Specific function of phosphoinositide 3-kinase beta in the control of DNA replication

The p110β Controls S-Phase Progression.

p110α regulates G1 entry and cell growth (1); both p110α and β regulate late G1 events and accelerate G1>S transition (6); however, no p110β-specific function has been described in cell division. To examine the potential p110β action in this process, we compared the division rates of NIH 3T3 stable cell lines expressing p110α or β active forms (Fig. 1A). Active p110β cells divided more rapidly (t_1/2 ≈18 h) than active p110α cells or controls (t_1/2 ≈24 h; Fig. 1B). In addition, although a small fraction of active p110α and β cells enter cell cycle after serum deprivation (6), only active p110β cells escaped cell contact inhibition in confluence (Fig. S1A). We also compared synchronous cell cycle progression in these cells. Cells were first serum-deprived (G0 arrest) and released by serum addition; using this protocol, NIH 3T3 cells reach S phase at ≈9 to 12 h postrelease (15). Active p110β cells were faster in terminating S phase than control or active p110α cells (Fig. 1C; Fig. S1B), as confirmed by calculation of S phase duration (4 ± 0.5 h for active p110β cells vs. 5.5–6 h for active p110α cells and ≈6 h for control cells); three distinct clones behaved similarly.

We also examined the consequences of reducing endogenous p110α and β activity using inactive K802R-p110α and K805R-p110β mutants (KR hereafter) (6). Expression of KR mutants in exponentially growing NIH 3T3 cells reduced PKB phosphorylation (pPKB, Fig. 1D) and affected cell division; we were unable to prepare stable KR-p110α or β lines. We expressed KR mutants by retroviral infection (95% efficiency), which yielded levels similar to endogenous p110 proteins (Fig. 1D). Cell division was significantly slower in KR-p110β cells (Fig. 1E), which remained in S phase for prolonged periods (Fig. 1F; Fig. S1C) and showed a longer S phase (≈6 h control cells; 6–6.5 h KR-p110α cells, ≈8 h for KR-p110β cells).

p110β expression did not vary appreciably throughout the cell cycle. We examined the consequences of reducing p110β expression using various shRNA and protocols in NIH 3T3 cells and human U2OS cells (Methods). Whereas efficient protocols for p110β deletion interfered with cell viability, partial p110β reduction permitted cell cycle progression studies. To reduce p110β expression in U2OS cells, we stably transfected pTER-shRNA vectors, which allow inducible shRNA expression (16). shRNA reduced p110α and β levels even before induction, but reduction was greater after doxycycline treatment (Fig. 1G). U2OS cells were synchronized at G1/S boundary by double thymidine block and examined S phase progression after release. We confirmed slower cell cycle entry in cells with reduced p110α or β levels (6); in addition, only the cells with reduced p110β levels remained in S phase for prolonged periods, showing a Gaussian peak at mid-S phase DNA content at 6–7 h postrelease (Fig. 1G).

We used the selective inhibitors PIK75 and TGX221 to inhibit p110α and β, respectively (17, 18). We confirmed inhibitor selectivity in NIH 3T3 cells (Fig. S2 A–D). Inhibition using 0.5 μM PIK resulted in complete blockade of S phase entry and triggered apoptosis (Fig. S2E), showing that p110α is needed for cell survival (19). p110α inhibition (0.08 μM PIK) near S phase permitted cell cycle entry (Fig. S2F) although it impaired G2/M entry, suggesting that p110α could be the isoform that acts in mitosis (1). This treatment nonetheless allowed S phase progression, as indicated by the increased proportion of S phase cells and displacement of the S phase population from near-G1 to near-G2 DNA content over the time course (Fig. S2F). In contrast, selective inhibition of p110β permitted G2/M entry but extended S phase compared with controls (Fig. S2F).

The p110β Activity Controls DNA Elongation.

