Mechanistic Analysis of a DNA Damage-Induced, PTEN-Dependent Size Checkpoint in Human Cells

Abstract

Following DNA damage, human cells undergo arrests in the G₁ and G₂ phases of the cell cycle and a simultaneous arrest in cell size. We previously demonstrated that the cell size arrest can be uncoupled from the cell cycle arrest by mutational inactivation of the PTEN tumor suppressor gene. Here we show that the cell size checkpoint is inducible by DNA-damaging chemotherapeutic agents as well as by ionizing radiation and is effectively regulated by PTEN but not by its oncogenic counterpart, PIK3CA. Mutational analysis of PTEN and pharmacological inhibition of Akt revealed that modulation of Akt phosphorylation is unnecessary for cell size checkpoint control. To discover putative PTEN regulators and/or effectors involved in size checkpoint control, we employed a novel endogenous epitope tagging (EET) approach, which revealed that endogenous PTEN interacts at the membrane with an actin-remodeling complex that includes actin, gelsolin, and EPLIN. Pharmacological inhibition of actin remodeling in PTEN^+/+ cells recapitulated the lack of size checkpoint control seen in PTEN^−/− cells. Taken together, these results provide further support for the existence of a DNA damage-inducible size checkpoint that is regulated by a major tumor suppressor, and they provide a novel Akt-independent mechanism by which PTEN controls cell size.

INTRODUCTION

A major focus of modern cancer research has been to determine the role of tumor suppressor gene pathways in the regulation of cell cycle arrest. The molecular mechanisms that enforce these cell cycle arrests are termed checkpoints and are enforced by several of the most commonly mutated tumor suppressors, including p53 and p16^INK4a. The study of checkpoint-dependent cell cycle arrest has focused primarily on the G₁/S and G₂/M cell cycle transitions. However, these arrests are almost invariably accompanied by a third, simultaneous arrest—an arrest in cell size.

The relationship between cell size arrest and the more conventional cell cycle arrests has not been investigated thoroughly, despite the fact that cancer cells are often aberrantly regulated in size. This phenotype is manifested in several clinical presentations, such as the formation of giant cells in many tumor types and the presence of unusually enlarged cells in tumor types such as hamartomas. Therefore, determination of the genetic and biochemical mechanisms that enforce cell size checkpoints is of fundamental importance in cancer biology.

Of the known tumor suppressor genes, the PTEN gene has been the most convincingly implicated in the control of mammalian cell size. Inherited mutations of PTEN cause a variety of related cancer predisposition syndromes collectively referred to as PTEN hamartoma syndrome (23), in which tumors are composed of enlarged cells. In Drosophila melanogaster, PTEN-deficient cells in the eye and wing are enlarged (16, 18, 24, 46). Additionally, cells and organs from conditional PTEN knockout (KO) mice are often oversized (2, 7, 14, 32). For example, tissue-specific deletion of PTEN in the mouse brain results in the formation of enlarged cells, leading to macrocephaly (2).

Human cells with targeted deletion of PTEN also have a notable size phenotype (34). After treatment with gamma irradiation, PTEN^+/+ cells arrest in the G₁ and G₂ phases of the cell cycle and simultaneously stop increasing in size. In contrast, otherwise isogenic PTEN^−/− cells also undergo cell cycle arrest but do not arrest their cell size. As such, PTEN^−/− cells arrested in either the G₁ or G₂ phases of the cell cycle continuously enlarge, ultimately reaching >20 times the size of their PTEN-proficient counterparts before detachment and death. Based on these data, we have proposed that PTEN controls a distinct radiation-induced cell size checkpoint that can be uncoupled from the radiation-induced G₁ and G₂ cell cycle arrests.

The mechanistic basis for the role of PTEN in cell size control remains mostly obscure. In mice, the large-cell phenotype is dependent on PDK1 and mTOR and independent of S6K (4, 5, 31). As most PTEN phenotypes are thought to occur via regulation of Akt activation, the effects of PTEN on cell size control are assumed to be dependent on this pathway as well. This assumption is based, in part, on the fact that the Akt kinase mTOR plays a known role in cell size regulation (21). However, whether Akt is an important effector of the PTEN cell size phenotype in mammalian cells has not been directly tested, due in part to technical difficulties in genetically inhibiting all three Akt isoforms simultaneously.

Examination of the cell size phenotypes of PTEN deficiency and the underlying molecular basis has substantial implications for understanding cell and cancer biology. Control of cell size has been almost entirely ignored from a mechanistic perspective, yet cell size is arguably one of the most obvious and important phenotypes in all of mammalian biology. Finally, although generally overlooked, an arrest in cell size is a critical component of cell cycle arrest. As most current anticancer agents function, at least in part, by inducing checkpoint-dependent cell cycle arrest, understanding the molecular basis of the accompanying cell size arrest will likely have implications for furthering our understanding of the molecular basis of cancer therapy. Here we describe investigations of the PTEN-dependent cell size checkpoint in human cells.

