STK19 is a DNA/RNA-binding protein critical for DNA damage repair and cell proliferation

Dating back to 1994, an MHC class III region gene (RP1 or G11) was reported to encode a 364-a.a. protein in several cell lines (the 41-kD isoform; Shen et al., 1994), but another group suggested that this gene encodes two proteins lacking the N-terminal 110 amino acids and with or without a 4-a.a. insertion (coding 254 or 258 a.a., named 29-kD STK19; Sargent et al., 1994). The encoded polypeptides showed limited sequence similarity to several proteins, one of which is the tyrosine kinase-transforming protein of Fujinami sarcoma virus (18.7% identity over 157 amino acids; Sargent et al., 1994). Both the 29- and 41-kD forms of the protein were reported to possess Mn²⁺-dependent protein kinase activity (Gomez-Escobar et al., 1998); hence, the protein was named STK19 (Yang et al., 2001).

However, the identity and functions of STK19 are highly debated. From 2012 to 2017, several groups reported a recurrent melanoma mutation in STK19, which was proposed to drive the progression of melanoma (Bonilla et al., 2016; Chang et al., 2016; Hodis et al., 2012). This mutation lies in the N-terminal region of 41-kD STK19 at D89 but outside the coding region of 29-kD STK19. In 2019, Yin et al. reported that the D89 mutation promoted NRAS phosphorylation and melanoma formation (Yin et al., 2019). Shortly after, Rodríguez-Martinez et al. claimed that STK19 is not an NRAS-dependent melanoma driver since 41-kD STK19 is not expressed, rendering the “D89 mutation” out of the coding region and functionally irrelevant in melanoma (Rodríguez-Martínez et al., 2020). Although no NRas-specific kinase activity of STK19 was detected, these data were unable to rule out STK19 as a kinase (Rodríguez-Martínez et al., 2020). In fact, several online databases (such as UniProt and The Human Protein Atlas) and papers published after the debate still refer to STK19 as a kinase protein (Asquith and Temme, 2020; Carrozza et al., 2021; Chelius et al., 2020; Chen et al., 2021; Duan et al., 2021; Garcia-Alvarez et al., 2021; Guo et al., 2021; Marino et al., 2022; Qiu et al., 2021).

To gain insights into the identity and molecular functions of STK19, we performed a combination of biochemical, structural, and cellular studies. Results showed that STK19 was highly unlikely to be a kinase but a DNA/RNA-binding protein. STK19 can form dimers and use the surface created by dimerization to engage double-stranded (ds) DNA/RNA. STK19 contributed to DNA damage repair (DDR) by recruiting many repair factors to damaged sites. Finally, STK19-downregulated cells displayed defects in mRNA expression, metabolism, and cell proliferation.

Compared with the 29-kD isoform, the 41-kD isoform contains an extra N-terminal 110 amino acids that are not required for kinase activity (Gomez-Escobar et al., 1998). In this work, we focused on the 29-kD STK19 isoform, which encodes a 254-a.a. protein. This protein contains two conserved domains, an N-terminal bipartite nuclear localization signal (NLS) and a C-terminal conserved putative kinase domain (Fig. S1 A). Pull-down showed that the NLS was necessary and sufficient for binding to Importin α1, an adaptor of the classical nuclear import (Fig. S1 B). Deletion of this NLS, however, only partially reduced the nuclear localization of STK19 in HeLa cells (Fig. S1 C), suggesting that additional factors are responsible for its nuclear localization.

Using an assay performed according to the reported conditions (Yin et al., 2019), the kinase activity of STK19 was not detected, either by western blot or by mass spectrometry (MS; Fig. S1 D and Data S1). In addition, no binding between ATP and STK19 was observed by isothermal titration analysis (Fig. S1 E). To understand the functions of STK19, we crystalized the putative kinase domain encompassing a.a. 25–254 (25-C) and determined its crystal structure at 1.65 Å resolution using selenomethionine derivative crystals and single wavelength anomalous diffraction (Table 1). Each asymmetric unit contains two copies of STK19 containing residues 32–254. The protein is folded with 10 α helices and nine small β strands that form three antiparallel β sheets (Fig. 1 A). At a glance, this is highly different from known catalytic domain structures of the kinase, which always contains two subdomains, an ATP-binding N-lobe and a substrate-binding C-lobe (Eswaran and Knapp, 2010; Huse and Kuriyan, 2002; Fig. 1 B). Three-dimensional structure blast using DALI revealed that the protein can be divided into three “winged” helix domains (WHDs; Fig. 1 C), but WHD has never been reported to possess kinase activity (Harami et al., 2013; Soler et al., 2007). Surface representation shows that STK19 does not contain a possible ATP-binding pocket (Fig. 1 D). The presumed “ATP-binding site” residue K203 (corresponding to K317 in 41-kD STK19) is located at a protruding loop but not in a concave pocket (Fig. 1 D). These biochemical and structural data suggest that STK19 is highly unlikely to be a kinase.

