Regulation of DNA damage response by trimeric G-proteins

Summary Upon sensing DNA double-strand breaks (DSBs), eukaryotic cells either die or repair DSBs via one of the two competing pathways, i.e., non-homologous end-joining (NHEJ) or homologous recombination (HR). We show that cell fate after DSBs hinges on GIV/Girdin, a guanine nucleotide-exchange modulator of heterotrimeric Giα•βγ protein. GIV suppresses HR by binding and sequestering BRCA1, a key coordinator of multiple steps within the HR pathway, away from DSBs; it does so using a C-terminal motif that binds BRCA1’s BRCT-modules via both phospho-dependent and -independent mechanisms. Using another non-overlapping C-terminal motif GIV binds and activates Gi and enhances the “free” Gβγ→PI-3-kinase→Akt pathway, which promotes survival and is known to suppress HR, favor NHEJ. Absence of GIV, or loss of either of its C-terminal motifs enhanced cell death upon genotoxic stress. Because GIV selectively binds other BRCT-containing proteins suggests that G-proteins may fine-tune sensing, repair, and survival after diverse types of DNA damage.


INTRODUCTION
Genomic integrity is under constant attack from extrinsic and intrinsic factors that induce DNA damage. 1 Damaged DNA must be repaired to maintain genomic integrity via processes that evolved by the cell type, collectively termed the DNA damage response (DDR). 2 DDRs are orchestrated by an incredibly complex network of proteins that sense and assess the type and extent of damage, decide between cell fates (death vs. repair), choose a repair pathway, and then initiate and complete the repair process. 3 For example, the DDR involves the activation of ATM kinase, a member of the phosphoinositide 3-kinase (PI3K)-related protein kinase family 4 which is rapidly recruited by the MRE11-RAD50-NBS1 (MRN) complex to chromatin. 5 Phosphorylation of a large number of substrates follows, which in turn activates cell cycle checkpoints and triggers the recruitment of repair factors to the DSBs. Positive feedback loops are orchestrated to amplify the signals, e.g., ATM phosphorylates the histone variant H2AX (resulting in the formation of the phosphorylated form called gH2AX), 6 which recruits additional ATM molecules and further accumulation of gH2AX. [7][8][9] Among the types of DNA damage, DNA double-strand breaks (DSBs) are the most cytotoxic lesions that threaten genomic integrity. 10 Failure to repair DSBs results in genomic instability and cell death. DNA repair can be achieved by different means that are commonly grouped into two broad, competing categories: 11 homologous recombination (HR) and non-homologous end-joining (NHEJ). 12 HR, which requires a homologous template to direct DNA repair, is generally believed to be a high-fidelity pathway. 13 By contrast, NHEJ directly seals broken ends; while some believe that repair by NHEJ is imprecise, we have shown that precision can indeed be achieved. 14 In fact, NHEJ offers an ideal balance of flexibility and accuracy when the damage to DNA is widespread with DSBs featuring diverse end structures. 15 Consequently, it is believed to represent the simplest and fastest mechanism to heal DSBs, 16 thus it is the most predominant DSB repair pathway within the majority of mammalian cells. Key molecular players in both pathways have been identified: 53BP1, first identified as a DNA damage checkpoint protein, and Breast cancer type 1 susceptibility protein (BRCA1), a well-known breast cancer tumor suppressor, 17 are at the center of molecular networks that coordinate NHEJ and HR, respectively.

