Structural and functional characterization of human SLFN14

Abstract

The Schlafen (SLFN) family of proteins are a group of DNA/RNA processing enzymes with emerging importance in human health and disease, where their functions are implicated in a variety of immunological and anti-tumor processes. Here, we present the cryo-electron microscopy structure of full-length human SLFN14, a member with antiviral activity and linked to an inherited bleeding disorder. SLFN14 is composed of an RNase domain, a SWADL domain, and a two-lobe helicase domain. SLFN14 exhibited strong RNase activity over different substrates, and the positively charged patches at the valley of the RNase domain, which contains the thrombocytopenia-related missense mutation sites, are crucial for binding oligonucleotides. SLFN14 lacks helicase activity, which can be attributed to the inability to bind ATP and the absence of positive charges at the canonical DNA-binding site of its RecA-like folds. SLFN14 is structurally similar to SLFN11, but differs from SLFN5 in the orientation of the helicase domain. Live-cell fluorescence resonance energy transfer (FRET) assays and AlphaFold2 analysis hinted that SLFN14 may adopt multiple conformations in cells. These results provide detailed structural and biochemical features of SLFN14, and greatly expand our knowledge of the functional diversity of the SLFN family.

Graphical Abstract

Figure 1.

Overall structure of full-length SLFN14. (A) Schematic representation showing the domain organization of full-length SLFN14. Elements are assigned divergent colors and the borders are indicated by residue numbers. (B) Cryo-EM reconstruction of the full-length SLFN14 dimer. The domains, motifs, and disease-related sites are specified with the indicated colors. (C) Disease-related mutations in the RNase domain of SLFN14. Involved residues are shown in ball-and-stick models. Secondary structural elements are indicated. (D) Disease-related mutation Thr853 in the helicase domain of SLFN14. The deleted portion of SLFN14 caused by the premature termination is colored gray.

Figure 2.

The dimeric state and interface of SLFN14. (A) The dimerization property of human subgroup III SLFNs (103−105 kDa as a monomer) was assayed by SEC on a Superose 6 column. The absorption peaks at 280 nm and corresponding molecular mass values for BSA monomer and BSA dimer are indicated. mAU, milli-absorption units. (B) Oligomeric state of SLFN14 in different NaCl concentrations analyzed by mass photometry. (C) Residues involved in the dimerization interface of the RNase domain shown as ball-and-stick models. (D) Intermolecular salt bridges between the RNase domain and SWADL domain. (E) Residues involved in the dimerization interface of the helicase domain.

Figure 3.

Structural and biochemical analysis of the SLFN14 RNase domain. (A) Structural overlay of the RNase domains of SLFN14, SLFN11 (Protein Data Bank accession code 7ZEL), SLFN13 (5YD0), SLFN5 (7PPJ), and SLFN12 (7EG1). The ranges of the central valley widths are indicated. (B) Surface electrostatic potential of the SLFN14 RNase domain (SLFN14-R). The four positively charged patches and involved residues are indicated. (C) Endoribonuclease activity of SLFN14-R on different nucleic acid substrates. (D, E) Dose- and time-dependent cleavage of SLFN14-R on tRNA (D) and total RNA (E). The indicated concentrations of SLFN14-R were incubated with tRNA^Ser, tRNA^Lys, or tRNA^Leu for 10/20/30 min. Total RNA was extracted from 293T cells, which mainly contains rRNAs. (F) Analysis of divalent cations as a cofactor of SLFN14-R for tRNA digestion. SLFN14 was incubated with tRNA^Ser for 30 min. Four different divalent cations were individually supplied at the indicated concentrations. (G−I) Cleavage assay of SLFN14-R and mutations on tRNA^Ser (G), total RNA (H), and small RNA (I). Total RNA and small RNA were extracted from 293T cells. Pa1−4_muts refer to composite alanine mutations of the positively charged residues from the four patches shown in (B), namely K38A/R39A (Pa1-mut), K213A/R214A (Pa2-mut), K218A/K219A/R223A/K261A (Pa3-mut), and R127A/R338A (Pa4-mut).

Figure 4.

Antiviral function of SLFN14. (A) Cleavage activity of SLFN14-R and the E206A mutant on synthetic RNA substrates derived from HIV and IAV genomes. SL denotes stem–loop. (B) Schematic of infection rate assay using VSV-G pseudotyped HIV (HIV^VSV-G). (C) HIV infection rates quantified by flow cytometry measuring GFP-positive cells (n = 3). (D  −  F) Extracellular viral RNA levels measured by qPCR: total vRNA (D), partially spliced vRNA (E), and unspliced vRNA (F) (n = 3). (G) Western blot analysis of SLFN14 variants and HIV p24 protein expression. Data in graphs are presented as means ± SD (n = 3). Statistical significance between SLFN14 WT and mutants was determined by Student's t-test (*P< 0.05; **P< 0.01; ns, not significant).

