The FANCI/FANCD2 complex links DNA damage response to R-loop regulation through SRSF1-mediated mRNA export

SUMMARY

Fanconi anemia (FA) is characterized by congenital abnormalities, bone marrow failure, and cancer susceptibility. The central FA protein complex FANCI/FANCD2 (ID2) is activated by monoubiquitination and recruits DNA repair proteins for interstrand crosslink (ICL) repair and replication fork protection. Defects in the FA pathway lead to R-loop accumulation, which contributes to genomic instability. Here, we report that the splicing factor SRSF1 and FANCD2 interact physically and act together to suppress R-loop formation via mRNA export regulation. We show that SRSF1 stimulates FANCD2 monoubiquitination in an RNA-dependent fashion. In turn, FANCD2 monoubiquitination proves crucial for the assembly of the SRSF1-NXF1 nuclear export complex and mRNA export. Importantly, several SRSF1 cancer-associated mutants fail to interact with FANCD2, leading to inefficient FANCD2 monoubiquitination, decreased mRNA export, and R-loop accumulation. We propose a model wherein SRSF1 and FANCD2 interaction links DNA damage response to the avoidance of pathogenic R-loops via regulation of mRNA export.

Graphical abstract

In brief

While the Fanconi anemia pathway has a well-characterized function in DNA crosslink repair, its newly uncovered role in regulating R-loop metabolism has not been fully understood. Olazabal-Herrero et al. reveal that FANCD2 functions in NXF1-SRSF1-mediated mRNA export regulation, preventing R-loop formation and subsequent genomic instability.

INTRODUCTION

Fanconi anemia (FA) is a rare genetic disorder characterized by congenital abnormalities, progressive bone marrow failure, and increased susceptibility to cancer, such as acute myeloid leukemia and squamous cell carcinoma. FA is a multigenic disorder with biallelic mutations in at least 23 different genes identified to date that cooperate in the FA pathway to repair DNA inter-strand crosslinks (ICLs).¹ Monoubiquitination of the central FA protein complex FANCI/FANCD2 (ID2) is a critical event for the recruitment of DNA repair proteins and replication fork protection. Cellular hypersensitivity to ICL agents, such as mitomycin C (MMC), is a hallmark of FA cells.

In addition to ICL repair, several FA factors, including FANCD2, FANCA, FANCM, BRCA1/FANCS, and BRCA2/FANCD1 have been linked to RNA metabolism and R-loop regulation.^2–8 Most R-loops are formed during transcription, when nascent RNA becomes stably paired with the DNA template, resulting in a three-stranded DNA:RNA hybrid that harbors a displaced DNA strand. Although R-loops are normal intermediates in a variety of physiologically relevant functions, such as mitochondrial DNA replication, transcription regulation, or class switch recombination of immunoglobulin genes, persistent R-loops can lead to serious conflicts with DNA replication, including replication fork stalling and collapse, resulting in genome instability.^9–11 In particular, transcription-replication conflicts have been postulated as an endogenous source of genotoxicity that contribute to the FA phenotype.^3,5,6 The functional interplay between FA proteins and R-loops remains to be mechanistically defined.

Disturbance of mRNA processing, export, or splicing has also been linked to R-loop-mediated genomic instability. SRSF1 (ASF/SF2) is a prototypical serine- and arginine-rich (SR) protein that functions in constitutive and alternative splicing. Moreover, SRSF1 plays a role as an export adapter involved in mRNA nuclear export through the nuclear export factor 1 (NXF1) pathway.^12–15 Aberrant expression of splicing factors such as SRSF1 has been linked to R-loop metabolism, with diminished SRSF1 leading to R-loop accumulation and concomitant genomic instability.¹⁶ In addition, alterations in splicing factors have been associated with cancer and the pathophysiology of myelodysplastic syndrome (MDS),¹⁷ which shares clinical features of FA.

In the present study, we uncover a novel role of FANCD2 and SRSF1 in the prevention of R-loop formation. We show that binding of SRSF1 to FANCD2 stimulates FANCD2 monoubiquitination, which is crucial for the licensing of signaling and DNA repair events in the FA pathway in an RNA-dependent manner. We provide evidence that SRSF1-FANCD2 interaction fulfills an important role in regulating NXF1-mediated mRNA export, thus preventing R-loop formation. Furthermore, cancer-associated SRSF1 mutants fail to interact with FANCD2, leading to inefficient FANCD2 monoubiquitination, impaired mRNA export, and accumulation of R-loops. Our work provides mechanistic insights into how the FA pathway serves a key regulatory role in coupling DNA damage response to R-loop regulation via the mRNA export apparatus and helps uncover potential vulnerabilities that can be targeted for therapeutic opportunities in cancer treatment.

RESULTS

Knockdown of the splicing factor SRSF1 results in an FA-like cellular phenotype

Alterations in splicing factors have been associated with cancer and MDS,^17,18 a clinical entity reminiscent of FA. Particularly, aberrant expression of SRSF1 has been linked to R-loop-mediated genomic instability.¹⁶ In addition, a high-throughput proteomics analysis has identified SRSF1 as a potential FANCD2 interactor.¹⁹ To study the relationship between splicing factors and the FA pathway, we depleted SRSF1 by using a small interfering RNA (siRNA) that is adept at reducing SRSF1 protein levels (Figure 1A). SRSF1 knockdown resulted in a significant decrease of FANCD2 monoubiquitination upon MMC-induced DNA damage (Figures 1A and S1A, lane 4) (p = 0.0201). To further validate the specificity of SRSF1 siRNA, we performed a rescue experiment in which we transiently expressed SRSF1-GFP (Figure 1A) or FLAG-SRSF1 (Figure S1A) in SRSF1 knockdown cells. We found that FANCD2 monoubiquitination is restored upon SRSF1 expression (Figures 1A and S1A, lane 6) (p = 0.0412). Conversely, qPCR and RNA sequencing (RNA-seq) experiments demonstrated that FANCD2 mRNA levels or isoforms were not altered (Figures 1B and S1B), indicating that SRSF1 regulates FANCD2 post-translationally. Importantly, depletion of other splicing factors (U2AF1, U2AF2, SRSF1, SRSF2, and SRSF6) did not result in diminished FANCD2 ubiquitination, highlighting the specificity of SRSF1 (Figure S1C). Because FANCD2 monoubiquitination is required for FANCD2 focus assembly, we examined the effect of SRSF1 silencing on FANCD2 foci after MMC treatment. As expected, we found diminished FANCD2 focus formation in MMC-treated SRSF1 knockdown cells (Figure 1C, left) (p = 0.0065). Importantly, MMC-induced FANCD2 foci were restored by SRSF1-GFP expression (Figure 1C, right) (p = 0.002). Next, we evaluated the effect of knocking down SRSF1 on MMC sensitivity, another hallmark of FA. Because SRSF1 depletion has been linked to decreased cell proliferation,²⁰ we first confirmed that SRSF1 knockdown did not significantly affect cell growth or the cell cycle under our experimental conditions (Figure S1D). Furthermore, using the Click-iT-EdU assay, we show that DNA replication is not significantly affected after SRSF1 depletion (Figure S1D). Using the crystal violet assay, we found that SRSF1 knockdown renders cells hypersensitive to MMC to a degree similar to that caused by FANCD2 knockdown (Figure 1D). Importantly, double knockdown (siSRSF1+siFANCD2) did not further increase MMC sensitivity (Figure 1D), indicating that FANCD2 and SRSF1 function epistatically in the cellular response to MMC-induced DNA damage. Finally, because FA cells experience chromosomal breaks^21,22 and cell-cycle arrest at the G2/M DNA damage checkpoint,^23,24 we directly monitored chromosome breaks and cell cycle progression following MMC treatment. We observed enhanced chromosome aberrations and G2/M arrest after SRSF1 depletion in MMC-treated samples, comparable with the impact of FANCD2 knockdown (Figures S1E and S1F). We exclude any potential indirect effect of SRSF1 knockdown on the observed phenotype by showing that SRSF1 knockdown has little or no impact on global gene expression or splicing (Figure S1G). Taken together, our data indicate that SRSF1 depletion exerts a strong negative impact on FA pathway activation and leads to an FA-like cellular phenotype.

Figure 1. Knockdown of the splicing factor SRSF1 results in an FA-like cellular phenotype.

(A) Immunoblot analysis of HeLa cells transfected with siRNA scrambled (siCTRL) or siRNA against SRSF1 (siSRSF1), untreated or treated with 1 μM MMC for 24 h. A rescue experiment was conducted in lanes 5 and 6 by co-transfection with siSRSF1 and SRSF1-GFP. Ku86 was used as a loading control (left). Densitometry analysis of the immunoblot shows the ratio of ubiquitinated to non-ubiquitinated FANCD2 (ub-FANCD2/FANCD2). The data represent the mean and SEM of three independent experiments. Data were analyzed using the unpaired t test (right).

(B) FANCD2 mRNA levels were measured by RT-qPCR after transfection with siCTRL or siSRSF1. The data represent the mean and SEM of three independent experiments, normalized to siCTRL. Data were analyzed using the unpaired t test.

(C) Immunofluorescence of MMC-induced FANCD2 foci in HeLa cells after transfection with siCTRL, siSRSF1 (left), or co-transfection of siSRSF1 and SRSF1-GFP (center). The graph shows the FANCD2 focus quantification. Data (mean and SEM) are representative of three independent analyses of at least 50 cells per slide. Data were analyzed using the unpaired t test (right).

(D) HeLa cells were treated with the indicated concentrations of MMC, and cell survival was analyzed by crystal violet after incubation at 37°C for 4 days. The percentages of surviving cells were normalized to an untreated control and are shown as the mean and SEM of three independent experiments. Data were analyzed using two-way ANOVA.

*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, non-significant. *, compared with siCTRL, #, compared with siSRSF1).

See also Figure S1.

SRSF1 physically interacts with the FANCI/FANCD2 complex

We conducted FANCD2-SRSF1 co-immunoprecipitation in FA-D2 mutant cells transduced with cDNA encoding wild-type FANCD2 (FA-D2 + FANCD2) or the FANCD2 ubiquitin-dead mutant (FA-D2 + K561R). Our results revealed that only wild-type FANCD2 co-immunoprecipitated SRSF1 in an MMC-enhanced manner (Figures 2A and 2B). Reciprocal co-immunoprecipitation of endogenous SRSF1 or FANCD2 in HEK293T cells or ectopically expressed FLAG-SRSF1 confirmed the interaction between SRSF1 and FANCD2 and the effect of MMC on complex formation (Figures 2B, S2A, and S2B). Co-immunoprecipitation of FANCD2-SRSF1 was not affected by benzonase treatment, which digests both DNA and RNA, indicating that SRSF1-FANCD2 interaction is not bridged by nucleic acid (Figure 2B). To ascertain whether FANCD2 and SRSF1 interact directly, we the tested purified FANCI-His/FANCD2-FLAG (ID2-FLAG) complex and His-SRSF1 by affinity pull-down using anti-FLAG resin treated with benzonase (Figure 2C). Importantly, SRSF1 readily associated with ID2, indicating that the interaction between SRSF1 and ID2 is direct (Figure 2C, lane 3). We conducted a deletion analysis followed by cell-based interaction assays to determine which region of SRSF1 binds FANCD2. We generated four FLAG-tagged SRSF1 deletion mutants: an amino-terminal fragment (1–91) corresponding to RNA recognition motif 1 (RRM1); the inter-linker (92–120), involved in SRSF1 interaction with NXF1;¹⁵ a fragment (121–197) encompassing RRM2 and the carboxy-terminal RS phosphorylation domain (198–247). Our FLAG immunoprecipitation revealed that the FANCD2 binding motif of SRSF1 lies within RRM1 (Figure S2C).