GIV is required for DNA damage response
We generated HeLa cells without GIV using CRISPR Cas9 and subsequently exposed them to Doxorubicin followed by several commonly used readouts of DDR (Figures 2A, S1A, and S1B). A mixture of À/À (henceforth, GIV KO) clones was pooled to recapitulate the clonal heterogeneity of parental HeLa cells (Figure S1B), and near-complete depletion of GIV (estimated $95% by band densitometry) was confirmed by immunoblotting ( Figure 2B). We chose HeLa cells because DDR has been extensively studied in this cell line 28,29 and because HeLa cells have defective p53. 30 The latter is relevant because GIV/CCDC88A aberrations (gene amplification) co-occur with defects in the tumor suppressor TP53 (TCGA pancancer iScience Article profile; cbioportal.org); $36% of tumors with aberrant CCDC88A expression was also associated with missense and truncating driver mutations in TP53. We chose Doxorubicin (henceforth, Dox) for inducing DNA damage because it is a widely used anthracycline anticancer agent and its impact on DNA integrity in HeLa cells has been mapped for each cell cycle with demonstrated reproducibility. 31 Compared to parental cells, fewer metabolically active GIV KO cells survived after a Dox challenge, as determined using an MTT assay ( Figures 2C and S1C), indicating that in the absence of GIV, cells show markedly reduced survival from cytotoxic lesions induced by Dox. GIV KO cells showed increased susceptibility also to two other cytotoxic drugs, Cisplatin and Etoposide ( Figures S1D and S1E). The lower IC50 values in the case of GIV KO cell lines for all 3 drugs ( Figure 2C) imply that GIV is required for surviving cytotoxic lesions induced by the most commonly used cytotoxic drugs.
Because cell cycle is a key determinant of the choice of repair pathway, next, we asked if GIV may impact one or more of the three checkpoints (G1/S, S phase, and G2/M) where cell cycle may be arrested in response to DNA damage. We found that Dox-challenged parental cells, as expected for cells with defective p53, escaped the G1/S checkpoint, 32 and instead, preferentially showed arrest in the S/G2 phase; however, GIV KO cells showed no such S phase arrest and instead arrested in the G2 phase (Figures 2D and S1F). Because chromosome duplication occurs during the "S phase" (the phase of DNA synthesis) and this phase surveys DNA for replication errors, 33 failure of GIV KO cells to arrest in the S phase indicates that this ''checkpoint'' is impaired (i.e., bypassed). Because irreparable DNA injury leads to the accumulation of mutations, which in turn may induce either apoptosis or necrosis, 34 next we analyzed cell death by flow cytometry using a combination of annexin V and propidium iodide (PI) staining. Compared to parental control cells, Dox challenge induced a significantly higher rate of cell death in GIV KO cells ( Figures 2E, 2F, and S1G), via both necrosis ( Figure 2E) and apoptosis (specifically, late apoptosis; Figure 2F).
To examine whether higher cell death was related to impaired repair activity and an accumulation of DNA strand breaks in GIV KO cells, genomic DNA was isolated from parental and GIV KO cells, with or without Dox challenge and the levels of strand breaks in the HPRT and POLB genes were compared using long amplicon qPCR (LA-qPCR) as described previously. 35 Strand breaks were measured for both the genes using a Poisson distribution, and the results were expressed as the lesion/10 kb genome. 36 A decreased level of the long amplicon PCR product (12.2 kb of the POLB or 10.4 kb region of the HPRT gene) would reflect a higher level of breaks; however, the amplification of a smaller fragment for each gene is expected to be similar for the samples, because of a lower probability of breaks within a shorter fragment. A higher level of DNA strand break was observed in the genomic DNA of GIV KO cells than in the DNA of parental controls ( Figures 2G, 2H, and S1H), indicating a role of GIV in DNA repair.
Reduced cell survival ( Figure 2C), cell-cycle arrest ( Figure 2D), higher cell death ( Figures 2E and 2F), and the accumulation of cytotoxic lesions ( Figures 2G and 2H) in GIV KO cells were also associated with reduced growth in anchorage-dependent clonogenic growth assays ( Figure S1I).
To test if the pro-survival functions of GIV in the setting of cytotoxic lesions are cell-type specific, we compared 2 other cell lines, the MDA-MB-231breast and DLD-1 colorectal cancer lines ( Figure S2A). We   Figure S1F. Data displayed as mean G SEM and one-way ANOVA using Tukey's multiple comparisons test was used to determine significance. (*; p % 0.05, **; p % 0.01; ns = not significant). (E and F) Bar graphs display the % necrotic (E) or apoptotic (early, EAC; late, LAC; or combined) cells after challenged with either Dox or vehicle control (DMSO), as assessed by annexin V staining and flow cytometry. See Figure S1G for the dot plot diagrams.
(G and H) Long amplicon qPCR (LA-QPCR) was used to evaluate genomic DNA SB levels in control vs. GIV KO cells. Representative gel showing PCRamplified fragments of the HPRT (G, top panel) and POLB (G, bottom panel) genes. Amplification of each large fragment (upper panels) was normalized to that of a small fragment of the corresponding gene (bottom panels), and the data were expressed as normalized (with short PCR amplicon) relative band intensity with the DMSO-treated (0 h) sample in each case arbitrarily set as unity and displayed as a bar graph in H. Full-length gels can be seen in Figure S1H. Data displayed as mean G SEM and one-way ANOVA to determine significance. (**; p % 0.01; ****; p % 0.0001; ns = not significant).
(I) Summary of the phenotype of cells with (parental; GIV +) or without GIV (GIV KO; GIV -). See also  37 We exposed these cells to Doxorubicin. Survival was significantly impaired in all the GIV KO cell lines ( Figures S2D-S2F), implying that our findings in HeLa cells may be broadly relevant in diverse cancers.
Taken together, these findings demonstrate that GIV is required for DNA repair; in cells without GIV, cell survival is reduced, S phase checkpoint is lost, and DNA repair is impaired, leading to the accumulation of mutations in POLB and HPRT (see Figure 2I). These findings suggest that genotoxic insult in the absence of GIV may lead to the accumulation of catastrophic amounts of mutations that may ultimately trigger cell death.
The C-terminus of GIV binds tandem BRCT modules of BRCA1 We next sought to validate the major BioID-predicted interaction of GIV, i.e., BRCA1. To determine if GIV and BRCA1 interact in cells, we carried out coimmunoprecipitation (Co-IP) assays and found that the two full-length endogenous proteins exist in the same immune complexes ( Figure 3A). BRCA1 features two prominent modules that mediate protein-protein interactions, an N-terminal RING domain, which functions as an E3 ubiquitin ligase, 38 and a C-terminal BRCT repeat domain, which functions as the phospho-protein binding module. 39 Pulldown assays using recombinant GST-tagged BRCA1-NT (RING) or CT (tandem BRCT repeats) proteins immobilized on Glutathione beads and lysates of HEK cells as a source of FLAG-tagged GIV showed that full-length GIV binds BRCT, but not the RING module ( Figure 3B). We noted that GIV is predicted to also interact with other BRCT-domain containing DDR pathway proteins, e.g., DNA Ligase IV (LIG4) and Mediator Of DNA Damage Checkpoint 1 (MDC1) [Human cell map, cell-map.org; a database of BioID proximity map of the HEK293 proteome; accessed on 01/06/2020] and with BARD1 [Bio-GRID, thebiogrid.org; accessed 09/05/2020]. Pulldown assays with these BRCT modules showed that GIV bound DNA Ligase and BARD1, but not MDC1 ( Figure 3B), suggesting that while GIV can promiscuously bind multiple DDR pathway proteins that contain the BRCT module, there may be a basis for selectivity within such apparent promiscuity. As a positive control for BRCT-binding protein, we tracked by immunoblotting the binding of BACH1 from the same lysates, which bound BRCA1's tandem BRCT module, as expected, 40 and to a lesser extent with DNA Ligase ( Figure 3B).
We next asked if the C-terminus of GIV can directly bind BRCA1; we focused on GIV's C-terminus (GIV-CT) because numerous studies have underscored the importance of GIV-CT as an unstructured and/or intrinsically disordered domain that scaffolds key proteins within major signaling cascades to mediate dynamic pathway crosstalk. 41,42 GST pulldown assays using recombinant His-tagged GIV-CT 1660-1870 and various GST-DDR pathway proteins showed that GIV's CT is sufficient to bind the C-terminal tandem BRCT domain of BRCA1 ( Figure 3C). Because we used purified recombinant proteins in this assay, we conclude the GIV,BRCA1 interaction observed in cells is direct. Using lysates from multiple different cell types as a source of GIV (Hs578T, Figure 3D; Cos7 and HeLa; Figures S3A and S3B) we further confirmed that endogenous full-length GIV binds the C-terminal tandem BRCT domain of BRCA1 (but not its RING domain) and weakly with BARD1. (G) Alignment of GIV's C-terminal sequence with known phosphopeptides that bind BRCA1, as confirmed by X-ray crystallography (PDB codes on the left). The consensus SxxF sequence is shown (evolutionary conservation of the SxxF motif and its relationship with other motifs on GIV-CT is shown in Figure S4).  Figure 3E) helped narrow the region within GIV that binds BRCA1. The longer GIV-CT fragments bound, but the shortest fragment (1790-1870) did not ( Figures 3E and 3F), indicating that the sequence of GIV that lies between aa 1660-1790 could be the key determinant of binding. A sequence alignment of this region on GIV against known interactors of BRCA1's tandem BRCT repeats revealed the presence of a canonical BRCT-binding phospho-peptide sequence of the consensus ''phosphoserine (pSer/pS)-x-x-Phenylalanine (Phe; F)'' ( Figure 3G). The structural basis for such binding has been resolved. 43 This newly identified putative BRCT-binding motif in GIV had three notable features: First, this motif ( 1716 SSDF 1719 ) is distinct from and farther downstream of GIV's Gai-modulatory motif (31 aa $1670-1690) ( Figure S4A), suggesting that they may be functionally independent. Second, the SxxF motif is evolutionarily conserved in higher vertebrates (birds and mammals) ( Figure S4A), suggesting that GIV could be a part of the complex regulatory capacities that evolved later. 44 Third, multiple independent studies have reported that the Ser in 1716 SSDF 1719 is phosphorylated ( Figure S4B), suggesting that GIV,BRCA1 complexes may be subject to phosphomodulation. Site-directed mutagenesis that destroys the consensus motif (by replacing Phe with Ala; F1719A) resulted in a loss of binding between GIV-CT and BRCA1 ( Figure 3H), thereby confirming that the putative BRCT-binding motif is functional and implicating it in the GIV,BRCA1 interaction. The independent nature of the BRCA1-binding and Gai-modulatory motif was confirmed in pulldown assays with full-length WT and mutant GIV proteins (Figures 3I and 3J); the BRCA1 binding-deficient F1719A mutant protein selectively lost binding to GST-BRCA1, but not GST-Gai3, and the well-characterized G-protein binding-deficient F1685A mutant protein 20,21 selectively lost binding to GST-Gai3, but not GST-BRCA1.
Collectively, these findings demonstrate that GIV binds BRCA1 via its C-terminally located BRCT-binding motif. This motif is sensitive to disruption via a single point mutation but specific enough that such mutation does not alter GIV's ability to bind Gai-proteins.