MATERIALS AND METHODS

Tissue culture.

HCT116 PTEN^+/+, PTEN^−/−, PIK3CA^WT/KO, and PIK3CA^KO/mut cells were created using human somatic cell gene targeting and were described previously (28, 34). HCT116^{FLAG-PTEN/FLAG-PTEN} cells were created by endogenous epitope tagging (EET) and described in a previous study (27). The glioblastoma multiforme (GBM) cell lines U87MG and SNB19 were obtained from ATCC and cultured as recommended.

Antibodies.

Primary antibodies were obtained from Cascade Bioscience (PTEN clone 6H2.1), Calbiochem (p53 clone DO-1), Cell Signaling [p-Akt S473, p-Akt T308 clone C31E5E, Akt, p-FoxO1(T24)/3a(T32), FoxO1 clone C29H4, FoxO3a clone 75D8, p-4E-BP1 clone 236B4, p-mTOR S2448, mTOR clone 7C10, p-TSC2 T1462, TSC2 clone D93F12, p-p70S6K S371, p-p70S6K T389, PARP clone 19F4], Santa Cruz Biotechnology (α-actin clone 1A4, β-actin clone C4, γ-actin clone 1-24, actin clone I-19, gelsolin clone C20, PTEN N19), Zymed Laboratories (E-cadherin clone HECD-1), Bethyl Laboratories (EPLIN A300), Neomarkers (α-tubulin clone DM1A), and Sigma (FLAG).

Flow cytometry.

Cells were fixed in 70% ethanol and stained in phosphate-buffered saline (PBS) containing 0.1% Triton X-100, 50 μg/ml RNase, and 50 μg/ml propidium iodide. DNA content was measured on a FACSort flow cytometer (Becton Dickinson), and data were analyzed using ModFit software (Verity Software House). At least 20,000 cells were analyzed for each sample.

Irradiation.

Subconfluent cell monolayers were irradiated using a J. L. Shepard Mark I ¹³⁷Cs irradiator at ∼2 Gy/min.

Cell sizing.

Cells were trypsinized in 0.5 ml, added to 0.5 ml of serum-containing medium, and further diluted in 10 ml of Isoton II. Cell diameters were determined using a Multisizer III Coulter Counter (Beckman Coulter). At least 10,000 cells were counted for each measurement.

Immunoblotting and immunoprecipitation.

Protein lysates for direct Western blotting were prepared in radioimmunoprecipitation (RIPA) buffer. Nuclear and cytoplasmic lysates used for FLAG purification were prepared using a modification of Dignam's nondetergent lysis method, described in reference 27 and references therein. Protein concentrations were determined using the bicinchoninic assay (Pierce).

For FLAG affinity purification, α-FLAG M2 beads (Sigma) were washed once with Tris-buffered saline (TBS) (150 mM NaCl) and then incubated with resuspended protein lysates derived from parental or FLAG epitope-tagged cells. Samples were incubated with rotation at 4°C for 1 h. Beads were then washed three times in TBS and packed into a Poly-Prep chromatography column (Bio-Rad). Bound proteins were eluted with 100 ng/μl 1× FLAG peptide (Sigma). Fractions were concentrated by trichloroacetic acid (TCA) precipitation, resuspended in sample buffer, and separated by SDS-PAGE. Subcellular fractions were prepared using a ProteoExtract native membrane protein extraction kit (Calbiochem).

PTEN protein complex purification and mass spectrometry.

PTEN immunoprecipitation was performed on protein lysates derived from HCT116 parental cells and their FLAG-PTEN-modified derivatives. Equivalent amounts of total protein were immunoprecipitated from both cell lines using FLAG-M2 beads, and bound proteins were eluted via competition with 1× FLAG peptide. Eluted proteins were then concentrated by TCA precipitation, separated by SDS-PAGE, and stained with GelCode blue stain reagent (Thermo Scientific). After destaining, the two gel lanes were divided into seven sections, reduced, carboxyamidomethylated, and digested with trypsin in gel. To identify proteins specifically present in immunoprecipitates from FLAG-PTEN-modified cells, the resulting peptides from each section were subjected to microcapillary reverse-phase high-pressure liquid chromatography (HPLC) directly coupled to the nanoelectrospray ionization source of a ThermoFisher LTQ-Orbitrap XL Velos hybrid mass spectrometer (liquid chromatography-tandem mass spectrometry [LC/MS-MS]). The Orbitrap repetitively surveyed an m/z range from 395 to 1,600, while data-dependent MS-MS spectra on the 10 most abundant ions in each survey scan were acquired in the linear ion trap. MS-MS spectra were acquired with relative collision energy of 30% and a 2.5-Da isolation width, and recurring ions were dynamically excluded for 60 s. Preliminary evaluation of peptide spectrum matches (PSMs) was facilitated using SEQUEST with a 30-ppm mass tolerance against the human subset of the Uniprot Knowledgebase. With a custom version of the Harvard Proteomics Browser Suite (Thermo Scientific), PSMs were accepted with a mass error of <3.0 ppm and score thresholds to attain an estimated false discovery rate of ∼1% using a reverse decoy database strategy.