To compare S phase progression rates more accurately we BrdU-labeled (1 h pulse) newly synthesized DNA in exponentially growing cells and collected cells at various times after BrdU deprivation. While most BrdU⁺ control and KR-p110α cells reached G2/M at 3 to 5 h, the majority of BrdU⁺ KR-p110β cells remained in S phase at 5 h (Fig. S3A). We examined the consequences of impaired p110β function on DNA elongation with the DNA combing assay (20, 21). We used PI3K inhibitors, as they permit p110α or β blockade in late G1 without affecting prior events. G0-synchronized NIH 3T3 cells were serum-released, treated with PIK 75 (0.08 μM) or TGX 221 (30 μM) at 7 h, BrdU-labeled (20 min) at 12 h, then collected to examine the replication profile (Fig. 2A). For each sample, we analyzed ≈30 MB of individual DNA fibers (>250 kb). TGX-treated cells showed 43% reduction in the length of BrdU tracks relative to controls, suggesting that p110β is required for normal replication fork progression; in contrast, elongation was not significantly affected by p110α inhibition (Fig. 2A). Median center-to-center distance between adjacent BrdU tracks, indicative of the initiation rate, was shorter in TGX- than in PIK-treated cells or in controls (Fig. 2B), consistent with cell activation of additional replication origins to compensate slow fork progression (21). The percentage of replication of individual DNA fibers was lower in TGX- (20.7%) than in control or PIK-treated cells (33.0 and 32.3%). These results suggest that p110β activity controls replication fork progression.

The p110β Is Located in the Nucleus and Controls PCNA Binding to Chromatin.

Since DNA replication occurs in the nucleus, we examined p110β localization. Subcellular fractionation (Methods) and immunofluorescence analysis showed that the majority of endogenous p110β, but not of α, concentrated in the NIH 3T3 cell nucleus (Fig. 2C). Both the nuclear p110β signal and the mainly cytosolic p110α signal decreased with selective shRNA (Fig. 2C; Fig. S3B). A similar distribution was observed in MEF, COS-7, HeLa and U2OS cells. These results indicate that p110β concentrates in the nucleus.

We examined the mechanisms by which p110β regulates replication. One of the first events required to initiate replication is MCM complex loading on origins (replication licensing, 8,13,22). We compared MCM loading to chromatin by cell fractionation on nuclear and chromatin extracts (23). Whereas in control cells, MCM 2/4 appeared on chromatin fractions in exponential growth, but not after GF starvation or in confluence, active p110α or β expression induced a similar and moderate enhancement of MCM 2/4 loading onto chromatin in starving and confluence conditions (Fig. S3C). Accordingly, KR-p110α or β expression induced a slight reduction in late G1 MCM loading (Fig. 3A). MCM loading onto chromatin is thus modulated to some extent by p110α and β, but is not selectively controlled by p110β.

The p110β Activity Regulates PCNA Loading onto Chromatin.

After replication origin activation, Polα binding to the ORC triggers primer DNA synthesis; elongation of DNA synthesis requires subsequent binding of PCNA that tethers the processive polymerases Polδ and ε to the DNA template (9, 14). In control synchronized NIH 3T3 cells, we observed PCNA appearance in chromatin extracts (22) as well as PCNA-Polδ association at ≈12 h after GF addition, at the onset of S phase (Fig. 3B). Active p110α cells behaved similarly; in contrast, active p110β expression accelerated PCNA binding to chromatin and PCNA-Polδ association (Fig. 3B). Moreover, expression of KR-p110β (Fig. 3C), reduction of p110β levels with shRNA (Fig. 3D; Fig. S3D) and p110β inhibition (Fig. S3 E and F) diminished PCNA loading onto chromatin as well as PCNA-Polδ association; interference with p110α only had a modest inhibitory effect. These data show that p110β controls PCNA binding to chromatin and to Polδ, providing a potential mechanism for DNA elongation impairment after interference with p110β function.

The p110β Activity Regulates p21^Cip Phosphorylation.

PCNA loads Pol δ and ε to the DNA template for efficient elongation; PCNA also binds p21^Cip through the same region, p21^Cip thus impairs PCNA association to Polδ/ε (22, 24). We examined PCNA-p21^Cip complex formation in cells with altered p110β activity. Whereas interference with p110α did not appreciably affect PCNA-p21^Cip complexes, active-p110β reduced (Fig. 3E) and inactive p110β (or p110β inhibition) increased PCNA-p21^Cip association (Fig. 3F, Fig. S4A). Phosphorylation of p21^Cip on T145 and Ser-146 phosphorylation (by PKB and PKC) regulates its dissociation from PCNA (25–28), nonetheless, in vivo T145 appears to be the critical residue (27, 28). We confirmed that T145 phosphorylation induced PCNA-p21^Cip dissociation in U2OS cells and NIH 3T3 cells (Fig. 4A); expression of the phosphomimetic D145-p21^Cip mutant reduced PCNA-p21^Cip association increasing PCNA binding to chromatin (Fig. 4A).