Figure 2. SRSF1 physically interacts with the ID2 complex.

(A) FA-D2+FANCD2 and FA-D2+K561R cells were left untreated or treated with 0.5 μM MMC overnight. Whole-cell extracts were prepared and immunoprecipitated with FANCD2 antibody. The immunoprecipitation products were analyzed for FANCD2 interaction with SRSF1 antibody.

(B) Whole-cell extracts from FA-D2 mutant cells and FA-D2+FANCD2-corrected cells were prepared and immunoprecipitated with immunoglobulin G (IgG) or SRSF1 antibody. The immunoprecipitation products were analyzed for SRSF1 interaction with FANCD2 antibody (above). HEK293T cells were transfected with mock or FLAG-SRSF1 plasmid and treated or not with 1 μM MMC for 24 h. Whole-cell extracts were treated with benzonase (50 units/mL) and immunoprecipitated with FLAG-M2 agarose. The immunoprecipitation products were analyzed for SRSF1 interaction with FANCD2 antibody (bottom).

(C) Purified FANCI-His/FANCD2-FLAG (ID2-FLAG) and His-SRSF1 were analyzed by SDS-PAGE and Coomassie blue (top). The ID2-FLAG protein complex was incubated alone or with His-SRSF1 and treated with benzonase. Protein complexes were captured on FLAG resin, and the different fractions were analyzed by SDS-PAGE. S, supernatant; W, wash; E, SDS eluate of the FLAG resin (bottom).

See also Figure S2.

Binding of SRSF1 to RNA enhances FANCD2 monoubiquitination

We have reported previously that, in addition to DNA, ID2 can also bind single-stranded RNA (ssRNA) and R-loops.⁶ Furthermore, ssRNA and R-loops can support robust ID2 monoubiquitination.^6,25 To better understand the role of SRSF1 and FANCD2 in RNA metabolism, we directly compared the DNA and RNA binding activity of SRSF1 using an electrophoretic mobility shift assay (EMSA). Consistent with previous studies,²⁶ we found that SRSF1 binds ssRNA (Figure 3A). Interestingly, we found little or no affinity of SRSF1 for single-stranded DNA (ssDNA). This observation, together with the fact that SRSF1 silencing resulted in decreased FANCD2 monoubiquitination (Figure 1A), prompted us to test SRSF1 for the ability to enhance FANCD2 monoubiquitination in the presence of ssRNA or ssDNA.^6,25,27 We used our reconstituted in vitro ubiquitination system containing the E2-E3 enzymatic pair UBE2T and FANCB-FANCL-FAAP100 (BL100) complex.²⁸ We found that SRSF1 enhances FANCD2 monoubiquitination in the presence of ssRNA (p = 0.0302, p = 0.0201) but not ssDNA (p = 0.7684, p = 0.5014) severalfold within 30 min of incubation (Figure 3B). Notably, when the incubation was extended to 90 min, both conditions (with or without SRSF1, ssDNA/ssRNA) reached their maximum or plateau levels of ubiquitinated FANCD2 (ub-FANCD2) (Figure S3A). Finally, to study whether SRSF1 binding to RNA is necessary for enhancing ID2 monoubiquitination, we used site-directed mutagenesis to generate an RNA mutant version of SRSF1 (FLAG-SRSF1^W134A),¹⁵ and we transiently expressed this mutant in HeLa cells after depletion of endogenous SRSF1 (Figure S3B). Our results showed that SRSF1^W134A did not enhance ID2 ubiquitination as efficiently as the wild type, suggesting that RNA binding by SRSF1 is critical for efficient ID2 ubiquitination (Figure S3B, lane 4). Taken together, these results reveal a specific role of SRSF1 in the enhancement of RNA-dependent monoubiquitination of FANCD2 by the BL100 E3 ligase complex.

Figure 3. Binding of SRSF1 to RNA stimulates FANCD2 monoubiquitination.

(A) His-tagged SRSF1 (100–400 nM) was incubated with radiolabeled ssRNA or ssDNA. The mobility shift of the RNA and DNA was analyzed by EMSA (left). Graph shows the quantification of the shifted nucleic acid substrate. The error bars represent the mean and SEM of data from three independent experiments.

(B) In vitro ubiquitination reaction of recombinant ID2 with either ssDNA or ssRNA substrates and with or without SRSF1 (200 nM and 400 nM) (left). The graph shows the ratio of ubiquitinated to non-ubiquitinated FANCD2 (ub-FANCD2/FANCD2). Each nucleic acid has been normalized to its nucleic acid control without SRSF1 (3 and 4 to 2; 6 and 7 to 5). The error bars represent the mean and ± SEM of data from three independent experiments. Statistics: unpaired t test (right).

*p < 0.05, **p < 0.01, ***p < 0.00, ****p < 0.0001.

See also Figure S3.

DNA damage-induced co-localization of SRSF1 and FANCD2 at a transcribed genomic site

Recent work has demonstrated the involvement of the FA pathway in R-loop regulation upon transcription and replication stress.^3,5,6 Furthermore, R-loop accumulation has been postulated as an endogenous source of genomic instability in FA cells.^3,5,6 We used the S9.6 antibody directed against RNA:DNA hybrids in slot-blot analysis to quantify intracellular R-loop levels. We used RNaseIII and RnaseT1 pretreatment, which eliminates double-stranded RNA (dsRNA) that could complicate data analysis and interpretation. We found that FA-D2 and FA-D2+K561R mutant cells exhibit increased R-loops compared with FA-D2+FANCD2 cells (Figure 4A) (p = 0.0236, p = 0.0073). As expected, pretreatment with RNaseH1, which specifically digests the RNA strands in RNA-DNA hybrids, led to a decreased signal in the slot blot analysis. Consistent with published results,^3,6,8,16 we observed R-loop accumulation in the nucleus of HeLa cells depleted of SRSF1 (p < 0.0001) or FANCD2 (p < 0.0001), as visualized by S9.6 immunofluorescence (Figures 4B and S4A). Importantly, double siSRSF1+si-FANCD2 knockdown did not further increase S9.6 nuclear intensity compared with siFANCD2 or siSRSF1 single knockdown (p < 0.0001), again suggesting that FANCD2 and SRSF1 act epistatically in the same pathway to prevent the R-loops. We verified that the S9.6 signal could be abolished by RNaseH1 treatment. Slot blot analysis was also conducted to complement our immunofluorescence experiments. Here, we observed a strong increase in R-loops in siFANCD2 (p = 0.0002), siSRSF1 (p = 0.0006), or double knockdown cells (p = 0.0003) (Figure 4C). Consistent with the R-loop increase, we observed a substantial increase in γH2AX foci (a DNA damage marker) in cells depleted of SRSF1 by immunoblot and immunofluorescence (Figure S4B) (p = 0.0147). We have reported previously that FANCD2 colocalizes with R-loops in actively transcribed genomic sites using the DNA damage at RNA transcription (DART) system⁶ (Figure 4D). We used the DART system to measure how transcription and DNA damage influence SRSF1 recruitment. Briefly, the DART assay uses the KillerRed (KR) fluorescent protein chromophore, which generates superoxide at a genome-integrated tetracycline response element (TRE) upon visible-light irradiation to induce focal oxidative DNA damage.²⁹ Here, the transcriptional status of the TRE locus can be modulated by expressing a fusion protein of KR fused to either a transcription activator (TA-KR) or transcription repressor (tetR-KR). As expected by its transcription-related functions, we found SRSF1 accumulation in active transcription (TA-Cherry) (Figure 4D). Interestingly, similar to FANCD2,⁶ we saw enrichment of SRSF1 at sites of active transcription and DNA damage (TA-KR), known to be an R-loop hotspot, but less in cells with only active transcription (TA-Cherry), only DNA damage (tetR-KR), or neither transcription nor DNA damage (tetR-Cherry) (Figure 4D). Moreover, foci of FANCD2 or GFP-FANCD2 and SRSF1 colocalized in TA-KR-positive cells (Figures 4E and S4C), highlighting that the effects of FANCD2 and SRSF1 on R-loops are pronounced at hypertranscription sites coupled to DNA damage. Additionally, siRNA depletion of endogenous FANCD2 and SRSF1 significantly increased S9.6 focus intensity in TA-KR-positive cells to a similar extent as depletion of our positive control RNA helicase Aquarius (AQR)¹¹ (Figures 4F and S4D). Collectively, our results support the idea that ID2 and SRSF1 cooperate at actively transcribed genomic sites to prevent R-loop accumulation.

Figure 4. SRSF1 and FANCD2 colocalize at DNA damage within an actively transcribed genomic site prone to R-loop formation.

(A) Genomic DNA (0.6 mg) from FA-D2, FA-D2+FANCD2, and FA-D2+K561R cells was extracted, left untreated or treated with RNase H at 37°C for 1 h, and analyzed with slot blot assay using the S9.6 antibody. Methylene blue staining was used as a loading control (left). Densitometry of the slot blot shows the S9.6 intensity normalized to FA-D2+FANCD2 (right).

(B) HeLa cells transfected with siCTRL, siSRSF1, siFANCD2, or siSRSF1+siFANCD2 double knockdown were subjected to immunofluorescence to visualize RNA:DNA hybrids using the S9.6 antibody. RNAseH1 treatment was used as a control for S9.6 antibody specificity. Samples were co-stained with nucleolin to subtract the nucleolar S9.6 signal (top). The scatterplot shows the quantification of S9.6 intensity per nucleus after subtraction of the nucleolar signal. At least 65 cells per slide were analyzed (bottom).

(C) Genomic DNA from samples in (A) (0.3 or 0.6 μg) was left untreated or treated with RNAseH1 at 37°C for 1 h and analyzed with S9.6 slot blot. Methylene blue staining was used as a loading control (left). Densitometry of the slot blot shows the S9.6 intensity normalized to siCTRL (right).

(D) Schematic of the DART system (left). Immunostaining of SRSF1 was done in four U2OS TRE cell lines expressing different fusion effector proteins in a DART assay (center). Quantification of the average SRSF1 foci intensity of at least 50 cells per condition was performed (right).

(E) TA-KR-transfected U2OS TRE cells showing the colocalization of FANCD2 and SRSF1.

(F) siRNA depletion of FANCD2, SRSF1, or AQR increased S9.6 focus intensity in a DART assay (left). Shown is quantification of the average S9.6 focus intensity of at least 50 cells per condition (right).

Data represent the mean and SEM of three independent experiments. Statistics were performed using the unpaired t test. *p < 0.05, **p < 0.01, ***p < 0.00, ****p < 0.0001.

See also Figure S4.

Suppression of MMC sensitivity in SRSF1-depleted cells via RNAseH1 expression

RNaseH1 expression is able to suppress genomic instability associated with SRSF1 depletion.¹⁶ Additionally, R-loops have been proposed to induce genome instability in FA cells as well^3,5,6 (Figure 4A). To determine whether the FA phenotype of SRSF1-depleted cells is due to R-loop formation, we analyzed HeLa cells co-transfected with siCTRL or siSRSF1 as well as RNase H1-GFP or GFP-N1 empty vector. As observed previously (Figure 1A), cells co-transfected with siSRSF1 and the empty vector displayed decreased levels of FANCD2 monoubiquitination (Figure 5A, lane 4) (p = 0.0004) and foci following MMC treatment (Figures 5B and S5) (p = 0.0184). We obtained the same result for cells transfected with both siSRSF1 and RNAseH1 in terms of FANCD2 ubiquitination (Figure 5A, lane 8) (p = 0.0177) and MMC-induced FANCD2 foci (Figures 5B and S5) (p = 0.0327). Interestingly, we found that RNAseH1 expression alleviates the MMC sensitivity observed in SRSF1 knockdown cells without affecting FANCD2 levels (Figure 5C). Overall, our data suggest that the MMC sensitivity of SRSF1-depleted cells is caused by excessive R-loop accumulation in these cells.

