Structure of Actin-related protein 8 and its contribution to nucleosome binding

Abstract

Nuclear actin-related proteins (Arps) are subunits of several chromatin remodelers, but their molecular functions within these complexes are unclear. We report the crystal structure of the INO80 complex subunit Arp8 in its ATP-bound form. Human Arp8 has several insertions in the conserved actin fold that explain its inability to polymerize. Most remarkably, one insertion wraps over the active site cleft and appears to rigidify the domain architecture, while active site features shared with actin suggest an allosterically controlled ATPase activity. Quantitative binding studies with nucleosomes and histone complexes reveal that Arp8 and the Arp8-Arp4-actin-HSA sub-complex of INO80 strongly prefer nucleosomes and H3-H4 tetramers over H2A-H2B dimers, suggesting that Arp8 functions as a nucleosome recognition module. In contrast, Arp4 prefers free (H3-H4)(2) over nucleosomes and may serve remodelers through binding to (dis)assembly intermediates in the remodeling reaction.

Figure 1.

Alignment and structure comparison of H. sapiens Arp8 with actin. (A) Structure-based amino acid sequence alignment of human Arp8 with actin reveals five sequence insertions and one deletion. (B) Classical view of ATP-actin (blue) with the nucleotide-binding cleft at the pointed and the target-binding cleft at the barbed end of the molecule. Sub-domains 1 and 2 comprise the outer or small domain while sub-domains 3 and 4 are annotated inner or large domain. (C) Structure of ATP-bound H. sapiens Arp8 (Δ1–33) (green) in the classical actin view with insertions depicted in brown. The basal actin fold of Arp8 can be recognized and is complemented by five insertions. Insertions I emanates from the DNase I-binding loop and covers actin’s pointed end like a lid. Insertion III aids in closing the nucleotide-binding cleft and insertion V adds a small α-helix to sub-domain 3. (D and E) The backside of Arp8 shows that insertion II adds a loop and one turn to the α-helix that separates sub-domains 1 and 2. Insertion IV emanates from the region of actin’s hydrophobic plug but could not be allocated to electron density.

Figure 2.

Close up of human Arp8’s active site. (A and D) Human Arp8 coordinates ATP similar to actin (see C) and E266^Arp8 could activate the ATPase akin to Q137^act. This would require an additional factor that binds to Arp8’s target-binding cleft. The H73^act sensor is replaced by R187^Arp8, which could in principle trigger a similar conformational change. (B and E) ATP in yeast Arp4 is more shielded compared to actin or Arp8. Also, T142^Arp4 is probably too far apart from the γ-phosphate for a putative ATPase activation. Hence, ATP seems to play an exclusively structural role in Arp4. (C and F) ATP in actin (pdb: 1YAG) is bound by residues in the P1 and P2 loops and activation of the ATPase is triggered by Q137^act, which comes into closer proximity to the γ-phosphate upon filament formation. After hydrolysis and release of the γ-phosphate S14 flips over to coordinate the β-phosphate making room for H73 of the sensor loop to occupy the now available space. This conformational change propagates via sub-domain 2 to the D-loop.

Figure 3.

ATPase activity of actin-related proteins 4 and 8. Low basal ATPase activity was found for yeast and human Arp8 but not for yeast Arp4. The Arp8–Arp4–actin-HSA sub-complex I of INO80 has a slightly higher activity than Arp8 but does not show any sign of efficient stimulation of the ATPase of either Arp8 or actin within this sub-complex. Arp4, H2A–H2B dimers, (H3–H4)₂ tetramers, DNA and nucleosomes have no measurable ATPase activity in comparison with the control reaction without proteins. No significant stimulation of the ATPase activity of Arp8 or sub-complex I was triggered by canonical nucleosomes and its constituents (Supplementary Figure S5).

Figure 4.

Solution structure of human Arp8. (A) Theoretical SAXS scattering curves calculated with CRYSOL for monomeric and dimeric human Arp8 show that Arp8 is a monomer in solution. (B) X-ray structure of human Arp8 (Δ1–33) docked into the ab initio SAXS model of full-length hArp8. The solution structure of human Arp8 provides extra density for insertion IV.

Figure 5.

Contribution of Arps to histone and nucleosome binding. Fluorescence based affinity measurements of H. sapiens Arp8 full-length (A), human Arp8 (Δ1–33) (B), S. cerevisiae Arp8 (C), yeast Arp4 (D), S. cerevisiae INO80 sub-complex I (Arp8–Arp4–actin-HSA) (E) to H2A–H2B dimers (blue data points), (H3–H4)₂ (green) and nucleosomes (red). Binding affinities and Hill coefficients are determined by titration of Arps or complex into fluorescently labelled histones or nucleosomes, and monitoring of fluorescence change over the titration series. Change of fluorescence (increase or decrease) upon substrate binding depends on alterations to the microenvironment of the attached fluorophore upon specific binding events.