We also examined whether p110β regulates T145 phosphorylation. Whereas in control cells T145 was phosphorylated near S phase entry, both p110β shRNA and KR-p110β expression reduced pT145-p21^Cip levels (Fig. 4B; Fig. S4 B–D). In these assays we observed that interference with p110β activity also resulted in greater p21^Cip expression levels. p21^Cip is degraded after its release from PCNA (29); the higher p21^Cip levels in cells with impaired p110β function might be due to stabilization of p21^Cip in complex with PCNA. Both KR-p110β and p110β shRNA expression increased p21^Cip protein stability (Fig. S5A), whereas active p110β reduced p21^cip stability (Fig. S5B). p110β activity is thus needed for p21^Cip phosphorylation and dissociation from PCNA.

The p110β Regulates Nuclear PKB.

The PI3K effector PKB phosphorylates T145-p21^Cip (27, 28). We confirmed that PKB phosphorylates T145-p21^Cip in vitro (Fig. S6A) and examined whether p110β regulates PKB-mediated T145-p21^Cip phosphorylation. We found that expression of KR-p110β (or p110β inhibition) reduced S phase PKB kinase activity in vitro (Fig. S6 A and B). Western blot analysis of pPKB in extracts from synchronized NIH 3T3 cells expressing KR-p110β or treated near S phase with TGX221 confirmed that p110β regulates S phase pPKB, whereas p110α inhibition had a lesser effect (Fig. 4C); results were similar in U2OS cells (Fig. S6C). As an alternative approach, we examined pPKB by immunofluorescence. At 1 h postserum addition (G1 phase) pPKB concentrated at the cell membrane and was reduced by KR-p110α (Fig. S7A), whereas in S phase pPKB concentrated in the nucleus and was notably reduced by KR-p110β and p110β inhibition (Fig. 4D; Fig. S7B). Cell fractionation confirmed TGX inhibition of S phase nuclear pPKB (Fig. S7C).

We examined other PKB substrates in S phase; GSK3β phosphorylation was reduced by p110β inhibition, whereas FKHRL1 phosphorylation was p110α activity-dependent (Fig. S7D), as is the case in G1 phase (6). WB using anti-pPKB substrate Ab showed that p110β inhibition reduced phosphorylation of some PKB substrates in S phase cells (such as p21^Cip, Fig. S7C), while others were p110α-regulated (Fig. S7D). Results were similar using S phase U2OS cells treated with PI3K inhibitors and then fractionated (Fig. 4E); this assay also showed that p110α inhibition affected mainly cytosolic substrates and p110β nuclear substrates, suggesting that p110α and β control distinct PKB pools. p110β thus governs nuclear S phase PKB activity. Since p110α is activated at the G1/S boundary (6), the early timing of phosphorylation of some PKB substrates or their cytosolic localization might determine a p110α activity requirement for phosphorylation.

Based on p110β regulation of S phase nuclear pPKB-mediated p21^Cip phosphorylation, expression of the phosphomimetic D145-p21^Cip mutant in cells with impaired p110β activity could replace p110β activity in S phase. BrdU labeling of newly-synthesized DNA in exponentially growing cells expressing KR-p110β alone or in combination with D145-p21^Cip showed that D145-p21^Cip expression accelerated S phase progression in KR-p110β cells (Fig. 4F). D145-p21^Cip expression also increased PCNA-Polδ association and reduced PCNA-p21^Cip complexes in KR-p110β cells (Fig. S8A). Accordingly, A145-p21^Cip expression corrected PCNA-Polδ complexes in active p110β cells (Fig. S8B). Thus, expression of phosphomimetic p21^Cip mutants corrects the S phase defects of cells with altered p110β activity.

PI3Kβ Protein Regulates PCNA Loading onto Chromatin.

The recently described conditional p110β^−/− mouse phenotype and that of inactive p110β knock-in mice (7, 30) indicate that p110β kinase activity regulates mouse growth and tumor development and also that p110β has a kinase-independent function in embryonic development. Kinase-independent functions often reflect the ability of a protein to associate a necessary partner, as is the case for PI3Kγ in the control of cardiac stress response (31). We examined whether p110β expression (independent of its kinase activity) regulates DNA elongation, studying the extent of PCNA binding to chromatin after p110β inhibition or p110β knockdown. To improve p110β deletion, we transfected cells with puromycin-shRNA-encoding vectors, selected them for 48 h and immediately analyzed these asynchronous cultures (synchronization requires longer culture times) before reduction of cell viability. Pulse–chase BrdU analysis in exponentially growing NIH 3T3 cells showed that p110β inhibition reduced S phase progression, but p110β knockdown had a greater effect in decelerating S phase (Fig. 5A). PCNA loading onto chromatin was also reduced by p110β or PKB inhibition, but was drastically diminished by p110β knockdown (Fig. 5B).