Figure 5.

Structural diversity of the SWADL domain. (A) Structural overlay for the SWADL domains of SLFN14 (cream), SLFN5 (PDB code 7PPJ, blue), and SLFN12 (7EG1, green). α_H1 of SLFN14 is included and colored gray. (B) Sequence alignment of the structurally variable N- and C-terminal regions of the SWADL domains. The missing connector loop with the RNase domain of SLFN14 is indicated by a dashed line. Numbers refer to the residues of SLFN14. (C, D) Structural differences in the SWADL domain between SLFN14 and SLFN5 (C), and between SLFN14 and SLFN12 (D). Structurally consistent regions are shown in transparent cartoons to highlight the difference. The termini of the SWADL domains and residues marking the borders of the variable regions are specified. PDE3A is shown in green transparent surface representation. The PDE3A-interacting region (PIR) of SLFN12 is indicated.

Figure 6.

Characterization of the SLFN14 helicase domain. (A) Structural comparison of SLFN14, UvrD, and Nsp13 for the helicase core. The helicase domain of SLFN is colored cream, and the helicase motifs are specified and differentially colored. The helicase cores of UvrD (PDB code 2IS4) and Nsp13 (6XEZ) are colored salmon and teal, respectively. The remaining parts of the UvrD and Nsp13 proteins are colored in gray. (B) Sequence alignment of the helicase motifs between subgroup III SLFNs and UvrD. The sequences of the motifs are colored as in (A). Numbers refer to the residues of SLFN14 at the start and end of each motif. Residues involved in nucleotide coordination and single-stranded substrate DNA binding for UvrD are indicated by rectangles and squares, respectively. (C, D) The two protruding loops in lobe-1 (C) and lobe-2 (D) of SLFN14 compared with UvrD and Nsp13. The unique region of SLFN14 on the protruding loop in lobe-1 is specified by ball-and-stick representation. (E) Comparison of SLFN14 and UvrD on the Walker A motif. The AMPPNP molecule bound to UvrD is shown in ball-and-stick models. The Walker A motif-equivalent region of SLFN14 is specified in purple. Note the clash between this region and the nucleotide. Parts of lobe-2 and the 2A domain of SLFN14 and UvrD were removed for clarity. (F) Surface electrostatic potential of SLFN14 and the SLFN11/ssDNA complex (7ZES). The canonical DNA/RNA binding areas are outlined by yellow dashed lines. (G) EMSA checking SLFN14H–ssDNA interaction. Migration of ssDNA is not retarded by increasing concentrations of SLFN14H. (H) Protein thermoshift (PTS) assay checking SLFN14H–ssDNA interaction. The T_m of SLFN14H remains unchanged in the presence of ssDNA.

Figure 7.

SLFN14 lacks helicase activity. (A) ATP hydrolysis assay showing that all subgroup III SLFNs lack the ATPase activity. ZIKV helicase and SARS-CoV-2 Nsp13 were used as positive controls. Error bars indicate the SD (n = 3). For each group, the average values of 620 nm absorbance indicating the amount of free phosphate at each time point are traced by line charts. (B) SLFN14 shows no ATPase activity in the presence of various nucleic acids. Nsp13 without nucleic acid was used as a control. Error bars indicate the SD (n = 3). Line charts are used to trace the data points for each group. (C) SLFNs have no unwinding activity on dsDNA substrates with 5′ overhangs. Error bars indicate the SD (n = 3). (D) Structural comparison between SLFN14 (cream) and AMPPNP-bound UvrD (PDB code 2IS4, salmon), showing that the long α_S7–α_H1 helix partly blocks the binding pocket of ATP. α_S7 of SLFN14 is colored orange. Part of the SWADL domain was removed for clarity. AMPPNP is shown in ball-and-stick models. (E) Structural comparison between SLFN5 (blue) and AMPPNP-bound UvrD (salmon). The SWADL domain part of SLFN5 is colored bright blue. (F) Structural comparison of SLFN14 and SLFN5 showing the opposite orientations of the helicase domain, as indicated by the arrows. (G) The different orientations of the helicase domains of SLFN14 and SLFN5. The AlphaFold model of SLFN5 was used because the helicase domain in the cryo-EM structure of SLFN5 is incomplete and the resolved region is in good accordance with the AlphaFold model. Note the positions of lobe-1 (pink) and lobe-2 (cream). The RNase and SWADL domains are colored dark and light gray, respectively.