PP2A functions downstream of ATM to regulate AKT

In a previous study from our laboratory, we demonstrated that ATM indirectly regulates the phosphorylation of AKT at S473 via an okadaic acid (OA) sensitive phosphatase [21]. It is known that several oncogenic DNA viruses such as SV40 and Polyoma interfere with PP2A activity in infected cells through the expression of small t-antigen and PyMT that block PP2A activity by replacing the B subunit [39,40,41] (Supplementary Fig. S1A). After examining the effect of an ATM kinase inhibitor (ATMi) in a panel of tumor cell lines including human U87 and U1242 glioma cells as well as HEK293 and HEK293T carcinoma cells, inhibition of AKT phosphorylation was observed across the board relative to untreated cells with the exception of HEK293T cells (Fig. S1B). HEK293T expresses SV40 t/T antigens, which inversely correlates with the lack of pAKT inhibition by ATMi, suggesting similar or overlapping mechanism of action. To build on this finding, we infected U87 and U1242 cells with retroviruses expressing PyMT or not (Fig. S1C, D). ATMi did not appear to reduce the levels of pAKT in glioma cells overexpressing PyMT and was unaffected by ionizing radiation (IR) [21, 42]. The presence of the PyMT gene in these cell populations was verified by PCR (Fig. S1E). We confirmed the ability of OA to increase the levels of phosphorylated AKT [21], this time in a nuclear extract with nM concentrations suggesting that PP2A is acting on AKT in the nucleus (Fig. S1F), perhaps as a complex in direct physical association with AKT [43]. Altogether, these results suggest that ATM regulates the phosphorylation of AKT indirectly via the inhibition of PP2A activity.

Generation of PR65 KO MEFs complemented with S401 mutant alleles

Ruediger et al. demonstrated that PR65 is an essential gene, since knocking out PR65 resulted in embryonic lethality [32]. We generated homozygous PR65 conditional knockout (CKO) mouse embryonic fibroblasts (MEFs) after breeding heterozygous 129S-PPP2R1Atm1.1Wltr/j mice to obtain 129S-PPP2R1Atm1.1Wltr/tm1.1Wltr (CKO/CKO) mice. Floxing out PR65 exon 5 and 6 generates PR65 KO MEFs (Fig. 1A). MEFs generated from 129S-PPP2R1Atm1.1Wltr/tm1.1Wltr mouse embryos were immortalized by passaging once per week [33], and infecting the cells with a lentivirus expressing human TERT. We generated two site-specific mutations at the S401 codon of PR65 using the retroviral vector pMIG-Aalpha WT as template [36]; a S401A mutant that cannot be phosphorylated and a S401D phospho-mimetic mutant. The MEFs were then separately infected with the retroviruses expressing Flag-tagged PR65 WT, S401A or a S401D mutant in addition to GFP (via an IRES). The endogenous PR65 alleles were then floxed out by infecting the cells with an adenovirus expressing Cre recombinase (Ad–Cre). Complete floxing was confirmed by PCR (Fig. 1B). MEFs that were not complemented with virus expressing PR65 did not survive after infection with Ad–Cre (Fig. 1C), in line with that PR65 is essential in the mouse [32]. Pooled MEFs from each infection were shown to be close to 100% GFP+ (Fig. 1D). Western blotting with anti-PR65 and -Flag antibodies showed that at the protein level all endogenous PR65 (Fig. 1E, left panel; lower band), was replaced with the slightly larger Flag-PR65 protein expressed from the retrovirus. Altogether, we generated a panel of isogenic WT, S401A, and S401D cell populations expressing PR65 at approximately the same levels as endogenous PR65 (compare band intensity of PR65 doublet (Fig. 1E, right panel; lane 1).