We also analyzed asynchronous cultures of p110β^−/− immortalized mouse embryonic fibroblasts (MEF) reconstituted with WT or KR-p110β (7). KR-p110β MEF progressed through S phase more slowly than WT p110β MEF, although p110β^−/− MEF showed the slowest S phase progression (Fig. 5A). KR-p110β MEF had less chromatin-bound PCNA than controls, but PCNA loading was lowest in p110β^−/− MEF (Fig. 5B). These results suggest that PCNA loading onto chromatin and in turn S phase progression rate is further regulated via a kinase-independent p110β function.

pPKB was little affected by p110β deletion in asynchronous cultures (7). To define whether p110β controls nuclear PKB in S phase in these MEF, we synchronized cells at the G1/S border and examined them after release. In WT p110β-reconstituted MEF, pPKB was found mainly in the nuclear fraction in S phase; KR-p110β MEF behaved similarly but had lower nuclear active pPKB levels (Fig. 5C). Both nuclear pPKB and PKB were undetectable in p110β^−/− MEF (Fig. 5C), indicating that p110β expression might control PKB nuclear entry. Cytosolic pPKB was more abundant in p110β^−/− MEF, but they expressed lower levels of PTEN (Fig. 5C); this might represent a compensatory mechanism for p110β deletion. We also analyzed nuclear/cytoplasmic distribution of pPKB and PKB in NIH 3T3 cells to further examine whether p110β deletion reduces not only nuclear phospho-PKB but also nuclear PKB, as in MEF. p110β shRNA diminished but did not completely eliminate nuclear PKB (Fig. S8C). These results do not demonstrate, but suggest that PKB nuclear entry is facilitated by p110β expression, an aspect that requires further study. In contrast, both p110β inhibition and p110β shRNA expression clearly reduced S phase nuclear pPKB (Fig. S8C), further confirming the function of p110β in control of nuclear PKB activity in S phase.

PI3Kβ Protein Associates PKB and PCNA.

To determine whether p110β-dependent PKB nuclear activity is due to direct association, we studied PKB-p110β complex formation in cytosolic and nuclear fractions. Cells were fractionated as described (32), since the method used earlier (Fig. 2C) (33) destroys protein–protein interactions. NIH 3T3 cells were cotransfected with HA-gagAKT and WT-p110α or -β, collected at 12 h post-G0 release, and examined. Although PKB and p110α associated in cytosol, this association was lower than that of PKB and p110β, and was not found in the nucleus, where only PKB-p110β complexes were observed (Fig. S9A). We also analyzed association of endogenous proteins in synchronized NIH 3T3 cells collected at 12 h postserum addition. WB analysis of the fractions confirmed that p110α was mainly cytosolic and p110β was more abundant in the nucleus (Fig. 5D). Although immunoprecipitation concentrated the scarce nuclear p110α protein, endogenous PKB associated mainly with p110β in the nuclear fraction (Fig. 5D).

To identify other nuclear proteins that regulate DNA replication and associate to p110β, we performed a pull-down assay using mammalian GST-p110β; we obtained a number of candidate proteins including PCNA. Immunoprecipitates of endogenous PCNA from nuclear extracts contained associated endogenous p110β but not p110α (Fig. 5E); results were similar in a reciprocal assay (Fig. S9B). To determine whether the selective association of PCNA with p110β was due to a p110β-specific structural feature or to its subcellular distribution, we inserted a nuclear localization signal (NLS) in p85 and cotransfected it with myc-WT-p110α or -β, which increased their nuclear localization. Both nuclear p110β and α associated with PCNA, although p110β association to PCNA was greater than that of nuclear p110α (Fig. S9C). Therefore, in addition to its subcellular distribution, p110β has a structural advantage for association to PCNA.

Here, we describe a role for p110β in replication fork elongation in mammalian cells, providing an example of elongation control by extracellular signal-regulated molecules. The nuclear localization and function of p110β resembles that of class IV PI3K, which are recruited to DNA damage sites and mediate cell responses as DNA repair (34). Although some cell cycle phenotypes were moderate (Fig. 1), complete p110β elimination interfered with cell survival, and p110β function was studied in partial p110β deletion conditions. p110β regulated DNA replication through kinase-dependent and -independent mechanisms. p110β associated with PKB, and p110β activity regulated nuclear PKB-mediated p21^Cip phosphorylation, PCNA release, PCNA binding to Polδ and replication elongation. Interference with p110α activity had a slight inhibitory effect on p21^Cip phosphorylation, and might partially compensate for p110β activity-dependent functions. In addition, p110β associated with PCNA and controlled PCNA loading onto chromatin in a kinase-independent manner. Since PCNA loading onto chromatin is essential for DNA duplication, this kinase-independent function explains the greater division defects in cells with reduced p110β expression. The role of p110β in DNA replication could contribute to cause the early lethality (E2–3, ref.4) of p110β-deficient mice.