Figure 5. RNase H1 overexpression rescues the MMC sensitivity observed in SRSF1-depleted cells.

(A) Immunoblot of HeLa cells co-transfected with siCTRL or siSRSF1 as well as RNase H1-GFP or GFP-N1 empty vector and untreated or treated with 1 μM MMC for 24 h. Ku86 was used as a loading control (left). Densitometry of the immunoblot shows ub-FANCD2/FANCD2. The data represent the mean and SEM of three independent experiments. Data were analyzed using the unpaired t test (right).

(B) Same cells from (A) were subjected to immunofluorescence using FANCD2 antibody. The graph shows the quantification of FANCD2 foci. Data (mean and SEM) are representative of three independent analyses of at least 50 cells per slide. Data were analyzed using the unpaired t test.

(C) HeLa cells were treated with the indicated concentrations of MMC, and cell survival was analyzed by crystal violet staining after incubation at 37°C for 4 days.

The percentages of surviving cells were normalized to the untreated control and are shown as the mean and SEM of three independent experiments.

Data were analyzed using two-way ANOVA. *, compared with siCTRL; #, compared with siSRSF1+RNAseH1. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

See also Figure S5.

RNA- and FANCD2-dependent interaction between SRSF1 and the mRNA export factor NXF1

SR proteins, such as SRSF1, bind the mRNA export factor NXF1 and act as adaptors for NXF1-dependent mRNA export.^12–15 Cells lacking functional mRNA export accumulate R-loops and R-loop-dependent genome instability phenotypes.^30–32 Consistent with this premise, S9.6 slot blot revealed increased R-loop levels upon NXF1 silencing (Figure S6A) (p = 0.0058). Given that FANCD2 and SRSF1 interact and impact R-loop metabolism, we surmised that FANCD2 might be involved in NXF1-SRSF1-mediated mRNA export. Importantly, we found, by anti-FLAG immunoprecipitation with cells expressing FLAG-SRSF1, that SRSF1, FANCD2, and NXF1 form a complex (Figure 6A). Consistent with previous results, we found that RNAse treatments abolished NXF1 and SRSF1 interaction¹⁴ but not SRSF1 and FANCD2 interaction (Figures 6A and S6B). In addition, in vitro FLAG immunoprecipitation, using purified ID2-FLAG and His-Trx-NXF1 proteins exclude any possible direct interaction between FANCD2 and NXF1 (Figure S6C). These results indicate that, whereas FANCD2 and SRSF1 interact directly, SRSF1-NXF1 interaction depends on RNA. Finally, we analyzed whether FANCD2 is required for NXF1-SRSF1 interaction by endogenous SRSF1 immunoprecipitation in FA-D2, FA-D2+FANCD2, and FA-D2+K561R cells. We found that SRSF1 and NXF1 only interacted in cells transduced with wild-type FANCD2 (Figure 6B), indicating that SRSF1 and NXF1 interact in a FANCD2-dependent fashion and that FANCD2 ubiquitination is important for this interaction. In line with this finding, we observed that SRSF1-NXF1 interaction is enhanced when promoting FANCD2 monoubiquitination by MMC treatment (Figure 6C). Altogether, these results support the contribution of FANCD2 in SRSF1-NXF1-mediated mRNA export.

Figure 6. FANCD2 monoubiquitination is necessary for NXF1-SRSF1-mediated mRNA export.

(A) HeLa cells were transfected with mock or FLAG-SRSF1 and treated or not with RNAse A. Whole-cell extracts were immunoprecipitated with FLAG-M2 agarose, and immunoprecipitation products were analyzed for SRSF1 interaction with FANCD2 and NXF1 antibodies.

(B) Whole-cell extracts from FA-D2 mutant cells, FA-D2+FANCD2, and FA-D2+K561R cells were prepared and immunoprecipitated with IgG or SRSF1 antibody. The immunoprecipitation products were analyzed for SRSF1 interaction with NXF1 antibody.

(C) Whole-cell extracts from FA-D2+FANCD2 cells were treated or not with increasing concentrations of MMC for 24 h and immunoprecipitated with IgG or SRSF1 antibody. The immunoprecipitation products were analyzed for SRSF1 interaction with NXF1 antibody.

(D) Schematic of the approach followed to detect binding of FANCD2, NXF1, and SRSF1 to mature RNA by appending MS2 trap (6× MS2 stem loops) at the end of the 3′ UTR of targets, creating a downstream chimeric RNA luciferase (left). Immunoblotting was used to detect RNA binding proteins (RBPs) by FLAG-MS2-BP immunoprecipitation. β-Actin was used as a control for loading and to assess specificity of the immunoprecipitation (IP).

(E) Immunoblot showing FLAG-MS2-BP immunoprecipitation in FA-D2 mutant, FA-D2+FANCD2, and FA-D2+K561R cells. β-Actin was used as control for loading and to assess the specificity of the IP.

(F) FA-D2 mutant, FA-D2+FANCD2, and FA-D2+K561R cells showing the translocation of poly(A) RNAs detected by RNA FISH (oligo-dT probe, red). Using image analyses, the poly(A) signal in the nucleus and cytoplasm was quantified to calculate the nuclear/cytoplasmic (N/C) ratio. Each circle in the graph represents the mean of the poly(A) N/C ratio of at least 15 cells, and the mean ( ± SEM) is also indicated. The data correspond to three independent experiments and were analyzed using the unpaired t test (right).

(G) Graph showing the N/C ratio of the mRNA targets identified by RNA-seq in FA-D2, FA-D2+FANCD2, and FA-D2+K561R cells. Each circle in the graph represents the N/C value of a specific mRNA target.

Data were analyzed using unpaired t test compared with FA-D2+FANCD2 wild-type cells. *p < 0.05, **p < 0.01, ***p < 0.00, ****p < 0.0001.

See also Figure S6.

FANCD2 monoubiquitination is necessary for NXF1-SRSF1-mediated mRNA export

Given the importance of FANCD2 in the SRSF1-NXF1 interaction, we asked whether FANCD2 is necessary for SRSF1-NXF1-mediated mRNA export. Because splicing is required for efficient mRNA export,^33,34 we first confirmed, by RNA-seq and rMATS, that FANCD2 depletion does not cause generalized splicing defects (Figure S6D). Next, we used affinity purification of MS2 aptamer 3′ UTR-tagged mRNA, which allows the systematic characterization of ribonucleoprotein (RNP) complexes formed on a given mRNA of interest to test whether FANCD2, SRSF1, and NXF1 bind in complex to mature mRNA (Figure 6D; STAR Methods).³⁵ To this end, we transfected HEK293T cells with pMS2-LUC-3′ UTR and performed anti-FLAG immunoprecipitation to isolate FLAG-MS2 bound mRNA. We identified FANCD2, SRSF1, and NXF1 in complex bound to mature mRNA, supporting their role in mRNA export (Figure 6D). Interestingly, FANCD2-deficient FA-D2 cells exhibited decreased SRSF1 and NXF1 recruitment to mature mRNA, suggesting that FANCD2 is important for the recruitment of SRSF1 to mRNA and subsequent coupling of NXF1 (Figure 6E). Importantly, FA-D2+K561R mutant cells also exhibit a reduction of SRSF1 and NXF1 association with mature mRNA. Transcript levels of luciferase were similar, indicating that an equal amount of mRNA was precipitated under each condition (Figure S6E). We validated the capture of FANCD2-SRSF1-NXF1 to mature mRNA by using an mRNA capture assay with oligo(dT) beads and UV crosslinking. FANCD2, SRSF1, and NXF1 were detected in UV-irradiated samples but not in non-crosslinked controls. PABP1 (poly(A) binding protein 1) was probed as a positive control and β-actin and SRSF2 as negative controls (Figures S6F and S6G). Together, we conclude that binding of FANCD2 to SRSF1 is important for the recruitment of the SRSF1-NXF1 complex to mRNA.

To directly probe whether FANCD2 depletion would have an impact on mRNA export, we used AF647-labeled oligo(dT)₃₀ to monitor poly(A) RNA localization by fluorescence in situ hybridization (FISH) in FA-D2, FA-D2+FANCD2, and FA-D2+K561R cells. As a control, cells were incubated with AF647-oligo(dA)₃₀ to demonstrate the specificity of the probe (Figure S6H). The results showed that, in FA-D2 (p < 0.0001) and FA-D2+K561R mutant cells (p < 0.0001), the nuclear/cytoplasmic (N/C) ratio of the poly(A) signal is higher than in FA-D2+FANCD2 cells, indicating stronger accumulation of poly(A) mRNA in the nucleus of mutant cells (Figure 6F). Amnis imaging flow cytometry showed a higher similarity score between poly(A) and DAPI in FA-D2 mutant cells, reflecting more nuclear poly(A) and, thus, confirming the result observed with RNA FISH (Figure S4I) (p = 0.0380). Interestingly, RNA FISH in FANCA-deficient cells also showed decreased mRNA export, demonstrating the importance of FANCD2 ubiquitination in mRNA export (Figure S6J) (p < 0.0001).

Next, we analyzed nuclear and cytoplasmic RNA from FA-D2, FA-D2+FANCD2, and FA-D2+K561R cells by RNA-seq, and we found mRNA accumulation in the nucleus of FA-D2 and FA-D2+K561R mutant cells compared with the wild type, thus confirming the result observed with RNA FISH (Figures 6G and S6K) (p(FA-D2) = 0.0103 and p(K561R) = 0.0004). Additionally, we selected the two top mRNA targets accumulated in the nucleus of FA-D2 cells (PNN [Pinin], a desmosome-associated protein) and DDX17 (DEAD-box helicase 17), and we compared the binding of FANCD2, SRSF1, and NXF1 with these mRNA targets in FA-D2 and FA-D2+FANCD2 cells by using RIP (RNA immunoprecipitation) (Figure S6L). ERP29 (endoplasmic reticulum protein 29), found to be unchanged between FA-D2 and FA-D2+FANCD2 cells in the RNA-seq, was used as a negative control. RIP showed that FANCD2 bound efficiently to PNN and DDX17 mRNAs, suggesting that they are targets for FANCD2-mediated export (Figure S6L, left). In addition, SRSF1 and NXF1 bind more efficiently to PNN and DDX17 mRNAs in the presence of FANCD2 (Figure S6L, center and right), confirming that FANCD2 is important for the recruitment of SRSF1-NXF1 to mRNA. Taken together, these results indicate that FANCD2 depletion or inefficient ubiquitination impairs mRNA export, suggesting that SRSF1 and FANCD2 cooperate in the prevention of R-loop formation by promoting mRNA export.

