The 3-Deoxy Analogue of α-GalCer: Disclosing the Role of the 4-Hydroxyl Group for CD1d-Mediated NKT Cell Activation

Abstract

KRN7000, or α-GalCer, is a potent agonist for natural killer T (NKT) cells. The 3-hydroxyl group of its phytosphingosine moiety is important for activating NKT cells, whereas its 4-hydroxyl group is perceived to be less crucial. To experimentally determine the role of the 4-hydroxyl group, we synthesized the 3-deoxy analogue of α-GalCer. It was found that 3-deoxy-α-GalCer induced potent cytokine responses from NKT cells, comparable to those of both α-GalCer and 4-deoxy-α-GalCer. This result and our docking studies suggest that the effects of an absence of the 3-hydroxyl group are compensated by the presence of a hydroxyl group at the C-4 position. Thus, we conclude that the 4-hydroxyl group of α-GalCer is as important to the mechanism of action as the 3-hydroxyl group and that the two hydroxyl groups could play individual and cooperative roles in orienting the glycolipid into the proper position in CD1d to be recognized by the T cell receptor of NKT cells.

KRN7000 (also commonly called α-GalCer, 1, Figure 1) is a structurally modified analogue of the marine natural product agelasphins.¹ This glycolipid is the first-defined agonist for natural killer T (NKT) cells, and it remains the most extensively studied compound for the exploration of NKT cell biology and pharmacology.²⁻⁶ α-GalCer activates NKT cells in a CD1d restricted manner; it first binds to the CD1d molecule of antigen-presenting cells, and the resulting α-GalCer/CD1d complex is then recognized by the conserved αβ T cell receptor (TCR) to form the ternary complex that leads to activation of the NKT cells. The activated NKT cells promptly secrete large amounts of T helper 1 and 2 (Th1 and Th2) cytokines, such as interferon-γ (IFN-γ) and interleukin-4 (IL-4), which play critical roles in the regulation of innate and adaptive immune responses.⁷⁻⁹

Figure 1.

Structures of glycolipids 1–4.

The recently achieved crystal structure of α-GalCer complexed with CD1d reveals that the acyl and phytosphingosine lipid chains of α-GalCer fit tightly into two hydrophobic pockets of the CD1d binding groove, whereas the galactose group protrudes from the CD1d cleft.¹⁰⁻¹³ There are several hydrogen-bonding interactions between the surface residues of CD1d and the hydroxyl groups of the galactose and the phytosphingosine base that appear to be crucial for maintaining α-GalCer in the correct position for recognition by the TCR of NKT cells.

The hydrogen-bonding network in the crystal structure is well matched with the previous results from the structure–activity relationships (SAR) studies of α-GalCer analogues.^6,14−18 SAR studies on the phytosphingosine moiety of α-GalCer have shown that the analogue lacking the 4-hydroxyl group on the phytosphingosine (4-deoxy-α-GalCer, 2, Figure 1) exhibits slightly reduced activity as compared to α-GalCer, whereas the analogue that lacks both 3- and 4-hydroxyl groups (3,4-dideoxy-α-GalCer, 3, Figure 1) was inactive.¹⁹⁻²² These SAR results, along with the observed hydrogen-bonding interactions of the 3-hydroxyl group of α-GalCer, with residues of both the CD1d (Asp-80) and the TCR (Arg-95), led to the general belief that the 3-hydroxyl group of phytosphingosine was crucial for activating NKT cells, whereas the 4-hydroxyl group was not crucial for activity.^16,20,23 However, the analogue lacking only the 3-hydroxyl group on the phytosphingosine (3-deoxy-α-GalCer, 4, Figure 1) had never been prepared and evaluated to ascertain the individual impacts of the 3- and 4-hydroxyl groups on NKT cell activation. This lack of potentially important SAR data prompted us to synthesize and evaluate the 3-deoxy analogue 4. Herein, we report our studies on this subject.

For the synthesis of the desired 3-deoxy analogue of α-GalCer 4, the known d-ribo-phytosphingosine-derived compound 5(24) was chosen as an ideal starting material given that its 2-amino and 4-hydroxyl groups were already suitably protected (Scheme 1). The primary hydroxyl group of 5 was selectively protected as its silyl ether to afford compound 6. To remove the C-3 hydroxyl group, alcohol 6 was first converted to bromide 7 with CBr₄ and PPh₃. The obtained bromide 7 was then reduced using NaBH₄ and NiCl₂ in EtOH. Under these reaction conditions, the azido group of 7 was concomitantly reduced to give amine 8 in good yield (85%).

Scheme 1. Synthesis of 3-Deoxy-α-GalCer 4.