We named the three WHDs WH1 (32–104), WH2 (120–179), and WH3 (180–254; Fig. 1 E). Each WHD comprises three α helices and three β strands in the canonical order α1-β1-α2-α3-β2-β3 (Fig. 1 F). The three WHDs are intimately packed and approximately arranged in a line, with WH1 in the middle (Fig. 1 E). WH1 and WH2 are connected by a linker helix (105–119), which forms substantial hydrophobic contacts with both WH1 and WH2 (Fig. S1 F). WH2 is followed by a long WH3 helix (α1) that sequentially lines up WH2, WH1, and WH3 (Fig. S1 G). These three domains are considerably different from each other, especially the α1 helix and the W1 wing (a loop between β2 and β3; Fig. 1 G). WHDs often play a role in DNA binding (Harami et al., 2013). Electrophoretic mobility shift assays (EMSA) were used to test whether STK19 is a DNA binder. The results showed that both full-length (FL) STK19 and its C-terminal domain (25-C) displayed substantial DNA-binding ability (Fig. 2 A). Using EMSA, the apparent binding affinity (Kd) between 25-C and DNA1 was determined to be 3.8 μM (Fig. 2 B). DNA longer than 25 bp bound to a similar extent, but binding of 10-bp dsDNA was hardly detectable (Fig. 2 C). STK19 barely bound to single-stranded DNA (Fig. 2 D) and was not dependent on a specific sequence (Fig. 2 E). Overall, these results showed that STK19 bound to dsDNA in a sequence-nonspecific manner.

The two chains in the asymmetric unit interact through WH3 and form an overall “V”-shaped architecture (Fig. 3 A). The dimer interface residues have low B factors as the inner core residues (Fig. 3 B) and have well-defined electron densities (Fig. 3 C), suggesting that the dimer may exist in solution. Dimerization is mediated by the third WHD (Fig. S2 A), and the dimer interface buries an area of ∼550 Ų in each monomer. At the heart of the interface, E206 and R250 formed two salt bridges with R250 and E206 in the other monomer, respectively (Fig. 3 D and Fig. S2 B). Additionally, L248 formed hydrophobic contact with Y204 in the other monomer (Fig. 3 D). The two monomers in the dimer are similar to each other and to the AlphaFold-predicted structure, except for two loop regions (Fig. S2, C and D). The purification yields of dimer interface mutants R250A, E206K, L248A, and Y204E decreased to varying degrees, which is unlikely if these were protein surface mutations (Fig. S2 E). The more severe the mutation interfered with dimer formation (E206K and R250A), the less the purification yield. The melting temperatures (Tm) were similar between WT and the mutants (except for R250A, no yield) when measured at a low concentration (no dimerization) by differential scanning fluorimetry (DSF, Fig. S2 F). At a high concentration, E206K and L248A were less stable but Y204E was similarly stable (Fig. S2 F). These results indicated that dimerization may be important for protein stability at high concentration and the purification yield.

In crosslinking experiments, odd oligomers were not observed in crosslinked products, and Y204E was less crosslinked (Fig. 3 E). Binding between MBP-tagged STK19 and untagged STK19 could be detected in pull-down assays, and the binding was reduced by the Y204E mutation (Fig. S2 G). Furthermore, increasing the concentration of 15-bp dsDNA in pull-down increased the binding of untagged WT STK19 but not the Y204E mutant (Fig. 3 F), arguing that dimerization is likely involved in DNA binding. Taken together, STK19 can form dimers in solution.