SRSF1 cancer-associated mutants are defective for FANCD2 interaction and monoubiquitination

Splicing factors have been found to be dysregulated in several types of hematological malignancies and cancer. In particular, SRSF1 is altered in many cancer types, such as breast, lung, and ovarian cancer and acute myeloid leukemia.^18,36–40 Using the Catalog Of Somatic Mutations in Cancer (COSMIC)⁴¹ database, which collects somatic mutations found in human tumors, we selected four different SRSF1 missense mutations linked to FA-associated cancers: E60D (acute myeloid leukemia), P89S (breast cancer), R97Q (serous ovarian cancer), and W134C (adult T cell lymphoma-leukemia). Figure S7A shows the mapping of these residues on the AlphaFold SRSF1 model. Using site-directed mutagenesis, we introduced each of these amino acid changes into SRSF1-GFP or FLAG-SRSF1, and we transiently expressed these mutants in HeLa cells after depletion of endogenous SRSF1. We found that the mutants, in contrast to the wild type, failed to restore mono-ub-FANCD2 levels (Figure 7A) (p(E60D) = 0.0347, p(P89S) = 0.0175, p(R97Q) = 0.0134, p(W134C) = 0.0277), MMC-induced FANCD2 foci (Figures 7B and S7B) (p(E60D) = 0.0430, p(P89S) = 0.0327, p(R97Q) = 0.0361, p(W134C) = 0.0356), and MMC resistance (Figure 7C), indicating that SRSF1 cancer-associated mutant expression cause an FA-like cellular phenotype. We excluded any potential indirect effect of the tested SRSF1 mutants on the observed phenotype by showing little or no change in global gene expression and splicing upon SRSF1 mutant expression (Figure S7C). We next analyzed whether these mutants interact with FANCD2 and NXF1 in cells by anti-FLAG immunoprecipitation. Our result revealed that the mutants are impaired for interaction with both FANCD2 and NXF1 compared with wild-type SRSF1 (Figure 7D). We next purified His-SRSF1 E60D and P89S proteins (Figure S7D) and tested them for ID2 interaction by in vitro FLAG immunoprecipitation. Consistent with cell-based assays, we found decreased interaction between SRSF1 E60D and P89S mutants and ID2 compared with the wild type (Figure 7E). Additionally, we found that, whereas E60D mutant bound ssRNA as efficiently as wild-type SRSF1, the P89S mutant failed to bind ssRNA (Figure 7F). Finally, we evaluated the ability of SRSF1 E60D and P89S mutant proteins to enhance FANCD2 monoubiquitination in vitro in the presence of ssRNA. We found decreased FANCD2 monoubiquitination with both mutants compared with the SRSF1 wild type (Figure 7G) (p(E60D) = 0.0117, p(P89S) = 0.0080). These results prompted us to ask whether the SRSF1 mutant proteins are impaired for mRNA export. As revealed by RNA FISH and RT-qPCR of the specific mRNA targets (PNN and DDX17) identified in our RNA-seq (Figure 6G), we found that cells expressing the SRSF1 mutants exhibit decreased mRNA export (Figures 7H and S7E) (p(E60D) < 0.0001, p(P89S) < 0.0001, p(R97Q) < 0.0001, p(W134C) < 0.0001). Accordingly, cells expressing the SRSF1 mutants exhibit enhanced R-loop accumulation (Figure 7I) (p(E60D) = 0.0191, p(P89S) = 0.0038, p(R97Q) = 0.0227, p(W134C) = 0.0494) compared with wild-type SRSF1. Of note, splicing analysis using RNA-seq and multivariate analysis of transcript splicing (rMATS) did not show general splicing changes among cell types depleted for endogenous SRSF1 and expressing FLAG-SRSF1 wild-type, E60D or P89S (Figure S7F), suggesting that expression of the mutants does not affect global splicing. These results demonstrate that SRSF1 E60D and P89S cancer-associated mutants exhibit impaired interaction with FANCD2, leading to inefficient FANCD2 monoubiquitination, impaired SRSF1-NXF1 complex formation, and consequential accumulation of R-loops. Altogether, our data suggest that FANCD2-SRSF1-NXF1-mediated mRNA export function may be disrupted in different tumor types.

Figure 7. SRSF1 cancer-associated mutants are defective for FANCD2 interaction and monoubiquitination.

(A) Immunoblot of HeLa cells co-transfected with siCTRL or siSRSF1 and mock, SRSF1-GFP wild type, or the indicated mutants and left untreated or treated with 1 μM MMC for 24 h. Ku86 was used as a loading control (left). Densitometry of the immunoblot shows ub-FANCD2/FANCD2. The data represent the mean and SEM of three independent experiments. Data were analyzed using the unpaired t test (right).

(B) The same cells from (A) were subjected to immunofluorescence using FANCD2 antibody. The graph shows the FANCD2 focus quantification. Data (mean and SEM) are representative of three independent analyses of at least 50 cells per slide. Data were analyzed using the unpaired t test compared with MMC-treated siCTRL.

(C) HeLa cells co-transfected as in (A) were treated with the indicated concentrations of MMC, and cell survival was analyzed by crystal violet after incubation at 37°C for 4 days. The percentages of surviving cells were normalized to the untreated control and are shown as the mean and SEM of three independent experiments. Data were analyzed using two-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. *, #, ^, ~, compared with siCTRL).

(D) HeLa cells were co-transfected with mock, FLAG-SRSF1 wild type, or mutants and treated with 1 μM MMC for 24 h. Whole-cell extracts were prepared and immunoprecipitated with FLAG M2 agarose. The immunoprecipitation products were analyzed for SRSF1 interaction with FANCD2 or NXF1 antibodies.

(E) Purified His-SRSF1 wild type or E60D or P89S mutants were incubated alone or with FANCI-His/FANCD2-FLAG protein complex and treated with benzonase. Protein complexes were captured on FLAG resin, and the different fractions were analyzed by SDS-PAGE.

(F) In vitro ubiquitination reaction of recombinant ID2 with ssRNA substrate and with His-SRSF1 wild type and E60D and P89S mutants (400 nM) (left). The graph shows ub-FANCD2/FANCD2 normalized to the control without SRSF1 (lane 2). The error bars represent the mean and ± SEM of data from three independent experiments. Statistics: unpaired t test (right).

(G) HeLa cells co-transfected with siCTRL, siNXF1, or siSRSF1 and mock, FLAG-SRSF1 wild type, or the indicated mutants, showing the translocation of poly(A) RNAs detected by RNA FISH (oligo(dT) probe, red) (left). Using image analyses, the poly(A) signal in the nucleus and cytoplasm was quantified to calculate the N/C ratio. Each circle in the graph represents the mean of the poly(A) N/C ratio of at least 15 cells, and the mean ( ± SEM) is also indicated. The data correspond to three independent experiments and were analyzed using the unpaired t test compared with siCTRL (right).

(H) Genomic DNA (0.6 μg) from HeLa cells co-transfected with siCTRL or siSRSF1 and mock, FLAG-SRSF1 wild type, or the indicated FLAG-SRSF1 mutants was extracted, treated with RNAseH1 at 37°C for 1 h or left untreated, and analyzed with S9.6 slot blot assay. Methylene blue was used as a loading control (left). Densitometry analysis of the slot blot shows the S9.6 intensity normalized to siCTRL. Results represent the mean and SEM of three independent experiments. Data were analyzed using unpaired t test, compared to siCTRL (right). *p < 0.05, **p < 0.01, ***p < 0.00, ****p < 0.0001.

See also Figure S7.

DISCUSSION

While the FA pathway has a well-characterized function in ICL repair, its recently described role in RNA metabolism remains to be defined mechanistically. In this study, we provide strong evidence that the FA pathway coordinates the prevention of R-loop formation and subsequent genomic stability through its interaction with SRSF1 via mRNA export regulation. Specifically, our results identify a physical and functional relationship between ID2 and SRSF1, a splicing factor also linked to R-loop regulation, genomic instability, and cancer.^16,18 First, we demonstrate that SRSF1 activates the FA pathway by binding to ID2 and stimulating its monoubiquitination in the presence of RNA. Furthermore, our data reveal that FANCD2 ubiquitination plays a crucial role in facilitating the formation of the NXF1-SRSF1 nuclear export complex and subsequent mRNA export. Importantly, we found that cancer-associated mutations in SRSF1 are impaired for interaction with FANCD2 and NXF1, leading to decreased FANCD2 monoubiquitination, inefficient mRNA export, accumulation of R-loops, and an FA-like cellular phenotype. Therefore, our work uncovers a novel function of the FA pathway in the prevention of R-loop-mediated genome instability through its involvement in mRNA export.

Although R-loops form naturally and have functions in normal cellular physiology, their persistence or aberrant accumulation can lead to increased replication stress and genomic instability.^9–11 R-loop-induced genomic instability has emerged as a significant contributor to human diseases, and several proteins involved in R-loop metabolism, including FA factors, are associated with cancer predisposition. We have shown recently that ID2 binding to RNA and R-loops can activate the FA pathway by promoting ID2 monoubiquitination.⁶ These findings emphasize the intricate relationship between RNA processing and DNA repair, reflecting the dual role of the FA pathway in these interconnected activities. However, the exact mechanistic role of ID2 in regulating R-loops has not been fully elucidated. Here, we first provide evidence of a collaborative interaction between FANCD2 and SRSF1 in RNA-related processes, including R-loop regulation. We demonstrate that direct binding of SRSF1 to ID2 activates the FA pathway by stimulating ID2 monoubiquitination in the presence of ssRNA but not ssDNA. Importantly, our study demonstrates the importance of SRSF1-RNA binding for the stimulation of FANCD2 ubiquitination. We used an SRSF1 RNA binding mutant, SRSF1^W134A, which allowed us to demonstrate that both FANCD2-SRSF1 interaction and the RNA-binding ability of SRSF1 are crucial for promoting FANCD2 ubiquitination. The enhancement of ID2 ubiquitination only with ssRNA is consistent with the known RNA-associated functions of SRSF1.⁴²Accordingly, we observed increased R-loops in cells depleted of both FANCD2 and SRSF1 as well as colocalization of FANCD2 and SRSF1 in R-loop enriched regions, and epistasis in phenotypic analyses of these and of the FA pathway.

In spite of crystal structure modeling that posits the interaction of ID2 and FANCD2 with a clamping role on double-stranded nucleic acid,⁴³ the general single-stranded nature of RNA versus double-stranded DNA does not appear to be a discriminating feature of the promotion of ID2 ubiquitination. Indeed, we and others have demonstrated previously that various forms of each can stimulate ubiquitination in vitro.^6,25,27 While, in cells, the interaction of SRSF1 and FANCD2 is enhanced under DNA-damaging conditions and by the presence of wild-type FANCD2 as opposed to its corresponding ubiquitin-dead mutant, such differential binding may be enhanced by chromatin localization, and even the ubiquitin-dead mutant form of FANCD2 may be found in some amount on chromatin. In our hands, non-ub-FANCD2 as part of ID2 certainly binds SRSF1 in in vitro assays. We would speculate that the specific nature of SRSF1 stimulation of ID2 ubiquitination with single-stranded RNA suggests structurally why there is deviation from that implied by the crystal structure model.

Consistent with the finding that SRSF1 activates the FA pathway, we show that SRSF1 depletion from human cells mirrors the FA phenotype, as evidenced by decreased MMC-induced FANCD2 monoubiquitination and foci and marked MMC sensitivity. Of note, SRSF1 does not have a global effect on FANCD2 ubiquitination, which is not completely abrogated by SRSF1knockdown or by mutant SRSF1 expression. This is not surprising, given that FANCD2-SRSF1 cooperate in actively transcribed regions and that FANCD2 is also involved in replication fork protection and activation of homology-directed repair (HDR)-related pathways. Interestingly, we found that RNAseH1overexpression, which partially restores MMC sensitivity in FANCD2-deficient cells,⁶ can restore resistance to MMC sensitivity in SRSF1-depleted cells but not FANCD2 monoubiquitination levels. These effects appear to be primarily due to the downstream removal of excess R-loops through RNAseH1, thus reducing DNA damage. While our findings do not demonstrate a direct relationship between the FA pathway and other splicing factors, such as SRSF2, SRSF6, U2AF1, and U2AF1, indicating the specificity of SRSF1 toward the FA pathway, a previous study has shown interaction of ID2 with SF3B1.⁴⁴ Further investigation is needed to understand the involvement of other splicing factors or RNA binding proteins in the function and regulation of the FA pathway.