Fig. 1

Generation of PR65 KO MEFs complemented with S401 mutants. A Targeting strategy for generating PR65 conditional knockout (CKO) MEFs. Expression of Cre recombinase in these PR65-CKO MEFs would generate PR65 KO MEFs by the deletion of exon 5 and 6. PR65-CKO MEFs were infected with retroviral vector pMIG-Aalpha, expressing Flag-tagged PR65 WT, S401A or S401D mutant in addition to GFP. This was followed by infection of cells with Ad-CMV-Cre recombinase to knock out the endogenous PR65 alleles. B Genomic DNA was isolated from PR65 KO MEFs (WT, S401A and S401D) and PCR amplified. PCR screening with primers P1, P2, and P3 shown in A generated a 263-bp product for the WT allele and a 417-bp product for the CKO allele. In cells where Cre was expressed, primer pair P2–P3 generated a 587-bp product for the KO PR65 allele. C PR65 is essential for the survival of MEFs. PR65 CKO MEFs were infected with an adenovirus expressing Cre recombinase. Cells were stained with crystal violet after one week to assess cell viability. PR65–CKO MEFs infected with Ad–CMV–Cre did not survive whereas MEFs complemented with PR65flox/E64G did. PR65flox/E64G MEFs express a known human tumor-associated PR65 mutant [32], constructed the same as S401A and S401D MEFs, and here used as control cells expressing mutant PR65. D Plasmid map of pMIG-Aalpha WT with mutations marked (top) and expression of GFP via an IRES in WT, S401A, S401D, and E64G MEFs with endogenous PR65 alleles KO. Scale bar is 10 μm. E Western blot verifying the Flag-PR65 expression of site-specific mutants and floxing of endogenous PR65 in MEFs. Left panel: explanation of the protein bands seen by western blotting. Right panel: whole cell extracts of GFP+ sorted PR65 WT, S401A and S401D cells infected with Ad-CMV-Cre or not were separated on an SDS-PAGE gel and analyzed by western blotting using anti-PR65 antibody, anti-Flag and anti-GAPDH (loading control) antibodies

ATM kinase phosphorylates PR65 in vitro

The PR65 S401 residue is part of an –S/Q-ATM kinase signature motif highly conserved in PR65’s among mammals including human and mouse and is a bona fide substrate for ATM phosphorylation [31]. To obtain an abundant source of co-expressed ATM and PR65 for further analysis and experimentation, we first attempted to isolate stably transfected HEK293 cells co-expressing mRuby2–PR65 and YFP–ATM after selection with G418 (expressed from YFP–ATM) and isolating cells expressing both YFP and Ruby. However, we were only able to recover YFP–ATM/S401A cells (Fig. S2A), suggesting that the overexpression of ATM with either S401D or WT was not tolerated.

To determine whether ATM phosphorylates PR65 in vitro and to confirm previously documented results [31], YFP–ATM was immunoprecipitated from HEK293 cells expressing YFP–ATM only (Fig. S2B) and utilized for in vitro kinase assays with [γ-32P] ATP and GST-PR65FL as substrate. The result shows that ATM kinase phosphorylates PR65 in vitro and that phosphorylation is significantly reduced in the presence of an ATMi (Fig. S2C, compare lanes 5 and 6). Similarly, GST-p531–100, a known substrate of ATM kinase at p53 S15 [44], was also phosphorylated and inhibited in the presence of the ATMi. This result confirms that ATM kinase is able to phosphorylate PR65 in vitro in line with a previous report [31].

A phospho-mimetic (S401D) PR65 mutant prevents the association with ATM, PP2A catalytic subunit and AKT

Goodarzi et al. showed that PP2A exerts its control over ATM by interacting via the PR65 subunit in a manner lost in irradiated cells [25]. To determine whether the S401D amino acid replacement affected the interaction with ATM, we immunoprecipitated over-expressed Flag-PR65 WT, -S401A, and -S401D, respectively, from transduced HEK293 cells. Flag-PR65 IP’s were then analyzed for the presence of endogenous ATM, the catalytic subunit of PP2A (PP2A-C), and AKT. We found an association of ATM, PP2A-C, and AKT with the WT and the nonphosphorylatable S401A but not with S401D (Fig. 2A). The interaction of ATM with PR65-WT and -S401A, but not -S401D, suggests that the PP2A Core enzyme (AC) is in a complex with ATM, in agreement with previous work [25]. That the S401D phospho-mimetic disrupts this complex is new information and fits well with the dissociation of ATM and PP2A complex after irradiation and the phosphorylation of PR65 S401 by ATM [25, 31]. That AKT is found complexed with PR65 is no surprise [43], but that S401D disrupts this interaction is intriguing, even though we cannot tell from this result whether ATM and AKT are in the same or separate PR65 complexes.

