Structure of the HCMV UL16-MICB complex elucidates select binding of a viral immunoevasin to diverse NKG2D ligands

Abstract

The activating immunoreceptor NKG2D promotes elimination of infected or malignant cells by cytotoxic lymphocytes through engagement of stress-induced MHC class I-related ligands. The human cytomegalovirus (HCMV)-encoded immunoevasin UL16 subverts NKG2D-mediated immune responses by retaining a select group of diverse NKG2D ligands inside the cell. We report here the crystal structure of UL16 in complex with the NKG2D ligand MICB at 1.8 A resolution, revealing the molecular basis for the promiscuous, but highly selective, binding of UL16 to unrelated NKG2D ligands. The immunoglobulin-like UL16 protein utilizes a three-stranded beta-sheet to engage the alpha-helical surface of the MHC class I-like MICB platform domain. Intriguingly, residues at the center of this beta-sheet mimic a central binding motif employed by the structurally unrelated C-type lectin-like NKG2D to facilitate engagement of diverse NKG2D ligands. Using surface plasmon resonance, we find that UL16 binds MICB, ULBP1, and ULBP2 with similar affinities that lie in the nanomolar range (12-66 nM). The ability of UL16 to bind its ligands depends critically on the presence of a glutamine (MICB) or closely related glutamate (ULBP1 and ULBP2) at position 169. An arginine residue at this position however, as found for example in MICA or ULBP3, would cause steric clashes with UL16 residues. The inability of UL16 to bind MICA and ULBP3 can therefore be attributed to single substitutions at key NKG2D ligand locations. This indicates that selective pressure exerted by viral immunoevasins such as UL16 contributed to the diversification of NKG2D ligands.

Figure 1. Structure of UL16 in complex with MICBpf.

(A), Ribbon drawing of the structure of UL16. The portion of UL16 belonging to the V-type immunoglobin superfamily fold is colored blue, and the N-terminal “plug” is colored red. Glycosylated asparagines (nitrogen atoms dark blue, oxygen atoms red) and attached N-acetylglucosamine residues (yellow) are shown as ball-and-stick models. The grey N-acetylglucosamine residue attached to Asn35 has high temperature factors and was therefore not included in the refinement. Disulfide bonds are shown in green. (B) Structure of the UL16-MICBpf complex. UL16 is colored as in (A), MICBpf is shown in orange. In order to visualize the native glycosylation of UL16, modeled glycans are shown in yellow as ball-and-stick models with a semitransparent surface. See Materials and Methods for details.

Figure 2. Kinetic and equilibrium SPR analyses of UL16 interactions with MICB.

(A,B) Kinetic analyses of UL16 binding to covalently immobilized MICB proteins comprising domains α1 and α2 only (MICBpf) (A), and domains α1, α2 and α3 (B). Each individual analyte concentration was injected twice and data are representative of at least two separate experiments with similar results. Double-referenced sensorgrams (shown in color) are overlaid with fits of a “1∶1 binding with mass transfer” model (black lines). Corresponding residual plots below the sensorgrams show the kinetic-fit range and absolute deviation (Δ) of data points from curve fit values. The red arrow and the red highlighted area of the sensorgram series indicate data used to determine averaged (AVG) equilibrium (Eq) response values (Eq-Response AVG) for equilibrium analysis. (C,D) Equilibrium analysis of UL16 binding to MICBpf (C) and MICB domains α1, α2 and α3 (D). Averaged equilibrium response values (red squares) are plotted against injected UL16 concentrations and fitted to a “1∶1 Langmuir isotherm” model (black line). The shaded boxes contain additional information about setup details (black font) and measured parameters from kinetic (blue font) and equilibrium analysis (red font).

Figure 3. Amino acid sequences of NKG2D ligands, NKG2D, and UL16.