Efficient export of mature mRNAs is critical for maintaining cellular homeostasis and genetic integrity. In yeast, defective tethering of transcribed DNA to the proximity of the nuclear pore complex (gene gating) has been linked to R-loop formation.⁴⁵ In addition, disturbance of mRNA export has been associated with R-loop-mediated genomic instability.^2,30,31 For instance, the THO/TREX nuclear complex prevents R-loop accumulation by promoting assembly and nuclear export of messenger RNPs (mRNPs),³⁰ and the TREX2/BRCA2 (BRCA2 being a bona fide FA gene) complex also suppresses R-loops by promoting mRNA export.² In addition to its role as a splicing regulator, SRSF1 participates in mRNA export by promoting the recruitment of the NXF1 nuclear export factor to mRNA.^12–15 Here, we demonstrate that FANCD2 is a component of the SRSF1-NXF1complex. Our findings reveal that FANCD2, SRSF1, and NXF1 form a multiprotein complex associated with mature mRNA. More importantly, we establish that the monoubiquitination of FANCD2 plays a critical role in the recruitment of SRSF1 to RNA, the subsequent coupling of NXF1, and the mediation of mRNA export. Supporting this notion, we found that cells expressing the FANCD2 K561R mutant protein exhibit increased R-loops,^5,6 emphasizing the importance of ubiquitination in this process. The functional connection between mRNA export defects and R-loop accumulation is still to be deciphered. It is plausible that, in the absence of FANCD2, the aberrant accumulation of mRNA molecules in the nucleoplasm may increase the chances of de novo rehybridization with unwound DNA at replication, leading to the formation of R-loops. Consistent with this notion, we found increased R-loops in NXF1-depleted cells as well.

There is growing evidence of an intimate link between RNA processing pathways and the cellular response to DNA damage. Our finding that ID2 ubiquitination is involved in activating both ICL repair and mRNA export suggests that, rather than being separable from canonical DNA repair activities, the prevention and resolution of R-loops are intertwined with repair and the avoidance of transcription-replication conflicts. Notably, many FA proteins that participate in ICL repair, such as FANCM, FANCA, or BRCA1/BRCA2, have also been implicated in the prevention of R-loop accumulation.^2–4,7 The role of these DNA repair proteins in preventing R-loop accumulation is crucial for the avoidance of transcription-replication conflicts and now appears to be dependent on the interactions of some of these proteins with mRNA processing factors. One of the most intriguing findings from our study is that MMC-induced DNA damage not only enhances the interaction between FANCD2 and SRSF1 but also promotes the FANCD2-dependent interaction between NXF1 and SRSF1.

Oncogenic mutations in genes encoding splicing factors were found to induce R-loops and replication stress.⁴⁶ Even though missense mutations in SRSF1 have been found in different cancers, there is still limited knowledge regarding how these cancer-associated mutations alter SRSF1 function. Importantly, our analysis of E60D, P89S, R97Q, and W134C SRSF1 cancer-related mutations furnishes evidence that they affect the interaction of SRSF1 with FANCD2 and NXF1 in cells and in vitro. Asa consequence, FANCD2 ubiquitination is decreased, leading to inefficient mRNA export, increased R-loops, and an FA-like cellular phenotype. It is important to emphasize that expression of the analyzed mutants does not induce global changes in splicing, suggesting that the mutations in question mainly affect the mRNA export function of SRSF1. SRSF1 is a splicing factor that modulates both constitutive and alternative pre-mRNA splicing. Our study does not exclude the involvement of the splicing activity of SRSF1 but raises the possibility of the usage of common factors in both pre-mRNA processing and mRNA export. Taken together, our findings highlight the importance of FANCD2 and SRSF1 in the proper regulation of NXF1-mediated mRNA export and R-loop prevention and suggest that this function may be dysregulated in the pathophysiology of sporadic cases of cancer. Therefore, we speculate that the ub-FANCD2-SRSF1-NXF1 axis could represent a promising therapeutic target, not only for FA but also for other bone marrow failure syndromes marked by aberrant RNA metabolism (dyskeratosis congenita, Diamond-Blackfan anemia, and Shwachman-Diamond syndrome) and multiple cancer types. We propose that targeting members of this complex could potentially be accomplished via downregulation of SRSF1 and/or NXF1 without affecting conventional FANCD2-associated functions in replication fork protection or homologous recombination (HR) activity.

Limitations of the study

In this study, we show that mRNA export dysregulation by an impaired FANCD2-SRSF1-NXF1 complex results in nuclear R-loop accumulation and genomic instability. While mRNA export defects have been associated with R-loop accumulation, the exact functional link between these two processes remains unknown. Although we surmise that aberrant accumulation of mRNA in the nucleus increases the probability of rehybridization with the DNA template to enhance R-loop formation, further investigation is needed to fully validate this model.

SRSF1 modulates both constitutive and alternative pre-mRNA splicing. Here, we demonstrate that SRSF1 knockdown does not affect the splicing of FA pathway gene transcripts such as FANCD2. However, we cannot exclude the possibility that SRSF1 depletion induces the alternative splicing of other transcripts of genes involved in mRNA export, thus impacting the mRNA export efficiency. Additionally, SRSF1 perturbation may affect splicing of transcripts of genes that function in genome maintenance. Finally, our description of this novel pathway of genome maintenance does not exclude the participation of other functions engendered by its component proteins. For example, we note that ubiquitination of FANCD2 is not completely abrogated by SRSF1 knockdown or expression of mutant SRSF1, and it is also clear that DNA also stimulates FANCD2 ubiquitination, exposing its role in HDR and replication fork protection. This observation may explain a dichotomy of lesions in non-transcribed versus transcribed regions of the genome.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Gary M. Kupfer (gary.kupfer@georgetown.edu).

Materials availability

Further information and requests for resources and reagents should be directed to and will be fulfilled by Gary M. Kupfer (gary.kupfer@georgetown.edu). Plasmids generated for these studies will be provided upon request.

Data and code availability

• RNA-seq data have been deposited at Gene Expression Omnibus (GEO) under accession number GSE212471 and are publicly available as of the date of publication.

• This paper does not report original code.

• All other data supporting the findings of this study are available from the corresponding authors upon reasonable request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mammalian cell lines

HeLa, HEK293T, and U2OS TRE cells (female) used in DART assay were grown in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen), supplemented with 10% fetal bovine serum (FBS) (Invitrogen), 100 U/ml penicillin, and 100 μg/mL streptomycin (Invitrogen). The SV40-transformed FANCD2-deficient PD20 cells (FA-D2, GM16756, Coriell Institute, male) and its derived FANCD2 wild-type (FA-D2+FANCD2), and ubiquitin mutant (FA-D2+K561R) complemented cells, as well as FANCA-deficient cells (FA-A, GM6914, Coriell Institute, male) and its derived FANCA wild-type cells (FA-D2+FANCA) were cultured in DMEM containing 15% FBS, 100 U/ml penicillin, and 100 μg/mL streptomycin. Mouse hybridoma cell line (HB-8730, ATCC, sex unknown) was used to produce S9.6 antibody. All the cells were maintained in dark at 37°C in a 5% CO₂ incubator.

Insect cell lines

Sf9 insect cells (Thermo Fisher) were used to produce initial baculovirus and High Five insect cells (Thermo Fisher) were used as the host for protein expression.

Bacteria strains

Rosetta (DE3) (Novagen) was used as the strain for protein expression and DH5a (New England Biolabs) was used as the strain for plasmid cloning.

METHOD DETAILS

Plasmids, cloning procedures and site-directed mutagenesis

Plasmids encoding pEGFP-N1-RNAseH1, pet28a-6xHis-SRSF1 and pet-32a-His-Trx-NXF1 were generously provided by Dr. Manley (Columbia University, New York, USA), Dr. Kurumizaka (The University of Tokyo, Tokyo, Japan), and Dr. Fontoura (University of Texas Southwestern Medical Center, Texas, USA) respectively. The pMS2-LUC-3′UTR plasmid used in mRNP immunoprecipitation was previously described.³⁵ pcDNA-Flag-SRSF1 was obtained from the laboratory of Dr. Honglin Chen (University of Hong Kong, China) through Addgene (Plasmid #99021).⁴⁸ TA-KR, tetR-KR, TA-Cherry, and tetR-Cherry plasmids were previously described.²⁹ To generate SRSF1-GFP plasmid, SRSF1 from Flag-SRSF1 was cloned into pEGFP-N1 backbone using the EcoRI and BamHI restriction sites. FANCD2 fused with C-terminal 3xFlag and FANCI with C-terminal 6xHis was subcloned into the same MacroBac 438A vector. All PCR amplifications were carried out using high fidelity Pfu UltraII fusion HS DNA polymerase (Stratagene). Flag-SRSF1, SRSF1-GFP, and pet28a-6xHis-SRSF1 point mutations were created using Q5 Site-Directed Mutagenesis Kit (New England Biolabs), according to manufacturer’s directions. See Table S1 for oligonucleotide sequence information. Finally, Flag-SRSF1 W134A and Flag-SRSF1 deletion mutants (1–91, 92–120, 121–197 and 198–248) were synthesized at Gene Universal Inc (Newark, DE). All the new constructs generated were subjected to DNA sequencing (Keck DNA Sequencing Facility, Yale University).

Transfections and mitomycin C treatment

Plasmid transfections were carried out with X-tremeGENE 9 transfection reagent (Roche Diagnostics), according to the manufacturer’s protocol.

50 nM SRSF1, SRSF2, SRSF6, U2AF1, U2AF2, FANCD2, SRSF6, AQR (siGENOME SMARTpool, Dharmacon, M-018672-01, M-019711-00, M-016067-01, M-012325-01, M-012380-01, M-016376-02, M-016067-01, M-022214-01), or NXF1 (Ambion, ID:134693) siRNA pool and non-targeting control siRNA (siCTRL) (Dharmacon, D-001206-13-05) were transfected using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions. A second transfection was performed next day after first transfection to ensure maximum knockdown.

For rescue experiments cells were first transfected with an siSRSF1 targeting the 3′UTR of SRSF1 (CCAACAAGATAGAGTATAA) for 48 h and next day transfected with mock or Flag-SRSF1 wild type or mutants, SRSF1-GFP wild type or mutants, or RNaseH1-GFP plasmid constructs for 24 h. SRSF1 siRNA targets the 3′UTR of SRSF1 and therefore does not affect expression of the SRSF1 ORF expression construct.

Cells were treated with MMC (Sigma) at 0.5 μM for PD20 cells and 1 μM for HeLa or HEK293T cells for 24 h.