Fig. 2

Phospho-mimetic S401D prevents the association of PR65 with ATM, AKT, and PP2A catalytic subunit. A Whole cell lysates of HEK293 cells stably expressing Flag-PR65 (WT, S401A and S401D) were immunoprecipitated with anti-Flag beads. Immunoprecipitates (left panel) and whole cell extracts (right panel) were separated on an SDS-PAGE gel and analyzed by western blotting using anti-Flag, anti-ATM, anti-AKT, and anti-PP2A-C antibodies. Fold protein levels relative to WT is shown above each panel. B MD simulations suggest that a conformational change occurs when S401 is phosphorylated. (Left) Top view of the in silico S401 phosphorylated PR65, after 50 ns, aligned to the original experimental structure (PDB ID 2IAE) with a root-mean square deviation (RMSD) of 8.05 Å. PR65 (pS401) is shown in magenta, with pS401 represented as stick figures. PR65 from protein database (PDB ID 21AE) is shown in green. (Right) Top view of the unphosphorylated control structure of PR65 subunit, after 50 ns, aligned to the original experimental structure taken from PDB ID 2IAE (RMSD = 6.51 Å). Unphosphorylated control PR65 is shown in cyan, and the original experimental structure shown in green. Molecular modeling studies also indicate that phosphorylation at S401 could cause a conformational change which could affect its association with a binding partner. C Two 50 ns molecular dynamics simulations were carried out using the S401 phosphorylated PR65 chain as well as the unphosphorylated structure. Comparing only the amino acid residues surrounding S401 (residues 375–454), the RMSD of the phosphorylated subunit and the unphosphorylated experimental structure was found to be 4.448 Å while the RMSD of the control and the unphosphorylated experimental structure was 2.55 Å. At the end of the simulation, the distance between α-carbons for residues S401 and A431 for the unphosphorylated original structure (PDB ID 2IAE), the S401 phosphorylated structure, and control structure were 10.873 Å (green), 15.321 Å (magenta) and 9.498 Å (cyan), respectively, with the first two structures shown (top panel). Initially, there was a readjustment phase in the phosphorylated MD simulation in the region surrounding the phosphorylated S401 residue after which it stabilized at 18 ns (bottom panel). Altogether, this result suggests that phosphorylation of PR65–S401 could affect the interaction with its binding partners and result in the dissociation of the holoenzyme

To support a critical role for S401 phosphorylation by ATM in the disruption of such physical interaction we performed molecular modeling studies based on the known PR65 crystal structure (PDB ID 2IAE). Two 50 ns molecular dynamics runs were carried out with phosphorylated S401 PR65 as well as the unphosphorylated structure (Fig. 2B, C). Our result suggests that phosphorylation at S401 induces a local movement of the α-helices on either side of the helix (residues 375–454) encompassing S401 ultimately causing a conformational change of the entire PR65 protein, particularly at its N-terminal portion. This conformational change results in the distortion of the horseshoe shape of the subunit by 4.448 Å as compared to the unphosphorylated structure (15.321–10.873). Altogether, the modeling suggests that phosphorylation at S401 could cause the dissociation of the physical interaction with its binding partners ATM and AKT.