GIV binds BRCA1 in both phospho-dependent and -independent modes via the same motif
We next asked how GIV binds BRCA1(BRCT). BRCT modules are known to bind ligands via two modes-(i) canonical, phospho-dependent (e.g., BACH1, CtIP, Abraxas) and (ii) non-canonical, phospho-independent (e.g., p53); 45 while the structural basis for the former has been resolved, 43 the latter remains unclear. Because bacterially expressed His-GIV-CT directly binds the tandem BRCA1-BRCT ( Figure 3C), the GIV,BRCA1 (BRCT) interaction appears phospho-independent. As positive controls for canonical phospho-dependent binding, we used BACH1 and CtIP, two bona fide binding partners of the BRCA1-BRCT module. Recombinant His-GIV-CT did not impact the canonical mode of binding of either BACH1 (Figure 4A) or CtIP ( Figure 4B) to BRCA1-BRCT, suggesting that unphosphorylated GIV binds BRCA1 at a site that is distinct from the interdomain cleft where BACH1 or CtIP are known to occupy. 43 Furthermore, binding of GIV to the tandem BRCT was enhanced $3to 5-fold in the presence of the most frequently occurring mutation in BRCA1, M1775R ( Figure 4C); this mutation is known to abrogate canonical mode of phosphopeptide binding by destroying a hydrophobic pocket that otherwise accommodates the Phe in the pSxxF consensus (see Figure S5A). 46 The unexpected increase in binding to the BRCA1-M1775R  iScience Article mutant was also observed in the case of p53, which is another direct and phospho-independent BRCA1(BRCT)-interacting partner 47-50 ( Figure 4D). The expected disruptive effect of this mutation could, however, be confirmed in the case of both BACH1 ( Figure S5B) and CtIP ( Figure S5C). These findings demonstrate that GIV binds BRCA1 via a non-canonical phospho-independent mechanism that is distinct from CtIP and BACH1.
Because $10 high-throughput (HTP) studies have confirmed that Ser 1716 within the BRCA1-binding motif of GIV is phosphorylated ( Figure S5C), presumably by one of the many DDR and cell-cycle regulatory kinases ( Figure S5D), we asked if the GIV,BRCA1 interaction is phosphomodulated. Phosphomimic (Ser 1716 /Asp; S1716D) and non-phosphorylatable (Ser 1716 /Ala; S1716A) mutants of GST-GIV-CT were generated, rationalized based on systematic peptide screening studies demonstrating that Glu/Asp-x-x-Phe peptides bind BRCT modules with $10-fold higher affinity. 51 Binding of BRCA1 was accentuated with GIV-S1716D mutant but restored to levels similar to WT in the case of GIV-S1716A mutant ( Figure 4E), indicating that the GIV,BRCA1 interaction may be phosphoenhanced and that the -OH group in Ser (which is absent in Ala; A) is not essential for the interaction. The phosphate group in the consensus pSxxF mediates polar interactions with S1655/G1656 in b1 and K1702 in a2 of BRCA1, 43 and a K1702M mutant has previously been shown to impair phospho-dependent canonical mode of binding. 44 We found that the observed phosphoenhanced GIV,BRCA1 in Figure 4E is virtually abrogated in the case of BRCA1-K1702M ( Figure 4F), indicating that upon phosphorylation at S1716, GIV may bind BRCA1 in a phospho-dependent canonical mode. Finally, in pulldown assays with the BRCA1-M1775R mutant, binding was inhibited to the phosphomimic GIV-S1716D mutant, but not to GIV-WT ( Figure 4G), likely via the obliteration of the binding pocket for the F1719, as has been reported in the canonical binding mode. 46 That the F1719 is also important for phospho-dependent binding was also confirmed; the addition of F1719A mutation to S1716D mutation disrupted binding to BRCA1 ( Figure 4H), indicating that the same BRCA1-binding motif participates in both modes of binding. Homology models of GIV,BRCA1 co-complexes ( Figure 4I; top), built using the solved structure of canonical BACH1,BRCA1 co-complex (PDB:1T29) as template further confirmed that phospho-dependent canonical mode of binding and disrupted binding when M1775 is mutated to R (Figure 4I; bottom) is compatible with the observed biochemical studies.
Taken together, these findings support the conclusion that GIV binds BRCA1 in two different modes: a noncanonical phospho-independent mode, the structural basis for which remains unknown ( Figure 4J; left), and a canonical phospho-dependent mode ( Figure 4J; right). Both modes of binding occur via the same motif in GIV.
Both GIV,BRCA1 and GIV,Gai interactions are required for DNA repair To dissect the role of the GIV,BRCA1 interaction, we rescued GIV KO HeLa cell lines with either GIV-WT or single-point specific mutants of GIV that either cannot bind BRCA1 (F1719A) or cannot bind/activate Gaiproteins (F1685A) and used them in the same phenotypic assays as before ( Figure 5A). First, we confirmed  iScience Article by immunoblotting that the G418-selected clones stably express physiologic amounts of GIV-WT/mutants at levels similar to endogenous ( Figure 5B). When challenged with Dox, cisplatin, or etoposide, survival, as determined using an MTT assay was significantly reduced in the cells expressing either mutant compared to GIV-WT ( Figures 5C and S6A-S6C). The lower IC50 values in the case of GIV mutant cell lines for all 3 drugs ( Figure 5C) imply that both functions of GIV, i.e., BRCA1-binding and G protein-binding/activating, are required for surviving cytotoxic lesions induced by commonly used cytotoxic drugs. Lower survival was associated with G2/M phase arrest in both mutant lines ( Figures 5D and S6A). The S phase checkpoint, however, was intact in cells expressing GIV-WT and GIV-F1685A mutant, but not in GIV-F1719A mutant (Figure 5D), indicating that the disruption of the GIV,BRCA1 interaction blocks the S phase checkpoint. Flow cytometry studies showed that cell death, both necrosis ( Figures 5E and S6B) and apoptosis (late apoptosis; LAC; Figures 5F and S6B), was significantly increased in both mutant-expressing lines compared to GIV-WT. The extent of death was higher in GIV-F1719A mutant lines, indicating that the disruption of the GIV,BRCA1 interaction is catastrophic. Consistently, the burden of mutations was increased in both mutant lines, but to a higher degree in GIV-F1719A mutant lines ( Figures 5G, 5H, and S6C).
Taken together, these results demonstrate that both functions of GIV (BRCA1-binding and Gai binding and activation) are important for GIV's role in mounting a DDR. The use of GIV KO cell lines rescued with WT or specific binding-deficient mutants further pinpointed the role of each function in the process ( Figure 5I; summarized in Table S3). The GIV,BRCA1 interaction was required for S phase checkpoint arrest, cell survival, and DNA repair. However, GIV's Gai-modulatory function was somehow important for cell survival and the efficiency of DNA repair.