The observation that STK19 bound to dsDNA irrespective of its sequence (Fig. 2 E) implied that it may interact with the acidic phosphate backbone of DNA. The electrostatic surface potential map shows a basic patch in STK19, mostly centered on the α1 helix of WH3 (α1^WH3; Fig. 4 A). However, this region only spans ∼25 Å, which is quite small compared with the ∼67 Å span of a 20-bp dsDNA (Mandelkern et al., 1981). In the protein homodimer, two basic patches are adjacent in space and the distance between Cα atoms of K186 is 58.9 Å (Fig. 4 B). In view that the dimerization interface mutant Y204E reduced DNA binding (Fig. S2 H), it was speculated that the dsDNA bound to the STK19 dimer rather than a monomer. Computer modeling showed that bending of dsDNA is likely required to bind STK19 (Fig. S2 I). Sequence alignment of STK19 from different species showed that K186, R192, and K201 in α1^WH3 are well conserved (Fig. 4 C). R192 is mostly buried and thus unlikely to be involved in DNA binding (Fig. 4 A). To prevent possible mutational compensation, the adjacent K190 was simultaneously mutated when performing the K186 mutation. Either the K201A or K186A/K190A mutation dramatically reduced STK19’s DNA-binding ability, and simultaneously mutating three lysine residues (3KA, K186A/K190A/K201A) totally abolished the DNA-binding activity (Fig. 4 D). Chromatin precipitation results showed that WT STK19 bound to isolated chromatin, but 3KA did not (Fig. 4 E). The above results indicate that STK19 may use two basic patches flanking the homodimer to bind dsDNA.

Immunofluorescence results showed that the overexpressed STK19 was mainly localized in the nucleus of HeLa cells (Fig. 4 F). The 3KA mutant, albeit containing an intact NLS, dramatically relocalized from the nucleus to the cytoplasm (Fig. 4 F). In fact, nuclear localization of STK19 was more dependent on its DNA-binding residues than its NLS (compare Fig. 4 F and Fig. S1 C). DSF showed that the thermal stability of 3KA was similar to that of WT at a low concentration but slightly increased at a high concentration (Fig. S2 F). The cytoplasmic localization of dimerization mutants, including Y204E, E206K, R250A, and L248A, was also notably increased (Fig. 4 F), further supporting the notion that STK19 dimerizes to bind DNA. Overall, both dimerization and DNA-binding mutants can affect STK19 localization and possibly its function.

To investigate the molecular functions of STK19, we sought to identify STK19 WT-interacting proteins by GST pull-down and MS analysis (Data S2). Proteins involved in nucleotide excision repair (NER) and mismatch repair (MMR) were significantly enriched by STK19 WT (Fig. 5 A), suggesting that STK19 may participate in MMR and NER. Tandem mass tag (TMT)–based quantitative proteomic showed that STK19 knockdown (KD) cells downregulated several proteins involved in MMR and NER (Fig. S3, A and B). Upon UVB treatment, which induces NER, both the mRNA and protein expression levels of STK19 increased in B16-F10, HeLa, and 293T cells (Fig. 5 B and Fig. S3 C). Depletion of STK19 (Cas9-mediated polyclonal KD, which depletes both 29-kD and barely detectable 41-kD) led to more pronounced γ∙H2AX expression, a marker of DNA damage, either by immunofluorescence or western blotting (Fig. 5, C and D). Re-expression of STK19 WT in the STK19 KD cells reduced the level of γ∙H2AX to a greater extent than re-expression of 3KA (Fig. 5 E), suggesting that the DNA-binding activity of STK19 is required for its NER function. In the presence of methylnitronitrosoguanidine (MNNG), a compound that triggers MMR, more H2AX foci were also observed in STK19 KD cells (Fig. S3 D). In contrast, after cells were treated with double-strand-break-inducing x-ray, neutral comet assay revealed hardly any tail moment difference between control siRNA and siSTK19 (Fig. S3 E). These results indicated that STK19 was required for single-stranded repair including NER and MMR but not double-stranded repair.

Using EMSA, no binding difference in STK19 binding was observed when DNA was treated with different doses of UVB (Fig. S3 F), suggesting that STK19 is probably not directly involved in the recognition of damaged DNA in NER. In the presence of UVB treatment, STK19 and DNA repair factors such as CSB, MSH2, RPA2, and PCNA were notably enriched in the chromatin faction to mediate DDR (Fig. 6 A). STK19 KD inhibited the chromatin recruitment of all these repair factors, although the expression of these proteins was not affected by STK19 KD (Fig. 6 A and Fig. S3 G). Chromatin recruitment of DDB1, a global genome NER (GG-NER) factor that participates in DNA lesion recognition, was not affected by UVB or STK19 KD. In UVB-treated HeLa cells expressing endogenous STK19, overexpression of STK19 WT weakly promoted the chromatin recruitment of upstream repair factors DDB1, MSH2, and RPA2, and notably enhanced chromatin recruitment of the downstream repair factor PCNA (recruiting polymerase for DNA synthesis; Fig. 6 B). In contrast, overexpression of 3KA did not enhance chromatin recruitment of these repair factors. These results suggested that STK19 likely bound to the periphery of damaged DNA to stabilize upstream DDR factors and recruit downstream DDR factors to facilitate NER.