Growth and radiation signaling are altered in S401 mutant cells

We then examined whether AKT signaling and growth was affected in response to insulin stimulation and radiation [21]. WT, S401D, and S401A MEFs were serum-starved prior to stimulation with insulin followed by western blotting for pAKT (S473) (Fig. S3). We found that S401D cells responded more robustly to insulin when pAKT levels were determined over the course of 60 min (Fig. S3A), which correlated with an elevated growth rate of S401D relative to WT and S401A cells (Fig. S3B). Similar to AKT signaling, ERK signaling is associated with prosurvival and increased cell growth after low radiation doses, which in part is regulated by ATM [45, 46]. When we examined ERK signaling in unstimulated cells we observed elevated pERK1/2 levels in S401A and more so in S401D cells compared with WT, suggesting that AKT and ERK signaling are positively affected when S401 is mutated, with S401D cells being affected the most (Fig. S3C). Furthermore, after a single radiation dose of 2 Gy, S401 mutant cells showed dampened responses when analyzed for both γ-H2AX and pAKT by western blotting (Fig. S3D). It is well-established that ATM phosphorylates H2AX at S139 (γ-H2AX) in response to radiation at the sites of DSBs [47]. Furthermore, PP2A modulates DSB repair via the dephosphorylation of γ-H2AX to reverse chromatin accessibility [30]. On examining the γ-H2AX levels in more detail, we found that in WT cells the levels of γ-H2AX peaked at 15–30 min post irradiation and then declined. However, H2AX phosphorylation was reduced in both S401A and S401D cells with the latter trending even higher than WT and S401A at 60 min (Fig. S3D). Similarly, pAKT levels increased after IR in WT and less so in S401A and S401D cells. These results suggest that after low dose radiation, S401 mutant cells have dampened pAKT and γ-H2AX levels likely as a result of malfunctioning DDR. In addition, S401D seems to overcome this initial blockade with a delayed response and increased γ-H2AX levels at later times. Altogether, we conclude that S401 mutant cells have aberrant responses to insulin and radiation yet distinct from each other, resulting in altered DDR signaling and proliferative activity compared with WT cells. We recognize that these experiments need further in-depth investigation and analyses.

PR65 mutants have impaired G2/M checkpoint resulting in aberrant mitosis and elevated levels of mitotic catastrophe

PP2A plays distinct roles at different stages of mitosis while associated with different B subunits [48, 49]. PP2A regulates mitotic entry and exit, as well as playing a role in the protection of centromeres by inhibiting PLK1 and Aurora A kinases [50,51,52,53,54]. Taking into consideration the importance of PP2A in mitosis and elsewhere in the cell cycle, we examined whether S401 mutant cells show any chromosomal aberration or perturbations in mitosis and cell cycle checkpoints. We found elevated basal and radiation-induced levels of chromosomal aberrations in S401 mutants (Fig. 3A). The mitotic index (cells undergoing mitosis/total number of cells) was reduced in WT cells after radiation but not in S401 mutant cells, suggesting that the radiation-induced G2/M check point is intact in WT cells, as expected, but not functioning properly in the mutants (Fig. 3B). Furthermore, S401 mutants had significantly increased numbers of basal and radiation-induced (2- to 3-fold) aberrant mitoses (Fig. 3C). To support this finding, live cell imaging of MEFs expressing a histone H2B-mCherry reporter revealed that the S401 mutant cells exhibited abnormal mitoses including the formation of micronuclei and elevated levels of mitotic catastrophe, and the formation of tetraploid cells whose frequency increased after exposure to radiation (Fig. 3D). Again, S401D cells showed more aberrations than S401A. Furthermore, in undamaged cells, the length of mitosis in S401 mutant cells was significantly longer than in WT cells (Fig. 3E). To examine mitosis in more detail, we then focused on exit from mitosis. PLK1 is known to be critical during G2/M entry as well as for successful chromosome separation and exit from mitosis [55]. Normally, pPLK1 (T210) is elevated in mitosis and upon exit is dephosphorylated by PP2A [56]. A nocodazole blockade synchronized the cell panel in mitosis and following washout (Fig. S4A), we collected cells for western blotting with anti-pPLK1 (T210) antibody (Fig. S4B). Mitotic S401 mutant cells showed elevated pPLK1 levels compared with WT at 1 h and exited mitosis faster than WT cells with S401D showing the most pronounced effect (Fig. S4C). Together, both mitotic entry and exit are abnormal in S401 mutant cells as reflected in significantly longer mitosis and clear signs of struggle to get through mitosis.