Site-directed mutagenesis.

Site-directed mutagenesis was performed using the Quikchange Kit (Stratagene) using PAGE-purified oligonucleotides (Integrated DNA Technologies) to introduce the indicated mutations.

Lentiviruses.

The pHR-SIN-PTEN was a gift from Nick Leslie. Constructs for stable depletion of gelsolin and EPLIN were obtained from Open Biosystems. A negative-control construct in the same vector system (pLKO.1) was obtained from Addgene. The lentiviral helper plasmids pHR′CMV8.2ΔR and pCMV-VSV-G were also obtained from Addgene. All plasmids were prepped, and their integrities were confirmed by restriction analysis. The integrity of each short hairpin RNA (shRNA insert) was confirmed by sequencing. Lentiviral packaging and infection were performed as described previously (28).

Confocal microscopy.

Cells were fixed with 3.7% formaldehyde for 10 min at room temperature and further permeabilized with 3.7% formaldehyde–0.1% Triton X-100 for 10 min at room temperature. After being washed with PBS three times, actin filaments were labeled and visualized with Alexa-phalloidin (Molecular Probes) using a Zeiss LSM 510 Meta with a 63× Zeiss PLAN Apo objective.

RESULTS

PTEN is required for the cell size arrest induced by both ionizing radiation and DNA-damaging chemotherapeutic drugs.

Treatment of human cells with ionizing radiation (IR) and DNA-damaging chemotherapeutics leads to senescence-like cell cycle arrest (11). During this cell cycle arrest, cells also stop increasing in size and mass (i.e., stop growing). We have previously shown that PTEN-deficient cells undergo a normal senescence-like cell cycle arrest after treatment with IR but fail to arrest in size (34). As such, we have proposed that PTEN regulates a novel, radiation-induced cell size checkpoint.

Our initial work focused exclusively on IR as an inducer of the PTEN-dependent cell size checkpoint (reference 34 and Fig. 1 A). In an effort to demonstrate the generalizability of this phenotype, we tested whether DNA-damaging chemotherapeutic drugs also induce the PTEN-dependent cell size checkpoint. HCT116 PTEN^+/+ and PTEN^−/− cells previously created by human somatic cell gene targeting were treated with the topoisomerase II inhibitor doxorubicin (0.2 μg/ml) for 6 days, a course of doxorubicin that induces senescence-like cell cycle arrest in HCT116 cells and does not cause apoptosis (56). The cell size profiles of treated cells were then measured using a Multisizer III, a specialized Coulter Counter designed to measure cell size (Fig. 1A). The cell cycle profiles were also measured using flow cytometry (Fig. 1B). Two independently derived isogenic clones of each genotype were tested to avoid the possibility of clone-specific artifacts. HCT116 PTEN^+/+ cells arrested at an average volume of 33,100 μm³. In contrast, otherwise isogenic HCT116 PTEN^−/− cells continued to enlarge and eventually arrested at an average volume of 52,900 μm³. As previously demonstrated for IR, this size phenotype was not secondary to a more primary effect on the cell cycle, as the flow cytometry profiles of doxorubicin-treated HCT116 PTEN^+/+ and PTEN^−/− cells were indistinguishable (Fig. 1B). Phase-contrast micrographs of doxorubicin-induced enlargement of PTEN^−/− cells are depicted in Fig. 1C.

Fig. 1.

DNA-damaging chemotherapeutics induce a PTEN-dependent cell size checkpoint in human cells. (A) HCT116 PTEN^+/+ and HCT116 PTEN^−/− cells created by human somatic cell gene targeting were measured using a Multisizer III during exponential growth or 6 days after treatment with IR (6 Gy), doxorubicin (0.2 μg/ml), and etoposide (5 μg/ml). (B) Cells treated as described above were stained with propidium iodide and studied by flow cytometry. (C) HCT116 PTEN^−/− cells treated as described above were imaged using phase-contrast microscopy.

