Table 1.
Data processing and structure refinement.
Fig 1.
Comparison of mArc and dArc domain structure.
The constructs used for structural studies on the individual dArc lobe domains are indicated.
Fig 2.
The crystal structure of dArc2-NL.
(A) The domain-swapped dimer observed in the crystal, with the three α-helices labeled. (B) The folded dimer encapsulates an extensive hydrophobic core, with no polar interactions connecting the two monomers. Residues are only labelled in subunit A, but also seen in subunit B. (C) Topology diagram of domain swapping. The location of the conformational change is indicated by red shading. (D) The electrostatic surface of the dimer. The α2 and α2’ have a positive surface potential, in contrast to the surface formed by α1, α3, α1’, and α3’. The protein is in the same orientations as in (A).
Fig 3.
The N-terminal region preceding dArc-NL packs into the hydrophobic core of the domain and leads to the formation of the capsid hexamer.
(A) The hexameric form of dArc2-NL observed in the capsid (PDB: 6TAQ; [21]), showing the electrostatic surface potential for half of the monomers. The canonical fold enables contact formation between the oppositely charged surfaces of each monomer. The N-terminal tail is showed in orange. (B) Residues contributing to the packing of the N-terminal tail (orange) into the capsid hexamer. Phe32 and Phe39 pack into two exposed pockets in the hydrophobic core. Further interactions are observed for Ser40, which hydrogen bonds directly with Lys78 and Ser79 in the α2 kink (yellow dashed lines).
Fig 4.
The dArc2-NL domain-swapped dimer resembles flaviviral coat proteins and DNA-binding proteins.
(A) The tetrameric coat protein of the Kunjin subtype West-Nile virus (WNc), where the longest helix of each monomer (analogous to α2 of dArc2-NL) contributes to a four-helix bundle interface (PDB: 1SFK [59]) (left). Middle: a single dimer of the tetramer (yellow/orange) overlaid with subunit A from dArc2-NL (grey). Right: the electrostatic surface potential of a WNc dimer, which resembles that of dArc2-NL. (B) Structural comparison between the dArc2-NL and similar domain-swapping proteins. Shown are the retroviral Dengue virus CA (green; PDB: 1R6R [57]) and the DNA binding dimers of (HMfb)2 histone (red; PDB: 5T5K [61]), a dimer of histones H3 and H4 (cyan; PDB: 5C3I [62]), TAFII transcription factor (blue; PDB: 1TAF [63]) and the foxhead domain of the FoxP2 transcription factor (yellow; PDB: 2A07 [64]). Each chain in a dimer is coloured with a different shade, and a dArc2-NL monomer is superimposed and shown in grey. (C) Domain swapping and conformational selection in the apoptosis-induced BAK protein. Shown on the left is the inactive monomeric form of BAK (PDB: 2IMT [65]), which has an orthogonal bundle fold similar to Arc-NL. Binding of a BH3 domain causes partial unfolding and opening of the hinge region (middle, PDB: 4U2U [66]), which leads to the formation of a membrane-binding domain-swapped dimer (right, PDB: 4U2V [66]). Panel C is based on [67]. The two chains in the BAK dimer are coloured grey and orange.
Fig 5.
Sequence conservation analysis of the central lobe region.
(A) Sequence variability in Arc N- and C-lobes (left and right, respectively). S is sequence entropy / variability within the seach results at each position (see Methods for details). Numbering follows dArc2. Red dashed lines: a search with dArc2 gave 220 homologues. Blue: a search with dArc1 and dArc2 resulted in a combined group of 250 homologues. Black: 699 sequences resulting from a search with dArc1, dArc2, and rat Arc. (B) Sequence logo for the region centered at Ser79 of dArc2-NL compared to corresponding residues from dArc1 and Rattus norvegicus homologues. While Gly is the most conserved residue at this position, starting searches with dArc1 and dArc2 indicates variability also at this position, unlike a search with rat Arc. (C) Mapping of conservation onto the dArc2-NL dimer. Blue corresponds to conserved and red to non-conserved sites.