Immunoblot analysis

Cells were lysed in buffer containing 300 mMNaCl, 1% Triton X-100, 5 mMEDTA, 2 mMNa₃VO₄,2 mMNa₄O₇P₂, 0.02% NaN₃, and 50 mM Tris, pH7.4, with complete EDTA-free protease inhibitor cocktail (Roche). The protein concentration was determined with the DC Protein Assay (Bio-Rad). Protein samples were separated by 7.5% SDS-PAGE (for Ub-FANCD2 blot) or 4–15% SDS–PAGE denaturing gels (Bio-Rad) and transferred onto a nitrocellulose membrane (Bio-Rad). Membranes were blocked in TBS-T (TBS+0.1% Tween 20) with 10% milk for 1 h and incubated with the indicated primary antibodies overnight at 4°C in TBS-T 2% milk. Membranes were washed three times in TBS-T and then incubated at room temperature for 1 h with horseradish peroxidase (HRP)-linked secondary antibody (GE Healthcare, 1:3000) in TBS-T 2% milk. Finally, membranes were washed three times in TBS-T and developed by chemiluminescence (Supersignal West Pico Kit or Supersignal West Femto kit, Pierce). Fiji software⁴⁹ was used to quantify intensity of bands in immunoblots. Primary antibodies used were FANCD2 (Santa Cruz, sc-20022, 1:1000), Ku86 (Santa Cruz, sc-5280, 1:5000), SRSF1 (ThermoFisher, 32–450, 1:500), GFP (Santa Cruz, sc-9996, 1:1000), RNaseH1 (Abnova, H00246243,1:500), NXF1 (Abcam, ab129160, 1:2000), β-actin (Cell Signaling Technology, 4970, 1:3000), Flag-M2-HRP (Sigma, A8592, 1:2000), SRSF6 (ThermoFisher, PA5-51810, 1:1000), SRSF2 (ThermoFisher, PA5-12402, 1:1000), U2AF1 (Cell Signaling Technology, 13705, 1:1000), U2AF2 (ThermoFisher, A303-666A, 1:2000), PABP1 (Cell Signaling Technology, 4992S, 1:2000), γH2AX (EMD, Millipore, 05–636, 1:500) and AQR (ABclonal, A80590, 1:1000).

Quantitative RT-PCR

Total cellular RNA was extracted using TRIzol (Invitrogen) according to manufacturer’s instructions, and 1 μg of RNA and random hexamers were used to synthesize cDNA using SuperScript II reverse transcription (ThermoFisher). PCR primers were purchased from Keck Oligo Synthesis Resource (Yale University) and used in PCRs with KAPA SYBR FAST qPCR Master Mix (Kapa Biosystems) on a CFX96 Real-Time System instrument (Bio-Rad). Relative gene expression was calculated using the ΔΔCq method and normalized to GAPDH or MS2 expression. Values are represented as the fold change compared to the control transfection values (siCTRL). Primer sequences are available in Table S2.

Immunofluorescence and confocal microscopy and image analysis

Cells, grown on sterile coverslips, were fixed with 3.7% formaldehyde in PBS for 10 min, permeabilized with 0.2% Triton X-100 in PBS for 5 min, blocked for 1 h in blocking solution (3% BSA in PBS), and incubated with the indicated primary antibodies diluted in blocking solution overnight at 4°C. Primary antibodies used were FANCD2 (Santa Cruz, sc-20022, 1:1000), γH2AX (EMD, Millipore, 05–636, 1:500), and SRSF1 (ThermoFisher, 32–450, 1:500).

For staining of RNA:DNA hybrids, S9.6 antibody was purified from the mouse BALB/c hybridoma cells (gift of Tae Hoon Kim, Yale University), using NAb Protein A/G Spin Kit (ThermoFisher) according to the manufacturer’s instructions. Cells, grown on sterile coverslips, were fixed with cold methanol for 10 min at −20°C, washed once with PBS, and incubated or not with RNase H1 and RNase III (New England Biolabs) for 1 h at 37°C. Then cells were blocked for 1 h in blocking solution (3% BSA in PBS) and co-stained with 1 μg/mL of S9.6 and nucleolin (Abcam, ab50279, 1:800) antibodies in blocking solution overnight at 4°C.

Cells were then washed with PBS and incubated with secondary antibodies (Alexa Fluor 488-conjugated anti-mouse IgG, Alexa Fluor 488-conjugated anti-rabbit IgG and Alexa Fluor 594-conjugated anti-mouse/rabbit IgG, ThermoFisher, 1:400); Cy5-conjugated goat anti-mouse IgG (Jackson ImmunoResearch, 1:800) for 1 h at room temperature. Coverslips were washed with PBS and mounted onto microscope slides using Vectashield mounting medium with DAPI (Vectorlabs). Samples were analyzed using Leica SP5 Confocal Microscope equipped Leica LAS-X software. A Fiji image processing package⁴⁹ was used for image analysis. The nuclear mean gray value for S9.6, after subtraction of nucleolar signal, was measured for each condition. In the case of FANCD2, the average of foci number per cell were quantified. Approximately 50 cells were analyzed for each condition, and statistical significance was determined by unpaired t test. Error bars represent SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = non-significant.

Cell survival assay

Cells were seeded into 6-well plates at 1×10⁴ cells/well and incubated at 37°C for 24 h. After 48 h of silencing, MMC was added at the indicated concentrations and incubated at 37°C for 4 days. Surviving cells were fixed in 10% methanol/10% acetic acid for 5 min and stained with crystal violet (1% in methanol, Sigma) for 3 min. Plates were then rinsed in water and allowed to dry, and the stain was dissolved with 0.1% SDS in methanol for 1 h at room temperature. Dye concentration was measured in a microplate reader (BioTek) by absorbance at 595 nm and normalized to untreated control to calculate the percentage of survival. Statistical significance was determined by two-way ANOVA. Error bars represent SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = non-significant.

Co-immunoprecipitation

FANCD2-deficient and proficient cells were cultured on 15 cm² dishes and treated with or without 0.5 μM MMC overnight prior to harvest. Cells were lysed in lysis buffer (300 mM NaCl, 1.0% Triton X-100, 5 mM EDTA, 2 mM Na₃VO₄, 2 mM Na₄O₇P₂, 0.02% NaN₃, and 50 mM Tris, pH 7.4) with complete EDTA-free protease inhibitor cocktail (Roche). Following a 10 s sonication, extracts were cleared by centrifugation at 14,000 rpm for 15 min at 4°C. Supernatants were saved, and equal amounts of extract (2 mg protein) were used for each immunoprecipitation. In some cases, the lysates were treated with 20 U DNase I (Roche), RNAse A (ThermoFisher), or RNase H1 (New England Biolabs) for 15 min at room temperature and then 15 min at 37°C. Then, 2 μg antibody FANCD2 (Bethyl Laboratories, A302-174A), SRSF1 (ThermoFisher, 32–450), normal rabbit, or mouse IgG, (Santa Cruz) was added. After incubation overnight at 4°C with rotation, protein A beads (GE Healthcare) were added for another 4 h. Beads were washed four times, and the pellets were subjected to immunoblot analysis.

For Flag immunoprecipitation, HeLa or HEK293T cells were cultured on 10 cm² dishes, transfected with the indicated plasmids overnight and treated or not with 1 μM MMC overnight prior to harvest. Cells were treated or not with benzonase (50 unit/ml) and lysed in NET-2 lysis buffer (40 mM Tris pH 7.4, 150 mM NaCl, 0.05% NP-40, 5 mM MgCl₂, 1 mM PMSF) with complete EDTA-free protease inhibitor cocktail (Roche). Following 10 s sonication, extracts were cleared by centrifugation at 14,000 rpm for 15 min at 4°C. Supernatants were saved, and equal amounts of extract (1 mg protein) were incubated with 20 μL of anti-Flag M2 agarose (Sigma) 1.5 h at 4°C with rotation. Finally, beads were washed four times, eluted using 200 μg/mL of 3xFlag peptides (Sigma), and subjected to immunoblot analysis.

Immunoprecipitation of mRNP using MS2-trap

To detect interaction of RNA binding proteins (RBPs) to mature mRNA 3′UTR, we carried out mRNP immunoprecipitation using MS2-trap as previously described.³⁵Briefly, HEK293T cells at 70–80% confluence in 10 cm dishes were transfected with 10 μg of each of the pMS2-LUC-3′UTR with X-tremeGENE 9 transfection reagent (Roche Diagnostics), according to manufacturer’s instructions. The pMS2-LUC-3′UTR vector scheme is shown in Figure 6D. Forty-eight hours later, cells were harvested, lysed in NET2 buffer (40 mM Tris pH 7.4, 150 mM NaCl, 0.05% NP-40, 5 mM MgCl₂, 1 mM PMSF) supplemented with 40 U/ml of RNAsin (NEB) and complete EDTA-free protease inhibitor cocktail (Roche) and subjected to Flag immunoprecipitation using anti-Flag-M2 agarose (Sigma) as described above. Input and immunoprecipitation (IP) were analyzed by immunoblotting as detailed earlier.

S9.6 slot-blot

Genomic DNA was extracted using DNeasy Blood & Tissue kit (Qiagen) following manufacturer’s protocol. After quantification, different concentrations of DNA (300 ng or 600 ng) were incubated with 1:200 dilutions of RNase T1 (ThermoFisher), RNase III (New England Biolabs), and/or RNaseH1 (New England Biolabs) for 1 h at 37°C to avoid S9.6 antibody artifacts.⁵⁰ 200 μL DNA solution was loaded to Hybond N+ nylon membrane (GE Life Sciences) presoaked with ddH₂O using the Bio-Dot SF Microfiltration Apparatus (Bio-Rad). The membrane was then crosslinked in the UV Stratalinker 2400 (Stratagene) at the ‘‘Auto Crosslink’’ setting (1200 mJx1000) and incubated with Methylene Blue as a loading control. The membrane was washed in ddH₂O, blocked in 10% milk in TBS-T for 1 h and incubated with S9.6 antibody diluted in TBST-T 2% milk overnight at 4°C to detect RNA:DNA hybrids. Membranes were washed three times in TBS-T and then incubated at room temperature for 1 h with secondary antibody in TBS-T 2% milk. Finally, membranes were washed three times in TBS-T and developed by chemiluminescence (Supersignal West Femto kit, Pierce). Fiji software⁴⁹ was used to quantify intensity of bands in immunoblots.

Damage At RNA transcription (DART) assay

The DART system has been described previously.²⁹ Briefly, the KillerRed (KR) or mCherry fusion protein expression vectors TA-KR, tetR-KR, TA-Cherry or tetR-Cherry in pBroad3 backbone were transfected into genetically engineered U2OS TRE cells in 35 mm glass-bottom dishes (MatTek) in darkness for 36–48 h. DNA damage was induced by exposing cells to light (15 W Sylvania cool white fluorescent bulb) for 30 min and then allowed to recovered for 1 h. For staining of endogenous SRSF1 foci in DART system, cells were fixed in 4% paraformaldehyde for 15 min at room temperature, permeabilized by 0.3% Triton X-100 in PBS for 10 min and blocked by 3% BSA for 1 h at room temperature. Cells were incubated with the indicated primary antibodies overnight at 4°C. Next day cells were incubated with secondary antibody for 1 h at room temperature. Images were acquired using the Olympus FV1000 confocal microscopy system. The foci intensity was directly measured by arbitrary unit from using Fiji.⁴⁹ 50 cells were analyzed, and statistical significance was determined by unpaired t test. Error bars represent SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, non-significant.

RNA Fluorescence in situ hybridization

Cells growing onto coverslips were washed with PBS +2 mM MgCl₂ and fixed for 10 min in 4% formaldehyde +2 mM MgCl₂. Cells were incubated with 0.1% Triton X-100 + 3% BSA for 1 h and 10% formamide in 2x SCC was added for 5 min. Cells were hybridized overnight at 37°C in hybridization buffer containing 10% dextran sulfate, 2 mM vanadyl-ribonucleoside complex, 0.02% RNAse-free BSA, 500 μg E.coli tRNA, 2x SSC, 10% formamide and 6.25 nM of AF647-labelled oligo(dT)₃₀ or AF647-labelled oligo(dA)₃₀ (negative control) (Integrated DNA Technology). The next day, cells were washed three times with 10% formamide in 2xSCC for 30 min at 30°C, nuclei were counterstained with DAPI, and coverslips were mounted in mounting medium. The nuclear/cytoplasmic ratio of at least 50 cells per condition was calculated using Fiji software.⁴⁹ Statistical significance was determined by unpaired t test. Error bars represent SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, non-significant.