GIV may inhibit homologous recombination and favor non-homologous end-joining
We next asked how GIV's ability to bind BRCA1 or activate Gai might impact the choice of the pathway for DNA repair. While gH2AX is responsible for the recruitment of many DNA maintenance and repair proteins to the damaged sites, including 53BP1 and RAD51, 52 the preferential accumulation of 53BP1 indicates NHEJ, whereas the preferential accumulation of Rad51 indicates HR 12 ( Figure 6A). BRCA1 favorably activates Rad51-mediated HR repair and actively inhibits 53BP1-mediated NHEJ repair. 53 We found that the nuclear accumulation of 53BP1, as determined by confocal microscopy, was higher in parental HeLa cells compared to GIV KO cells (Figures 6B; left; 6C). By contrast, nuclear accumulation of Rad51 was much more pronounced in GIV KO compared to parental cells (Figures 6B; right; 6C 0 ). These findings indicate that NHEJ is the preferred choice for repair in cells with GIV, but HR is favored in the absence of GIV. This preference of HR over NHEJ in GIV KO cells was reversed in KO cells rescued with GIV-WT but could not be rescued by mutant GIV proteins that could not bind BRCA1 or modulate Gai proteins ( Figures 6D and  6E-6E 0 ). DSBs were increased in GIV KO cells and in cells expressing either of the GIV mutants, as determined by gH2AX staining; this is consistent with the prior long amplicon PCR studies assessing the burden of mutations ( Figures 2G, 2H, 5G, and 5H).
That GIV is required for NHEJ was further confirmed by live cell imaging using parental and GIV KO HeLa ( Figures 6F-6H) and MDA-MB-231cells ( Figures S7A and S7B) stably expressing a fluorescent reporter of  Taken together, these findings suggest that GIV and its BRCA1-binding and Gai-modulatory functional modules may influence the choice of DDR; when GIV is present and its two functional modules are intact, 53BP1 is preferentially recruited to DSBs, in the detriment of Rad51, indicating that NHEJ may be preferred over HR. It is also noteworthy that the mutational burden is increased despite the DNA damage-induced accumulation of nuclear Rad51, which suggests that HR is initiated successfully, but may not be as effective as NHEJ. The latter offers an ideal balance of flexibility and accuracy in the setting of widespread DSBs with diverse end structures. 15 GIV translocates to the nucleus after DNA damage, inhibits the colocalization of BRCA1 with double-strand breaks Because BRCA1 is a nucleocytoplasmic shuttling protein 55 and it is nuclear BRCA1 that augments DNA repair 56 and cell-cycle checkpoints, 57 we asked if suppressed HR in cells with GIV, or those with functionally intact modules in GIV stemmed from the mis-localization of BRCA1. We determined the localization of GIV and BRCA1 by confocal immunofluorescence and found that the Dox challenge was associated with the nuclear localization of GIV (see Parental cells; Figure 6I, top-left). Compared to parental control cells, nuclear localization of BRCA1 was more prominent in GIV KO cells ( Figure S7C), where BRCA1 colocalized with gH2AX (see Figure 6I, bottom; see Figure 6K for colocalization index), indicating that the nuclear localization of BRCA1 to sites of DSBs may be suppressed by GIV.
To discern which functional module of GIV may be important for the nuclear localization of GIV and/or suppression of the nuclear localization of BRCA1, we carried out similar assays in stable cell lines expressing GIV-WT or mutant. DNA damage-dependent shuttling of GIV to the nucleus was observed in the case of GIV-WT and GIV-F1719A, but not GIV-F1685A (see Figure 6J, top), indicating that GIV's ability to shuttle into the nucleus after DNA damage does not depend on its interaction with BRCA1, but requires a functionally intact Gai-modulatory function. We observed prominent nuclear localization of BRCA1 only in the GIV-F1719A mutant line ( Figure S7D), where it colocalized with gH2AX (see Figure 6J, bottom). Colocalization coefficient of BRCA1 with gH2AX across all cell lines showed that colocalization was greatest in the absence of GIV (GIV KO cells; Figures 6K and S7C) or when the GIV,BRCA1 interaction is impaired (F1719A; Figure S7D), indicating that the GIV,BRCA1 interaction is required for the observed inhibitory effect of GIV on the nuclear localization of BRCA1. Nuclear-cytosol fractionation studies also showed that the nuclear pool of BRCA1 and Rad51 in Dox-challenged HeLa cells is increased in the absence of GIV ( Figure S7E).
Taken together, these findings demonstrate that GIV, like BRCA1, is a nucleocytoplasmic shuttling protein; shuttling is independent of its BRCA1-binding function but depends on its Gai-modulatory function. The GIV,BRCA1 interaction appears to be primarily responsible for sequestering BRCA1 away from DSBs. Localization of BRCA1 at sites of DSBs is not only impaired in the case of GIV-WT expressing cells, in which GIV shuttles into the nucleus upon DNA damage, but also impaired in GIV-F1685A mutant cells ( Figures 6H  and S7D), in which GIV fails to localize to the nucleus. This indicates that the inhibitory GIV,BRCA1 interaction may occur in the nucleus as well as in the cytoplasm, and is in keeping with our BioID studies revealing BRCA1 as a candidate interactor of GIV in both nuclear and cytosolic compartments ( Figure 1C).

