Molecular architecture and conservation of an immature human endogenous retrovirus

Abstract

A significant part of the human genome consists of endogenous retroviruses sequences. Human endogenous retrovirus K (HERV-K) is the most recently acquired endogenous retrovirus, is activated and expressed in many cancers and amyotrophic lateral sclerosis and possibly contributes to the aging process. To understand the molecular architecture of endogenous retroviruses, we determined the structure of immature HERV-K from native virus-like particles (VLPs) using cryo-electron tomography and subtomogram averaging (cryoET STA). The HERV-K VLPs show a greater distance between the viral membrane and immature capsid lattice, correlating with the presence of additional peptides, SP1 and p15, between the capsid (CA) and matrix (MA) proteins compared to the other retroviruses. The resulting cryoET STA map of the immature HERV-K capsid at 3.2 Å resolution shows a hexamer unit oligomerized through a 6-helix bundle which is further stabilized by a small molecule in the same way as the IP6 in immature HIV-1 capsid. The HERV-K immature CA hexamer assembles into the immature lattice via highly conserved dimmer and trimer interfaces, whose interactions were further detailed through all-atom molecular dynamics simulations and supported by mutational studies. A large conformational change mediated by the flexible linker between the N-terminal and the C-terminal domains of CA occurs between the immature and the mature HERV-K capsid protein, analogous to HIV-1. Comparison between HERV-K and other retroviral immature capsid structures reveals a highly conserved mechanism for the assembly and maturation of retroviruses across genera and evolutionary time.

Figure 1 |

CryoEM analysis of immature HERV-K virus-like particles (VLPs). (a) Schematic representation of Gag domains. (b-d) Representative cryoEM images of purified immature HERV-K VLPs at low (b), medium (c) and high (d) magnifications. (e) The mean (red line) and distribution (gray bars) of the diameter (red arrow in d) of HERV-K VLPs, compared with the mean diameters of other retroviruses (blue lines). (f) The mean (purple line) and distribution (gray bars) of the distance between lipid bilayer and capsid lattice (purple in d) of HERV-KVLPs, compared with the mean distance of other retroviruses (blue lines).

Figure 2 |. CryoET STA of immature HERV-K VLPs.

(a) CryoET STA map of immature HERV-K CA hexamer colored by local resolution (red to blue). (b) A slab of HERV-K CA hexamer map overlapped with the atomic model (rainbow colors blue to red, N-terminus to C-terminus. (c) The atomic model of the immature HERV-K CA monomer (NTD in blue, CTD in orange, h12 in red) with α-helices labeled. (d) The atomic model of the immature HERV-K hexamer viewed from the side and top with one monomer colored in blue and orange. (e) A central slice of the HERV-K hexamer density map, showing IP6 density with an IP6 fitted (red arrow) and two Lys rings (K166 and K240), in addition to the hydrophobic core of the 6HB (M244 and I248). (f-g) Density slabs at the top (f, top dashed box in e) and bottom (g, bottom dashed box in e) of the 6HB, with the protruding side chains shown as sticks. (h) The dimer (blue circle) and trimer (red circle) interfaces between adjacent immature HERV-K hexamers. (i) A detailed view of the trimer interface involving helix 4 with the contributing side chains shown as sticks. (j) A detailed view of the dimer interface involving helix 9 with the involved side chains shown as sticks.

Figure 3 |. MD simulations and mutational validation of immature HERV-K CA intermolecular interactions.

(a-c) Snapshots of the inter-hexamer trimer interface involving helix 4 with the contributing side chains shown as sticks. Residues in (b) and (c) show stable interactions. (d) A snapshot of the inter-hexamer dimer interface involving helix 9 with the contributing side chains. (e-f) Views of the 6HB with the contributing side chains of the hydrophobic core (M244 and I248) (e) and the K166 and K240 rings with the IP6 molecule (f). In each snapshot, protein is represented as cartoon (NTD in blue, CTD in orange). The amino acid side chains are colored based on the heavy-atom RMSF computed throughout the MD simulation. (g) Western-blot of VLP assembly carrying trimer interface mutants of HERV-K Gag.

Figure 4 |. Comparison between immature and mature HERV-K CA.

(a) Overlay of the immature HERV-K structure (NTD blue, CTD orange, h12 red) with mature HERV-K (PDB: 6SSM), aligned to the NTD (left), CTD (middle), and NTD/CTD separately (right). (b-c) Arrangement of HERV-K immature (b) and mature (c) lattices. NTD is colored in blue/cyan, CTD in orange/red. (d-e) Detailed views of the dimer (top) and trimer (bottom) interface in the immature (d) and mature (e) lattices.

Figure 5 |. Conservation among the immature retroviruses.

(a) Overlay of the immature CA monomer structures of HERV-K (NTD blue, CTD orange, h12 red) with those of other retroviruses (grey) aligned to the NTD, shown are M-PMV (PDB: 6HWI, beta-retrovirus), RSV (PDB: 5A9E, alpha-retrovirus), HIV-1 (PDB: 7ASL, lentivirus), MLV (PDB: 6HWW, gamma-retrovirus). (b) Hexamer structures of M-PMV, RSV, HIV-1 and MLV. (c) Structure based phylogenetic tree. (d-f) Conservation of immature CA intermolecular interfaces, compared with HIV-1 (hexamer interface, d), RSV (trimer interface, e) and MLV (dimer interface, f). HERV-K structure is colored in blue and orange, other retroviruses in gray.