1.5x10⁶ cells/condition were harvested by trypsinization and same protocol was used for RNA FISH in cell suspension. After counterstaining with DAPI, cells were filtered for flow cytometry analysis using Amnis Imagestream-X MarkII imaging flow cytometer. Fluorescent images were collected (6000 events per condition) and analyzed using nuclear localization wizard (IDEAS software, Amnis) to quantify colocalization of poly(A) and DAPI measured by similarity. Statistical significance was determined by the unpaired t test. Error bars represent SEM, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = non-significant.

Protein expression and purification

The human ID2 complex was purified from High Five insect cells co-infected with baculoviruses that produce 6xHis-FANCI and 3xFlag FANCD2 as previously described.^6,25,51 Briefly, cells were lysed in buffer A (25 mM Tris-Cl pH 7.5, 10% Glycerol, 0.5 mM EDTA, 100 mM KCl, 0.01% IGEPAL, 1 mM DTT, 1 mM PMSF and protease inhibitors (5 mg/mL each of leupeptin, chymotrypsinogen, aprotinin, and pep-statin), cleared by ultracentrifugation for 1 h at 100,000 g, and purified using anti-FLAG M2 Affinity Agarose Gel (Sigma). The resin was poured into a column, washed with Buffer B (25 mM Tris-HCl pH 7.5, 10% glycerol, 0.5 mM EDTA, 75 mM KCl, 0.01% IGEPAL, 1 mM DTT), followed by protein elution with 0.3 mg/mL 3xFlag peptide in 10 mL buffer B. The eluate was mixed with 1.5 mL of nickel-NTA resin (QIAGEN) for 1 h in Buffer B with 20 mM imidazole. The resin was poured into a column and washed with Buffer B containing 20 mM imidazole, followed by protein elution with 300 mM imidazole in 10 mL buffer B. The eluate was concentrated to 0.5mL in an Ultra-4 centrifugal filter unit with 100 kD cutoff (Amicon) and subjected to size exclusion chromatography on a Superdex200 Increase GL column (GE) on an AKTA FPLC system in Buffer C (25 mM Tris-HCl pH 7.5, 10% glycerol, 0.5 mM EDTA, 150 mM KCl, 0.01% IGEPAL, 1 mM DTT). Fractions containing FANCI-FANCD2 heterodimer (~300kD) were pooled, concentrated in a 15 mL 100 kDA Centrifugal Filter (Amicon), and stored at −80°C.

FANCB-FANCL-FAAP100 (BL100) complex with N-terminally Flag-tagged FANCB was expressed in insect cells and purified as described previously.⁴⁷ Briefly, the BL100 complex, with N-terminally Flag-tagged FANCB, was expressed in insect cells using the multi-bacmid provided by Andrew Deans (University of Melbourne, Australia). Purification involved FLAG tag affinity chromatography and MiniQ (GE Healthcare) ion-exchange chromatography. A final step of size exclusion in Superdex 200 was incorporated using buffer A (20 mM Tris-HCl, pH 7.5, 10% glycerol, 0.5 mM EDTA, 0.01% Igepal, 1 mM DTT, and 150 mM KCl). HA-Ubiquitin and UBE1 were purchased from Boston Biochem.

The human SRSF1 wild type and mutant proteins were prepared using expression plasmids provided by the Kurumizaka laboratory (The University of Tokyo, Tokyo, Japan). The human NXF1 protein was prepared using an expression plasmid provided by Fontoura laboratory (University of Texas Southwestern Medical Center, Texas, USA). In brief, cultures of Escherichia coli strain Rosetta (DE3) (Novagen) transformed with pET-28a-6xHis-SRSF1 (WT, E60D or P89S) or pet-32a-His-Trx-NXF1 in LB medium were induced with 0.2 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) at OD₆₀₀ of 0.6 and incubated overnight at 16°C. Cells were harvested by centrifugation and lysed in Buffer B [50 mM Tris HCl, pH 8.0, 10% glycerol, 0.5 M KCl, 10 mM Imidazole, 1 mM DTT, 1 mM PMSF, and complete protease inhibitor cocktail (Sigma)]. The lysate was clarified by ultracentrifugation, followed by incubation with Nickel-NTA affinity chromatography (Qiagen) in Buffer B in presence of 2 mM ATP and 4 mM MgCl₂. The beads were washed with buffer W (50 mM Tris HCl, pH 8.0, 10% Glycerol, 0.5 M KCl, 20 mM Imidazole, 1 mM DTT, 2 mM ATP and 4 mM MgCl₂), followed by elution with Buffer E [50 mM Tris HCl, pH 8.0, 10% Glycerol, 0.1 M KCl, 200 mM Imidazole, 1 mM DTT, and complete protease inhibitor cocktail (Sigma)]. The eluted protein was further purified by ion-exchange chromatography using Hitrap-Q-HP (1mL) and size exclusion chromatography on Superdex 200 10/300 Increase GL column (Cytvia, GE) on an Akta FPLC system in Buffer C without imidazole. Peak fractions with monomer size for SRSF1 or NXF1 were collected and concentrated using 10 kDa (SRSF1) or 30 kDa (NXF1) MWCO Centrifugal Filter (Amicon), aliquoted, and stored at −80°C. Same procedure was used for the purification of SRSF1-E60D and SRSF1-P89S.

In vitro immunoprecipitation

6xHis-SRSF1 wild type, E60D or P89S, or His-Trx-NXF1 (1 μg) and 6xHis-FANCI/3xFlag-FANCD2 complex (1 μg) were incubated in 30 μL reaction buffer (25 mM Tris-HCl pH 7.5, 10% glycerol, 0.5 mM EDTA, 0.01% Igepal, 1 mM DTT, and 150 mM KCl) on ice for 30 min, and then with 20 μL anti-Flag M2 affinity resin (Sigma). After gentle mixing at 4°C for 1 h, the resin was washed three times with 30 μL of buffer and eluted using 500 μg/mL of 3xFlag peptides (Sigma). The supernatant (S), last wash (W), and SDS eluate (E) (10 μL each) were analyzed by 4–15% SDS-PAGE and Oriole Fluorescent Gel stain (Bio-Rad).

DNA/RNA mobility shift assay

6xHis-SRSF1 wild type, E60D, or P89S (100–400 nM) were incubated with radiolabeled ssDNA (2.4 μM nucleotides) or ssRNA (2.4 μM nucleotides) (Integrated DNA Technology) in 10 μL reaction buffer (25 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, 100 μg/mL BSA, 1.5 mM MgCl₂) at 37°C for 10 min. The reaction mixtures were resolved in 10% polyacrylamide gels in TBE buffer (40 mM Tris-HCl, pH 8.3, 45 mM boric acid and 1 mM EDTA) at 4°C. After gel drying, the radiolabeled DNA or RNA species were visualized and quantified by phosphorimaging analysis. Signal of the non-shifted substrate in each lane was quantified using Fiji⁴⁹ in order to calculate the percentage of the shifted substrates. Means were calculated from 3 replicates and error bars represent SEM. Oligonucleotide sequences are available in Table S3.

ID2 in vitro ubiquitination assay

Purified ID2 complex (0.2 μM) was incubated with 6xHis-SRSF1 wild type (200 nM or 400 nM), E60D (400 nM), or P89S (400 nM) for 30 min at 4 °C. Next, the protein mixture was incubated with 64-mer ssDNA or ssRNA (0.2 μM) (Integrated DNA Technology) in the reaction buffer (50 mM Tris-HCl, pH 7.4, 25 mM KCl, 4 mM MgCl₂, 2 mM ATP, 0.01 mg/mL BSA and 0.5 mM DTT) at 25 °C for 10 min, followed by the addition of 0.1 μM UBE1 (Boston Biochem), 0.3 μM UBE2T, 0.2 μM BL100, and 32 μM HA-ubiquitin (Boston Biochem), and then incubated for 30 min or 90 min at 25 °C in a final volume of 12.5 μL. After mixing with 6 μL of 8% SDS, 240 mM Tris-HCl, pH 6.8, 40% glycerol, 5% BME, and 0.04% bromophenol blue, samples were analyzed by 6% SDS-PAGE gel. Gels were stained with Oriole Fluorescent Gel stain (Bio-Rad) and analyzed using Fiji software.⁴⁹ Oligonucleotide sequences are available in Table S3.

Cellular fractionation

Cellular fractionation was performed as described in.⁵² Briefly, equivalent number of cells for the indicated cell lines were harvested and washed once in 1XPBS (Phosphate buffer saline). Cells were then lysed in lysis buffer (10 mM Tris pH 7.4, 150 mM NaCl, 0.15% Igepal CA-360) followed by incubation on ice for 5 min. Cell lysate was then layered on 2.5 volume of ice-cold sucrose buffer (10 mM Tris pH 7.4, 150 mM NaCl, 24% sucrose) and centrifuged at 3500xg for 10 min at 4°C. This results in a pellet (Nuclear) and supernatant (cytoplasmic). The nuclear pellet was washed once in 1 x PBS containing 0.5 mM EDTA and was resuspended in 1 mL Trizol (Ambion). Cytoplasmic fraction was cleared once by centrifugation at 14,000xg for 1 min and 200 μL of extract was used per mL of Trizol (Ambion). To control for variability and for normalization throughout RNA preparation 1 μL of 1:100 dilution of external RNA spike in mix1 (ERCC; Ambion; Cat # 4456740) was added to nuclear and cytoplasmic fraction in Trizol before proceeding to RNA extraction as per manufacturers protocol.

In vivo UV crosslinking and isolation of poly-A mRNA bound proteins

mRNP capture assay was performed as described previously.⁵³ Cells were UV-crosslinked and scraped from 150 mm dish and lysed in lysis buffer (10 mM Tris pH 7.5, 60 mM NaCl, 5 mM MgCl₂, 0.5 mM EDTA, 0.1 mM EGTA, 0.2% NP-40, and 1 mM DTT) with complete EDTA-free protease inhibitor cocktail (Roche). Following 10 s sonication, extracts were cleared by centrifugation at 14,000 rpm for 15 min at 4°C. Supernatants were saved, and equal amounts of extracts were incubated with 25 μL of magnetic oligo(dT) beads (New England Biolabs) 1 h at room temperature with rotation. Finally, captured mRNP was eluted from beads with elution buffer (10mM Tris pH 7.5, 1mM EDTA, complete EDTA-free protease inhibitor cocktail and 4 μL of RNAse A/T1 enzyme mix (Thermo Scientific)) for 30 min at 37°C. The elute was precipitated with 20% TCA and the pellet was washed with ice-cold acetone and resus-pended in SDS-PAGE loading buffer for western blotting.

RNA immunoprecipitation (RIP)-RT-qPCR

RIP was performed as previously described.⁵⁴ Briefly, cells were lysed in RIP Buffer (150 mM KCl, 25 mM Tris pH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% NP-40 substitute) with freshly added RNase inhibitor (100 U/mL) and protease inhibitor, sonicated and centrifuged to pellet insoluble cellular debris. Supernatants were saved and equal amounts of extract (2 mg protein) were used for each immunoprecipitation. Then, 2 μg antibody FANCD2 (Bethyl Laboratories, A302-174A), SRSF1 (Thermo Fisher, 32–450), NXF1 (Abcam, ab129160) or normal rabbit, or mouse IgG, (Santa Cruz) was added. After incubation overnight at 4°C with rotation, protein A beads (GE Healthcare) were added for another 4 h. RIP reactions were then washed three times with RIP Buffer and subjected to WB analysis or resuspended in Trizol for RNA extraction and RT-qPCR analysis as previously described. Relative RIP enrichments were calculated by normalizing across biological replicates by normalizing for input expression levels.

