The crucial role of single-stranded DNA binding in enhancing sensitivity to DNA-damaging agents for Schlafen 11 and Schlafen 13

Summary Schlafen (SLFN) 11 enhances cellular sensitivity to various DNA-damaging anticancer agents. Among the human SLFNs (SLFN5/11/12/13/14), SLFN11 is unique in its drug sensitivity and ability to block replication under DNA damage. In biochemical analysis, SLFN11 binds single-stranded DNA (ssDNA), and this binding is enhanced by the dephosphorylation of SLFN11. In this study, human cell-based assays demonstrated that a point mutation at the ssDNA-binding site of SLFN11 or a constitutive phosphorylation mutant abolished SLFN11-dependent drug sensitivity. Additionally, we discovered that nuclear SLFN13 with a point mutation mimicking the DNA-binding site of SLFN11 was recruited to chromatin, blocked replication, and enhanced drug sensitivity. Through generating multiple mutants and structure analyses of SLFN11 and SLFN13, we identified protein phosphatase 2A as a binding partner of SLFN11 and the putative binding motif in SLFN11. These findings provide crucial insights into the unique characteristics of SLFN11, contributing to a better understanding of its mechanisms.


INTRODUCTION
DNA-damaging anticancer agents (DDAs), such as platinum derivatives and topoisomerase I and II inhibitors, have been applied in the clinical setting for decades.However, the clinical application of predictive biomarkers for responses to DDAs has not yet been established.Schlafen 11 (SLFN11) has attracted attention as a candidate because its mRNA expression level is highly correlated with sensitivity to various DDAs in multiple cancer cell line databases. 1,24][5][6][7] Moreover, the clinical implications of SLFN11 have been validated in multiple cancers, including breast, 8 ovary, [9][10][11] stomach, 12 bladder, 13 lung, 14,15 esophagus, 16 medulloblastoma, 17 head and neck, 18 and prostate. 19lfn genes (Slfn1-14) evolved rapidly after the branching of boreoeutheria.20 Among the 5 human SLFNs (SLFN5, 11, 12, 13, 14), SLFN5, 11, 13, and 14 commonly harbor a nuclease domain at the N-terminus, an SWAVDL motif in the middle region, and a putative helicase/ATPase domain with Walker A/B motifs at the C-terminus 21 (Table S1).Despite the strong conservation of sequences and structures among SLFNs, SLFN11 is unique due to its highly significant correlation with sensitivity to DDAs.22 Although the mechanisms of SLFN11-mediated cell killing are not fully understood, two functions of SLFN11 are most likely.][24][25] Dephosphorylation at the N-domain (S219 and T230) and the C-domain (S753) of SLFN11 is important for SLFN11-dependent tRNA cleavage.23 Notably, the N-terminal nuclease active site residues E209, E214, and K216 in SLFN11 are well conserved in human SLFN11, -12, À13, and -14.[26][27][28][29] Conversely, human SLFN5 lacks the K216 residue and has no endonuclease activity.26 The other is the unique function of SLFN11 that blocks replication independently of the ataxia telangiectasia and Rad3-related (ATR) activation (i.e., S-phase checkpoint) following replication stress.7 SLFN11 has replication protein A1 (RPA1) binding domain in the C-terminus 3 and is recruited to replication forks via RPA1/2/3 complex (RPA)-coated ssDNA under replication stress conditions.7 Stressed replication forks harbor RPA-coated ssDNA gaps to which SLFN11 binds, inducing permanent replication blockage (i.e., until the cell dies).SLFN11 with a Walker B motif mutation (E669Q, a putative ATPase-dead) can still be recruited to stressed replication forks, but lacks the replication-blocking and drug-sensitizing capabilities.Recent cryo-electron microscopy and biochemical analyses have revealed that SLFN11 preferentially binds ssDNA rather than double-stranded DNA (dsDNA), in a dimeric state.30 The results showed that SLFN11 (K652) is the ssDNA binding site, and dephosphorylation at S753 is critical for ssDNA binding.However, the relationship between ssDNA binding and drug sensitivity remains unclear.Furthermore, the reasons why SLFN11 is unique compared with other SLFNs have not been clarified.
In this study, we searched for the domains, amino acid residues, and interactors that determine the uniqueness of SLFN11.Using human cellbased analyses, we found that the ssDNA binding capability of SLFN11 is crucial for its functions (chromatin binding, replication block, drug sensitivity).Then, we explored the possibility of converting SLFN13 into a protein resembling SLFN11 with minimal modifications.We identified protein phosphatase 2A (PP2A) as a binding partner of SLFN11 and the binding domain-associated SLFN11-dependent drug sensitivity.