To confirm and extend these results, we repeated these experiments with the topoisomerase II inhibitor etoposide. We previously demonstrated that this dose of etoposide induces senescence-like cell cycle arrest in HCT116 cells without concomitant apoptosis (56). After 6 days of treatment, HCT116 PTEN^+/+ cells arrested at an average volume of 42,000 μm³ (Fig. 1A), whereas otherwise isogenic HCT116 PTEN^−/− cells continued to enlarge and eventually arrested at an average volume of 89,300 μm³. As with IR and doxorubicin, the size phenotype was not secondary to a more primary effect on cell cycle, as the flow cytometry profiles of etoposide-treated HCT116 PTEN^+/+ and PTEN^−/− cells were indistinguishable (Fig. 1B). Micrographs of etoposide-induced enlargement of PTEN^−/− cells are depicted in Fig. 1C. Taken together, these data, which were obtained using two different topoisomerase II inhibitors, demonstrate that PTEN controls a size checkpoint that is inducible not only by IR but also by several commonly used DNA-damaging chemotherapeutic drugs.

Restoration of size checkpoint control in PTEN^−/− cells via lenti-PTEN infection.

Despite the use of multiple independently derived PTEN^+/+ and PTEN^−/− clones, it remained a formal possibility that differences in cell size following DNA damage may stem from clone-specific artifacts unrelated to PTEN. To investigate this possibility, we tested whether ectopic reexpression of PTEN restored cell size checkpoint control to HCT116 PTEN^−/− cells. We obtained a lenti-PTEN construct (gift from Nick Leslie), created infectious lentivirus, and infected HCT116 PTEN^−/− cells as described in Materials and Methods. Infection of PTEN^−/− cells with lenti-PTEN but not with the lentiviral vector alone led to reexpression of PTEN protein in these cells (Fig. 2 A).

Fig. 2.

Reconstitution of the cell size checkpoint in PTEN^−/− cells via infection with lenti-PTEN. HCT116 PTEN^−/− cells were infected with lentiviral vector alone (pHR-SIN) or with lenti-PTEN (pHR-SIN-PTEN). (A) Three days after infection, cells were harvested in RIPA buffer and studied by Western blotting to document expression of PTEN protein. (B) Infected cells were treated with 6 Gy IR and cultured for 6 days, and their sizes were measured using a Multisizer III.

Next, infected cells were exposed to 6 Gy IR and cultured for 6 days before cell size determination using a Multisizer III. As expected, HCT116 PTEN^−/− cells infected with the lentiviral vector alone were unable to a undergo cell size arrest and enlarged dramatically to a postirradiation average cell volume of 69,100 μm³ (Fig. 2B). In contrast, infection of HCT116 PTEN^−/− cells with lenti-PTEN led to a virtually complete restoration of cell size checkpoint control, as evidenced by a postirradiation average cell volume of 10,700 μm³ (Fig. 2B). These data provide formal confirmation of the role of PTEN in cell size checkpoint control. Furthermore, these findings validate lenti-PTEN as a reagent that can restore the cell size checkpoint to PTEN^−/− cells.

PTEN regulates size checkpoint control in GBM cells that contain naturally occurring mutations of PTEN.

We next tested whether human cancer cell lines with naturally occurring mutations of PTEN were deficient in the DNA damage-inducible size checkpoint. For these experiments, we focused our studies on glioblastoma multiforme (GBM) cell lines, since mutations and deletions of PTEN are common in GBM (>40%; reference 57). In particular, we studied two different PTEN-deficient human GBM cell lines: U87MG cells, which harbor a 49-bp deletion that leads to a frameshift mutation and an absence of PTEN protein expression (37), and SNB19 cells, which harbor an insertion of two T residues in exon 7 leading to a frameshift mutation and a complete absence of PTEN protein expression (25). Both cell lines harbor loss of heterozygosity of the remaining wild-type allele of PTEN. Initially, U87MG and SNB19 cells were infected with lenti-PTEN or with vector alone, and expression of PTEN was confirmed by Western blotting (Fig. 3 A). Infected cell lines were then treated with doxorubicin or etoposide and cultured for 5 days. The resultant cell size was then measured using a Multisizer III (Fig. 3B to D). Of note, IR was not used in any of these experiments, since GBM cell lines are notoriously radioresistant. Cells infected with lentiviral vector alone continued to enlarge after treatment with doxorubicin and etoposide. In contrast, cells infected with lenti-PTEN arrested in size, reflecting restoration of cell size checkpoint control. This size phenotype was not due to differences in polyploidization between PTEN-proficient and PTEN-deficient cells. Expression of PTEN in U87MG cells appeared to rescue U87MG cells from doxorubicin- and etoposide-induced cytotoxicity (compare the fragmented cell populations in the Multisizer data depicted in Fig. 3B). This result is consistent with previous observations that PTEN expression protects cells from DNA damage-induced cytotoxicity (34). Taken together, these data generalize our previous findings and demonstrate that two different GBM cell lines with naturally occurring PTEN mutations are deficient in PTEN-dependent size checkpoint control.