GIV's Gai-modulatory function activates Akt, BRCA1-binding function triggers S-phase checkpoint
Because the choice of DNA damage repair pathway is fine-tuned by a network of kinases (e.g., ATM, ATR, Akt, and so forth) and the signaling cascades they initiate, 58,59 we asked how GIV and its functional modules may impact these pathways. More specifically, we focused on two key readouts rationalized by our observations: (i) Akt phosphorylation, because GIV is a bona fide enhancer of Akt phosphorylation 60,61 and does so via its Gai-modulatory function, 20 Figures 7D and 7E), albeit more significantly impaired in the latter, but phosphoSMC1 was specifically impaired in cells expressing the GIV-F1719A mutant ( Figures 7D and 7F). These findings show that the BRCA1-binding function of GIV is critical for the initiation of Akt signaling upon DNA damage, as well as for the activation of the ATM/ pSMC1 pathway for S-phase checkpoint signaling. The Gai-modulatory function of GIV, however, was specifically responsible for enhancing Akt signals after DNA damage.
We asked if the previously delineated Gi / ''free'' Gbg release / Class 1 PI3K signaling axis triggered by GIV's Gai-modulatory function may be essential. 20 To this end, we first assessed the extent of the activation of Gai in cells after DNA damage by using a conformation-sensing antibody that specifically recognizes GTP-bound (active) conformation of Gai1-3 ( Figure 7G; top), 63 and more importantly, recognize GIV-dependent G protein activation in cells. 42 We found that DNA damage was associated with the activation of Gai in parental cells, but that such activation was virtually lost in GIV KO cells ( Figure 7G; bottom). To dissect if Akt activation is mediated via the ''free'' Gbg/Class 1 PI3K signaling axis, we used the commonly used small molecule Gbg inhibitor, Gallein ( Figure 7H; top), and it's an inactive isomer, Fluorescein (negative control). 64 We found that Gallein, but not Fluorescein inhibited DNA damageinduced Akt phosphorylation in parental control cells, reducing it to the levels observed in GIV KO cells ( Figure 7H; bottom). These findings indicate that Akt signaling induced after DNA damage occurs in part via GIV-dependent Gi activation.
Taken together, our findings support the following working model for how GIV may influence the choice of repair pathway after DNA damage, favoring NHEJ over HR ( Figure 7I). Using a set of single-point mutants and chemical inhibitors of G protein signaling, we charted the mechanisms that allow GIV to accomplish such a goal via two parallel pathways (see Figure 7J). One pathway is mediated by GIV's ability to bind and sequester BRCA1 in the cytoplasmic pool, and thereby reducing its ability to localize to DSBs, suppress HR, and activate S phase checkpoint cascades. Another is GIV's ability to bind and activate Gi and enhance Akt signaling, which further skews the choice of repair pathway toward NHEJ, while actively suppressing HR.