Reagents and conditions: (a) TBDPSCl, Et₃N, CH₂Cl₂, room temperature, 99%. (b) CBr₄, PPh₃, CH₂Cl₂, 40 °C, 72%. (c) NaBH₄, NiCl₂, EtOH, room temperature, 85%. (d) Bu₄NF, THF, room temperature, 80%. (e) Hexacosanoic acid, EDCI, DMAP, CH₂Cl₂, room temperature, 78%. (f) Compound 12, AgClO₄, SnCl₂, 4 Å molecular sieve, THF, −10 °C, ca. 10%. (g) TfN₃, K₂CO₃, CuSO₄, MeOH, CH₂Cl₂, H₂O, room temperature, 79%. (h) Compound 15, Bu₄NBr, N,N-tetramethylurea, CH₂Cl₂, room temperature, 83%. (i) PPh₃, benzene/H₂O (10:1), 60 °C, 12 h. (j) Hexacosanoic acid, EDCI, DMAP, CH₂Cl₂, room temperature, 54% from 14. (k) H₂, Pd(OH)₂, EtOH/CH₂Cl₂ (3:1), room temperature, 72%.

After successful removal of the C-3 hydroxyl group of ribo-phytosphingosine, the silyl protecting group in 8 was removed with Bu₄NF to furnish amino alcohol 9. Acylation of 9 with hexacosanoic acid and EDCI gave ceramide 10. For selective α-glycosyl bond formation, we employed the Mukaiyama glycosylation reaction²⁵ involving the reactive galactosyl fluoride 12 as the glycosyl donor. This reaction provided the desired α-galactoside 11, but the yield was very low (ca. 10%). Attempts to increase the yield of 11 by varying the glycosyl donor and reaction conditions were not successful.

The low yield of glycosylation product 11 could be attributed to the fact that the amide group of 10 diminishes the nucleophilicity of the primary hydroxyl group through hydrogen bonding.²⁶ An alternative approach was therefore devised in which an azido group was used in place of the amide during the glycosylation. The free amine group of 9 was first reconverted to an azide 13 via a modified diazo transfer reaction.²⁷ When the in situ-generated galactosyl bromide 15(28) was employed as a galactosyl donor, the glycosylation of 13 was accomplished efficiently to give α-galactoside 14 as the only identifiable anomer in good yield (83%). Staudinger reduction of the azido group of 14 followed by condensation of the resulting amine with hexacosanoic acid led to the formation of 11. This route required two more steps than the original but proved significantly more efficient and practical for the preparation of 11. Finally, the global deprotection of the benzyl ethers by hydrogenolysis afforded the desired 3-deoxy analogue 4.

For a preliminary biological evaluation of 3-deoxy analogue 4, we used CD1d-specific NKT cell hybridoma cells (DN32.D3) that are immortalized NKT cells and that tend to produce IL-2 upon stimulation. The parent α-GalCer 1 and its 4-deoxy analogue 2 were also tested for comparison.^29,30 As shown in Figure 2a, 4-deoxy analogue 2 displayed comparable stimulatory effects to α-GalCer 1, confirming previous reports that the activity of 4-deoxy analogue 2 does not differ significantly from that of α-GalCer.¹⁹ Contrary to the expectation, however, the 3-deoxy analogue 4 was slightly more effective than α-GalCer at promoting IL-2 production. Because IL-2 secretion is dependent on antigen-loaded CD1d,³¹ these results indicate that the 3-deoxy analogue 4 efficiently binds to CD1d and triggers TCR signaling in a manner similar to α-GalCer.

Figure 2.

Biological evaluation of deoxy analogues. (a) IL-2 secretion by DN32.D3 NKT hybridoma cells. IL-2 production was measured from cocultured supernatants of NKT hybridoma DN32.D3 and mouse CD1d transfected RBL cells after 16 h of culture. Representative data of two individual experiments are expressed as the mean ± SD of duplicates. (b) IFN-γ and IL-4 secretion by mouse splenocytes. Cytokine production was measured after 72 h of culture. Results are expressed as relative activity. Representative data of two individual experiments are expressed as means ± SDs of triplicates. The statistical significance of the difference in secretion levels was determined by Student's t test. **p < 0.01.

To determine whether the 3-deoxy analogue 4 can stimulate cytokine release from intact NKT cells, we measured the levels of IFN-γ and IL-4 in in vitro culture supernatants of mouse splenocytes stimulated with 8 ng/mL of compounds 1, 2, and 4. Figure 2b shows the relative IFN-γ and IL-4 production levels of 2 and 4 when compared with those of 1. As expected, the NKT stimulation activity of 4-deoxy analogue 2 did not significantly differ from that of α-GalCer 1. Our evaluation showed that 3-deoxy analogue 4 could induce NKT cell cytokine responses similarly to 1. However, compound 4 seemed to bias cytokine secretion toward the Th2 response; 3-deoxy analogue 4 showed a comparable stimulatory effect on IL-4 production as α-GalCer 1, whereas it promoted smaller amounts of IFN-γ production as compared to 1.