In agreement with the stabilizing/recruiting function, STK19 pull-down experiments showed that STK19 interacted with all these DDR factors with or without UVB treatment (Fig. 6 C). Even without UVB treatment to prevent possible UV-induced crosslinking, DNase did not abolish their interaction (Fig. S3 H), possibly due to DNA protection by the complexes formed by STK19 and other DDR factors, or that once formed, the STK19 complexes were stable even without DNA. The 3KA mutant bound to DDB1, MSH2, and PCNA, suggesting that STK19 can form DNA-independent complexes with these proteins. On the other hand, the interaction with CSB (lesion recognition) and RPA2 (binds the undamaged strand) was reduced by 3KA mutation (or DNase) since 3KA cannot be recruited (by dsDNA adjacent to the lesion) to form a complex with lesion-localized CSB and RPA2. The recruitment of WT STK19 may corecruit the bound DDR factors DDB1, MSH2, and PCNA to form a larger repair complex. STK19 further induced PCNA degradation even in the absence of UVB, but such degradation was not observed in chromatin precipitation experiments with shorter assay time. In addition, STK19 directly bound Escherichia coli–purified PCNA and RPA2 in pull-down experiments, and the binding was potentiated by dsDNA (Fig. 6 D). These results suggest that STK19 may interact with DNA and different DDR factors directly or indirectly, thereby stabilizing/recruiting these factors to damaged DNA.

Except for DDR proteins, STK19 interacted with many proteins involved in RNA transcription, splicing, translation, and degradation (Fig. S4 A), and interaction with these RNA metabolism proteins was reduced by the 3KA mutation (Fig. S4 B). EMSA results showed that STK19 bound similarly to dsDNA and dsRNA (Fig. 7 A). In cells, HeLa-expressed GST-STK19 bound much more RNA than DNA (Fig. 7 B). Furthermore, RNA binding also relied on the DNA-binding residues (Fig. 7 C), suggesting a shared binding interface. TMT proteomics by Gene Set Enrichment Analysis (GSEA) revealed that many proteins, especially those involved in metabolic pathways, were downregulated after STK19 siRNA treatment (Fig. 7, D and E). Six downregulated proteins were tested by quantitative PCR (qPCR) analysis, and the results showed that their mRNA levels were all downregulated to varying degrees (Fig. 7 F). Among these proteins, PTBP2 and RBM47 regulate multiple aspects of RNA metabolism, such as splicing, stability, and translation (Shivalingappa et al., 2021; Stern et al., 2009). These results suggest that STK19 binds dsRNA and regulates the expression levels of some mRNAs, which may (partially) explain the observed proteomic changes in GSEA analysis.

GSEA analysis by Cas9-gRNA-mediated STK19 KD also revealed the downregulation of many metabolic pathways (Fig. S4 C). Succinate dehydrogenase complex subunit A (SDHA), a component in the TCA cycle and oxidative phosphorylation, was significantly downregulated in STK19 KD cells (Fig. S4 D). Further transfection of STK19 WT rescued the expression level of SDHA. Consistent with the upregulation of apoptotic pathway in GSEA (Fig. S4 C), flow cytometry results showed an increase in apoptosis after STK19 KD (Fig. S4 E). Western blotting revealed obvious p-JNK activation but no changes in the levels of apoptotic mediators, including BCL-2; FL and cleaved caspase-3; and FL and cleaved caspase-8 (Fig. S4, F and G). Increased BAX levels were observed in STK19-KD HCT116 cells but not HeLa cells. Significant reduction of S phase was observed only in one of the gRNA-treated HeLa cells, but not in siRNA-treated HCT116 cells (Fig. S4, H and I). These results suggest that STK19 may regulate other cellular processes such as metabolism and apoptosis, possibly via a mRNA regulation mechanism.

