SIK2 kinase synthetic lethality is driven by spindle assembly defects in FANCA‐deficient cells

The Fanconi anemia (FA) pathway safeguards genomic stability through cell cycle regulation and DNA damage repair. The canonical tumor suppressive role of FA proteins in the repair of DNA damage during interphase is well established, but their function in mitosis is incompletely understood. Here, we performed a kinome‐wide synthetic lethality screen in FANCA −/− fibroblasts, which revealed multiple mitotic kinases as necessary for survival of FANCA‐deficient cells. Among these kinases, we identified the depletion of the centrosome kinase SIK2 as synthetic lethal upon loss of FANCA. We found that FANCA colocalizes with SIK2 at multiple mitotic structures and regulates the activity of SIK2 at centrosomes. Furthermore, we found that loss of FANCA exacerbates cell cycle defects induced by pharmacological inhibition of SIK2, including impaired G2‐M transition, delayed mitotic progression, and cytokinesis failure. In addition, we showed that inhibition of SIK2 abrogates nocodazole‐induced prometaphase arrest, suggesting a novel role for SIK2 in the spindle assembly checkpoint. Together, these findings demonstrate that FANCA‐deficient cells are dependent upon SIK2 for survival, supporting a preclinical rationale for targeting of SIK2 in FA‐disrupted cancers.

The Fanconi anemia (FA) pathway safeguards genomic stability through cell cycle regulation and DNA damage repair. The canonical tumor suppressive role of FA proteins in the repair of DNA damage during interphase is well established, but their function in mitosis is incompletely understood. Here, we performed a kinome-wide synthetic lethality screen in FANCA À/À fibroblasts, which revealed multiple mitotic kinases as necessary for survival of FANCA-deficient cells. Among these kinases, we identified the depletion of the centrosome kinase SIK2 as synthetic lethal upon loss of FANCA. We found that FANCA colocalizes with SIK2 at multiple mitotic structures and regulates the activity of SIK2 at centrosomes. Furthermore, we found that loss of FANCA exacerbates cell cycle defects induced by pharmacological inhibition of SIK2, including impaired G2-M transition, delayed mitotic progression, and cytokinesis failure. In addition, we showed that inhibition of SIK2 abrogates nocodazole-induced prometaphase arrest, suggesting a novel role for SIK2 in the spindle assembly checkpoint. Together, these findings demonstrate that FANCA-deficient cells are dependent upon SIK2 for survival, supporting a preclinical rationale for targeting of SIK2 in FA-disrupted cancers.

Introduction
Fanconi anemia (FA) is a complex, heritable bone marrow failure disorder characterized by genomic instability and cancer predisposition. Canonical FA pathway signaling is essential for DNA damage repair in response to different genotoxic insults [1,2]. Defective FA signaling impairs homologous recombination, resulting in genomic instability and increased risk of malignancy. Micronucleation and chromosome abnormalities in cells with dysfunctional FA pathway suggest that FA signaling provides an additional layer of protection from cancer by ensuring error-free chromosome segregation during mitosis. We and others have also implicated FA signaling throughout cell cycle, including roles in centrosome function [3,4], spindle assembly checkpoint (SAC) [4,5], resolution of ultrafine anaphase bridges [6], and cytokinesis [7,8].
The most commonly mutated gene in FA is FANCA [9,10], one of the members forming the FA core complex, which recognizes interstrand crosslinks and induces subsequent DNA repair [11][12][13][14]. Individually, FANCA is capable of catalyzing single-strand annealing and strand exchange as an alternative repair mechanism [15]. Beyond its role in interphase, FANCA regulates mitotic centrosome function and the SAC [3,5]. Previous studies have demonstrated that FANCA-deficient cells exhibit centrosome abnormalities, chromosomal bridges, micronuclei, and multinucleation [3,5]. Together, this evidence suggests diverse roles for FANCA throughout the cell cycle, including the regulation of mitosis.
Germline biallelic loss of FA genes is relatively rare, but germline heterozygous FA gene mutations are more common, with a subset causing haploinsufficiency, which that predisposes to cancer [16][17][18][19][20][21]. Several studies have reported monoallelic germline mutation of FANCA in familial breast cancers, suggesting that FANCA may also be associated with breast cancer susceptibility [22][23][24][25]. Furthermore, the Cancer Genome Atlas (TCGA) reports somatic disruption of the FA pathway in 30% of all cancers (Table 1) [26,27]. Germline loss of FA pathway function renders FA patients highly susceptible to lethal toxicities from front-line chemotherapy regimens [28]. Thus, novel synthetic lethal approaches are needed to synergize the anti-cancer effect of standard chemotherapeutics in these patients. Additionally, the high frequency of somatic mutations within the FA pathway in cancers of the general population broadens the clinical significance of such approaches. Most notably, the use of PARP inhibitors in the treatment of BRCA1-and BRCA2-deficient tumors exemplifies successful synthetic lethal targeting of the FA pathway [29][30][31].