DISCUSSION
Cellular decision-making in response to any stressful insult is mediated by a web of spatiotemporally segregated events within the intracellular signaling networks, often requiring crosstalk between unlikely pathways. The major discovery we report here is such an unexpected crosstalk that is orchestrated via a versatile multi-modular signal transducer, GIV/Girdin. There are three notable takeaways from this study.
First, this work ushers a new player in DNA repair. Although GIV entered the field of cancer biology more than a decade ago, and quickly came to be known as a pro-oncogenic protein that coupled G protein signaling with unlikely pathways [reviewed in 24 ], its role inside the nucleus remained unknown. Although predicted to have nuclear localization signals (NLS ; Table S4), how GIV shuttles into the nucleus remains iScience Article unresolved. Regardless, what emerged using specific single-point mutants is that GIV inhibits HR by sequestering BRCA1, suppressing its localization to DSBs.
Second, one of the most unexpected observations was that GIV uses the same short linear motif (SLIM) located within its C-terminus to bind the C-terminal tandem-BRCT modules of BRCA1 in both canonical (phospho-dependent) and non-canonical (phospho-independent) modes. Although both modes of BRCT-binding have been recognized in other instances, 45 the versatility of dual-mode binding via the same motif is unprecedented. However, these findings are in keeping with the fact that GIV-CT is an intrinsically disordered protein (IDP) 41,65 comprised of distinct SLIMs, of which the BRCT-binding motif described here is an example (see Figure S3A). SLIMs enable GIV to couple G protein signaling to a myriad of molecular sensors, of both the outside of the cell (i.e., receptors; [reviewed in 66 ]) or its interior. 67 Because IDPs that fold/unfold on demand expose/hide SLIMs, which in turn imparts plasticity to protein-protein interaction networks during signal transduction, 68 GIV may do something similar in couple G protein signaling to DDR. Given this degree of versatility of the BRCT-binding SLIM in GIV, and the additional BRCT interactors we found here (to DNA Lig IV and BARD1), it is more likely than not that this SLIM binds other players within the DDR pathways. By scaffolding G proteins to BRCT-modules in BRCA1 (and presumably other DDR proteins) GIV may serve as a point of convergence for coordinating signaling events and generating pathway crosstalk upon DNA damage.
Third, this work provides a direct mechanistic link between DDR and trimeric G proteins; the latter is one of the major pervasive signaling hubs in eukaryotic cells that was notably absent from the field of DNA repair. Although multiple peripheral components within the GPCR/G-protein signaling system have been found to indirectly influence DNA damage and/or repair, 23 who/what might activate G proteins on endomembranes was unknown. We demonstrated that trimeric Gi proteins are activated upon DNA damage and that such activation requires GIV's Gai-modulatory motif. That the GIV/Gai pathway activates Akt signaling helps explain the hitherto elusive origin of Akt signaling during DDR. 59 That GIV favors NHEJ over HR and activates Akt signaling during DDR is in keeping with the previously described role of Akt signaling in inhibiting HR and promoting NHEJ. 59,[69][70][71][72] Limitations of the study Although how GIV binds BRCA1 was studied at greater depth, how exactly GIV may inhibit the shuttling/ localization of BRCA1 remains unresolved. Because nuclear import of BRCA1 and its retention requires BARD1, 73 whereas nuclear export requires p53, 74 GIV may either inhibit the BARD1,BRCA1 interplay or augment the actions of p53. Although we could gain some insight into the mechanism of GIV,BRCA1 interaction, the stoichiometry of this complex in cells remains unknown. Neither do we know the mechanism of how GIV binds DNA Lig IV and BARD1. Because phospho-peptide,BRCT interactions are generally believed to be exclusive, GIV's interaction with DNA Lig IV and BARD1 we observed here are likely to be phospho-independent because they were observed using bacterially expressed recombinant proteins. Which DDR proteins bind GIV, and which do not, may be dictated by the residues flanking the SLIM, as shown in other instances; 75 additional mutagenesis or structural studies are required to understand the selectivity and specificity of GIV,BRCT interactions.
In closing, damage to the genome can have catastrophic consequences, including cytotoxicity, accelerated aging, and predisposition to cancers. Our findings, which revealed a hitherto unknown link between a major hub in DNA repair (i.e., BRCA1) and a signaling hub of paramount importance in just about all aspects of modern medicine (trimeric G proteins) open new avenues for the development of therapeutic strategies.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:   Transfection was carried out using Genejuice (Novagen) for DNA plasmids following the manufacturers' protocols. Hela cell lines stably expressing GIV constructs (WT, F1685A, and F1719A) were selected after transfection in the presence of 800 mg/mL G418 for 6 weeks. The resultant multiclonal pool was subsequently maintained in the presence of 500 mg/mL G418. GIV expression was verified independently by immunoblotting using anti-GIV antibody. Whole-cell lysates were prepared after washing cells with cold PBS prior to resuspending and boiling them in sample buffer. Lysates used as a source of proteins in immunoprecipitation or pulldown assays were prepared by resuspending cells in Tx-100 lysis buffer [20 mM HEPES, pH 7.2, 5 mM Mg-acetate, 125 mM K-acetate, 0.4% Triton X-100, 1 mM DTT, supplemented with sodium orthovanadate (500 mM), phosphatase (Sigma) and protease (Roche) inhibitor cocktails], after which they were passed through a 28G needle at 4 C, and cleared (10,000 3 g for 10 min) before use in subsequent experiments.