RESULTS
The single-stranded DNA binding site (K652) and the dephosphorylation of S753 are critical for Schlafen 11-dependent drug sensitivity SLFN13 shares the highest amino acid conservation with SLFN11 (78% identity and 83% similarity; Table S2).However, in contrast to SLFN11, SLFN13 expression is not correlated with drug sensitivity, although the expression levels of both proteins vary extensively (Figure S1).Among the functional domains and sites of SLFN11, the dimer surface sites (R82, K591, Y722), nuclease activity sites (E209, E214, K216, Y234, D252), and SWAVDL, Walker A, and Walker B motifs are perfectly conserved between SLFN11 and SLFN13 (Figures 1A, S2A, and S2B).However, one ssDNA binding site (K652) and two phosphorylation sites (T230 and S753) of SLFN11 are not conserved in SLFN13 (Figure 1A).Given that S753 dephosphorylation is important for activating the ssDNA binding ability of SLFN11, we speculated that K652 and S753 in SLFN11 may serve as determinants for its unique functions.Hence, we designed various constructs expressing SLFN11 with or without mutations at K652, S753, or both (Figure 1B).
The K652E and K652D mutations eliminate the positive charge of K652.The S753P mutation mimics the constitutively dephosphorylated state, whereas S753D mimics the constitutively phosphorylated state.We generated stable cells expressing (over 80%) wild-type or mutant SLFN11 in K562 human leukemia cells, which normally have undetectable levels of endogenous SLFN11 (Figures 1C and S3A).The S753P SLFN11 construct was expressed at lower levels compared to the other SLFN11 mutants (K652E, K652D, S753D, and K652E/S753P).A cell viability assay confirmed that the S753P SLFN11 mutant conferred sensitivity to camptothecin (CPT), similar to the wild-type SLFN11, whereas the other SLFN11 mutants (K652E, K652D, S753D, and K652E/S753P) did not (Figure 1D).The differences in cell death were morphologically validated by microscopy (Figure S3B) and the detection of cleaved poly(ADP-ribose) polymerase 1 (PARP1) and cleaved caspase 3 (Figure S3C), which are both produced upon apoptosis.These results suggested that the ssDNA binding site (K652) and the dephosphorylation of S753 in SLFN11 are critical for SLFN11-dependent drug sensitivity.

Chromatin binding is required for the functions of Schlafen 11
The SLFN11-dependent replication block is a hallmark of its functionality. 7To examine the effects of the SLFN11 mutants, we performed an EdU-labeled cell cycle analysis.As expected, CPT suppressed replication regardless of SLFN11 because it activates ATR (S-phase checkpoint) (Figure 2A, compare Control vs. CPT).The addition of an ATR inhibitor restored the replication in SLFN11-negative cells (Figure 2A, Vector), but exhibited little efficacy in wild-type SLFN11-expressing cells (Figure 2A, wild-type SLFN11).The S753P SLFN11 mutant maintained the replication block, whereas the other SLFN11 mutants (E669Q, K652E, K652D, S753D, and K652E/S753P) restored replication under the CPT+ATR inhibitor treatment conditions (Figure 2A).
Next, to examine the chromatin binding capabilities of these SLFN11 mutants, we measured the levels of chromatin-bound SLFN11 (Figure 2B).As reported, both wild-type and E669Q SLFN11 were recruited to chromatin following CPT treatment.Likewise, the S753P SLFN11 mutant exhibited chromatin recruitment following CPT treatment.In contrast, the remaining SLFN11 mutants (K652E, K652D, S753D, and K652E/S753P) were not further recruited to chromatin following CPT treatment.The increased background levels of these mutants could be attributed to their higher expression levels relative to S753P SLFN11 (Figure 1C).To verify these findings, we performed an immunofluorescence analysis with pre-extraction, which enabled us to observe chromatin-bound proteins exclusively (Figures 2C and S4).As reported previously, wild-type SLFN11 was observed at both the nuclear periphery and inner nucleus, with the colocalization of RPA2, upon CPT treatment.The S753P SLFN11 mutant also exhibited patterns similar to those of wild-type SLFN11, but the other SLFN11 mutants (K652E, K652D, S753D, and K652E/S753P) did not.Collectively, these results indicated that the ssDNA binding site (K652) and the dephosphorylation of S753 within SLFN11 are critical for its functions related to chromatin binding, replication blockage, and drug sensitivity.