shRNA library screen
Primary fibroblasts derived from a FANCA À/À patient were used for the screen. FANCA-deficient patient fibroblasts (FANCA À/À ) and their corrected counterparts (FANCA + ) were plated (two replicates/genotype) to 50% confluency in 15 cm plates (Falcon, Glendale, AZ, USA). The next day (day 1), cells were transduced per manufacturer's protocol with a lentiviral library (MISSIONÒ LentiExpress TM Human Kinases; Sigma-Aldrich) that encodes $ 5000 unique shRNA sequences targeting 513 human kinase genes (4-10 shRNA clones per gene), as well as nontargeting controls. Cells were transduced at a multiplicity of infection (MOI) of~0.5. A nontransduced plate served as a negative control. On day 3, cells were selected with 1 µgÁmL À1 puromycin media for 5 days and then allowed to grow for 2 weeks without puromycin. Cells were then pelleted and sent to Sigma for deep sequencing.
To determine the effect of each shRNA clone on cell viability, four raw copy numbers were determined for each clone representing two FANCA À/À replicates (1 and 2) and two FANCA + replicates (3 and 4). Four ratios were calculated from the raw data (1/3, 1/4, 2/3, and 2/4). Kinases with at least one shRNA showing a five-fold copy number fold change (increase or decrease) between FANCA + and FANCA À/À cells in all four comparisons were considered significant. A ratio of < 1 was considered as synthetic lethality. Genes with at least two shRNA sequences showing a ratio < 1 with at least one of them showing five-fold difference were chosen as potential targets for further analysis. Raw data were also processed using the Panther gene classification database to confirm significant results. Functional enrichment analysis was performed by using g:Profiler (ELIXIR) to map hits to biological processes, and pathway enrichment maps were created by Cytoscap [39]. For candidate hits, the z-score of each shRNA affecting cell viability in FANCA deficiency was calculated by comparing the deviation of the shRNA frequency in FANCA À/À fibroblast from its corrected counterpart.

MTT, CellTiter-Glo, trypan blue exclusion, and colony formation assays
For MTT assay, 8000 cells per well were plated on a 96-well plate. After indicated treatment, 100 µL fresh medium containing 10 µL of 5 mgÁmL À1 MTT (Sigma-Aldrich) was added and incubated at 37°C for 2 h. DMSO was used to dissolve the crystals formed, and the absorbance was measured at 570 nm through a microplate reader (VersaMax, Molecular Devices, San Jose, CA, USA).
For CellTiter-Glo cell viability assays (Promega, Madison, WI, USA), 2000 cells per well were plated in a 96-well plate. After the indicated treatment, the CellTiter-Glo reagent was added according to the manufacturer's instructions, and the luminescent signal was measured using Synergy H4 microplate reader (BioTek, Winooski, VT, USA).
For trypan blue exclusion assay, 10 000 cells per well were plated on a 24-well format, and after being treated with indicated conditions, they were trypsinized and stained with 0.4% trypan blue solution (Corning, Glendale, AZ, USA). The cell number was then counted with a hemocytometer.
For colony formation assays, 500 cells per well were seeded on a six-well plate and ARN-3236 was added at the indicated concentrations. After 10-12 days, the cells were fixed and stained with Giemsa solution (Merck, Kenilworth, NJ, USA) containing 50% methanol. Colonies were then manually counted.

Immunoblotting
Cell extracts were prepared by incubating in M-PER TM Mammalian Protein Extraction Reagent (Thermo Fisher, Waltham, MA, USA) with protease (Complete Mini, EDTA-free; Roche, Indianapolis, IN, USA) and phosphatase inhibitors (Pierce Phosphatase Inhibitor Mini tablets; Thermo Scientific) on ice. Lysates were then centrifuged at top speed in a microcentrifuge for 10 min to remove cellular debris. Protein concentrations were measured using Pierce TM BCA protein assay kit (Thermo Fisher), and samples were diluted with LDS sample buffer (NuPAGE; Life technologies, Carlsbad, CA, USA). Proteins were separated by electrophoresis on 4-12% Bis-Tris gels (Invitrogen) and transferred to PVDF membranes (Millipore, Burlington, MA, USA). After blocking in 5% nonfat milk, the membranes were probed with primary antibodies at 4°C overnight and then the corresponding secondary antibodies (IRDye, LI-COR) for another hour at room temperature before the images were captured by Odyssey CLx imager (LI-COR). IMAGE STUDIO 2.1 software was used to quantify protein levels by measuring the relative fluorescence intensities of bands (normalized against loading control). The primary antibodies used included rabbit anti-Akt (Cell Signaling Technology, Danvers, MA, USA), rabbit antiphospho-Akt (Cell Signaling Technology, Danvers, MA, USA), rabbit anti-Aurora B (Abcam, Cambridge, UK), mouse anti-b actin (Sigma), rabbit anti-GAPDH (Cell Signaling Technology), rabbit anti-Cyclin B1 (Cell Signaling Technology), rabbit anti-DDK (Cell Signaling Technology), rabbit anti-FANCA (Abcam), rabbit anti-SIK2 (Cell Signaling Technology), and rabbit anti-pS358 SIK2 (Kinexus, Vancouver, Canada).