Fig 6.
Crystal structures of the dArc1 and dArc2 C-lobes.
(A) dArc1-CL. (B) dArc2-CL. (C) The two structures, which deviate with an all-atom RMSD of 0.48 Å, superimposed. (D) Residues contributing to the dimer interface in dArc-CL. dArc1-CL residues are marked in blue, and residues of dArc2-CL are marked in orange. Variable residues are indicated in italics. All residues contributing to the dimer interface are conserved, with the exception of A125 (dArc1) which corresponds to S112 (dArc2). Polar interactions are shown with red dashed lines. (D) A comparison of the dArc-CL crystal structures with the same domains in dArc capsids. Both the dArc1 and dArc2 C-lobes closely resemble their counterparts in the capsids, with an all-atom RMSD of 1.75 Å2 and 1.13 Å2, respectively.
Fig 7.
The structure of dArc-CL resembles that of mArc and retroviral capsid proteins.
(A) Structures similar to dArc1-CL. dArc1-CL (grey) is shown superimposed with crystal structures of the rat Arc C-lobe (yellow; PDB: 4X3X) [12], HIV CA-CTD (green; PDB:1A43) [80], bovine leukemia virus (BLV) C-terminal domain (black; PDB:4PH0) [81], the rous sarcoma virus (RSV) C-terminal domain crystallized at pH 4.6 (purple; PDB: 3G21) [82], and the C-terminal domain of the Ty3 retrotransposon capsid (cyan; PDB: 6R23) [83]. Also shown are the scoring criteria obtained from the Dali server. (B) Comparison of CT dimerization. Shown are homodimers of the structural homologues in (A) and dArc2-CL, as calculated by PISA [39] from the crystalline states, apart from the BSV-CTD, which was not dimeric. Buried surface area (BSA) of each interface is shown below each structure. Note that even though a homodimer is predicted for the rat Arc-CL, this domain is monomeric in solution, and the predicted dimer is arranged differently from dArc.
Fig 8.
Comparison to the dArc1 crystal structure.
Left: top view of the dArc1 dimer (gray); the two monomers are highlighted by ellipsoids in light gray. The superposed structures on the CL dimer and the two NL domains are indicated, and include dArc2-NL monomer (pink; this work), dArc1-CL dimer (blue; this work), dArc2-CL dimer (orange; this work), and hArc-NL (green) complexed with the Stg ligand peptide (red) [17]. Right, the same structures viewed from the side of the dArc1 dimer.
Fig 9.
Solution structures of dArc lobe domains.
(A) SAXS data for dArc lobes in solution (left) and distance distribution plots (right). (B) Ab initio models (grey spheres) of dArc lobes. The models are superimposed with the following structures (shown as cartoons): dArc2-NL dimer (overlaid on dArc1-NL), dArc1-CL dimer (on dArc1-CL), a dimer of dArc2-NL dimers (on dArc2-NL), and dArc2-CL dimer (on dArc2-CL). (C) Fit of the SAXS data for dArc2-NL (dots) with the possible tetramer of dArc2-NL seen in panel (D). (D) Tetrameric assembly of the West Nile virus protein C (yellow) and an aligned structure of two dArc2-NL dimers (red) showing a possible tetrameric structure. (E) SEC-MALS for human and Drosophila N-lobes. (F) CD data for dArc and hArc lobes. (G) SEC-MALS for human and Drosophila C-lobes.
Table 2.
Dimensions and oligomeric state for different dArc constructs.
Fig 10.
(A) Sequence alignment between dArc1-NL and dArc2-NL. (B) The homology model of dArc1-NL displays contrasting electrostatic surface potential, where the highly positive character of dArc2-NL along α2 and α2’ (Fig 2) is replaced with a more modest surface potential. (C) Additional monomer-monomer interactions observed in the dArc1-NL model, not observed in the dArc2-NL crystal structure. Polar interactions are shown with purple dashed lines.
Fig 11.
Binding of human and Drosophila Arc N-lobes to a Stargazin peptide.