We were unable to obtain Cas9-gRNA-mediated STK19 single-colony knockout cells (HeLa, 293T, and B16-F10) because these knockout cells barely proliferated. Polyclonal STK19 KD cells were viable and always re-expressed STK19, typically 3–4 days after KD (Fig. S5 A). The proliferation of polyclonal STK19 KD cells was indeed much lower than that of control KD cells, especially considering the re-expression of STK19 in this assay (Fig. 8 A). Transiently transfected WT notably enhanced the growth of STK19 KD cells (Fig. 8 B). 3KA and the dimerization mutants, which were completely and partially cytoplasm mislocalized, were ineffective and barely effective in rescuing cell proliferation, respectively (Fig. 8 B). Analysis of The Cancer Genome Atlas (TCGA) database showed that STK19 was overexpressed in colorectal cancer (CRC) compared with normal colon tissues (Fig. 8 C). Moreover, the Kaplan–Meier survival analysis indicated that STK19 expression was negatively correlated with overall survival and progression-free survival in CRC patients (Fig. 8, D and E). The colony formation and proliferation of the human CRC cell line HCT116 were dramatically inhibited by STK19 silencing mediated by two siRNAs (Fig. 8, F and G). Proliferation by STK19 siRNA KD cells can be enhanced by transfection of STK19 WT, but not 3KA (Fig. S5 B). Taken together, these results suggested that STK19 was critical for cell proliferation by binding DNA/RNA and may play a role in CRC.

In the first report that suggested STK19 is a kinase, the authors mutated K186, K201, and K203 of STK19 and found that the kinase activity was inhibited (Gomez-Escobar et al., 1998). Our results showed that these mutations are not active site mutations but protein surface mutations that should affect DNA binding and result in cytoplasmic mislocalization. Furthermore, HA-tag immunoprecipitated STK19 was used to demonstrate the kinase activity of STK19 (Gimple and Wang, 2019; Gomez-Escobar et al., 1998; Yin et al., 2020), but we found that many kinases bound to STK19 in cells (Data S2). The experiments of Rodríguez-Martinez et al. at best can rule out the possibility of STK19 being an NRas-specific kinase, but it remained possible that STK19 is a kinase. Our structural and biochemical analysis firmly established that STK19 is highly unlikely to be a kinase but a DNA/RNA-binding protein. To avoid further mistakes and to prevent the development of more kinase inhibitors for STK19 (Qian et al., 2020; Yin et al., 2019), we renamed this protein as TWH19 (Tandem Winged Helix protein previously known as STK19) hereafter and urge timely correction to its annotations.

WHD usually uses α3 and W1 to interact with DNA or protein (Harami et al., 2013). The α3 of WH3 is mostly buried by WH1, and the W1 of WH3 is involved in protein dimerization (Fig. S2 A). Instead, WH3 uses its α1 to interact with DNA. In WH1 and WH2, the α3 and W1 elements are generally exposed, rendering them possible for protein–protein interactions. Some of the TWH19-interacting proteins such as RPA2 and PCNA may bind to WH1 or WH2 or both domains. Therefore, the architecture of TWH19 may allow it to serve as a multivalent scaffold for DNA and DDR factors. Nonetheless, it is unlikely that TWH19 directly interacted with every DDR factor, but its interaction with some DDR factors may be mediated through DNA or direct-interacting DDR factors. The multivalence of TWH19 can be further increased by its dimerization, which is supported by multiple lines of evidence including crystal structure b-factors, purification yield, crosslinking, and pull-down. Dimer interface mutants partially mislocalized TWH19 and barely rescued the proliferation of TWH19 KD cells, suggesting that TWH19 functions as dimers in cells.

Two previous genomic screening studies indicated that TWH19 is involved in transcription-coupled NER (TCNER) (Boeing et al., 2016), a subtype of NER that is activated during transcription blockade, or NER (Olivieri et al., 2020), but whether TWH19 participates in other repair pathways is not tested. Here, we showed that TWH19 played a role in both NER and MMR but not double-stranded repair. This should not be surprising since many DDR proteins such as RPA2 and PCNA are shared in multiple DNA repair pathways (Pearl et al., 2015). TWH19 is possibly involved in both GG-NER and TCNER since it interacted with both the GG-NER player DDB1 and the TCNER player CSB (Iovine et al., 2011; Tiwari et al., 2021). In TCNER, TWH19 may bind to dsDNA adjacent to the damage site after upstream factors CSB and RPA2 bind to DNA lesions, preventing their dissociation and recruiting downstream factors such as PCNA (Fig. 6). PCNA promotes the processivity of DNA polymerase during DNA repair and is degraded after damage repair to prevent genomic instability (Cazzalini et al., 2014). Recruitment of MSH2, an MMR factor, can activate E3 ubiquitin ligases CRL4 to degrade repair factors after repair in NER (Tanaka et al., 2017), but whether PCNA is degraded by CRL4 is unclear. TWH19 may additionally regulate NER and MMR by regulating the expression level of several repair factors in these pathways (Fig. S3 B), but further validation is needed.