Fig. 3.

Restoration of cell size checkpoint control in PTEN-deficient human GBM cell lines via infection with lenti-PTEN. U87MG and SNB19 cells were infected with lentiviral vector alone and lenti-PTEN. (A) Infected cells were then harvested in RIPA buffer 3 days after infection and studied by Western blotting to document expression of PTEN protein. Infected U87MG cells (B) and SNB19 cells (C) were treated with doxorubicin (0.2 μg/ml) and etoposide (5 μg/ml) and cultured for 5 days, and their sizes were measured using a Multisizer III. (D) Quantification of size differences induced by doxorubicin and etoposide in cells infected with lentiviral vector alone or lenti-PTEN. (E) SNB19 cells were treated with temozolomide (100 μM) and cultured for 5 days, and their sizes were measured using a Multisizer III.

While these data are intriguing, neither doxorubicin nor etoposide is used clinically for treatment of GBM, and therefore, these data have questionable clinical relevance. To determine whether PTEN may modulate cell size control in GBM cells arrested by a more clinically relevant chemotherapeutic drug, we tested temozolomide, an alkylating agent that is a standard-of-care upfront treatment for GBM. SNB19 cells that had been preinfected with either lentiviral vector alone or lenti-PTEN were treated with temozolomide and then cultured for 5 days. The sizes of these treated cells were measured using a Multisizer III (Fig. 3E). Cells infected with lentiviral vector alone continued to enlarge after treatment with temozolomide. In contrast, cells infected with lenti-PTEN arrested in size, reflecting restoration of cell size checkpoint control. These data implicate PTEN in the control of GBM cell size arrest that was induced by a clinically relevant chemotherapeutic drug.

Oncogenic PIK3CA fails to efficiently modulate cell size checkpoint control.

We wondered whether abrogation of the radiation-induced cell size checkpoint was a generalizable feature of activation of PI3K signaling. To test this, we studied PIK3CA gene-targeted derivatives of HCT116 cells, which harbor an endogenous heterozygous oncogenic mutation in the catalytic domain of PIK3CA (H1047R). Human somatic cell gene targeting technology was used to create derivatives of HCT116 cells in which either the mutant allele or the wild-type allele of PIK3CA had been deleted (28). Parental HCT116 cells and derivatives lacking either the wild-type or mutant allele of PIK3CA were treated with 6 Gy IR and examined 6 days after irradiation (Fig. 4). In contrast to HCT116 PTEN^−/− cells, each of the three otherwise isogenic PIK3CA gene-targeted cell lines was able to efficiently arrest its cell size (Fig. 4A), despite the ability of oncogenic PIK3CA to regulate the phosphorylation state and activity of Akt in these cells (Fig. 4B). These data indicated that unlike PTEN, PIK3CA appears not to be involved in regulation of the IR-induced cell size checkpoint. In addition, these results suggested that the ability of PTEN to regulate intracellular levels of PIP₂ and PIP₃ is not its only biochemical activity required for cell size checkpoint control.

Fig. 4.

Inability of PIK3CA to efficiently regulate cell size checkpoint control. (A) HCT116 parental cells and PTEN and PIK3CA gene-targeted derivatives lacking either the wild-type allele of PIK3CA (PIK3CA^KO/mut) or the oncogenic allele (PIK3CA^WT/KO) were treated with 6 Gy IR and cultured for 6 days, and their sizes were measured using a Multisizer III. (B) Cells were harvested in RIPA buffer 24 h after treatment with DNA-damaging agents, and Western blotting was performed with the indicated antibodies.

The lipid phosphatase activity of PTEN is necessary for cell size checkpoint control.

The fact that lenti-PTEN was able to restore cell size checkpoint control to PTEN-deficient human cells provided us with an experimental system for testing the effect of PTEN mutations on cell size checkpoint control. Initially, we used site-directed mutagenesis to introduce 11 different tumor-derived mutations into the known functional domains of PTEN (Fig. 5 A; also described in Materials and Methods). The origins of the mutations and their previously determined effects on PTEN lipid phosphatase activity are listed in Fig. 5D. The constructs were then packaged into infectious lentivirus and used to infect HCT116 PTEN^−/− cells. Western blotting was performed to confirm expression of PTEN and to measure the effects of mutant PTEN proteins on modulation of p-Akt (Fig. 5B). In addition, infected cells were treated with 6 Gy IR and cultured for 6 days. The cell size was then measured using a Multisizer III (Fig. 5C).