(A) Sequence alignment of NKG2D ligands. Sequences of the α1α2-platform domains of NKG2D ligands MICA*01, MICB*02, ULBP1, ULBP2, ULBP3, ULBP4, ULBP5 and ULBP6 are included in the alignment. The alternative RAET nomenclature of ULBP proteins is indicated. Secondary structure elements as observed in the structure of MICBpf in complex with UL16 were assigned by DSSP and are represented with cylinders (helices) and arrows (β-strands) below the alignment. Helices are named as described in . Residues shaded in blue contact UL16 in the UL16-MICBpf complex. Residues shaded in salmon contact the salmon-colored NKG2D monomer (Figures 5A, B) in the MICA-NKG2D and ULBP3-NKG2D complex structures ,. Residues shaded in green contact the green NKG2D monomer (Figures 5A, B) in the MICA-NKG2D and ULBP3-NKG2D complex structures. Residues marked with a red triangle indicate substitutions between MICA and MICB in regions that contact UL16 in the MICBpf-UL16 complex. The ULBP5 residue boxed in cyan was recently shown to be the major determinant for diminished binding to NKG2D and UL16 . Disulfide bridges and corresponding cysteines are represented with magenta lines. Gaps are indicated by (−). (B) Structural mimicry of UL16. Shown are relevant portions of the sequences of the green human NKG2D monomer , (Figures 5A, B) and UL16. Secondary structure elements as observed in the structure of MICBpf in complex with UL16 and MICA in complex with NKG2D , respectively, were assigned and represented as described in panel A. The five residues marked with numbered black boxes below the sequence define the central binding motif that engages MICBpf or, in the case of NKG2D, MICA , in a similar manner (Figure 5C). Residues with the same number superimpose in space, although they are located in different regions in the protein sequences. Residues shaded in yellow and orange form contacts with MICA in the case of NKG2D and with MICBpf in the case of UL16, respectively. NKG2D residues in red contact ULBP3 in the ULBP3-NKG2D complex . Residues that augment the central binding motif, performing similar functions in the UL16-MICBpf and NKG2D-MICA complexes are marked with filled light red boxes below the sequence. An example is shown in Figure 5C. Disulfide bridges are represented with magenta lines. Hexagons mark the seven UL16 asparagine residues linked with glycans as observed in the UL16-MICBpf complex.

Figure 4. Interaction between UL16 and MICBpf.

(A), Ribbon tracing of the complex using the color code from Figure 1. Also shown in the lower right corner of panel A is a schematic representation of the “saddle on horseback” arrangement between UL16 and MICB. (B–D), The three major contact regions A, B and C of the complex. Nitrogen, oxygen and sulfur atoms are colored blue, red, and yellow, respectively. Hydrogen bonds and salt bridges are represented with dashed green and red lines, respectively, and hydrophobic contacts (distance<4.0 Å) are shown as dashed magenta lines. The dashed blue line indicates π-π interactions of two arginine guanidinium groups. Water molecules are shown as red spheres.

Figure 5. Comparison of the UL16-MICBpf and NKG2D-MICA complex structures.

In all panels, the two NKG2D monomers are shown in salmon and green, whereas UL16 and MICBpf are colored blue and orange, respectively. (A), Superposition of the UL16-MICBpf complex onto the MICA-NKG2D complex . MICA, which is very similar to MICB, is not shown for clarity. (B), Ribbon drawings of the α1α2-platform domains of MICA (left side, yellow) and MICB (right side, orange), with their molecular surfaces outlined in grey. Surface-exposed areas of residues that are buried upon complex formation with NKG2D and UL16, respectively, are colored using the color scheme of panel (A). MICB/MICA residues 155, 158, 159, 162 and 163, which contact both UL16 and NKG2D in a similar manner are shown in darker green and blue shading, respectively. (C), Structural mimicry of UL16. Close-up view of the core region of the structures shown in panel (A) with UL16 residues Ile63, Lys123, Tyr125, Leu110, Leu114 that superimpose with chemically equivalent NKG2D residues Ile182, Lys197, Tyr199, Met184 and Tyr152. Side chain atoms, UL16-MICBpf contacts, and water molecules are colored as described in the legend to Figure 4.

Figure 6. Selectivity of NKG2D ligand binding by UL16.

(A), The α1α2-platform domain of NKG2D-bound MICA was superimposed onto MICBpf, but only the MICA side chains Arg170 (pink) and Arg169 (magenta) are shown. The α1α2-domain of NKG2D-bound ULBP3 was also superposed onto MICB, and only the ULBP3 side chains of Arg169 (green) and Asp170 (light green) are shown. Cages surrounding the two arginines of MICA and ULBP3 at position 169 depict the area that these side chains would require in a space-filling model. In both cases, the arginine side chains would clash with UL16 residues. (B), The α1α2-platform domain of NKG2D-bound MICA (yellow) and ULBP3 (red), respectively, was superimposed onto MICBpf (orange). The side chains of alanine (present in MICBpf and MICA) and arginine (present in ULBP3) at position 162 are shown. Also shown are the Met184 side chains of both the MICA-bound (white) and ULBP3-bound (green) NKG2D monomers, both of which correspond to the green NKG2D monomer in Figures 5A and 5B. Conformational changes of the L2-loop of MICA-bound NKG2D displaces Met184 and allows for the accommodation of Arg162 in ULBP3-bound NKG2D. In UL16, the rigid DEB sheet does not allow for a similar conformational adjustment, and ULBP3 residue Arg162 would therefore clash with UL16 residues.