Tumor cells are particularly sensitive to DDR inhibitors (O’Connor, 2015). Radiotherapy and chemotherapy induce DNA damage to drive cancer cells into apoptosis and senescence (Ou and Schumacher, 2018). Several PARP inhibitors have been licensed to treat cancers (Gourley et al., 2019). Here, the DDR protein TWH19 was also shown to be critical for the proliferation and, to a lesser extent, the apoptosis of human cells. As TWH19 is amplified or overexpressed in several cancer types (AACR Project GENIE Consortium, 2017; Uhlen et al., 2017), it may play an oncogenic role in affected cancer and serve as a new anticancer target. Another striking observation in cells depleted of TWH19 (via either siRNA or gRNA) was the downregulation of metabolic pathways, which may explain the decrease in cell proliferation. These KD observations are consistent with the DDR function of TWH19, whereby cells respond to unrepaired DNA damage (possibly via JNK activation; Li and Chen, 2018; Milanese et al., 2019; Surova and Zhivotovsky, 2013; Williams and Schumacher, 2016). However, the cell cycle, which is often arrested in the presence of unrepaired DNA damage (Fousteri and Mullenders, 2008), was barely altered by TWH19 KD (Fig. S4, H and I), suggesting that the KD phenotypes may not be sorely due to DNA damage responses. Alternatively, TWH19 may regulate these cellular processes and cell fate via a DDR-independent route, i.e., binding to RNA and regulating mRNA levels. Further studies are needed to elucidate how TWH19 exerts its molecular functions to mediate these diverse cellular processes.

The DNA constructs and antibodies used in this paper are listed in Table S2.

Human STK19 (accession # NM_004197.2), STK19-C (25–254), 3KA (STK19^{K186A, K190A, K201A}) RPA2, PCNA, or H4 were cloned into a psyno-1 vector expressing an N-terminal cleavable 6His-MBP fusion to facilitate its protein expression. His-STK19-WT and its Y204E, E206K, or L248A mutations were cloned into a psyno-1 vector expressing an N-terminal 6His. All proteins were expressed in the E. coli BL21 (DE3) strain and grown in LB broth medium. Protein expression was induced by the addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside, and the culture was grown overnight at 18°C. The protein was purified by Ni-NTA beads and eluted with a buffer containing 30 mM Tris, pH 8.0, 300 mM NaCl, 300 mM imidazole, and 10% glycerol. The eluate was collected, added with 2 mM DTT, and further purified by size exclusion chromatography using a Superdex 200 Increase gel filtration column (GE Healthcare) in 20 mM Tris, pH 8.0, 200 mM NaCl, 2 mM DTT, and 10% glycerol.

Crystals of selenomethionine (SeMet)-labeled STK19 (25-C) were grown using the hanging-drop vapor-diffusion method at 16°C by mixing an equal volume of STK19-C (10 mg/ml) and crystallization buffer. The crystallization conditions contained 0.1 M 2-(N-morpholino)ethanesulfonic acid, pH 5.5, 0.2 M NaCl, and 2.0 M (NH₄)₂SO₄. All crystals were cryo-protected using 20% (vol/vol) glycerol-supplemented crystallization buffer and flash-frozen in liquid nitrogen. X-ray diffraction data were collected at the Shanghai Synchrotron Radiation Facility beamline BL17U1. The data collection statistics are given in Table 1.

The structure was solved by selenium single-wavelength anomalous dispersion using the program Autobuild and manually built using the program COOT. Refinement was performed using the program Refmac. Data at intervals of 50.00–1.65 Å resolution were used, and at the end of the refinement, the R-value was 0.149 (R_free = 0.194) for all reflections. Coordinates were deposited PDB with accession codes 7XRB.

To assess different interactions, maltose binding protein (MBP)-tagged proteins were immobilized on dextrin beads (Smart Lifesciences). Soluble proteins at the indicated concentrations were incubated with the immobilized proteins in a total volume of 1 ml for 2 h at 4°C. After three washing steps, bound proteins were separated by 12% SDS/PAGE and visualized by Coomassie Blue staining. Each experiment was repeated at least twice and checked for consistency. The pull-down buffer contained 20 mM Tris, pH 8.0, 200 mM NaCl, 10% glycerol, 0.005% Triton X-100, and 5 mM DTT.

DNA (25 μM, 2 μl) was incubated with increasing amounts of proteins in a buffer (50 mM NaCl, 20 mM Tris8.0, 10% glycerol, 1 mM DTT, and 2 mM MgCl₂) for 30 min on ice (total 20 μl). After incubation, all samples were loaded onto an 8% nondenaturing PAGE gel. Gel electrophoresis was performed with cold TBE buffer (180 mM Tris/boric acid buffer, pH 7.6, and 2 mM EDTA) at 80 V for 60 min. Gels were stained with ethanol bromide and photographed with a gel documentation system (GenoSens 1880).