Forced nuclear expression of Schlafen 13 cannot mimic the phenotypes of Schlafen 11
Having identified the crucial sites for SLFN11's functions, we next explored the possibility of converting SLFN13 into SLFN11 by genetic modification.SLFN11 is dominantly expressed in the nucleus, as verified in various cells and tissues, 31 while SLFN13 is localized in the cytoplasm. 28hrough computational analyses using cNLS Mapper 32 and NLStradamus, 33 we identified a robust NLS at the N-terminus of SLFN11, in addition to the previously reported NLS at the C-terminus 22 (Figures 3A, 3B, and S5A).Conversely, only a minimal NLS was detected in SLFN13 (Figures 3A, 3B, and S5A).
Because SLFN11 exerts its functions on chromatin (Figures 1 and 2), we attempted to overexpress SLFN13 in the nucleus.Accordingly, we constructed SLFN13-expression plasmids with a FLAG tag at the N-terminus (Flag-SLFN13) or the C-terminus (SLFN13-Flag) or with a FLAG tag and NLS at the N-terminus (NLS-SLFN13) or the C-terminus (SLFN13-NLS) (Figure 3C).We generated stable K562 cells, which normally have undetectable levels of endogenous SLFN13, expressing these SLFN13 constructs (Figures 3D and S5B).A fractionation analysis demonstrated that NLS-SLFN13 exhibited the highest expression in the nucleus (Figure 3E).Cellular viability assays showed that wild-type SLFN11, • Dimer interface sites ■ Nuclease activity sites but none of the SLFN13-carrying cells, conferred sensitivity to CPT, even though the expression level of NLS-SLFN13 in the nuclear fraction exceeded that of wild-type SLFN11 (Figures 3F and S5C).Moreover, upon CPT treatment, SLFN11, but none of the modified SLFN13s, was recruited to chromatin (Figure 3G).These results indicated that nuclear-localized SLFN13 per se cannot mimic the drug sensitivity and chromatin binding capabilities of SLFN11.
The single-stranded DNA binding site enables Schlafen 13 to recapitulate the functions of Schlafen 11 Having found that K652E SLFN11 lost its chromatin-binding ability (Figures 1 and 2), we then introduced the E652K mutation into NLS-SLFN13.We also introduced the P753S mutation, mimicking the phosphorylation site of SLFN11 (S753), into NLS-SLFN13 and E652K  NLS-SLFN13.We generated stable K562 cells expressing wild-type or mutant NLS-SLFN13 (Figures 4A and S6A).Viability assays revealed that the E652K NLS-SLFN13 had mild sensitivity to CPT, although the effect was not as strong as that of wild-type SLFN11 (Figure 4B).As expected, P753S NLS-SLFN13 did not show drug sensitivity (Figure 4B).Interestingly, E652K/P753S NLS-SLFN13 lost the acquired drug sensitivity of E652K NLS-SLFN13 (Figure 4B).The differences in cell death were morphologically validated by microscopy (Figure S6B) and immunoblotting against cleaved PARP1 and cleaved caspase 3 (Figure 4C).Cell cycle analysis revealed that E652K NLS-SLFN13 acquired the replication-block function, whereas E652K/P753S NLS-SLFN13 lost this capability (Figure 4D).Furthermore, E652K NLS-SLFN13 was recruited to chromatin following CPT treatment, whereas E652K/P753S NLS-SLFN13 was not (Figure 4E).These results indicated that NLS-SLFN13 can mimic the functions of SLFN11 when it acquires the ssDNA binding site (E652K).However, E652K NLS-SLFN13 loses these acquired functions if the amino acid residue at position 753 becomes phosphorylatable.
Protein phosphatase 2A (PP2A) is a binding partner of Schlafen 11 Since the P753S mutation nullified the chromatin-binding function in the E652K NLS-SLFN13 mutant, whereas SLFN11 harbors an intrinsic S753 site, we speculated that either (1) phosphatases for S753 in SLFN11 may be unable to access P753S in SLFN13 owing to differences in the local protein structure, or (2) SLFN11 may utilize specific phosphatases for S753 that are not functional with SLFN13 (Figure 5A).
To examine the former possibility, we generated two NLS-SLFN13 constructs with portions of the SLFN11 amino acid sequence, as illustrated in Figure S7A.The N-lobe chimera NLS-SLFN13 (E652K/P753S) contains the SLFN11 amino acid sequence from residues 577-731, while the C-lobe chimera NLS-SLFN13 (E652K/P753S) has the SLFN11 amino acid sequence from residues 731-901.Although we generated K562 cells stably overexpressing these plasmids, neither the N-lobe nor C-lobe mutant gained drug sensitivity (Figures S7B and S7C), indicating that the protein structure surrounding P753S in SLFN13 is not likely to cause the impaired dephosphorylation.
Then, to investigate the latter possibility, we performed a protein structure analysis focused on the short linear motif (SLiM) of protein phosphatases (PPs). 34As the PP1 catalytic subunit g (PPP1CC) reportedly binds SLFN11 and dephosphorylates three phosphorylation sites (S219, T230, and S753), we found two SLiMs of the PP1 catalytic subunit (PP1C) 5B). 35,36The two motifs were localized on the same surface of the dimerized SLFN11 (S753), implying that PP1 can access SLFN11 (S753).Furthermore, we determined that SLFN11, but not SLFN13, contains a SLiM for the PP2A regulatory B56 subunit (PP2A-B56) [(L/F/ M)-x-x-(I/V/L)-x-E] on this surface (Figure 5B). 37This region is annotated as a disordered region, suggesting the presence of a binding protein.