Complementary DNA and shRNA.

pSG5-p110αCAAX (active p110α), pSG5-HA-wt-PKB and -gag-PKB were described (5, 35). pCEF2-hp110βCAAX (active p110β) was a gift of Dr. Murga (Centro de Biología Molecular/CSIC, Madrid, Spain). PcDNA-Myc -WT and p21^Cip mutants were donated by Dr. Rössig (28). pcDNA Myc-S146A/T145A double mutant was generated using Quick Change Site-Directed mutagenesis (Stratagene). Myc-K802R-hp110α and myc-K805R-hp110β mutants were subcloned into pSG5 and pRV-IRES-GFP for retroviral infection (6). We used several specific short hairpin RNA (shRNA) directed to human or murine p110 sequences, each assay was performed at least with two shRNA, with similar results. These shRNA (6) were subcloned in pBluescript/U6 or in pTER vector; we used control shRNA that did not reduce p110α or β expression. We also used Pik3cb shRNA (Origene; Fig. 5). To prepare NLS-p85, the PKKKRKV sequence was inserted 3′ of the p85 sequence.

Cell Lines, Cell Culture, and Retroviral Transduction.

Active p110α and active p110β NIH 3T3 cells lines were described (6). KR-p110α and β mutations were transduced by transient transfection or retroviral infection. We generated pTER-p110α or pTER-p110β U2OS clones according to manufacturer's protocol (Invitrogen); shRNA expression was induced for 2 days (p110α) or 5 days (p110β) in medium plus doxycycline (6 μg/mL, Sigma). NIH 3T3 murine fibroblasts, U20S and COS7 cells were cultured as described (6). For retrovirus production, Phoenix cells were transfected using JetPei-NaCl (Qbiogene). MEF were donated by Drs. Zhao and Roberts (7) (Dana Farber Cancer Institute, Boston, MA).

Cell Cycle, BrdU Labeling, Immunofluorescence, and Dynamic Molecular Combing.

Immunofluorescence and NIH 3T3 G0 synchronization were as reported (15). Briefly, cells were incubated in serum-free medium (19 h) and released by serum addition. Cell cycle distribution was examined by DNA staining with propidium iodide and analyzed by flow cytometry (Beckman-Coulter) using Multicycle AV (Phoenix Flow Systems). Cells were synchronized at G1/S by double thymidine block (6) or using aphidicolin (22). To determine cell division time (t_1/2), cells were seeded at similar densities and counted at 24 h intervals. S phase duration was calculated considering t_1/2 (mean of n = 6) and the proportion of cells in S phase in exponential growth (mean of n = 12). S phase progression rates were examined in exponentially growing cultures incubated with 20 μM bromodeoxyuridine (BrdU; 1 h), chased at different times and stained with BrdU-FITC Ab (BD Biosciences), then examined by three-dimensional FACS.

For dynamic molecular combing, synchronized NIH 3T3 cells were treated with 0.08 μM PIK75 or 30 μM TGX221 at 7 h postserum addition; 20 min before harvest (12 h postserum addition), cells were treated with 20 μM BrdU. After harvest, cells were embedded in LMP agarose plugs (3 × 10⁶ cells/plug) and DNA fibers were purified and stretched on silanized coverslips as described (21). BrdU tracks were detected with rat monoclonal Ab (clone BU1/75; AbCys) and an Alexa 488-conjugated secondary Ab (Molecular Probes). DNA fibers were counterstained with mouse anti-ssDNA (MAB3034, Chemicon) and Alexa 546-secondary Ab (Molecular Probes). Signals were analyzed with MetaMorph.

Statistical analyses were performed using StatView 512+ (Calabasas, CA). Gel bands and fluorescence intensity were quantitated with ImageJ software. Statistical significance was calculated using Student's t test. For DNA combing, statistical analysis was performed with GraphPad Prism 5.0 (GraphPad Software).

For description of antibodies and reagents, cell lysis, subcellular fractionation, Western blotting, immunoprecipitation, and kinase assays, see SI Methods.