Fig. 3

PR65 mutants accumulate mitotic aberrations leading to mitotic catastrophe and radiosensitization. A–C S401 mutant cells have defective G2/M checkpoint. MEFs were exposed to 5 Gy of ionizing radiation, fixed after 24 h and stained with DAPI to label nuclei. Mitotic cells and total number of cells were counted. The mitotic index was calculated by dividing the number of cells undergoing mitosis in a population to the number of cells not undergoing mitosis. Five separate fields were assessed for each group Scale bar is 10 μm. B Mitotic indices were quantified from A. Statistical analysis was carried out using unpaired, two-tailed t tests. Error bars; mean ± SEM. p values expressed as *(p < 0.05) were considered significant; NS, not significant (p > 0.05). WT: untreated vs IR; p = 0.0251. C Aberrant mitoses were quantified from A. Five separate fields were assessed for each group (n = 5). Error bars; mean ± SEM. Statistical analysis was carried out using unpaired, two-tailed t tests. p values expressed as *(p < 0.05) were considered significant; NS, not significant (p > 0.05). IR: WT vs S401D; p = 0.0228 D MEFs were infected with a virus expressing H2B–mCherry to monitor chromatin structure in mitosis. MEFs were exposed to 5 Gy and subjected to live-cell video imaging for 8 h on the Zeiss Cell Observer Spinning Disc confocal microscope. Representative still images (D and quantified mitotic length E are shown. Scale bar is 10 μm. F S401 mutant cells are more radiosensitive than wild-type MEFs. MEFs were exposed to 1, 2 or 5 Gy and radiosurvival determined at 10 days. Surviving fractions were determined by crystal violet staining and colony counting. Data points, surviving cells plotted as fraction of control (- irradiation). The results are presented as mean ± SEM, (n = 4). Statistical analysis was carried out with longitudinal ANOVA on the survival fractions. p values expressed as *(p < 0.05) were considered significant. WT vs. D, p = 0.0302

To get better insight into what occurs during S-phase we then examined the recovery of collapsed replication forks in the cell panel by determining pRPA32 (S4/8) and γ-H2AX levels after the release from a hydroxyurea (HU) blockade (Fig. S4D, top panel). A previous study showed that PP2A is critical for the dephosphorylation of pRPA32, cell cycle reentry and efficient DSB repair presumably by HRR [57]. We found that S401 mutant cells had reduced γ-H2AX levels after HU release, which peaked at ~ 1 h post washout (Fig. S4D, bottom left panel). Perhaps more importantly, S401D cells showed a delay in the removal of pRPA32 (Fig. S4D, bottom right panel) indicative of an alteration in PP2A activity during fork recovery. In addition, a colony-forming radiosurvival experiment demonstrated that S401D cells were also significantly more radiosensitive than WT cells, with S401A trending more radiosensitive (Fig. 3F). The same pattern of sensitivity was seen when the cell panel was exposed to camptothecin (CPT), a topoisomerase I inhibitor, known to produce DSBs preferentially in S and G2 (Fig. S5).

Altogether, increased chromosomal aberrations seen in untreated S401 mutant cells (and more so after radiation) are suggestive of impaired G2/M checkpoint and aberrant mitosis (entry, duration, and exit), resulting in elevated levels of mitotic catastrophe and increased sensitivity to DNA damage. In addition, replication fork recovery and DSB repair were impaired, especially in the S401D cells. It was however unexpected to see that the impact on survival was relatively minor considering the major negative effects on the G2/M checkpoint and increased mitotic struggle seen in S401 mutant cells.

S401 mutants have aberrant DSB repair

In addition to reversing changes in chromatin structure caused by H2AX phosphorylation, PP2A is believed to facilitate NHEJ by dephosphorylating KU70/80 and DNA-PKcs, which are critical participants in NHEJ [14], and for promoting their release from DNA [58]. Furthermore, PP2A restores kinase activity of phosphorylated, inactive DNA-PKcs in vitro to regulate NHEJ [59]. To determine how S401 alterations might affect DSB repair, we then carried out in vivo as well as in vitro end joining assays. First, in a reporter assay, WT, S401A and S401D mutant cells were transfected with a linearized pCSCMV:tdTomato plasmid with incompatible restriction endonuclease DNA ends unable to ligate. Once the plasmid is repaired and circularized, the red fluorescence protein (RFP) expressed is a measure of end joining activity (Fig. 4A). Post transfection, cells were collected at various times and RFP+ cells quantified. At 16 h the S401A cells had almost twice the number of RFP+ cells relative to WT whereas the S401D cells had significantly less than the WT cells throughout (Fig. 4B, top panel). To examine early repair, we quantified the circularized plasmid at 1 and 4 h after transfection and determined end joining yield by quantitative PCR (Fig. 4B, bottom panel). The PCR result is in agreement with what we observed with the reporter assay at early times, i.e., S401D cells have compromised end joining. S401A cells probably did not have enough time to translate RFP into showing elevated repair over WT at 4–8 h that was seen at 16 h by FACS.