Fig. 5.

Mutational analysis of the PTEN-dependent cell size checkpoint. (A) Eleven different missense mutations were engineered into lenti-PTEN using site-directed mutagenesis, as described in Materials and Methods. The constructs were then packaged into infectious lentivirus and used to infect HCT116 PTEN^−/− cells. WT, wild type; V, vector. (B) Cells were harvested 3 days after infection in RIPA buffer and studied by Western blot analysis to evaluate levels of PTEN, p-Akt (S473), and total Akt protein. (C) Cells were treated with 6 Gy IR and cultured for 6 days, and their sizes were measured using a Multisizer III. (D) Tumor type of origin, previously determined lipid phosphatase activity with reference, and ability to regulate p-Akt and the cell size checkpoint are shown. The three mutant proteins that were deficient in the cell size checkpoint are denoted with boxes.

Three of the 11 mutations are known to disrupt the lipid phosphatase activity of PTEN (D24Y, C124S, and G129E). As expected, these mutants were unable to downregulate levels of p-Akt in PTEN-deficient cells (Fig. 5B and D). Similarly, these three mutant proteins were completely unable to restore size checkpoint control to HCT116 PTEN^−/− cells (Fig. 5C and D). Based on these data, we concluded that the lipid phosphatase activity of PTEN is necessary for effective PTEN-dependent cell size checkpoint control.

The PTEN Y138L mutant is deficient in protein phosphatase activity but retains wild-type lipid phosphatase activity (9). Therefore, this mutation is particularly useful for evaluating the effect of protein phosphatase activity on PTEN-related phenotypes. As expected, PTEN Y138L downregulated the p-Akt levels in HCT116 PTEN^−/− cells similarly to wild-type PTEN (Fig. 5B). Furthermore, PTEN Y138L efficiently restored cell size checkpoint activity to HCT116 PTEN^−/− cells (Fig. 5C). Therefore, we concluded that the protein phosphatase activity of PTEN is dispensable for the control of the DNA damage-inducible cell size checkpoint.

Mutations in the amino terminus of PTEN uncouple lipid phosphatase activity and cell size regulation from control of Akt phosphorylation.

Of the 11 mutations tested, PTEN Y16C was particularly intriguing. This mutant protein, which was previously reported to have wild-type lipid phosphatase activity, restored cell size checkpoint control to HCT116 PTEN^−/− cells similarly to wild-type PTEN but failed to downregulate p-Akt levels (Fig. 5B and C and reference 15). This dichotomy suggests that the ability of PTEN to modulate p-Akt levels is not required for cell size checkpoint control.

Next, we created an additional seven missense mutations and two deletions in the amino terminus of PTEN. The biochemical and phenotypic properties of several of these mutations have been previously reported (10). These nine additional mutant proteins were tested for their abilities to regulate levels of p-Akt and for their abilities to regulate the DNA damage-inducible size checkpoint (Fig. 6 A to C). Of the nine mutations, Δ1-15 and Δ1-25 were defective in both assays and were therefore not evaluated further. As with PTEN Y16C, each of the additional seven missense mutations in the amino terminus of PTEN restored cell size checkpoint control to HCT116 PTEN^−/− cells similarly to wild-type PTEN (Fig. 6B). However, PTEN R11A, R14A, F21A, L23F, and L25A were each deficient in their ability to downregulate the levels of p-Akt in HCT116 PTEN^−/− cells (similar to results for PTEN Y16C, most likely due to the deleterious effects of these mutations on the expression of PTEN). Taken together, these data provide strong evidence that the Y16C mutation is not an outlier and that missense mutations in the amino terminus of PTEN uncouple the ability to control the radiation-induced cell size checkpoint from the ability to regulate p-Akt levels.

Fig. 6.

Mutations in the amino terminus of PTEN uncouple control of Akt phosphorylation from control of cell size. Nine different missense mutations were engineered into the amino terminus of PTEN using site-directed mutagenesis, as described in Materials and Methods. The constructs were then packaged into infectious lentivirus and used to infect HCT116 PTEN^−/− cells. Of note, the data for two of the mutations, Δ1-15 and Δ1-25, are described in the text but are not depicted in this figure. (A) Cells were harvested 3 days after infection in RIPA buffer and studied by Western blot analysis to evaluate levels of PTEN, p-Akt (S473), p-Akt (T308), and total Akt protein. WT, wild type; V, vector. (B) Cells were treated with 6 Gy IR and cultured for 6 days, and their sizes were measured using a Multisizer III. (C) Ability of proteins to regulate p-Akt and cell size checkpoint is shown. The six mutant proteins that were proficient in regulating the cell size checkpoint but deficient in regulating p-Akt are denoted with boxes.