Cell cycle analysis and phospho-Histone H3 (Ser10) staining
After indicated conditions with ARN-3236, cells were pelleted and fixed with 70% ethanol at 4°C overnight. For cell cycle analysis, FxCycle PI/RNase staining solution (Invitrogen) was used to stain the cells at 4°C overnight. For phospho-H3 staining, cell pellets were washed with PBS once and resuspended in PBS containing 0.25% Triton X-100 on ice for 15 min. The pellets were then stained with rabbit anti-phospho-Histone H3 antibody (1 : 1000; Abcam) in 1% BSA at 4°C overnight. After washing with PBS, the cell pellets were probed with anti-rabbit Alexa Fluor 488 secondary antibody (Invitrogen) for 1 h. The pellets were counterstained with propidium iodide (25 lgÁmL À1 ) before analysis by flow cytometry (FACSCanto II; BD Biosciences, Franklin Lakes, NJ, USA).

Live-cell imaging
Twenty thousand cells per quadrant were plated in a 35mm Hi-Q4 culture dish (Ibidi, Planegg, Germany). ARN-3236 was added, and time-lapse phase-contrast images (4 z-sections, 2 lm each) were acquired every 2 min for a total of 24 h using a BioStation IM-Q time-lapse imaging system (Nikon, Tokyo, Japan) fitted with a temperature-controlled growth chamber. Video analysis was done through NIS-ELEMENTS VIEWER 4.50 imaging software (Nikon), and Imaris (Bitplane, Z€ urich, Switzerland) was used to export the video.

Co-immunoprecipitation
Immunoprecipitation was performed using Pierce Crosslink Immunoprecipitation Kit (Thermo Scientific) according to the manufacturer's instruction with modifications. In brief, approximately 1 9 10 7 cells were lysed using lysis/wash buffer, and the lysates were precleared with control agarose resin for an hour at 4°C. Cell lysates were collected after centrifugation and added to Protein A/G Plus Agarose resin premixed with antibody in coupling buffer. The mixtures were incubated with agitation at 4°C overnight. The resin was then washed sequentially with lysis/wash buffer and TBS. The bound antigens were eluted by mixing the resin with sample buffer and incubated at 95°C for 5 min. The eluate was collected after centrifugation at 7200 g for 2 min.
The fluorescence signals were captured by a Deltavision Ultra microscope using 609 or 1009 lenses (GE Healthcare, Chicago, IL, USA). Images were acquired with z-section of 0.2 lm each and deconvolved using Softworx (GE healthcare). All images were processed with Imaris (Bitplane). Images in figures represent individual z-sections of deconvolved stacks. The 'interpolate zoom' function in Imaris was used to smooth images.

Double thymidine block and RO3306 induced G2 arrest
Cells were treated with 2.5 mM thymidine solution for 24 h. The cells were then released for 8 h and blocked again with 2.5 mM thymidine for another 16 h. Cells were released, and cell lysates were collected at different time points as indicated. For RO3306 treatment, 9 µM RO3306 was added to cells grown on coverslips for 20 h. After extensive washing, the cells were released and ARN-3236 was added for 4 h. The cells were fixed and processed as mentioned in fluorescence microscopy. The coverslips were stained with Alexa Fluor 594 phalloidin (Life technologies) for 2 h and then mounted.

Statistical analysis
GRAPHPAD PRISM8 (San Diego, CA, USA) was used to perform the statistical analyses, and P value < 0.05 was considered as significance. The exact P values with significance are denoted on the graph.