The protein phosphatase 2A binding motif in Schlafen 11 contributes to the Schlafen 11-dependent functions
To examine the effect of the PPA2-B56 binding motif, we introduced single or multiple point mutations (L358A, L361A, E363A, E363D and L358A/L361A/E363D) within the motif and obtained cell lines overexpressing these SLFN11 mutants (Figure 6A).In cell viability assays, the triple L358A/L361A/E363D SLFN11 mutant attenuated drug sensitivity, compared to wild-type SLFN11 and the other single SLFN11 mutants (Figure 6B).Moreover, the triple L358A/L361A/E363D SLFN11 mutant showed reduced replication-blocking capability, in contrast to wildtype SLFN11 (Figure 6C).These results indicated that the PP2A-B56 binding motif is a regulatory site for the SLFN11-dependent drug sensitivity and replication blockage.

DISCUSSION
This is the first study to demonstrate that the ssDNA binding site (K652) and dephosphorylation at S753 in SLFN11 are critical for chromatin binding, replication blockage, and drug sensitivity in human cells.We also revealed that nuclear SLFN13 can mimic the functions of SLFN11, as long as SLFN13 possesses the ssDNA binding site (E652K).Moreover, we clarified the interaction between PP2A and SLFN11, and identified the putative PP2A binding motif as a regulatory domain of SLFN11.Our study provides important insights into the unique functions and characteristics of SLFN11.

