Structural basis of cowpox evasion of NKG2D immunosurveillance

Abstract

AbstractNKG2D is a key component of cytotoxic antitumor and antiviral responses. Multiple viruses evade NKG2D recognition by blocking NKG2D ligand expression on infected cells. In contrast, cowpox virus targets NKG2D directly by encoding a secreted antagonist, Orthopoxvirus MHC Class I-like Protein (OMCP). We have previously reported that OMCP also binds to the orphan receptor FcRL5 on innate B cells. Here, we demonstrate that mammalian-derived, glycosylated OMCP binds NKG2D but not FcRL5. Cowpox viruses either lacking OMCP, or expressing an NKG2D-binding deficient mutant, are significantly attenuated in wild type and FcRL5-deficient mice but not NKG2D-deficient mice, demonstrating that OMCP is critical in subverting NKG2D-mediated immunityin vivo. Next we determined the structure of OMCP bound to human NKG2D. Despite a structure similar to that of host NKG2D ligands, OMCP uses a drastically different orientation for NKG2D binding. The re-orientation of OMCP is associated with dramatically higher affinity for human NKG2D and the targeted interface is highly conserved in mammalian NKG2Ds, increasing the zoonotic potential of cowpox virus. We also show that cell surface presented OMCP can trigger NKG2D effector functions equivalently to host NKG2D ligands, demonstrating that NKG2D-mediated signaling requires clustering but is insensitive to binding orientation. Thus, in contrast to TCR/MHC interactions, the docking topology of NKG2D with its ligands does not appear to regulate its activation.Author SummaryVirally infected or tumor-transformed cells display NKG2D ligands (NKG2DLs) on their cell surface, which activates NKG2D-bearing lymphocytes to kill the transformed cell. Pathogens are known to counter this by blocking NKG2DL expression and/or surface display. In contrast, some tumor cells cleave endogenous NKG2DLs creating soluble NKG2D antagonists. Unlike other viral pathogens, cowpox virus uses a strategy analogous to cancer cells by targeting NKG2D directly with a soluble, high affinity NKG2D-antagonist named OMCP. We determined that OMCP’s virulencein vivois attributed to blocking NKG2D-mediated NK cell responses with no apparent effect due to binding to other receptors or cell types. We have also determined the crystal structure of cowpox OMCP bound to human NKG2D, revealing that despite conservation of the ligand scaffolding with host NKG2DLs, the viral protein is engaged with a radically altered orientation compared to all host NKG2DLs. Our structure provides key insight into how OMCP binds with an ∼5,000-fold increased affinity compared to human NKG2DLs and show that the OMCP binding site is exceptionally conserved among primates and rodents, suggesting that the ability of OMCP to recognize this conserved interface contributes to the broad zoonotic potential of cowpox virus. Finally, we show that cell membrane-anchored OMCP can trigger equivalent NKG2D-mediated killing as host NKG2DLs, demonstrating that NKG2D signaling is insensitive to ligand binding orientation.