Identification of kinases crucial for survival of FANCA-deficient cells
Primary patient-derived FANCA À/À fibroblasts and their gene-corrected counterparts (FANCA + ) were used to carry out a lentivirus-based shRNA kinome screen (Fig. 1A). FANCA À/À and FANCA + cells were transduced with a kinome-wide lentiviral library containing approximately 5000 distinct shRNAs targeting over 500 human kinases. Cells were transduced with an MOI of 0.5 so that each cell should receive no more than one shRNA clone. Kinases with at least one shRNA with a 5-fold copy number change between FANCA + and FANCA À/À cells were deemed significant. Using this criterion, we identified 87 genes as essential to FANCA-deficient cell survival (i.e., the copy number of at least one shRNA targeting these kinases was five-fold less in FANCA À/À cells relative to FANCA + cells). Functional enrichment analysis was performed by using g:Profiler (ELIXIR) to map hits to biological processes, and pathway enrichment maps were created by Cytoscape (Fig. 1B) [39]. The identified gene list and their corresponding pathway enrichment analysis are shown in Data Set S1. Among these hits were multiple DNA damage response (DDR) kinases previously identified by others as FA synthetic lethal targets, including ATR [41], CHK1 [42,43], and CDK12 [44] (Fig. 1D,E). These independent findings consistently demonstrate the functional significance of FA genes in the DDR pathway and serve as internal validation of our experimental approach (Fig. 1A). In addition, ontology analysis showed enrichment of myriad biological process clusters, including cell cycle regulation, implying an increased reliance on mitotic governance in FANCA À/À cells (Fig. 1B,E). Z-score analysis of shRNA copy number revealed that most genes identified as synthetic lethal by five-fold exclusion clustered together with z-scores < À1.5 ( Fig. 1C; Data set S1). Hits with known mitotic functions included SIK2, which, like FANCA, has been specifically implicated in centrosome maintenance and cytokinesis [5,8,35,36]. Because of this functional overlap, we reasoned that further characterization of the synthetic lethal interaction between FANCA and SIK2 may provide mechanistic insight into the mitotic function of FANCA.
To validate that FANCA À/À cells are dependent upon SIK2 for survival, a pool of two screenindependent SIK2-targeting siRNAs was used to knock down SIK2 in FANCA À/À patient fibroblasts, as well as their gene-corrected counterparts ( Fig. 2A). FANCA À/À cells exhibited significantly decreased viability following SIK2 knockdown relative to FANCA + cells, as measured by MTT and trypan blue exclusion assays (Fig. 2B,C). To determine the effect of pharma- Workflow of kinome screen targeting > 500 human kinases conducted in FANCA À/À patient fibroblasts and their gene correction counterpart. Patient fibroblasts (FANCA À/À ) and the corrected fibroblasts (FANCA + ) were transduced with the lentivirus-based shRNA pool (> 5000 clones), cultured in puromycin for 5 days to select for transduced cells, and then allowed to grow for 2 weeks in puromycin-free media. The cells were collected, and the copy number of individual clones was analyzed by highthroughput sequencing. Two biological replicates for each FANCA genotype were used. (B) Node plot categorizing gene hits into significantly enriched biological processes (P < 0.01). The detailed gene list of each biological process is shown in Data Set S1. (C) Scatter Plot of z-score for each shRNA in the screen. Select synthetic lethal hits of interest (e.g., PLK1, SIK2, ATR, and Chk1) are indicated as enlarged, colored dots. The z-score for SIK2 is À1.53. (D) Representative candidate genes were grouped and categorized according to their functional roles in mitosis and several other biological processes. The values represented by the four columns in the heat map were generated by calculating the ratio of the copy number of each shRNA in one FANCA À/À replicate (samples #1 and #2) to that of the shRNA copy number in one FANCA + replicate (samples #3 and #4). Thus, the four columns represent the four ratios generated (i.e., sample#1/ sample#3, sample#1/sample#4, sample#2/sample#3, sample#2/sample#4). (E) A diagram demonstrating the link between cell cycle and other cellular pathways as well as how screen-identified candidate genes perform their cell cycle-related functions.
cological inhibition of SIK2 on FANCA À/À cell survival, we employed the SIK2-specific small molecule inhibitor ARN-3236. Following ARN-3236 treatment, SIK2 inhibition was confirmed by western blot analysis of SIK2 serine-358 phosphorylation (SIK2 pS358), a marker of SIK2 activity [38,45]. Indeed, ARN-3236-treated patient fibroblasts and HeLa cells exhibited reduced levels of SIK2 pS358 relative to DMSOtreated controls ( Fig. 2D and Fig. S1A). In addition, phosphorylation of AKT, a SIK2 downstream target, was also reduced in HeLa cells treated with ARN-3236, corroborating its inhibition of SIK2 kinase activity (Fig. S1B). ARN-3236 treatment significantly reduced colony formation of FANCA À/À cells relative to FANCA + cells (Fig. 2E,F). To validate this result in an independent cell type, we assessed colony formation of HeLa cells stably expressing FANCA-targeting shRNA or nontargeting shRNA following treatment with ARN-3236 (Fig. 2G). Similarly, ARN-3236 more severely decreased cell viability and colony formation in FANCA-knockdown cells relative to control cells ( Fig. 2H-J). We also established stable FAKO HeLa cells using CRISPR gene editing (Fig. 3A) and demonstrated that concomitant knockdown of SIK2 significantly decreased viability in HeLa-FAKO compared to HeLa CRISPR control (Fig. 3B,C). Consistently, HeLa-FAKO cells were hypersensitive to ARN-3236 (Fig. 3D). Since FA patients display a high risk of childhood acute myeloid leukemia (AML), we stably knocked down FANCA in the AML cell line THP-1 to assess sensitivity to SIK2 inhibition in a relevant cancer type. Importantly, knockdown of FANCA significantly sensitized THP-1 cells to ARN-3236 (Fig. 3E,F). Together, these findings demonstrate that FANCA-deficient cells are specifically sensitive to depletion or inhibition of SIK2.

Loss of FANCA exacerbates cell cycle defects induced by SIK2 inhibition
SIK2 depletion has been shown to arrest cells in G2 and block mitotic exit [35,36]. To determine at which stage of the cell cycle SIK2 inhibition impacts FANCA À/À cells, we analyzed cell cycle profiles of FANCA À/À vs FANCA + cells following ARN-3236 exposure. ARN-3236 treatment caused a small, but significant, increase in the percentage of FANCA + cells with 4N DNA content. This ARN-3236-induced 4N accumulation was dramatically increased in FANCA À/À cells (Fig. 4A,B). Phospho-histone H3 (Ser10), a mitotic marker, was significantly decreased in cells treated with ARN-3236 (Fig. 4C), suggesting that cells were blocked from entering mitosis by SIK2 inhibition. Notably, phospho-H3 was more significantly reduced in FANCA À/À cells (Fig. 4C). Consistently, live-cell imaging revealed that SIK2 inhibition reduced the number of cells entering mitosis, with this number being more significantly reduced in FANCA À/À fibroblasts and FANCA-knockdown HeLa cells relative to their respective FANCA + counterparts (Fig. 4D,E). Of ARN-3236-treated HeLa cells that did enter mitosis, quantification of time from cell rounding to flattening revealed that mitotic exit was delayed, with this delay being significantly prolonged in FANCA-knockdown cells (Fig. 4F,G).

