic ligands. It has been well established that modifying positions on sialic acid can modulate the selectivity and affinity of Siglec ligands.³² A productive strategy to identify ligands has involved synthesizing and screening libraries of glycan analogs.^33–39 While high affinity ligands have been developed for many Siglecs,^{3, 40–41} suitable ligands are currently unavailable for Siglec-8 or its closest murine functional ortholog Siglec-F, the latter also being prominently expressed on eosinophils. In view of the therapeutic potential of targeting eosinophils and mast cells we set out to develop selective ligands for Siglec-8 and -F. We describe here a ligand of Siglec-8 and -F capable of targeting Siglec-8 and -F positive cells, and in vivo targeting of murine eosinophils.

While initial attempts to identify ligands for Siglec-8 and -F by screening established sialoside analog libraries with amide-linked substituents at C-5/−9 yielded no ‘hits’, we obtained promising evidence of binding of Siglec-8 and -F to a focused panel of C-9 sulfonamide analogs (1-17 α2–3, Scheme 1, Figure 1 and S2). Based on these results we chemoenzymatically synthesized an expanded library of sulfonamide-based C-9 modified sialoside analogs (1-78 α2–3; 79-156 α2–6) (Scheme 1). Although, Siglec-8 and -F are known to bind 6′-O-sulfo Neu5Acα2–3Galβ1–4GlcNAc⁴² and 6′-O-sulfo Sialyl-Lewis-X⁴³ we constructed the library based off C-9 modification of the synthetically more accessible non-sulfated Neu5Acα2–3- and Neu5Acα2–6-Galβ1–4GlcNAc scaffolds. Briefly, the 9-NH₂ trisaccharides (C) and (D) were synthesized and purified on ~50 mg scales by reacting CMP-9-NH₂-Neu5Ac⁴⁴ (A) with Galβ1–4GlcNAc-ethyl azide⁴⁵ (B) using either P. multocida α2,3-sialyltransferase⁴⁶ or P. damsela α2,6-sialyltransferase⁴⁷, respectively. The sulfonamide analogs were synthesized from C and D in parallel two-step one pot reactions (0.5 mg scale) by reacting the 9-NH₂ groups with a panel of substituted sulfonyl chlorides (~2 eq. RSO₂Cl, see Table S1). Completed reactions (TLC) were placed in a vacuum desiccator to remove the methanol. After quenching excess RSO₂Cl with H₂O (pH 9–10), the ethyl azide aglycones were reduced using trimethyl phosphine (PMe₃, 2 eq.) to generate an ethyl amine linker. The reactions were concentrated under reduced pressure then the products were diluted to 0.1 mM (H₂O) without further purification for printing on the glycan microarray. In total, the library consists of 156 sulfonamide sialoside analogs. The glycans were printed directly on amine-reactive N-hydroxy succinimide (NHS)-activated glass slides (Schott Nexterion® Slide-H).³³ Neu5Acα2–3Galβ1–4GlcNAc (C1) which is the parent scaffold used to construct the analog library, and 6′-O-sulfo Neu5Acα2–3Galβ1–4GlcNAc (C2) which is the known Siglec-8 and -F ligand were included as controls. The printed slides were washed to remove excess synthetic reagents.

Scheme 1. Chemo-enzymatic synthesis of sulfonamide analogs.^a.

^aReagents and conditions: (i) Pasteurella multocida α2,3-sialyltransferase; (ii) Photobacterium damsela α2,6-sialyltransferase; (iii) RSO₂Cl (1-156, see Table S1), DIEA (5 eq.), CH₃OH; (iv) PMe₃ (2 eq.), THF, H₂O (pH 9).

Figure 1.

Binding of recombinant Siglec-8 to the α2,3-sialyl sulfonamide analog array. Arrays were screened to identify C-9 substituents of sialic acid that show increased binding compared to control glycan C1. Human Siglec-8 COMP (10 μg/ml) was pre-complexed with anti-penta-histidine IgG-Alexa Fluor 488 (5 μg/ml). The complexed proteins were overlaid onto the array. After incubation the slides were washed then scanned for fluorescence. Analogs 1-78 correspond to the groups listed in Table S1. Shown is mean fluorescence intensity of Siglec-8 binding. Each glycan was printed at 100, 20, 4, 0.8, and 0.16 μM in 4 replicates each (increasing concentration from left to right). The controls are Neu5Acα2–3Galβ1–4GlcNAc (C1) and 6′-O-sulfo Neu5Acα2–3Galβ1–4GlcNAc (C2).