RNA-seq analysis for differential gene expression, differential isoform expression and differential splicing events

Total cellular RNA extracted using TRIzol (Invitrogen) was used to isolate polyadenylated RNA and preparation of strand-specific Illumina-compatible cDNA libraries (dUTP protocol) and sequenced on Novaseq platform (paired end 150 bp, 100 million reads). Reads were processed with Cutadapt to remove adapters and aligned with STAR aligner (PMID: 23104886). Differential expression analysis for both genes and isoform level quantification was performed using the Cuffdiff algorithm of the Cufflinks package (PMID: 22383036). Log₂ fold changes of (−1.5)-(1.5) and q-value <0.05 were used as cut offs to determine significant changes between experiment and control samples. Differential splicing was assessed using rMATS (Replicates Multivariate Analysis of Transcript Splicing). Differentially spliced events were determined by calculating the difference in their average inclusion levels (ΔPSI), using a cutoff of IncDifferences (−0.1)-(0.1) and FDR<0.05. Events were classified as: skipped exon (SE), alternative 3′ splice site (A3SS), alternative 5′ splice site (5SS), retained intron (IR), and mutually exclusive exon (MXE). For N/C ratio calculation, nuclear and cytoplasmic fractions were spiked with same quantity ERCC spike-in controls and RPKM for each transcript was normalized using ERCC standard curve calculated from ERCC abundance.⁵² Normalized values were used for calculation of N/C ratio for each sample.

Cell cycle

For analysis of the cell cycle distribution, cells were analyzed by harvesting the medium containing floating cells together with adherent cells. Cells were subsequently washed with PBS and fixed in 70% ice-cold ethanol. Cells were counterstained with propidium iodide (PI, Sigma) diluted in RNase-containing PBS. PI positive cells were analyzed using FCS Express 2 (De Novo Software).

Click-IT-EdU

For analysis of cell proliferation, cells, grown on coverslips, were labeled with 10 μM EdU for 1 h before fixation and processed for Click-iT EdU reaction following manufacturer’s protocol (Alexa Fluor 647 Azide Dye, ThermoFisher). The cells were then mounted onto microscope slides using Vectashield mounting medium with DAPI (Vectorlabs). Samples were analyzed using Leica SP5 Confocal Microscope equipped Leica LAS-X software. The percentage of positive EdU cells was calculated using a Fiji image processing package.⁴⁹ Approximately 400 cells were analyzed for each condition, and statistical significance was determined by unpaired t test. Error bars represent SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = non-significant.

Metaphase spreads

HEK293T cells were treated with MMC (25 ng/mL) for 24 h, arrested with Colcemid (Roche) at a final concentration of 100 ng/mL during the last 3 h of the treatments, harvested by trypsinization, swollen in 75 mM KCl for 30 min at 37 °C and fixed in methanol:- acetic acid 3:1. Aliquots of the cellular suspension were dropped onto microscope slides to obtain chromosome spreads, which were subsequently G-banded by trypsin and Wright’s stain (GTW). Image acquisition was performed on a Leica DM6000 B microscope equipped with a 100× immersion oil objective (NA 1.32). Quantification analysis was performed using CitoVision (Leica Biosystems). Approximately 70 cells were analyzed for each condition, and statistical significance was determined by unpaired t test. Error bars represent SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = non-significant.

QUANTIFICATION AND STATISTICAL ANALYSES

Statistical analysis was performed using Prism 8 (GraphPad). Differences in survival assays were analyzed by two-way ANOVA. Statistical differences in all other cases were determined by unpaired t test. Statistical differences denoted in the Figures were determined by comparison to the relevant control-treated samples unless otherwise indicated. In all figure legends, n represents the number of independent biological replicates. All quantitative data were presented as mean and SEM. In all cases: * = p < 0.05; ** = p < 0.01; *** = p < 0.001, **** = p < 0.0001, ns = non-significant.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FANCD2	Santa Cruz	Cat# sc-20022; RRID:AB_2278211
FANCD2	Bethyl Laboratories	Cat# A302-174A: RRID:AB_1659803
Ku86	Santa Cruz	Cat# sc-5280: RRID:AB_672929
SRSF1	Thermo Fisher	Cat# 32-450; RRID:AB_2533079
GFP	Santa Cruz	Cat# sc-9996; RRID:AB_627695
RNase H1	Abnova	Cat# H00246243-M01; RRID:AB_530236
NXF1	Abcam	Cat# ab129160; RRID:AB_11142853
b-Actin	Cell Signaling Technology	Cat# 4970; RRID:AB_2223172
Flag M2-HRP	Sigma	Cat# A8592; RRID:AB_439702
SRSF6	Thermo Fisher	Cat# PA5-41810; RRID:AB_2607949
γH2AX	EMD, Millipore	Cat# 05-636 RRID:AB_309864
Mouse monoclonal S9.6	This paper. Purified from mouse BALB/c cells	N/A
Nucleolin	Abcam	Cat# Ab50279; RRID:AB_881762
SRSF2	Thermo Fisher	Cat# PA5-12402; RRID:AB_2184941
U2AF1	Cell Signaling Technology	Cat# 13705, RRID:AB_2798297
U2AF2	Thermo Fisher	Cat# A303-666A, RRID:AB_11203192
PABP1	Cell Signaling Technology	Cat# 4992, RRID:AB_10693595
Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	Thermo Fisher	Cat# A-11001; RRID:AB_2534069
Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	Thermo Fisher	Cat# A-11008; RRID:AB_143165
Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 594	Thermo Fisher	Cat# A-11005; RRID:AB_2534073
Cy5 Goat Anti-Mouse IgG (H + L)	Jackson ImmunoResearch	Cat# 115-175-166; RRID:AB_2338714
ECL Rabbit IgG, HRP-linked whole Ab (from donkey)	GE Healthcare	Cat# NA934; RRID:AB_772206
ECL Mouse IgG, HRP-linked whole Ab (from sheep)	GE Healthcare	Cat# NA931; RRID:AB_772210
Bacterial and virus strains
Rosetta (DE3) Competent cells	Novagen	Cat# 70954
NEB 5-alpha Competent E. coli (High Efficiency)	NEB	Cat# C2987I
Chemicals, peptides, and recombinant proteins
Mitomycin C (MMC)	Sigma	Cat# M4287
Crystal Violet	Sigma	Cat# C0775
X-tremeGENE 9	Roche	Cat# 06365787001
Lipofectamine RNAiMAX	Invitrogen	Cat# 13778150
3x Flag peptide	APExBIO	Cat# A6001
anti-FLAG M2 Affinity Agarose Gel	Sigma	Cat# A2220
Ni-NTA Agarose	QIAGEN	Cat# 30250
Imidazole	Sigma	Cat# I5513
Recombinant human FANCI-FANCD2	This paper/Liang et al., 20196	N/A
Recombinant human SRSF1 wild type	This paper	N/A
Recombinant human SRSF1 E60D	This paper	N/A
Recombinant human SRSF1 P89S	This paper	N/A
Recombinant human BL100	Liang et al., 201947	N/A
Recombinant human UBE2T	Longerich et al., 201425	N/A
Reombinant human NXF1	This paper	N/A
UBE1	Boston Biochem	Cat# E-305
HA-ubiquitin	Boston Biochem	Cat# U-110
Benzonase	Sigma	Cat# E1014
RNase H	NEB	Cat# M0297S
ShortCut RNase III	NEB	Cat# M0245S
RNase T1	Thermo Fisher	Cat# EN0541,
DNase I	Roche	Cat# 04716728001
RNase A	Thermo Fisher	Cat# EN0531
RNase Inhibitor (RNAsin)	NEB	Cat# M0314S
Oriole Fluorescent Gel stain	Bio-Rad	Cat# 1610496
Protein A Sepharose	GE Healthcare	Cat# 17-0780-01
Vectashield mounting medium with DAPI	Vectorlabs	Cat# H-1200
Critical commercial assays
Q5 Site-Directed Mutagenesis Kit	NEB	Cat# E0554S
NAb Protein A/G Spin Kit	Thermo Fisher	Cat# 89980
Propidium Iodide	Sigma	Cat# P4864
Click-IT-EdU (Alexa Fluor 647-Azide Dye)	Thermo Fisher	Cat# C10340
KaryoMAX Colcemid Solution	Thermo Fisher	Cat# 15212012
ERCC	Ambion	Cat#
Deposited data
RNA-seq data	This paper	GSE212471
Experimental models: Cell lines
Human: PD20 (FA-D2) cells	Coriell Institute	Cat# GM16756
Human: PD20 FANCD2 complemented cell lines	This paper	N/A
Human: PD20 (FA-A) cells	This paper	N/A
Human: PD20 (FA-D2) cells	This paper	N/A
Human: PD20 FANCA complemented	This paper	N/A
cell lines
HEK293T	ATCC	Cat# CRL-3216
Human: U2OS TRE cells	Lan et al., 201429	N/A
Mouse: S9.6 hybridoma cell line	ATCC	Cat# HB-8730
Insect: Sf9 cells	Thermo Fisher	Cat# 11496015
Insect: High Five cells	Thermo Fisher	Cat# B85502
Oligonucleotides
SRSF1 siRNA (CCAACAAGATAGAGTATAA)	Dharmacon	N/A
FANCD2 siRNA	Dharmacon	M-016376-02
SRSF6 siRNA	Dharmacon	M-016067-01
NXF1 siRNA	Ambion	ID:134693
AQR siRNA	Dharmacon	M-022214-01
SRSF2 siRNA	Dharmacon	M-019711-00
U2AF1 siRNA	Dharmacon	M-012325-01
U2AF2 siRNA	Dharmacon	M-012380-01
SRSF1 siRNA	Dharmacon	M-018672-01
non-targeting control siRNA	Dharmacon	D-001206-13-05
AF647-labelled oligo(dT)30	Integrated DNA Technology	N/A
AF647-labelled oligo(dA)30	Integrated DNA Technology	N/A
Oligonucleotide sequences used in site-directed mutagenesis (refer to Table S1)	This paper	N/A
Oligonucleotides used in RT-qPCR (refer to Table S2)	This paper	N/A
Oligonucleotides used for in vitro binding assay (Refer to Table S3)	This paper	N/A
Recombinant DNA
pEGFP-N1-RNaseH1	Gift of Dr. James Manley, Columbia University	N/A
TA-KR, tetR-KR, TA-Cherry and tetR-Cherry plasmids	Lan et al., 201429	N/A
Flag-SRSF1	Huang et al., 201748	RRID: Addgene_99021
SRSF1-GFP	This paper	N/A
pet28a-6xHis-SRSF1	Gift of Dr. Hitoshi Kurumizaka, The University of Tokyo	N/A
pMS2-LUC-3′UTR	Karmakar et al., 202235	N/A
pFastBac[6xHis-TEV-FANCI]	Longerich et al., 201425	N/A
pFastBac[3xFlag-FANCD2]	Longerich et al., 201425	N/A
Pet32a-His-Trx-NXF1	Gift of Dr. Fontoura (University of Texas Southwestern Medical Center).	N/A
Software and algorithms
FiJi	FiJi	Schindelin et al., 201249
Prism 8	GraphPad	RRID:SCR_002798
FCS Express 7	De Novo Software	RRID:SCR_016431
IDEAS software	Amnis (EMD Millipore)	N/A
CytoVision	Leica Biosystems	N/A

Highlights.

• SRSF1 activates the FA pathway by binding and stimulating FANCD2 ubiquitination

• FANCD2 ubiquitination is crucial for the formation of NXF1-SRSF1 export complex

• FANCD2 monoubiquitination is critical for NXF1-mediated mRNA export

• SRSF1 cancer mutants fail to bind FANCD2 and NXF1, leading to inefficient mRNA export