FANCA and SIK2 colocalize at mitotic structures
To determine whether FANCA and SIK2 have a functional relationship, we examined their immunolocalization during mitosis. Individually, both FANCA and SIK2 have been found to localize to the centrosome [4,35]. Using immunofluorescence, we found that FANCA colocalized with SIK2 at the centrosome during prometaphase and metaphase (Fig. 5A). We also observed previously unreported localization of SIK2 to  kinetochores and the central spindle. Interestingly, kinetochore-associated SIK2 colocalized with FANCA, and central spindle-associated SIK2 partially overlapped with FANCA (Fig. 5B). Localization of FANCA and SIK2 to mitotic kinetochores and central spindles was validated by three unique antibodies per protein (Figs S4, S6 and S7).
To determine whether FANCA interacts with SIK2, we immunoprecipitated FANCA from FANCA + patient fibroblasts and HEK293T cells expressing DDK-tagged SIK2. In both conditions, SIK2 coimmunoprecipitated with FANCA (Fig. 5C,E and  Fig. S2). These results were confirmed by reciprocal co-immunoprecipitations (Fig. 5D,F). Together, these results provide evidence of a physical interaction between FANCA and SIK2.

Phosphorylation of SIK2 at serine-358 is decreased in FANCA-deficient cells
We next investigated the effect of FANCA loss on SIK2 function. SIK2 is known to localize to centrosomes during early mitosis to facilitate bipolar spindle assembly [35]. We therefore utilized quantitative immunofluorescence to examine SIK2 recruitment to mitotic centrosomes in FANCA À/À cells (Fig. 6A). Total SIK2 accumulated at prophase and prometaphase centrosomes with equal intensity in FANCA À/À and FANCA + fibroblasts (Fig. 6C). However, SIK2 pS358 intensity was significantly reduced in FANCA À/ À prometaphase centrosomes relative to FANCA + counterparts (Fig. 6A,B). Normalizing the SIK2 pS358 intensity to total SIK2 intensity per centrosome also indicated that FANCA À/À fibroblasts had significantly lower SIK2 pS358 to total SIK2 ratio in prometaphase centrosomes (Fig. 6D). Consistent with these findings, western blot analysis of whole cell lysates prepared from two FA patient-derived FANCA À/À fibroblasts vs isogenic FANCA-complemented fibroblasts revealed reduced levels of SIK2 pS358 in FANCA-deficient cells, while total SIK2 levels were unaffected by FANCA status (Fig. 6F). Likewise, SIK2 pS358 levels were significantly reduced in NHDFs transfected with FANCA-targeting siRNAs relative to cells transfected with nontargeting siRNA, while total SIK2 levels were unchanged (Fig. 6F, right panel). To further confirm reduction of SIK2 activity in FANCA À/À fibroblasts, we quantified the immunofluorescence of C-Nap1, a downstream target of SIK2, which is displaced from the centrosome after phosphorylation by SIK2 [35]. Thus, increased C-Nap1 intensity at the centrosome indicates reduced SIK2 activity. We found that FANCA À/À fibroblasts exhibited significantly higher C-Nap1 intensity on prometaphase centrosomes than FANCA + fibroblasts, further indicating that FANCA deficiency results in impaired SIK2 activity (Fig. 6E  and Fig. S3C).

SIK2 pS358 and FANCA exhibit dynamic patterns of colocalization throughout mitosis
To further characterize mitotic functions of SIK2 and explore its functional relationship with FANCA, we first assessed whether their expression was cell cycledependent. Total SIK2 and SIK2 pS358 levels were assessed relative to the mitotic marker, cyclin B1. Western blot analysis of thymidine block-synchronized HeLa cells revealed that while total SIK2 levels remained constant, an increase in SIK2 S358 phosphorylation occurred simultaneously with the surge in cyclin B1 during mitosis (Fig. S3D). Likewise, we observed increased levels of SIK2 pS358 in HeLa cells arrested in mitosis by treatment with nocodazole and taxol relative to untreated cells (Fig. S3E). While SIK2 pS358 demonstrated an increase in expression during mitosis, FANCA expression remained constant throughout the cell cycle (Fig. S3F). Additionally, western blot analysis demonstrated that the expression levels of FANCA and SIK2 pS358 were steady over the course of mitosis (Fig. 7E,F).  Next, we examined localization of SIK2 pS358 and FANCA during mitosis via immunofluorescence in HeLa cells. FANCA associated with the centrosomes and kinetochores throughout mitosis, with intensity varying from prometaphase to anaphase (Fig. 7A,C). Similarly, we also observed localization of SIK2 pS358 at centrosomes and kinetochores throughout prometaphase to anaphase with varying intensities (Fig. 7B,D). Quantification of FANCA localization at centrosomes (Fig. 7G) showed decreased intensity during metaphase, and its intensity at kinetochores (Fig. 7H) also decreased from prometaphase toward metaphase and anaphase. SIK2 pS358 intensity at centrosomes decreased in anaphase compared to prometaphase and metaphase (Fig. 7I), whereas the intensity at kinetochores decreased progressively from prometaphase to metaphase and anaphase (Fig. 7J). In addition to the individual, dynamic changes in SIK2 pS358 and FANCA at these mitotic structures, we also observed through co-immunostaining that a portion of FANCA colocalized with SIK2 pS358 at mitotic centrosomes, kinetochores, and the central spindle (Fig. S4A). Of note, kinetochore association of both FANCA and SIK2 pS358 peaked at prometaphase, the stage of maximal SAC activity (Fig. 7H,J). These findings demonstrate novel subcellular localization patterns for FANCA and SIK2 and suggest stage-specific functions at these mitotic structures.