Biotin proximity labeling
BioID was performed as previously described. 80 Briefly, HEK293T were plated 24 hrs prior to transfection with mycBirA-tagged GIV construct. Thirty hours post transfection, cells were treated with 50 mM biotin (dissolved in culture media) for 16 hrs. Cells were then rinsed two times with PBS and lysed by resuspending in lysis buffer (50 mM Tris, pH 7.4, 500 mM NaCl, 0.4% SDS, 1 mM dithiothreitol, 2% Triton X-100, and 13 Complete protease inhibitor) and sonication in a bath sonicator. Cell lysates were then cleared by centrifugation at 20,000 X g for 20 mins and supernatant was then collected and incubated with streptavidin magnetic beads overnight at 4 C. After incubation, beads were washed twice with 2% SDS, once with wash buffer 1 (0.1% deoxycholate, 1% Triton X-100, 500 mM NaCl, 1 mM EDTA, and 50 mM HEPES, pH 7.5), followed with once wash using wash buffer 2 (250 mM LiCl, 0.5% NP-40, 0.5% deoxycholate, 1 mM EDTA, and 10 mM Tris, pH 8.0), and once with 50 mM Tris pH 8.0. Biotinylated complexes were then eluted using sample buffer containing excess biotin and heating at 100 C. Prior to mass spectrometry identification, eluted samples were run on SDS-PAGE and proteins were extracted by in gel digest.