STK19 or its mutants were cloned into pcDNA3.1(+) or pEBG. HEK293T, HeLa, and B16-F10 cell lines were cultured in DMEM supplemented with 10% fetal bovine serum standard (Newzerum) at 37°C. HEK293T cells were transfected with polyethylenimine (PEI) transfection reagent (Invitrogen). HeLa cells were transfected with liposome transfection reagent (Yeasen). The cells were cultured at different times after transfection and treated differently. For example, cells were harvested and lysed directly for western blot and exposure (Clinx Science Instruments).

Samples were fixed with 4% formaldehyde in PBS for 15 min and permeabilized with 0.2% Triton X-100 (Sangon Biotech) in PBS for 15 min. After blocking with 5% BSA in PBS for 1 h, samples were incubated with primary antibodies overnight at 4°C. After washing, the samples were incubated with fluorescently tagged secondary antibodies for 1 h at room temperature. Confocal images were acquired by ZEN blue software under a confocal microscope (LSM 880 with airyscan; Zeiss) with a 63×/1.40 oil-immersion objective lens (Zeiss) at room temperature. The fluorochromes used were Alexa Fluor 546 and DAPI. Images were prepared in Fiji and Adobe Photoshop. γ∙H2AX foci numbers were quantified using the Analyze Particles plugin in ImageJ.

Cells were harvested and lysed in lysis buffer (P0013; Beyotime Biotechnology) containing 1% protease inhibitors (B14002; Bimake) with or without DNase/RNase. Glutathione beads (Smart Lifesciences) and the supernatant from cell lysates were mixed and incubated at 4 °C for 6 h. The beads were then washed with lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 0.5% NP-40) four times, and the bound proteins were analyzed with the respective antibodies.

HeLa, 293T, or B16-F10 cells were irradiated with UVB at the indicated doses. To avoid UVB absorption by the medium, the cells were washed with PBS prior to irradiation, and fresh medium was replaced thereafter. The cells were then returned to the incubator and allowed to recover for 2 h before harvesting or fixing.

Cells were harvested and resuspended in lysis buffer A containing 10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucrose, 10% glycerol, 1 mM DTT, 0.1% Triton X-100, and protease inhibitor cocktail. After centrifuging at 4°C for 4 min at 1,500 g, the supernatant was collected as the cytoplasm. The pellet was gently resuspended in buffer B (3 mM EDTA, 0.2 mM EGTA, and 1 mM DTT) with protease inhibitor and placed on ice for 10 min. Cells were then centrifuged at 4°C for 4 min at a speed of 2,000 g and the pellet was collected as the chromatin fraction.

For CRISPR-Cas9-mediated gene knockout, gRNA sequences (5′-GTC​CTC​ACC​GTC​CGA​GAT​GC-3′ or 5′-CCC​ATC​GTG​CTG​AGG​AGC​CA-3′) were introduced into the V2T construct via PCR. Constructs encoding Cas9 and gRNA were cotransfected with viral packaging plasmids (V2T: pMDLg/pRRE:pRes-Rev:pCMV-VSVG = 3:1:1:1) into HEK293T cells. After 48 h, the viral supernatants were collected, filtered, and added to 6 μg/ml polybrene (YEASEN) before infecting HeLa cells. After 48 h of infection, the target cells were cultured in selection medium containing 8 μg/ml puromycin (BBI Life Sciences). Since these cells always repressed STK19 in the absence of puromycin, the efficiency of KD was enforced by puromycin selection before being used in any experiment. A portion of the puromycin-treated cells was used to determine the level of KD by immunoblotting.

Total RNA was extracted with a SteadyPure universal RNA extraction kit (Accurate Biotechnology) as described in the manufacturer’s manual. RNA was reverse-transcribed into cDNA and analyzed by RT-qPCR using a SYBR Green qPCR kit (Accurate Biotechnology). Three independent replicates were analyzed per sample and the relative gene expression was normalized to the internal tubulin (TUBB) control using the 2^−ΔΔCt method. Error bars represent the SEM of three biological replicates. Two-tailed unpaired Student’s t test was used in all groups. *P < 0.05; **P < 0.01; ***P < 0.001. The qPCR primers used are listed in Table S3.