SIK2 inhibition induces polynucleation
Given our observation that SIK2 pS358 localizes to the kinetochore, we wondered whether SIK2 has a function in the SAC [46]. To address this question, we synchronized HeLa cells by arresting in G2 via treatment with the CDK1-inhibitor RO3306 and then released the cells into nocodazole-containing media to induce SAC arrest (Fig. 8A). After 4 h of nocodazole treatment, cells were additionally treated with ARN-3236 (or DMSO control) and observed via time-lapse imaging. While mitotic arrest was maintained in control cells, ARN-3236 induced SAC escape, resulting in severe multinucleation (Fig. 8A,B). This finding suggests that SIK2 activity is required for SAC function. In support of this notion, western blot analysis demonstrated that ARN-3236 treatment of nocodazolearrested cells decreased levels of cyclin B1, as well as SAC proteins BUB1 and Aurora B, indicating silencing of the SAC (Fig. S4B).
Another potential source of ARN-3236-induced polyploidy is cytokinesis failure, which has previously been observed in SIK2-depleted cells [35,36]. Through live imaging, we monitored mitotic progression of HeLa cells treated with ARN-3236 and observed that these cells initiated cleavage furrow ingression but failed to trigger abscission (Fig. 8C,D). Such cytokinesis failure resulted in formation of bi-or multinucleated cells (Fig. 8E).
To determine whether ARN-3236-induced polynucleation was enhanced by FANCA deficiency, patient fibroblasts were arrested in G2 using RO3306, then released into ARN-3236-or DMSO-containing media, and allowed to proceed through mitosis. After sufficient time to complete cell division, fibroblasts were fixed, and nucleation was assessed via fluorescence microscopy. While FANCA + and FANCA À/À fibroblasts exhibited a striking increase in bi-and multinucleated cells upon treatment with ARN-3236, both forms The intensity was compared between FANCA + and FANCA À/À fibroblasts in prophase and prometaphase. Plot shows data pooled from three independent experiments (mean AE SEM; P value was calculated by one-way ANOVA). For FANCA + vs FANCA À/À prometaphase cells, P = 0.0073. At least 29 centrosomes were analyzed per condition. (C) Quantification of SIK2 intensity on centrosomes of FANCA À/À primary fibroblasts and its corrected counterpart. Plot shows data pooled from three independent experiments (mean AE SEM; P value was calculated by one-way ANOVA). At least 29 centrosomes were analyzed per condition. (D) Quantification of the intensity ratio SIK2 pS358/ SIK2 in FANCA À/À primary fibroblast prometaphase cells. The ratio plot shows representative data pooled from two independent experiments (mean AE SEM; P = 0.0009, as calculated by one-way ANOVA). At least 28 centrosomes were analyzed per condition. (E) Quantification of C-Nap1 intensity on centrosomes of FANCA À/À primary fibroblast and its corrected counterpart. Plot shows data pooled from three independent experiments (mean AE SEM; P = 0.037, as calculated by unpaired t-test). At least 47 centrosomes were analyzed per condition. (F) Representative western blot showing reduced levels of SIK2 pS358 in FANCA-deficient cells. Two different pairs of primary FANCA-deficient patient fibroblasts and their gene-corrected counterparts were used. NHDF cells with siRNA-mediated FANCA knockdown (90% knockdown in siFANCA1 and 57% knockdown in siFANCA2) exhibited lower levels of SIK2 pS358 (30-44% less) when compared to nontargeting control siRNA-transfected NHDF cells, while total SIK2 levels were unchanged. The quantification was average of two independent experiments. of polynucleation were significantly higher in FANCA À/À cells (Fig. 8F,G).