In gel digest
Protein digest and mass spectrometry was perform as previously described. 81 Briefly, the gel slices were cut into 1 mm 3 1 mm cubes, destained 3 times by first washing with 100 ml of 100 mM ammonium bicarbonate for 15 minutes, followed by the addition of equal volume acetonitrile (ACN) for 15 minutes. The supernatant was collected, and samples were dried using a speedvac. Samples were then reduced by mixing with 200 mL of 100 mM ammonium bicarbonate-10 mM DTT and incubated at 56 C for 30 minutes. The liquid was removed and 200 ml of 100 mM ammonium bicarbonate-55mM iodoacetamide was added to gel pieces and incubated covered at room temperature for 20 minutes. After the removal of the supernatant and one wash with 100 mM ammonium bicarbonate for 15 minutes, equal volume of ACN was added to dehydrate the gel pieces. The solution was then removed, and samples were dried in a SpeedVac. For digestion, enough solution of ice-cold trypsin (0.01 mg/ml) in 50 mM ammonium bicarbonate was added to cover the gel pieces and set on ice for 30 min. After complete rehydration, the excess trypsin solution was removed, replaced with fresh 50 mM ammonium bicarbonate, and left overnight at 37 C. The peptides were extracted twice by the addition of 50 mL of 0.2% formic acid and 5% ACN and vortex mixing at room temperature for 30 min. The supernatant was removed and saved. A total of 50 mL of 50% ACN-0.2% formic acid was added to the sample, and vortexed again at room temperature for 30 min. The supernatant was removed and combined with the supernatant from the first extraction. The combined extractions are analyzed directly by liquid chromatography (LC) in combination with tandem mass spectroscopy (MS/MS) using electrospray ionization.

LC-MS analysis
Trypsin-digested peptides were analyzed by ultra-high-pressure liquid chromatography (UPLC) coupled with tandem mass spectroscopy (LC-MS/MS) using nano-spray ionization. The nanospray ionization experiments were performed using a Orbitrap fusion Lumos hybrid mass spectrometer (Thermo) interfaced with nano-scale reversed-phase UPLC (Thermo Dionex UltiMateä 3000 RSLC nano System) using a 25 cm, 75-micron ID glass capillary packed with 1.7-mm C18 (130) BEH TM beads (Waters corporation). Peptides were eluted from the C18 column into the mass spectrometer using a linear gradient (5-80%) of ACN (Acetonitrile) at a flow rate of 375 mL/min for 1h. iScience Article For CoIP assays, cells lysates (as prepared above) was incubated with capture antibodies for 3 hours at 4 C, followed by the addition of Protein A or Protein G beads to capture antibody bound protein-protein complexes. Bound proteins were then eluted through boiling at 100 C in sample buffer.

Quantitative immunoblotting
For immunoblotting, protein samples were boiled in Laemmli sample buffer, separated by SDS-PAGE and transferred onto 0.4 mm PVDF membrane (Millipore) prior to blotting. Post transfer, membranes were blocked using 5% Non-fat milk or 5% BSA dissolved in PBS. Primary antibodies were prepared in blocking buffer containing 0.1% Tween-20 and incubated with blots, rocking overnight at 4 C. After incubation, blots were incubated with secondary antibodies for one hour at room temperature, washed, and imaged using a dual-color Li-Cor Odyssey imaging system.

Immunofluorescence and confocal microscopy, image analysis
Cells were fixed using À20 C methanol (or 4 C paraformaldehyde, PFA) for 20 to 30 min, rinse with PBS, and then permeabilized for 1h using blocking/permeabilization buffer (0.4% Triton X-100 and 2 mg/mL BSA dissolved in PBS). Primary antibody and secondary antibody were diluted in blocking buffer and incubated with cells for 1 hr each. Coverslips were mounted using Prolong Gold (Invitrogen) and imaged using a Leica SPE CTR4000 confocal microscope.

Anchorage-dependent colony formation assay
Anchorage-dependent growth was monitored as described previously. 21,82 Briefly, anchorage-dependent growth was monitored on solid (plastic) surface. Approximately 2,000 parental or GIV-KO HeLa cells or GIV-KO cells stably expressing WT GIV, GIV F1685A, or GIV F1719A were plated in 6-well plates and incubated in 5% CO2 at 37 C for $2 weeks in 2% FBS growth media in the presence of 10 nM Dox. After every three days, media were changed with fresh media containing 10 nM Dox. Colonies were then stained with 0.005% crystal violet for 1 hr. Entire plate surface area was scored for colonies and each treatment was done in triplicate and repeated thrice.

Cell cycle and apoptosis analyses
Cell cycle analysis and apoptotic cell quantification was performed using the Guava cell cycle reagent (Millipore Sigma) or the annexin V/propidium iodide (PI) staining kit (Thermo Fisher Scientific), respectively, according to the manufacturer's instructions. Cells were quantified on a BD LSR II flow cytometer and analyzed using FlowJo software (FlowJo, Ashland, OR, USA).

Long amplicon PCR
Genomic DNA extraction was performed using the genomic-tip 20/G kit (Qiagen, Cat no. 10223, with corresponding buffer sets) per the manufacturer's directions. This kit has the advantage of minimizing DNA oxidation during the isolation steps, and thus it can be used reliably for isolation of high molecular weight DNA with excellent template integrity to detect endogenous DNA damage using LA-qPCR. After precise quantitation of the DNA by Pico Green (Invitrogen Cat no. P7589) in a 96-well black-bottomed plate, the genomic DNA (500 ng) was digested with the E. coli enzymes Fpg and Nei (New England Biolabs) in reaction volume of 50 mL using Buffer 1 from NEB (with 1 mM MgCl2) as the common buffer to induce strand breaks at the sites of the unrepaired oxidized base lesion. Gene-specific LA-qPCR analyses for measuring DNA damage were performed using Long Amp Taq DNA polymerase (New England Biolabs, Cat no MO323S). The numbers of cycles and DNA concentrations were standardized in each case before the actual ll OPEN ACCESS