Pharmacological inhibition of Akt kinase activity fails to restore size checkpoint control to HCT116 PTEN^−/− cells.

Since the Akt pathway has been previously implicated in the control of cell size, our mutational analysis data that suggested that Akt was not a required effector of the PTEN-dependent cell size checkpoint were surprising. To more directly test the hypothesis that Akt activity is unnecessary for cell size checkpoint control, we employed MK2206, a recently developed submicromolar pharmacological inhibitor of all Akt isoforms (5 nM for Akt1, 12 nM for Akt2, and 65 nM for Akt3) that is currently in phase II clinical trials (39). MK2206 is an allosteric Akt inhibitor that inhibits the folding of Akt proteins and, therefore, abolishes the ability of Akt to be recruited to the plasma membrane and be activated by phosphorylation.

Initially, we examined the effectiveness of MK2206 in regulating the activation state of Akt. HCT116 PTEN^−/− cells were treated with MK2206 or LY294002 for 2 h, and then protein lysates were prepared and analyzed by Western blotting. As depicted in Fig. 7 A, MK2206 treatment led to a dramatic reduction in levels of p-Akt at both S473 and T308, as well as of the Akt substrate p-FoxO1/3a. These effects were more pronounced than the effects of LY294002 and occurred at substantially lower concentrations. In contrast, MK2206 had little (if any) effect on the phosphorylation state of members of the mTOR pathway, consistent with data that have been reported by Samuels et al. (45). Next, we tested whether inhibition of activated Akt using MK2206 restored cell size control to HCT116 PTEN^−/− cells. HCT116 PTEN^−/− cells were treated with 6 Gy IR in the presence or absence of 2 μM MK2206 and cultured for 3 days in the presence of drug, which led to no overt toxicity. Cell size was then measured using a Multisizer III. Pharmacological inhibition of Akt failed to restore cell size checkpoint control to PTEN-deficient cells (Fig. 7B). To further confirm that Akt was not involved in the radiation-induced cell size checkpoint, HCT116 PTEN^+/+ cells were transiently transfected with a myr-Akt expression construct (Fig. 8). Despite expression of p-Akt (as documented by Western blot in Fig. 8A), there was no effect on the integrity of the radiation-induced cell size checkpoint (Fig. 8B). Taken together, these data confirm that Akt is not a required PTEN effector for cell size checkpoint control.

Fig. 7.

Pharmacological inhibition of Akt fails to restore the cell size checkpoint to PTEN^−/− cells. (A) HCT116 PTEN^−/− cells were treated with 0.5, 1, or 2 μM MK2206 and 10, 50, and 100 μM LY294002 for 2 h, harvested in RIPA buffer, and studied by Western blotting. (B) HCT116 PTEN^−/− cells were pretreated with 2 μM MK2206 for 1 h, irradiated (6 Gy), cultured in the presence of MK2206 for 3 days, and measured using a Multisizer III.

Fig. 8.

Expression of myr-Akt fails to abrogate the DNA damage-induced cell size checkpoint in HCT116 PTEN^+/+ cells. HCT116 PTEN^+/+ cells were transiently transfected with myr-Akt expression vector or vector alone. (A) 48 h after transfection, cells were harvested in RIPA buffer and levels of p-Akt and total Akt were measured by Western blotting. (B) Cells were treated with 6 Gy IR, cultured for 3 days, and measured using a Multisizer III.

Identification of novel putative PTEN effectors via endogenous epitope tagging (EET).

Since the ability of PTEN to regulate Akt phosphorylation is unnecessary for regulation of the PTEN-dependent cell size checkpoint, we sought to identify novel effectors of this checkpoint. In particular, we hypothesized that PTEN interacts with one or several PIP₂- or PIP₃-regulated proteins in order to regulate cell size checkpoint control.

Since identification of PTEN-interacting proteins has proven to be very difficult due, in part, to issues of ectopic overexpression (3), we developed a new technology (27), termed endogenous epitope tagging (EET). This technique allows us to efficiently add a short epitope tag to the endogenous allele of genes in cultured human cells in order to avoid ectopic overexpression of epitope-tagged transgenes while still exploiting high-efficiency affinity reagents for protein complex purification. In a proof-of-principle experiment for this approach, we described the creation of HCT116 cell lines in which the amino termini of both PTEN alleles were modified with the addition of a 1× FLAG tag (27). Here, we employed these cells to identify novel PTEN-interacting proteins.