The sulfonamide analog array (1-156) was screened against fluorescently labeled recombinant Siglec-8 COMP (Cartilage Oligomeric Matrix Protein)⁴⁸ or Fc and Siglec-F Fc chimeras to identify substituents that show increased binding compared to C1 (Figures 1, S1, and S2). As expected, Siglec-8 and -F bound the 6′-O sulfated C2 (down to 4 μM) however neither bound C1. Binding of the Siglecs to the analogs was markedly different. Siglec-8 bound to only a few of the α2–3 analogs while Siglec-F bound strongly to many (1-78, Figure 1 and S2). Each Siglec bound to only a few of the α2,6 analogs (79-156, Figure S1 and S2). Siglec-8 showed robust binding to several ligands (12, 23, 61, 74). We selected 12 as a lead hit since it showed strong binding down to 0.8 μM. It comprises Neu5Ac modified at C-9 with a 2-naphthyl sulfonyl group (^NSANeu5Ac, NSA = 2-naphthyl sulfonamide). To further optimize this hit we also synthesized 158 (^NSANeu5Acα2–3[6SO₄]Galβ1–4GlcNAc, 6′-O-sulfo NSANeu5Ac), which combines the ^NSANeu5Ac with the preferred scaffold C2 (Scheme S1).

We next used a cell binding assay to evaluate if liposomes displaying ligands C2, 12, or 158 could bind Chinese hamster ovary (CHO) cells expressing Siglec-8 or -F (Figure 2 and S3).⁴⁹ Accordingly, ligands C2, 12, and 158 were synthesized, purified then coupled to NHS activated PEGylated lipid to generate the conjugates 159–161 (Scheme S1–S3). ‘Targeted’ liposomes containing these glyco-PEG-lipid conjugates (2 mol%) and fluorescent (Alexa Fluor® 647) lipids were formulated as described in the supporting information. Fluorescent ‘non-targeted’ liposomes without Siglec ligand were used as a control. Siglec-8 and -F expressing CHO cells were incubated with targeted (159-161) or non-targeted liposomes for one hour, washed then analyzed for binding of fluorescent liposomes by flow cytometry (Figure 2 and S3). Liposomes formulated with either 160 or 161 strongly bound Siglec-F expressing cells with slight preference for 161. In contrast Siglec-8 expressing CHO cells only bound liposomes containing 161. In contrast, liposomes displaying the natural ligand C2 did not bind to either Siglec-8 or -F expressing CHO cell (Figure S3). This result demonstrates that both the C-9 2-naphthyl sulfonamide and 6′-O-sulfation of 161 are required to support binding of liposomes to Siglec-8 expressing cells. Accordingly, we proceeded to further evaluate liposomes decorated with 161.

Figure 2.

In vitro binding/uptake of fluorescent targeted liposomes displaying 160 or 161 to CHO cells expressing Siglec-8 and -F. Cells were treated with 20 μM fluorescent targeted liposomes (2 mol% of 160 or 161) at 37 °C for 1 hour. Liposome binding was assessed by flow cytometry. Low binding to some cells reflects loss of Siglec expression (see Figure S4).

To assess the degree of specificity for Siglec-F/8, the targeted (161) liposomes were then assessed against a panel of cell lines expressing several human and murine Siglecs. Cells were incubated with fluorescent targeted (161, 2 mol%) or non-targeted liposomes and analyzed by flow cytometry as described above. For murine Siglecs, targeted liposomes (161) bound strongly to and were highly selective for Siglec-F (Figure 3). For the human Siglecs, the targeted liposomes showed strong binding to Siglec-8 and to a lesser extent to Siglec-3, −5, and −9. We further tested if targeted liposomes (161) bound bone marrow derived eosinophils (bmEos) generated from wild-type (Siglec F⁺/8⁻) and various transgenic (Siglec F⁺/8⁺, Siglec F⁻/8⁺, and Siglec F⁻/8⁻)^50–51 mice (Figure S5). In all cases the targeted liposomes bound to cells expressing Siglec-F and/or −8, However, no appreciable binding was observed with bmEos from Siglec F⁻/8⁻ mice. These results demonstrate that liposomes decorated with 161 selectively bind to Siglec-8 and -F expressing cells.

Figure 3.

Flow cytometry analysis of in vitro binding/uptake of targeted (161, 2 mol%) or non-targeted (no ligand) fluorescent liposomes to cells expressing human and murine Siglecs. Binding is shown as mean fluorescence intensity (MFI) ± SEM (n = 3).