SIK2 is a synthetic lethal target of multiple FA genes
In addition to FANCA, other FA family members, such as FANCC, BRCA1, and BRCA2, have been shown to participate in mitosis [4,[47][48][49]. We therefore reasoned that SIK2 inhibition may be synthetic lethal to cells deficient in other FA genes. To address this possibility, we knocked down FANCC, BRCA1, and BRCA2 in HeLa cells using gene-specific siRNAs (Fig. 9A,D). We then tested their responses to SIK2 inhibition and showed that depletion of FANCC, BRCA1, or BRCA2 resulted in hypersensitivity to ARN-3236 treatment compared to the nontargeting siRNA control (Fig. 9B,C,E,F). Because loss of BRCA2 is associated with susceptibility to breast cancer, we generated two clonal cell lines with stable BRCA2 knockdown in the breast cancer cell line MDA-MB231to assess synthetic lethality with SIK2 in a relevant cancer type. We found that siRNAmediated depletion of SIK2 significantly reduced cell survival and viability of both MDA-MB231-shBRCA2 cell lines compared to control MDA-MB231 cells (Fig. 9G-J). These results suggest that SIK2 is a synthetic lethal target of multiple FA pathway genes.

Discussion
Strict regulation of mitosis is essential to the preservation of genomic integrity during cell proliferation. Mitotic errors may result in chromosome misalignment, segregation defects, or faulty cell division, potentially culminating in genomic instability and subsequent malignancy [50]. FA pathway signaling has been reported to affect S phase progression through DNA damage response, but its direct participation in mitosis is less studied [51]. Previous studies demonstrated that individual FA proteins exhibit unique localization patterns during mitosis, which is suggestive of distinct mitotic functions [4]. FANCA, a fundamental member of the FA core complex, has also been shown to interact with non-FA proteins in order to carry out DNA repair responses [52,53]. FANCA was shown to affect centrosome integrity [3] and cytokinesis [54]. Both of these functions are critical to the progression and fidelity of mitosis; however, the underlying involvement of FANCA in these processes remains elusive. Underscoring the importance of FANCA function in mitosis, our kinome screen for synthetic lethal FANCA targets identified numerous candidates known to regulate mitosis. The spectrum of their functions span mitotic entry, SAC, and cytokinesis, strongly suggesting that mitotic processes are particularly vulnerable in FA pathway-deficient cells. As a result, the simultaneous disruption of mitotic and FA signaling pathways may result in a more pronounced inhibition of cell growth. Here, we show that FANCA and SIK2 are interacting partners that colocalize at multiple mitotic structures. We found that phosphorylation of SIK2 S358, which correlates with SIK2 activity [45], is associated with cyclin B1 surge during mitotic progression and that FANCA is required for optimal SIK2 S538 phosphorylation at centrosomes. In addition, we demonstrate that SIK2 inhibition affects mitosis in multiple ways, not only blocking mitotic entry and impairing cytokinesis [35][36][37], but also through compromise of the SAC.
The segregation of sister chromatids during mitosis is an important process to ensure proper distribution of genetic materials among daughter cells [55]. During metaphase, the SAC acts to ensure that each chro- matid is stably attached to microtubules emanating from opposite poles before proceeding to anaphase. The kinetochore is the prime apparatus for spindle attachment, upon which sensors and effectors for SAC regulation assemble [56,57]. We have previously shown that the FA signaling pathway regulates SAC activity and that FA-deficient cells exhibit a weakened SAC (Fig. S5) [4,5]. Here, we showed that SIK2 and FANCA colocalize on several mitotic structures, including centrosomes, kinetochores, and the central spindle in a stage-specific manner. Interestingly, the trend of both SIK2 pS358 and FANCA on kinetochores shows similar changes during the mitotic progression from high in prometaphase to low in anaphase, which is temporally consistent with the respective activation and satisfaction of the SAC. It is therefore conceivable that SIK2 and FANCA signaling pathways may have overlapping functions in the SAC and mitotic progression.
SIK2 is also known to play a role in cytokinesis [35,36]. Cytokinesis initiates during anaphase when the sister chromosomes separate and the   microtubules rearrange to form the central spindle [58]. The central spindle is crucial for determining the plane of cleavage furrow where ingression occurs [59,60]. Aberrant central spindle arrangement impairs furrow induction and, subsequently, abscission [61,62]. We showed that SIK2 pS358 is enriched at the central spindle, suggesting that SIK2 may play a functional role at this key structure. As SIK2 inhibition resulted in decreased Aurora B expression, it is also possible that SIK2 may indirectly regulate cytokinesis through Aurora B, which is known to regulate central spindle formation and dynamics [63][64][65]. Importantly, FANCA deficiency is also known to induce cytokinesis failure [8]. Thus, functional overlap of FANCA and SIK2 in these key steps of mitosis may be the basis of their synthetic lethal interaction. Germline biallelic loss of FA genes is rare: The incident rate of FA is approximately one out of 100 000 births in the United States [66]. However, recent studies have shown that the frequency of heterozygous FA gene mutations in cancer is underestimated [67,68]. Somatic FA gene mutations have been found in different types of non-FA-associated malignancies with remarkable frequency [69]. As mentioned above, somatic disruptions within the FA pathway are also present in 30% of cancers documented in the TCGA PanCancer Atlas Studies (Table 1) [26,27,70,71]. Therefore, synthetic lethal targets of the FA pathway are broadly significant to precision therapies for the general population, as well as FA patients. Until now, synthetic lethal targeting of the FA pathway has mainly focused on its role in DDR, identifying targets such as ATM, CHK1, FEN1, and most famously PARP1 [41,42,[72][73][74][75]. Beyond their roles in interphase DDR, recent work has demonstrated that FANCA, FANCC, BRCA1, and BRCA2 also have important functions in mitosis [4,5,48]. However, little attention has been given to synthetic lethal interactions between the FA pathway and known mitotic regulatory proteins. Thus, the mitotic dysregulation associated with FA pathway deficiency represents a potentially unexploited therapeutic target. AML is one of the most common cancers to develop in FA patients, and recent studies demonstrated efficacy of the pan-SIK inhibitor, YKL-05-099, in the suppression of AML in mouse models [69,76]. These findings suggested a potential role for SIK2 inhibition in the setting of FAassociated AML and thus warranted further examination [69]. Here, we demonstrate that loss of at least two FA core complex components (FANCA and FANCC) and two downstream FA genes (BRCA1 and BRCA2) is synthetic lethal upon SIK2 inhibition. These findings provide support for further assessment of the therapeutic efficacy of SIK2 inhibition as a single agent or in combination for the treatment of FAassociated malignancies.

