To the editor:

In a recent issue of Nature Medicine, Schön et al. characterized efomycine M (Efo-M) isolated from Streptomyces BS1261 as a specific inhibitor of selectins1. The authors showed that Efo-M at low micromolar concentrations is able to block E- and P-selectin interactions with sialyl Lewisx (sLex)-bearing ligands, for example, carcinoembryonic antigen (CEA). Moreover, Efo-M inhibited selectin-mediated leukocyte rolling in vivo and substantially improved the clinical symptoms of two in vivo inflammatory skin models using transplanted human skin biopsies. Thus, Efo-M was reported to function as a pan-selectin antagonist with promising therapeutic potential.

Following the hypothesis by Schön et al., we performed extended in vitro and in vivo investigations with Efo-M. Although in principle we could confirm the anti-inflammatory profile of Efo-M, our findings suggest a mode of action that is separate from pan-selectin inhibition. As shown in Figure 1a, highly purified Efo-M (Supplementary Fig. 1 online) does not compete with the binding of 3H-labeled, polyvalent sLex (3H–p-sLex) to E-, L- and P-selectins, even at concentrations of 100 μM in scintillation proximity assays (SPAs)2. In contrast, an excess of unlabeled p-sLex as well as fucoidan (Supplementary Fig. 2 online) clearly inhibited the binding of 3H–p-sLex, whereas monovalent sLex competed with 3H–p-Lex only in binding to E-selectin (Supplementary Fig. 3 online), as described previously3. As the experimental setup of SPAs described in Figure 1a allows the quantification of binding only to the glycan binding site of selectins, we repeated the experiment with 3H-labeled Efo-M (Fig. 1b), which should allow the detection of Efo-M binding to a glycan-independent binding site. However, no specific (for example, Ca2+-dependent and selectin-specific) binding of 3H-labeled Efo-M to L- and P-selectin could be detected. Moreover, measuring direct interactions of Efo-M to selectins in ELISA (Fig. 1c) as well as in highly sensitive Biacore experiments revealed only a weak binding of Efo-M to L-selectin at higher concentrations, and no specific binding of Efo-M to E- and P-selectin in this type of assay was detectable (Fig. 1d).