The above biological results suggest that the 3-deoxy analogue 4 can be favorably accommodated within the CD1d binding groove. To understand the binding mode of 4, we performed molecular modeling studies on the interaction of analogues 2 and 4 with CD1d and the NKT TCR.^32,33 The X-ray crystallographic structure of ternary complex α-GalCer/hCD1d/TCR (PDB code 2PO6)¹² was used for this docking analysis. The best possible geometry of the analogue within the CD1d/TCR complex was searched using Surflex-Dock. The docking scores of compounds were in partial agreement with the biological data. The 3-deoxy analogue 4 gave a better docking score than either α-GalCer 1 or its 4-deoxy analogue 2 (see Figure S1 in the Supporting Information), although analogue 4 forms fewer hydrogen bonds with the CD1d/TCR complex than 1 or 2 (see Figure S2 in the Supporting Information). In our docking model, the 4-deoxy analogue 2 occupies virtually the same position and forms the same set of hydrogen bonds as α-GalCer in its crystalline complex with hCD1d/TCR (see Figure S3 in the Supporting Information). However, 3-deoxy analogue 4 sits considerably deeper in the groove than α-GalCer 1 as shown in Figure 3a. The galactose headgroup is laterally shifted approximately 1.3 Å toward the center of the binding groove.³⁴ The absence of a hydroxyl group in the 3-position seems to cause the sphingoid base to shift itself lower in the pocket to establish new hydrogen bonds with CD1d/TCR complex (Figure 3b).

Figure 3.

Comparison of the docking model of 3-deoxy-α-GalCer 4 (green backbone) with the crystal structure of α-GalCer 1 (pink backbone) within the hCD1d/TCR (Protein Data Bank code 2PO6). The key amino acid residues are shown in blue (TCR residues) and orange (CD1d residues). (a) α-GalCer 1 (shown in pink) and 3-deoxy-α-GalCer 4 (shown in green) are overlaid. The galactose headgroup of 3-deoxy-α-GalCer 4 is shifted down (ca. 1.3 Å) as compared to that of α-GalCer 1. H-bonds between α-GalCer 1 and hCD1d/TCR are indicated by pink dashed lines. (b) Docking model of 3-deoxy-α-GalCer 4. H-bonds are shown as dark blue solid lines.

Although the computational modeling studies do not provide conclusive proof, our model offers a view of how 3-deoxy analogue 4 may bind CD1d and present its galactose epitope to the TCR of NKT cells. Throughout these molecular modeling and biological studies, the effect of the absence of the 3-hydroxyl group appears to be compensated by the presence of a hydroxyl group at the C-4 position. Thus, the 3- and 4-hydroxyl groups of phytosphingosine are both individually effective at orienting the glycolipid in CD1d and inducing NKT cells responses, although the position of the hydroxyl group affects how the glycolipid binds to CD1d.

In summary, we have prepared the α-GalCer analogue lacking the 3-hydroxyl group on the phytosphingosine (3-deoxy-α-GalCer, 4) to ascertain the individual impacts of the 3- and 4-hydroxyl groups on NKT cell activation. Contrary to the previous perception, 3-deoxy-α-GalCer was found to be similarly effective at inducing NKT cells responses as both α-GalCer and 4-deoxy-α-GalCer. Together with our docking studies, this result suggests that the 4-hydroxyl group of α-GalCer is as important as the 3-hydroxyl group and that the two hydroxyl groups could play both individual and cooperative roles in orienting the glycolipid into the proper position in CD1d to be recognized by the TCR of NKT cells. We believe that 3-deoxy-α-GalCer represents an important new tool for improving the understanding of glycolipid–CD1d interactions and the NKT response. Additionally, it could provide guidance regarding the development of nonstereotypical immunostimulating agents that are structurally distinct from the typical phytosphingosine-containing galactosylceramide.

Supporting Information Available

Figures showing the docking scores of compounds 1, 2, and 4 (Figure S1), the hydrogen-bonding modes of compounds 2 and 4 (Figure S2), and the docking conformation of compound 2 superimposed with α-GalCer (1) (Figure S3), detailed information on the experimental procedures, analytical data for all new compounds, and copies of ¹H and ¹³C NMR spectra for selected compounds. This material is available free of charge via the Internet at http://pubs.acs.org.

This work was supported by a National Research Foundation of Korea grant funded by the Korea government (MEST) (Grant 20100027763).