After GST pull-down with or without DNase/RNase, the proteins that interact with STK19 were boiled on beads and subsequently collected for in-gel digestion. After separation by SDS-PAGE, the gel was stained using colloidal Coomassie G-250 and the gel band was cut into 1 mm³ small pieces. The gel particles were washed with 50% ethanol and water and then dehydrated in 100% acetonitrile until the gel particles became opaque white in color. After drying, the gel was reduced by 5 mM DTT at 56°C for 1 h and alkylated with 15 mM iodoacetamide in the dark at room temperature for an additional 30 min. Then, the gel was digested with trypsin at 37°C overnight and the tryptic peptides were extracted by a decreasing gradient of trifluoroacetic acid and an increasing gradient of acetonitrile. After being desalted using C18 ZipTip, the peptides were analyzed using an EASY-nLC 1000 instrument coupled to a Q Exactive plus mass spectrometer.

To delineate the proteomic profile after STK19 KD, TMT-based quantitative proteomics was carried out. Three biological replicates of STK19 negative control (NC) and KD cells were first lysed in radioimmunoprecipitation assay buffer (150 mM NaCl, 50 mM Tris, 1% NP-40 [vol/vol], 0.5% sodium deoxycholate [wt/vol], pH = 7.5) containing protease and phosphatase inhibitors, followed by 3 min of sonication (3 s on and 10 s off, amplitude 25%). The lysate was centrifuged at 20,000 g for 10 min and the supernatant was collected. The protein concentration was determined using the Bradford assay. 50 µg of protein from each sample was reduced with 10 mM tris(2-carboxyethyl)phosphine at 56°C for 1 h and alkylated with 20 mM iodoacetamide for 30 min to block the free cysteine residues. After protein precipitation by methanol/chloroform, trypsin digestion was carried out overnight at 37°C. Peptides were labeled with TMT according to the TMT 10plex reagent instructions (Thermo Fisher Scientific). After a 1-h incubation at room temperature, 5% hydroxylamine was added to quench the reaction. Peptides labeled with nine different TMT labels were mixed, dried, and desalted. Then, the desalted TMT-labeled peptides were fractionated by HPLC (Agilent 1260; Agilent), and the obtained 120 fractions were combined into 20 parts. The 20 combined fractions were desalted with C18 ZipTip and analyzed by an EASY-nLC 1200 instrument coupled to a Q Exactive HF-X mass spectrometer.

All the raw files were searched against the Swiss-Prot human protein sequence database in MaxQuant (version 1.6). TMT isobaric labeling-based relative quantification algorithms were used for protein quantification.

All results were repeated at least three times. Statistical analyses were performed using GraphPad Prism 8 software. Comparisons between groups were analyzed by unpaired two-tailed Student’s t test where indicated. The statistical parameters (including n, SEM, and SD) are shown in the figure legends. Data distribution was assumed to be normal, but this was not formally tested. *P < 0.05; **P < 0.01; ***P <0.001. P values < 0.05 are considered to be significant.

The associated supplemental files contain five figures, three tables, and two Excel files. Fig. S1 shows that STK19 has no kinase activity. Fig. S2 shows that STK19 can form a homodimer to bind dsDNA. Fig. S3 shows that STK19 is involved in DDR. Fig. S4 shows that STK19 is implicated in multiple cellular pathways. Fig. S5 shows that STK19 is involved in cell proliferation. Table S1 contains different DNA sequences used in this study. Table S2 contains a summary of antibodies and plasmids used in this work. Table S3 is the list of primer sequences used for qPCR. Data S1 contains the MS data about kinase activity detection of STK19 on NRas. Data S2 shows the list of proteins bound to STK19 and TMT MS data.

All the MS raw data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX (Ma et al., 2019) partner repository with the dataset identifier PXD034123. The data are available from the corresponding author upon reasonable request.

We thank the staff from the BL17B1/BL18U1/BL19U1/BL19U2/BL01B beamline of the National Facility for Protein Science in Shanghai at Shanghai Synchrotron Radiation Facility for assistance during data collection (Zhang et al., 2019).

This work was supported by the National Key R&D Program of China (No. 2018YFC2000305), and the National Natural Science Foundation of China (No. 82273850, 31870826, 81570060, and 82073221).

Author contributions: Q. Sun conceived the project. Q. Sun, Y. Li, Y. Gong, and L. Dai prepared the manuscript. Y. Li, Y. Gong, Y. Zhou, Y. Xiao, and W. Huang performed all the experiments. Q. Sun, L. Dai, and Y. Zhao supervised the project. S. Zhang, Y. Tu, and Q. Zhou provided relevant resources.