To evaluate targeting cells in vivo, we investigated whether Siglec-F expressing eosinophils in the spleen of normal mice could be labeled with fluorescent liposomes containing 161 (Figure 4A, S6, and S7). To this end, wild type (WT) mice were intravenously given fluorescent targeted (161) or non-targeted liposomes. After 1 and 24 hours, the spleens were harvested and the binding/uptake of liposomes by splenocytes was analyzed by flow cytometry. Populations of eosinophils (CD11b⁺CCR3⁺), B (CD19⁺) and T (CD4⁺/CD8⁺) cells were identified using antibodies to surface markers of each cell type (Figure 4A). Analysis of the gated cell populations revealed that the targeted liposomes strongly labeled Siglec-F⁺ eosinophils in vivo. There was no appreciable binding of targeted liposomes to any other cell type, however, a small amount of labeling of B cells was observed (Figure 4A and S6 and S7). Since Siglec-G is a B cell Siglec, and we had observed a slight but significant labeling of Siglec-G expressing CHO cells not evident in Figure 3, we compared in vitro labeling of splenocytes from WT and Siglec-G knockout (KO) mice (Figure S8), and found that the weak binding to B cells is Siglec-G mediated.

Figure 4.

(A) Targeted liposomes (161) bind eosinophils in vivo. WT mice were intravenously given either targeted (161, 2 mol%, n=3) or non-targeted (no ligand, n=3) fluorescent liposomes. After one hour, splenocytes were harvested, stained with antibodies and analyzed by flow cytometry. In vivo binding of targeted (black line) or non-targeted liposomes (grey) to eosinophils (CD11b⁺CCR3⁺, left) or B/T cells (CD4⁺CD8⁺CD19⁺, right) were overlaid. Representative overlays are shown. See also Figure S6. (B) Targeted liposomes (161) do not alter eosinophil frequency in vivo. WT mice (n=4) were intravenously given PBS (white), non-targeted liposomes (grey), targeted liposomes (161, 5 mol%) (black), or anti-Siglec-F (red). Eosinophil frequencies in the blood were analyzed 1- and 3-days post injection and determined by dividing cells that were CD11b⁺CCR3⁺ by live immune cells (PI⁻CD45⁺). *** P <0.001; N.S., not significant (P >0.05) determined by 1-way ANOVA followed by Tukey’s test.

Lastly, previous work demonstrated that anti-Siglec-F and −8 antibodies deplete eosinophils in vivo.^52–55 Moreover, glycan ligand mediated ligation of Siglec-8 was also observed to increase death of human eosinophils in vitro.^55–56 To determine if the targeted liposomes (161) impacted eosinophil survival in vivo, WT mice were intravenously injected with either anti-Siglec-F antibody⁵⁷ or liposomes containing 5 mol% 161 (Figure 4B). Blood eosinophil frequency was then determined 1- and 3-days post injection by flow cytometry. Eosinophils were identified as CD11b⁺CCR3⁺ and their frequency was determined by comparison to total live peripheral blood leukocytes (CD45⁺PI⁻). As seen by others,⁵² anti-Siglec-F potently decreased eosinophil frequencies at both time points. In contrast, there was no significant change observed in eosinophil frequencies in mice injected with ligand targeted liposomes relative to control mice. The results suggest that anti-Siglec-F may deplete eosinophils in vivo by an immune dependent mechanism rather than ligation dependent signaling of Siglec-F. Furthermore, they suggest that targeted (161) liposomes could serve as a vehicle for delivery of agents to eosinophils without causing their depletion.

In summary, we have described the successful development of a sulfonamide-based glycan analog ligand, 6′-O-sulfo ^NSANeu5Ac that binds both Siglec-8 and -F. The sulfonamide substituent was identified by screening a glycan analog microarray against recombinant Siglec-8 and -F. The best ligand combined the C-9 sulfonamide with the preferred sulfated glycan scaffold (^NSANeu5Acα2–3[6-SO₄]Galβ1–4GlcNAc). Targeted liposomes showed strong binding to Siglec-F and −8 expressing cells. Evaluation of the targeted liposomes in mice showed effective in vivo targeting of eosinophils without significantly affecting eosinophil frequency. We envisage that incorporating this novel ligand of Siglec-8 and -F into nanoparticle delivery systems will allow targeting of eosinophils and modulation of immune cell responses.