Do mice possess ''Slfn11''?
The establishment of a mouse model for SLFN11 is an important stride toward the development of this target in clinical and basic sciences, particularly because SLFN11 has attracted attention as a candidate for overcoming drug resistance.Mice express several Slfn genes, including Slfn1, -2, -3, -4, -5, -8, -9, and -10, with only Slfn5 being a shared gene between humans and mice. 39An ortholog of human SLFN11 has not been identified in the mouse genome.According to the phylogeny of Slfn genes based on a Bayesian analysis of amino acid sequences, mouse Slfn8 and Slfn9 are closer to each other and to human SLFN13 than to human SLFN11. 20Hence, it has been assumed that there is no ''Slfn11'' in mice.However, we successfully converted SLFN13 into a protein resembling SLFN11 by introducing an NLS and an ssDNA binding site (K652) (Figure 3).Based on our findings and considering the additional functional sites and domains of SLFN11, we explored the possibility that mouse Slfn8/Slfn9 may exhibit SLFN11-like characteristics (Table S1).The mouse Slfn8/Slfn9 proteins are localized in the nucleus, 40 and the phosphorylation site corresponding to SLFN11 (S753) is proline (P).Hence, mouse Slfn8/Slfn9 are likely to be constitutively active.The ssDNA binding site corresponding to SLFN11 (K652) is lysine (K) in mouse Slfn8/Slfn9, and thus mouse Slfn8/Slfn9 are likely to be capable of binding ssDNA.Just timely, during the review process, Alvi et al. reported that the mouse Slfn8 and Slfn9 genes share orthologous function with human SLFN11. 41w does human Schlafen 11 exhibit its uniqueness?
Our study revealed that the ssDNA binding ability is required for the unique functions of SLFN11.Human SLFN13 and SLFN12 lack the SLFN11 (K652)-relevant ssDNA binding site, 42,43 while SLFN5 harbors this site and the SLFN11 (S753)-relevant site is proline (P), and shows potent ssDNA and dsDNA binding ability (Table S1).However, the DNA-binding ability is related to the N-terminus, and the protein lacks dimerization sites (Table S1). 26Human SLFN14 shares 45% amino acid identity with SLFN11 (Table S2).Human SLFN14 harbors an SLFN11-relevant ssDNA binding site and the SLFN11 (S753)-relevant site is proline (P) (Table S1).Hence, SLFN14 could potently mimic SLFN11.However, SLFN14 expression is very low across $700 cell lines in the GDSC database (https://discover.nci.nih.gov/rsconnect/cellminercdb/), and thus its functions may have been overlooked.
We found that functionally active SLFN11 or SLFN13 showed lower levels of overexpression compared to their nonfunctional counterparts (Figures 1C and 4A).Nevertheless, cells overexpressing S753P SLFN11 remained viable without apparent phenotypic changes, suggesting that cells maintain the constitutively active SLFN11 below a certain level (Figure 1).Because S753D SLFN11 lost its SLFN11 function, dephosphorylation at S753 is apparently necessary to activate SLFN11 (Figure 1).Interestingly, this potential phosphorylation site is exclusive to primates (macaques, chimpanzees, and humans) (Table S1), implying that primates have acquired an SLFN11 ON/OFF switching system with the S753 site.We speculate that constitutively active SLFN11 might be harmful under specific conditions.For example, during B cell development SLFN11 expression is downregulated at the germinal center, where DNA damage and high proliferation occur. 44Additionally, highly proliferative embryonic stem cells turn off SLFN11 expression, whereas hematopoietic stem cells with lower proliferation rates express elevated levels of SLFN11. 45Under such highly proliferative conditions, constitutively active SLFN11 could be harmful.To control the switching system, SLFN11 must employ specific kinases and/or phosphatases, possibly including PP2A (Figure 5).Consistent with these findings, we demonstrated that the SLiM sequence of PP2A-B56 is conserved exclusively in primates (macaques, chimpanzees, and humans) and hedgehogs (Table S1).Interestingly, only human SLFN13, but no other species of SLFN13 (even in macaques or chimpanzees), lacks the DNA binding site (K652) (Table S1).If human SLFN13 had retained K652, it would likely function similarly to SLFN11.Strikingly, a most recent article revealed that Caenorhabditis elegans conserves SLFN-like domains. 46Accordingly, the evolution of the SLFN family may need to be revised based on functional sites and domains.
S N P S F N I P T SLFN13 I I E N P P I N