Figure 1: No specific Efomycine M binding to E-, L- and P-selectins.
figure 1

(a) Binding of 3H-p-sLeX (glycoconjugates (p-sLeX, poly[N-(2-hydroxyethyl)acrylamide] with 20% molar ratio sLeX) to selectin fusion proteins was tested in a bead-based scintillation proximity assay (SPA) format similar to one previously described2. Protein-A coated SPA-beads (1 μg/μl), E-, L- or P-selectin fusion proteins (each 0.5 ng/μl) and test compounds (10 μM unlabeled p-sLex or Efo-M at the indicated concentrations) were mixed in a 96-well microplate. 3H–p-sLeX (1 nCi/μl) was added, the microplate was sealed and scintillation was measured after 30 min incubation in a Topcount instrument. Final buffer conditions were 25 mM HEPES, pH 7.4, 137 mM NaCl, 2.7 mM KCl, 0.5 mM CaCl2 and 0.1% bovine serum albumin. All assay steps were performed at 20 °C in a total volume of 100 μl. Addition of EDTA (10 mM) abolished binding of 3H–p-sLeX to immobilized selectins. Data are mean ± s.d. from triplicate values. (b) 3H–Efo-M was coincubated with selectin fusion proteins or a control immunoglobulin in a SPA as described in a. (c) Binding to P-Sel-Ig was monitored in a cell-free ELISA with horseradish peroxidase–conjugated Streptavidin. Plastic-bound anti-human Ig (0.5 μg/ml) was incubated with P-selectin–Ig fusion proteins (1 μg/ml) and binding of biotinylated p-sLeX was determined. Two different batches of highly purified Efo-M (Charge A, Charge B) did not inhibit this interaction, whereas fucoidan (starting with 200 μg/ml, 1:3 dilutions) or unlabeled p-sLeX (data not shown) reduced the binding to background levels at higher concentrations. (d) Binding of E-, P- and L-selectin–Ig fusion proteins immobilized on Protein-A beads to chip surface–bound p-sLeX was determined in the absence and in the presence of titrated amounts of Efo-M in BiaCore analysis. (e) Inhibition of contact hypersensitivity following prolonged exposure to Efo-M. Experiments were carried out in accordance with the German Tierschutzgesetz (Animal Protection law) along the guidelines of the Landesamt für Arbeitsschutz, Gesundheit und technische Sicherheit (LAGetSI) of the Land Berlin, Germany. Female NMRI mice (24–28 g) were sensitized and challenged with DNFB according to standard methods. Twenty-four hours before challenge, mice received intraperitoneal injections of Efo-M (5.0 mg/kg body weight per dose in DMSO/ethanol/transcutol). Twenty-four hours after DNFB challenge, ear thickness (e) was determined and mice were killed for measurement of neutrophil and granulocyte infiltration (Supplementary Fig. 8). Data show mean ± s.d. n = 10 per group. *P < 0.05 (treatment versus DNFB challenge). Prednisolone served as a positive control for anti-inflammatory treatment. Serum levels of Efo-M after systemic administration are shown in Supplementary Figure 9. (f–h) Efo-M interferes with lymphocyte viability/activation. (f) Spleen cells from mice (in e) were stimulated ex vivo with LPS (100 ng/ml for 18 h or left unstimulated (Unstim.). TNF-α in the supernatants was determined with specific ELISA (triplicate values). (g) Human PBMCs were activated with CD2-, CD3-, CD28-coated beads for 24 h in the absence or presence of Efo-M. Cell viability and proliferation was assessed (y-axis, 530/590 nm arbitrary units (AU), triplicate values) with AlarmarBlue. Unstimulated, stimulated and cells incubated with the same concentration of the Efo-M solvent DMSO (0.3%) served as controls. (h) Purified, unstimulated and stimulated human B (LPS, 100 ng/ml) and T cells (CD2-, CD3-, CD28-coated beads) were incubated with titrated amounts of Efo-M in 96-well plates. After 18 h, T cells were stained with propidium iodide (PI) and Annexin-V and analyzed in a FACScalibur. For DMSO, stimulated and unstimulated samples, the mean value of B and T cells is given. Results were reproduced in at least two independent experiments.

Full size image

In their original study and a subsequent publication4, Schön et al. showed a model-based alignment of Efo-M and sLex structures, suggesting a comparable orientation of hydroxyl groups in Efo-M and sLex, which are crucial for the adhesive function of selectin ligands. Our analysis using crystallized Efo-M and a liquid-state conformation analysis of Efo-M by nuclear magnetic resonance (NMR) spectroscopy provided no evidence that sLex and Efo-M share a three-point pharmacophore when binding to the selectin surface (Supplementary Fig. 4 online). These results as well as the lack of binding of Efo-M to selectins under different experimental conditions suggest an alternative explanation for the findings of Schön et al. Although we observed anti-inflammatory activity in our in vivo models as well (Fig. 1e), Efo-M clearly exhibited selectin-independent effects on immune cells at low micromolar concentrations. Ex vivo–stimulated spleen cells from dinitrofluorbenzene-challenged and Efo-M–treated mice produced less tumor necrosis factor (TNF)-α (Fig. 1f). Moreover, in the presence of Efo-M (10 μM), activated mouse spleen cells (data not shown) and human peripheral blood mononuclear cells (PBMCs) showed a 50% reduction in cell viability (Fig. 1g), and lipopolysaccharide (LPS) induced secretion of TNF-α (Supplementary Fig. 5 online). At concentrations of >10 μM, Efo-M induced apoptosis in more than 30% of purified human B and T cells (Fig. 1h) and inhibited the maturation of dendritic cells (Supplementary Fig. 6 online). Efo-M was cytotoxic in PBMC cultures at low micromolar concentrations (50% inhibitory concentration (IC50), 4 μM), indicating again the substantial inhibition of cell viability caused by Efo-M, whereas we observed induction of apoptosis in human endothelial cells with higher concentrations of Efo-M (IC50, 27 μM; Supplementary Fig. 7 online) and no apoptosis in human neutrophils (data not shown). Together, our findings suggest that a selectin-antagonizing function is probably not responsible for the observed in vivo results by Schön et al. and raise questions regarding how Efo-M inhibits inflammatory processes in the skin.

Note: Supplementary information is available at the Nature Medicine website.