Guanine-containing ssDNA and RNA induce dimeric and tetrameric structural forms of SAMHD1

Abstract

The dNTPase activity of tetrameric SAM and HD domain containing deoxynucleoside triphosphate triphosphohydrolase 1 (SAMHD1) plays a critical role in cellular dNTP regulation. SAMHD1 also associates with stalled DNA replication forks, DNA repair foci, ssRNA and telomeres. The above functions require nucleic acid binding by SAMHD1, which may be modulated by its oligomeric state. Here we establish in cryo-EM and biochemical studies that the guanine-specific A1 activator site of each SAMHD1 monomer is used to target the enzyme to guanine nucleotides within single-stranded (ss) DNA and RNA. Remarkably, nucleic acid strands containing a single guanine base induce dimeric SAMHD1, while two or more guanines with ∼20 nucleotide spacing induce a tetrameric form. A cryo-EM structure of ssRNA-bound tetrameric SAMHD1 shows how ssRNA strands bridge two SAMHD1 dimers and stabilize the structure. This ssRNA-bound tetramer is inactive with respect to dNTPase and RNase activity.

Graphical Abstract

Figure 1.

A1 site confers binding specificity for guanine and xanthine bases in ssDNA. (A) Guanine-specific hydrogen bonding network formed in the A1 site of wild-type SAMHD1 (PDB: 6TXC). (B) Xanthine-specific hydrogen bonding network formed in the A1 site of SAMHD1 D137N (PDB: 6TXA). (C) Binding of wild-type SAMHD1 to a series of 5′FAM-labeled 40mer ssDNA oligonucleotides (50 nM) consisting of a dT₃₈ homopolymer with nucleotides (N) dT, dG, dA or dX (deoxyxanthosine) in position two. Error bars indicate standard error of three independent replicate measurements at each SAMHD1 concentration. (D) Binding of SAMHD1 D137N to the same series of oligonucleotides as in panel C. Error bars indicate standard error of three independent replicate measurements at each SAMHD1 concentration. Error bars indicate standard error of three independent replicate measurements at each nucleotide concentration. (E) Displacement of the dG-containing 40mer (0.5 μM) from wild-type SAMHD1 (1 μM) with GTP (black) or XTP (xanthosine triphosphate, pink). (F) Displacement of the dX-containing 40mer (0.5 μM) from SAMHD1 D137N (1 μM) with GTP (black) or XTP (pink). Error bars indicate standard error of three independent replicate measurements at each nucleotide concentration.

Figure 2.

dG positional and oligonucleotide length effects on binding. (A) Binding isotherms of SAMHD1 to a series of 5′FAM-labeled ssDNA 40mers containing a single dG nucleotide in varying positions in a dT homopolymer backbone (50 nM each). (B) Plot of K_0.5 (left) and maximum anisotropy (A_max, right) for the ssDNA binding isotherms in (A) (black) and the ssRNA binding isotherms in Supplemental Figure S4B (pink). (C) Binding of SAMHD1 to a series of 5′FAM-labeled ssDNA oligonucleotides consisting of a single dG base at the 5′ end followed by dT homopolymer of varying lengths (n = 9 to 29). (D) K_0.5 (left) and maximum anisotropy (A_max, right) observed for the different homopolymer lengths used in panel (C). All Error bars for both K_0.5 and A_max indicate standard errors as determined by least-squares regression fit to the Hill equation (Eq. 2).

Figure 3.

Single G-bases near the 5′ end of ssDNA and ssRNA induce SAMHD1 dimerization. All crosslinking experiments were carried out as follows: SAMHD1 (1 μM) was incubated under each condition for 10 min, then crosslinked with 50 mM glutaraldehyde. All gels included dedicated marker lanes for SAMHD1 alone, SAMHD1 and 50 μM GTP, and SAMHD1 and 100 μM dGTPαS to confirm monomer (M), dimer (D) and tetramer (T) species. (A) Oligomeric states of SAMHD1 induced by 1 μM 5′FAM-dTdNdT₃₈ (where N = dT, dG or dA). (B) Binding of SAMHD1 to 5′FAM-dTdGdT₃₈ (1 μM) indicates a stoichiometry of two SAMHD1 monomers per DNA strand. (C) Oligomeric states induced by 1 μM 5′FAM-UNU₃₈ (where N = U, G or A). (D) Binding of SAMHD1 to 5′FAM-UGU₃₈ (1 μM) indicates a stoichiometry of one SAMHD1 monomer per RNA strand. (E) Oligomeric states induced by 5′FAM-labeled dT homopolymer ssDNA 40mers (1 μM) as a function of the position of the dG base within the sequence. (F) Oligomeric states induced by dT homopolymers of increasing lengths with a single dG base in position 1 ([DNA] = 1 μM).

Figure 4.