Fig. 4

DSB repair is abnormally high in S401A cells while being suppressed in S401D. A Scheme for in vivo end joining (EJ) assay. Linearized Tomato expression plasmid was transfected into MEFs followed by quantification of RFP+ cells (Nexcelom Cellometer) and PCR. B Quantification of EJ by RFP flow cytometry (top panel). S401A cells have super-active EJ at the later times after transfection whereas S401D cells have suppressed EJ throughout. qRT-PCR assay was carried out to determine early EJ (bottom panel). The y axis denotes the relative levels of repaired DNA (n = 3). EJ in S401D cells was suppressed by 75% at 4 h in line with the results by RFP flow. C In vitro end joining. End joining of an internally labeled (*) plasmid substrate with partially complementary ends in MEF extracts (left panel). Following incubation for the indicated times, substrates were cut with BstXI and TaqI and analyzed on sequencing gels (middle panel). Gap-filling products (42-base fragment) and resection-based products (34-base fragment) were quantified. The 24- and 16-base fragments are the corresponding head-to-head end joining products. Graph shows sum of the major 34-base fragments in each case (right panel). D Mutant S401 cells show aberrant IR-induced foci with delayed disappearance of 53BP1 and γ-H2AX foci in S401 mutant cells (left panel). MEFs were seeded in a chambered slide. Cells were exposed to 2 Gy of ionizing irradiation and fixed after 0.5 and 5 h. Cells were immuno-stained with anti-53BP1 and anti-γ-H2AX and counterstained with DAPI (blue) to label nuclei. Representative images 53BP1 (green), γ-H2AX (red) and DAPI (blue). Images were acquired on Zeiss LSM 710 confocal microscope at × 63 power. Repair foci remain significantly longer in S401A and -D (esp. D) than in WT cells suggesting aberrant DSB repair. Delayed disappearance of 53BP1 (top, right panel) and γ-H2AX (bottom, right panel) foci in S401 cells post irradiation. Five fields were assessed in at least 2 independent experiments. Error bars: mean ± SEM. Statistical analysis was carried out using unpaired, two-tailed t tests. p values are shown as ***(p < 0.005) is considered significant. Scale bar is 10 μm. E S401A and D cells are impaired in homologous recombination using the DR-GFP assay (top, left panel). PR65-DR-GFP (WT, S401A and S401D) cells were infected with Ad-SceI (bottom, right panel) or not (top, right panel) and 72 h later, cell populations analyzed by GFP flow cytometry after incubation in 0.1% Triton-X-100 in PBS. The results (bottom, left panel) are presented as mean ± SEM (30,000 events), n = 3, * p < 0.05 relative to WT Scale bar is 10 μm. F Reduced Rad51 and pRPA foci in S401 mutants post-CPT treatment. MEFs were treated with 100 nM of CPT for 6 h, fixed and immunostained with anti-pRPA32(S4/S8) and anti-Rad51 antibodies, and counterstained with DAPI (blue). Representative images Rad51 (red), pRPA (green) and DAPI (blue) images acquired on a Zeiss LSM 710 confocal microscope at 63 × power (left panel). Quantification of Rad51 and pRPA32 (S4/8) foci (right panels). Rad51 foci per cell were counted in 5 separate fields in at least two independent experiments. Error bars; mean ± SEM. Statistical analysis was carried out using unpaired, two-tailed t tests. p values expressed as *(p < 0.05), ***(p < 0.005), and #(p < 0.001) were considered significant. For Rad51 foci, WT vs. S401A; p = 0.0008, WT vs. S401D; p = 0.0011. For pRPA foci, WT vs. S401D; p = 0.0443

