Structural and functional insights into human Tudor-SN, a key component linking RNA interference and editing

Abstract

Human Tudor-SN is involved in the degradation of hyper-edited inosine-containing microRNA precursors, thus linking the pathways of RNA interference and editing. Tudor-SN contains four tandem repeats of staphylococcal nuclease-like domains (SN1-SN4) followed by a tudor and C-terminal SN domain (SN5). Here, we showed that Tudor-SN requires tandem repeats of SN domains for its RNA binding and cleavage activity. The crystal structure of a 64-kD truncated form of human Tudor-SN further shows that the four domains, SN3, SN4, tudor and SN5, assemble into a crescent-shaped structure. A concave basic surface formed jointly by SN3 and SN4 domains is likely involved in RNA binding, where citrate ions are bound at the putative RNase active sites. Additional modeling studies provide a structural basis for Tudor-SN's preference in cleaving RNA containing multiple I.U wobble-paired sequences. Collectively, these results suggest that tandem repeats of SN domains in Tudor-SN function as a clamp to capture RNA substrates.

Figure 1.

Different constructs of Tudor-SN, and their DNase and RNase activity assays. (A) Six constructs of Tudor-SN were prepared. (B) The purity of Tudor-SN proteins was assayed by 10% SDS–PAGE. (C) Plasmid digestion assays show that TSN, TSN-90 and TSN-70 had detectable DNase activity (14% to 3%), whereas TSN-64, TSN-50 and TSN-25 had residual activity (1%). (D) Tudor-SN truncated mutants were incubated with 20-bp RNAs (5′-end labeled on top strand) containing the wobble base-paired IIUI/UIUU sequence or Watson–Crick base-paired AAUA/UAUU sequence, in a reaction buffer for 8 h at 20°C. TSN and TSN-90 cleaved IIUI-dsRNA with the highest activities (5 and 15%), whereas TSN-70, TSN-64, TSN-50 and TSN-25 had no detectable activities (marked by -). The intensity of the cleavage product gel band of the 10-mer RNA (marked by *) was quantified and the substrate cleavage percentage was estimated, listed at the bottom of the gel. Undetectable cleavage was marked by -.

Figure 2.

Nitrocellulose filter-binding assays between truncated Tudor-SN mutants and RNA. (A) The binding assays show that TSN, TSN-90, TSN-70 and TSN64 all bind the 20-bp IIUI-dsRNA with comparable affinity, whereas TSN-50 and TSN-25 cannot bind RNA. (B) The filter-binding assays between Tudor-SN proteins and the 20-bp AAUA-dsRNA show that the truncated proteins containing more tandem repeats of SN domains bind AAUA-dsRNA better. TSN-50 and TSN-25 did not bind AAUA-dsRNA. (C) SDS–PAGE analysis of purified TSN-SN34. (D) The filter binding assays between TSN-SN34 and RNA. (E) The summary of the apparent dissociation constants (Kd_app) between Tudor-SN proteins and 20-bp IIUI- and AAUA-dsRNAs.

Figure 3.

Crystal structure of TSN-64. (A) The ribbon model of TSN-64 bound with four citrate ions. The four domains, SN3, SN4, Tudor and SN5, are rainbow colored from blue (N-terminus) to red (C-terminus). A loop between SN4 and SN5 (residues 635–644) is disordered and is displayed as a dotted line. (B) A schematic diagram representing the domain arrangement in TSN-64. The tudor domain is inserted in SN5 and packed between SN4 and SN5. (C) The electrostatic potential, mapped onto the solvent-accessible surfaces of TSN-64, calculated by APBS (32). The color scale was set from −5 kT/e (red) to +5 kT/e (blue). The molecular surfaces of tudor and SN5 are more acidic, whereas those of SN3 and SN4 are more basic.

Figure 4.

Structural comparison between staphylococcal nuclease (SMN) and SN domains in TSN-64. (A) Superposition of staphylococcal nuclease (PDB entry: 1EY0 (34)) and SN3 domain. SN3 contains extra long loops, marked by red dashed circles. (B) Superposition of SMN and SN4. SN4 contains an extra pair of β-strands and an α-helix, delineated by the blue circles. (C) Superposition of SMN and SN5. (D) The stereo view of the putative active site of SN3. A citrate ion (labeled as CIT) bound at the active site, overlapped with the pdTp bound at the active site of SMN [PDB accession code: 2ENB (35)]. (E) The stereo view of the putative active site of SN4. A citrate ion is bound next to the pdTp bound in SMN. (F) The putative active site in SN5. Most residues are hydrophobic, not appropriate for metal-ion binding or RNA hydrolysis.

Figure 5.

Sequence comparison of SN domains and structural-based sequence alignment between SMN and SN domains in Tudor-SN. (A) The sequence identities between different SN domains in Tudor-SN show that SN1 shares the highest sequence identity with SN3 (28.1%), while SN2 shares the highest sequence identity with SN4 (26.9%). (B) Sequence alignment between SMN and Tudor-SN. The secondary structures of SMN and TSN-64 are displayed at the top and bottom of the sequences, respectively. As SN1 shares higher sequence identity with SN3, and SN2 shares higher sequence identity with SN4, SN1-SN2 domains are aligned with SN3–SN4.

Figure 6.

Structural model of TSN-64 bound with a double-stranded RNA. (A) Two different views of the model show that a dsRNA may bind snugly at the concave basic surface in TSN-64 to SN3 and SN4 domains. (B) A pseudo 2-fold symmetry axis is identified between SN3 and SN4 domains. The structural model was built by aligning the dyad axis between a 16-bp RNA [PDB accession code: 1DI2 (36)] to the dyad axis between SN3 and SN4. Two stars mark the likely location of putative active sites in SN3 and SN4, based on the pdTp-binding site in staphylococcal nuclease. (C) The part of double-stranded RNA containing mis-paired I·U or U·I sequences may have an open conformation. The phosphate backbones may displace from a canonical A-form conformation so that they can be bound near the active sites in the SN3 and SN4 domains.