Two or more G bases in ssDNA and ssRNA with ∼20 nt spacing induce SAMHD1 tetramerization. GAXL crosslinking experiments were carried out as in Figure 3. (A) Oligomeric states of SAMHD1 induced by ssDNA₃₂. Lanes from left to right reflect serial two-fold dilutions in the range 10–0.3 μM. (B) Oligomeric states of SAMHD1 induced by ssRNA₃₂. Lanes from left to right reflect serial two-fold dilutions of each NA in the range 10 to 0.31 μM. (C) Stoichiometric binding of SAMHD1 to ssDNA₃₂ (1 μM). (D) Stoichiometric binding of SAMHD1 to ssRNA₃₂. (E) Oligomeric states induced by a 40mer ssDNA with a single 5′ G base fixed in position 1 and a second G base positioned at the indicated spacings (n) (5′FAM-dGdT_ndGdT_x). (F) Oligomeric states induced by a 40mer ssRNA with a single 5′ G base fixed in position 1 and a second G base positioned at the indicated spacings (n) (5′FAM-GU_nGU_x).

Figure 5.

CryoEM analysis of ssRNA₃₂ complexes with SAMHD1. (A) Overview of particles observed in cryo-EM study and proposed mechanism of T* formation. Abbreviations: D (dimer), T*_op, T*_cl (open and closed conformational states of the ssRNA bound tetramer). For low resolution species, (D and D·D) SAMHD1 dimers were rigidly docked into the density using the fitmap command in ChimeraX. For both conformations of T*, density corresponding to RNA is colored violet. From the cryo-EM densities, the dimensions of the tetramer and dimer forms are 80 Å × 110 Å and 50 Å × 80 Å, respectively. The Chimera map thresholds for the shown overall tetramer densities for T*_cl and T*_op are 0.9 and 0.3, respectively. The PDB and EMD accession numbers for the RNA complex T*_cl are PDB ID 8TDV and EMD-41174; the corresponding numbers for T*_op are PDB ID 8TDW and EMD-41175. (B) Comparison of structure of the canonical dNTP-saturated tetramer induced by dGTPαS (T, PDB: 7UJN) and the structure of the ssRNA₃₂-bound tetramer (T*_cl). The RNA was deleted from T*_cl to facilitate comparisons with T. (C) Dimer-dimer interface interactions in the T complex with dGTPαS. A charged hydrogen bonding network involving the interfacial residues D361, H364, and R372 is observed. (D) Dimer-dimer interface interactions in the T*_cl. complex with ssRNA₃₂. A significant displacement of the two interfacial helices disrupts the hydrogen bonding of residues D361, H364 and R372. In the T*_cl complex, the shown residues were built into the cryo-EM density map (EMD-41174).

Figure 6.

Molecular basis for ssRNA₃₂ binding (T*_cl). (A) Cryo-EM model for T*_cl complex. The segmented map density for the RNA (red) is shown at 5 σ using a threshold of 0.005 for the unsharpened map. The segmented map density for the protein (grey) is displayed using a threshold of 0.008 for the unsharpened map. (B) Cryo-EM model for the T*_cl complex with map density for the protein removed. The RNA (red) is displayed with the same parameters as in panel A and the perspective in the right panel is rotated by 180°. (C) Guanosine nucleotide modeled in the A1 site of chains A (purple) and B (pink), where it forms a hydrogen bond network with D137, Q142 and R145. The map density is contoured at 4 σ. (D) Guanosine nucleotide modeled in the A1′ site. The map density is contoured at 1.5 σ. (E) Cationic residues located near the bound RNA strands in T*_cl. Residues K116, K332, R333, K336, R352, K354, R371, R372, R451, K455, R559 are positioned to potentially interact with the flexible sugar-phosphate backbone of ssRNA₃₂. The dNTPase active sites require a 90° rotation for visualization and their approximate locations are indicated with asterisks. The cryo-EM density map for the RNA complex T*_cl is deposited in the Electron Microscopy Data Bank (EMDB) under accession code EMD-41174 (T*_cl); the cryo-EM derived structural model for T*_cl is deposited in the Protein Data Bank (PDB) under accession code PDB ID 8TDV.

Figure 7.

SAMHD1 interacts with long mRNA molecules as an ensemble of dimers and tetramers. Transmission electron micrographs (TEM) shown at 150,000X magnification using a 1% uranyl formate negative stain. The 50 nm scale bar applies to all images. (A) TEM image of in vitro transcribed 2 kb RNA (10 ng/μl) alone (top) and in the presence of 100 nM SAMHD1 (bottom). The mixed dimeric (D) and tetrameric (T*) states of SAMHD1 observed in the images were confirmed by the GAXL method (gel lane on the right). (B) SAMHD1 (100 nM) bound to ssRNA₃₂ (100 nM). The largely tetrameric (T*) state of SAMHD1 observed in the images was confirmed by the GAXL method (gel lane on the right). (C) Proposed model for SAMHD1 binding to long RNA involves G binding to A1 sites and the formation of enzyme dimers. Where the spacing of guanine residues is correct, dimeric units can coalesce to form tetramer.