To determine the possible effect of PP2A-mediated dephosphorylation on end joining in vitro, a 32P-oligonucleotide ligated into a plasmid substrate bearing partially complementary 5′ overhangs was incubated in whole-cell extracts from WT, S401A, and S401D cells for 1.5 or 6 h (Fig. 4C). As in our previous work [60, 61], the predominant 34-base head-to-tail product as well as the 16-base head-to-head product reflects 3′ resection and annealing at a 4-base micro-homology (Fig. 4C, left panel). We observed dramatically reduced end joining activity with the extract from S401D, whereas the S401A extract markedly enhanced it and yielded an additional 42-base product corresponding to gap filling without resection (Fig. 4C, middle and right panels). Thus, while the critical PP2A dephosphorylation targets are currently unknown these results show that using S401 mutants as surrogates for mimicking (de)phosphorylated S401 is important for end joining by reporter as well as in cell-free in vitro assays. In line with these results, S401D cells showed abnormal, extensive and more persistent 53BP1 foci after radiation compared with WT (Fig. 4D, left panel). 53BP1 is critical for directing DSB repair towards NHEJ at the expense of HRR [62]. 53BP1 foci co-localized with γ-H2AX, suggesting that NHEJ is significantly altered and delayed in S401D cells. Furthermore, S401A cells had fewer but more pronounced foci at 5 h than either WT or S401D (Fig. 4D, right panel). Altogether, S401 mutant cells have significantly altered end joining activity in quality as well as temporal execution, yet S401D and S401A differed in that the latter showed super-active end joining whereas S401D was severely compromised.

S401 mutants have impaired HRR

Since NHEJ and HRR are in competition in S/G2 of the cell cycle and we observed irregular end joining in the S401 mutants, we then examined HRR in the cell panel. The PP2A holoenzyme associated with the B55 subunit is implicated in HRR via the modulation of ATM phosphorylation and the G1/S checkpoint [63], as well as the dephosphorylation of pRPA32 mediated by PP2A after replication fork recovery, which might impact HRR [57]. To determine the importance of the S401 residue in HRR, we generated CKO/CKO MEFs with a chromosomally integrated DR-GFP repair reporter [38], followed by infection with either WT, S401A and S401D retrovirus and floxing of the endogenous PR65 gene with Ad-CMV-Cre (Fig. S6) as before (see Fig. 1B, E). We noticed that HRR was significantly reduced in S401A and S401D cells represented by lower GFP+ events (Fig. 4E). During HRR, single stranded DNA generated during DNA resection is first coated with RPA and then replaced by RAD51 during Holliday junction formation and resolution critical for efficient HRR [14, 64]. Therefore, we treated the cells with CPT and then co-stained for pRPA32 (S4/8) and RAD51 (Fig. 4F). Both S401A and S401D cells had ~ 40% less RAD51 foci compared with WT cells (Fig. 4F, right-top panel) with pRPA also trending lower in S401D relative to WT and S401A (Fig. 4F, right-bottom panel). This finding is in line with the result with the DR-GFP reporter and the conclusion that HRR is impaired in S401A and S401D cells relative to WT. Altogether, it is quite clear from our results so far that S401 mutant cells have rewired DDR and show aberrant DSB repair. Surprisingly, however, these mutants display fairly minor effects on cell survival after radiation and CPT treatment. This is particularly striking with the S401D cells since they are compromised in end joining as well as HRR.

Increased aberrant chromosomal DSB repair and translocation in S401 mutants

Alternative end joining (Alt-EJ, now referred to as Theta-mediated end joining; TMEJ) and other micro-homology repair pathways are believed to become engaged when c-NHEJ (classical end joining) and/or HRR are not fully functional or being disengaged [65,66,67]. However, TMEJ is highly mutagenic because of the promiscuous activity imposed by DNA polymerase Theta (Polθ)-mediated chromosomal repair and rearrangements through short (2–4 bp) stretches of micro-homology in resected DNA ends. To assess TMEJ in our PR65 cell panel we examined DSB repair quality after CRISPR/Cas9-mediated DNA cleavage at the Rosa26 locus and by chromosomal translocation between the Rosa26 and H3f3b loci on Chr. 6 and 11, respectively [66, 67]. The fact that the Rosa26 cleavage site is positioned in the middle of an XbaI site allows for the elimination of any DNA resealing products without indels, which if present would destroy the XbaI site, as well as remove DNA that was not cut by CRISPR/Cas9 (Fig. 5). As expected, XbaI-resistant PCR fragments remained after cleavage with Cas9–Rosa26/H3f3b gRNAs expressed but not without when examining events happening at the Rosa26 site (Fig. 5A). Cells not expressing Cas9–Rosa26/H3f3b gRNAs (Hygro) did not have any XbaI-resistant DNA as expected.

Fig. 5