Conclusions
Synergistic approaches for the treatment of FAinduced cancers are a critical need to reduce the toxic burden of standard chemotherapies in FA patients. Synthetic lethal targeting of the FA pathway is also significant to novel precision therapy approaches for the general population, in which 30% of cancers bear FA pathway mutations. Through an unbiased kinomewide shRNA screen, this work identified SIK2 as a synthetic lethal target of FANCA. Our findings suggest a model in which suboptimal SIK2 function in FANCA À/À cells contributes to cell cycle dysregulation and mitotic defects, with further weakening of the FANCA-SIK2 signaling axis via depletion or inhibition of SIK2 causing synthetic lethality.
ESP and DWC supervised the project. RS, AS, and SDR performed pathway analysis. KKC, ZAS, AS, DKM, RS, ESP, and DWC prepared the manuscript.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Depletion or inhibition of SIK2 is synthetic lethal in FAKO HeLa cells. (A) Representative Western blot demonstrating reduced SIK2 pS358 levels in HeLa cells treated with ARN-3236 (1.5 lm) vs DMSO control for 48 h. An average 43% decrease of SIK2 pS358 from three independent experiments was seen. (B) Representative Western blot demonstrating reduced AKT S473 phosphorylation in FANCA-corrected patient fibroblasts (left panel) and HeLa cells (right panel) upon ARN-3236 treatment (2 lm and 1.5 lm, respectively). The AKT pS473 was 48% less (average of three independent experiments) in ARNtreated patient fibroblasts and was 37% less (average of two independent experiments) in ARN-treated HeLa cells.  Fig. S3. Specificity of SIK2 pS358 antibody and expression of SIK2 pS358 during mitosis. (A) Total SIK2 and SIK2 pS358 antibodies colocalize at CREST foci, centrosomes, and central spindle. Scale bars represent 2 lm. (B) Phosphatase treatment of lysates from patient fibroblasts and HeLa cells abolishes SIK2 pS358 signal. (C) Immunofluorescence staining of C-Nap1 to centrosome in FANCA + (upper panel) and FANCA À/À (lower panel) patient fibroblasts in prometaphase (pericentrin-green, C-Nap1-red, DAPI-blue). Scale bars represent 2 lm. (D) Increase in SIK2 pS358 expression after thymidine release corresponding to the increase in cyclin B1. Relative fold change in SIK2 pS358 was significantly higher at the peak of cyclin B1 surge than at the point of lowest cyclin B1. Mean AE SEM of three independent experiments were shown. P value was calculated by unpaired t-test. (E) Representative Western blots showing seven-fold and four-fold increase, respectively in SIK2 phosphorylation in HeLa cells that were arrested in mitosis by 15h treatment with nocodazole (100 nM) or taxol (100 nM) relative to unsynchronized control. Three independent experiments were conducted. (F) Representative Western blots showing comparable FANCA expression in HeLa cells arrested in mitosis by 15-h treatment with nocodazole (100 nM) or taxol (100 nM) relative to unsynchronized control. Three independent experiments were done. For FANCA + vs FANCA À/À MNHN fibroblasts treated with 100 nM taxol, P = 0.003. For FANCA + vs FANCA À/À JRAST fibroblasts treated with 100 nM taxol, P < 0.0001. Fig. S6. Validation of SIK2 localization. Representative immunofluorescence images show localization of SIK2 to mitotic centromeres (A, B) and central spindle (C, D) using with two unique antibodies (Abcam and Cell Signaling). Scale bars for whole cell images represent 1 lm. Scale bars for images of enlarged regions represent 500nm. Fig. S7. Validation of FANCA localization. (A-C) Representative immunofluorescence images show the localization of FANCA to mitotic centromeres, spindle, and central spindle using an additional FANCA antibody (Rabbit polyclonal; Abcam). Scale bars for whole cell images represent 1 lm. Scale bars for images of enlarged regions represent 500 nm. Data Set S1 Supplementary Material.