Figure 1 .Figure 2 .
Figure 1.The ssDNA binding site (K652) and the phosphorylation site (S753) regulate SLFN11-dependent drug sensitivity (A) Pairwise amino acid sequence alignment.Residues are colored according to the percentage of identity (red = match, orange = more conserved, yellow = less conserved, white = mismatch).(B) List of SLFN11 mutants used in this study.(C) Representative immunoblots prepared from K562 cells expressing WT and mutant SLFN11 proteins.The indicated antibodies were used.(D) Viability curves of K562 cell lines treated with various concentrations of CPT for 72 h.Error bars represent means G standard deviations (n = 3).***p < 0.0001 (Dunnett test); n.s., not significant.See also Figures S1-S3.

Figure 2 .
Figure 2. Continued (C) Representative immunofluorescence confocal microscopy images of chromatin-bound Flag-tagged SLFN11 (green), RPA2 (red), and DAPI (blue) in K562 cells expressing the indicated genes.Cells were treated with 100 nM CPT for 4 h.See also Figure S4.See also Figure S4.

Figure 3 .
Figure 3. Forced expression of SLFN13 in the nucleus is not sufficient to confer SLFN11 functions on SLFN13 (A) Schematic diagram of the SLFN proteins.(B) Prediction of NLS signals in SLFN11 and SLFN13 with two algorithms (cNLS Mapper and NLStradamus).Scores of 7-8 indicate partial nuclear localization, 3-5 indicate localization in both the nucleus and cytoplasm, and 1-2 indicate cytoplasmic localization.(C) Schematics of the SLFN13 mutants generated in this study.(D and E) Blots of whole cell lysates (D) or cytoplasmic and nuclear fractions (E) from K562 cells expressing various constructs.The indicated antibodies were used.(F) Viability curves of the K562 cell lines.Experiments were performed and data were plotted as in Figure 1D.Error bars represent means G standard deviations (n = 3).***p < 0.0001 (Dunnett test); n.s., not significant.(G) Blots of the chromatin-bound fraction of K562 cells expressing various constructs.Experiments were performed and data were plotted as in Figure 2B.See also Figure S5.

Figure 4 .Figure 5 .
Figure 4. SLFN13 gains SLFN11-mimicking functions upon introducing the ssDNA binding site (E652K) but loses the functions if residue 753 is phosphorylatable (A) Blots of whole cell lysates from K562 cell lines expressing various constructs.The indicated antibodies were used.(B) Viability curves of K562 cell lines.Experiments were performed and data were plotted as in Figure 1D.Error bars represent means G standard deviations (n = 3).*p < 0.05, ***p < 0.0001 (Dunnett test); n.s., not significant.(C) Blots of whole cell lysates from K562 cell lines expressing various constructs after 250 nM CPT treatment for 24 h.The indicated antibodies were used.(D) Flow cytometry cell cycle data.Data are representative of two independent experiments.Experiments were performed and data were plotted as in Figure 2A.(E) Blots of the chromatin-bound fraction from K562 cells expressing various constructs.Experiments were performed and data were plotted as in Figure 2B.See also Figure S6.

Figure 5 .
Figure 5. Continued (D) Co-IP of E652K/P753S SLFN13 mutant and endogenous PP2A-B56g in E652K/P753S SLFN13-expressing K562 cells, detected by the same method as in panel C. (E) Representative images of PLA foci between PP2A-B56g and SLFN11 or E652K/P753S SLFN13.Cells were treated with 100 nM CPT for 4 h.PLA in empty vectorexpressing cells served as a negative control.PLA between Flag and SLFN11 or SLFN13 was used as a positive control.The scale bar represents 20 mm.

Figure 6 .
Figure 6.Putative PP2A binding motif contributes to the functions of SLFN11 (A) Upper: The SLFN11 mutation strategy.Lower: Western blots of the SLFN11 mutants.(B) Viability curves of various K562 cell lines.Experiments were performed and data were plotted as in Figure 1D.Error bars represent means G standard deviations (n = 3).***p < 0.0001 (Dunnett test); n.s., not significant.(C) Cell cycle flow cytometry data.Data are representative of two independent experiments.Experiments were performed and data were plotted as in Figure 2A.