Purification and mass spectrometric identification of PTEN-interacting proteins are described in detail in Materials and Methods. In brief, protein lysates were prepared from HCT116^{FLAG-PTEN/FLAG-PTEN} cells and an equivalent number of negative-control HCT116 parental cells and applied to a FLAG-M2 affinity column, and bound proteins were eluted using 1× FLAG peptide. The proteins were separated by SDS-PAGE (Fig. 9 A), and the protein composition of the eluents was determined using tandem mass spectrometry. Proteins present specifically in FLAG immunoprecipitates from HCT116^{FLAG-PTEN/FLAG-PTEN} cells but not in immunoprecipitates from HCT116 parental cells are listed in Fig. 9B.

Fig. 9.

Identification of novel putative PTEN-interacting proteins using endogenous epitope tagging (EET). HCT116 cells in which both endogenous alleles of PTEN have been modified via the addition of amino-terminal 1× FLAG tags (HCT116^{FLAG-PTEN/FLAG-PTEN}) have been previously described (27). HCT116 parental (+/+) and HCT116^{FLAG-PTEN/FLAG-PTEN} (F/F) cells were cultured, nondetergent lysis was performed, and FLAG immunoprecipitation was performed, all as described in Materials and Methods. (A) Immunoprecipitates were concentrated by TCA precipitation, separated by SDS-PAGE, and visualized after staining with Coomassie brilliant blue (CBB). PTEN and a coimmunoprecipitating protein later determined to be actin are denoted with arrows. As expected, several nonspecific immunoprecipitating proteins are also present in both lanes. (B) The gel lanes depicted in panel A were subjected to high-resolution mass spectrometry, as described in Materials and Methods. Proteins with peptide counts of >4 in immunoprecipitates from epitope-tagged cells but absent in immunoprecipitates from parental cells are listed.

As expected, the endogenous FLAG-PTEN fusion protein was the most prominent differentially immunoprecipitated protein. Other proteins that were present specifically in immunoprecipitates from FLAG-PTEN cells included actin and its remodeling proteins gelsolin and EPLIN. Actin was sufficiently abundant to be visible in the Coomassie brilliant blue (CBB)-stained gel (Fig. 9A). Notably, gelsolin is regulated by PIP₂ (50).

Endogenous PTEN interacts and colocalizes with an endogenous PIP₂-regulated actin depolymerization complex.

To confirm these putative endogenous interactions, immunoprecipitation and Western blot analyses were performed. PTEN was immunoprecipitated from FLAG-PTEN cells using FLAG-M2 beads, and Western blotting was performed with antibodies for gelsolin, EPLIN, and the three major actin isoforms. As depicted in Fig. 10 A and 10B, immunoprecipitation of endogenous PTEN led to coimmunoprecipitation of endogenous β-actin, γ-actin, gelsolin, and EPLIN. Subcellular fractionation experiments demonstrated that the plasma membrane was the only cellular compartment in which each of these proteins was present, suggesting that the interactions were likely to occur at the cell membrane (Fig. 10C). Subsequent immunoprecipitation and Western blot analyses of subcellular fractions confirmed that these interactions occur at the plasma membrane (Fig. 10D). These experiments also demonstrated that the interaction between PTEN, actin, gelsolin, and EPLIN was insensitive to oxidation state, a known regulator of PTEN (Fig. 10D; reference 35). The interaction between PTEN and actin was further confirmed by immunoprecipitation (IP)/Western blotting using anti-PTEN antibodies in genetically unmodified HCT116, LN229, and 293T cells (Fig. 10E).

Fig. 10.

Interaction of endogenous PTEN with an actin remodeling complex. (A and B) HCT116 parental (+/+) and HCT116^{FLAG-PTEN/FLAG-PTEN} (F/F) cells were cultured, and nondetergent lysis followed by FLAG immunoprecipitation (M2-IP) was performed. Immunoprecipitates were concentrated by TCA precipitation and separated by SDS-PAGE, and Western blotting was performed with the antibodies indicated. (C) Cytoplasmic, membrane, and nuclear fractions of HCT116^{FLAG-PTEN/FLAG-PTEN} cells were prepared and studied by Western blotting with the antibodies indicated. (D) Membrane fractions from HCT116 parental (+/+) and HCT116^{FLAG-PTEN/FLAG-PTEN} (F/F) cells were subjected to immunoprecipitation with FLAG-M2 beads and studied by Western blotting with the antibodies indicated. Immunoprecipitations were performed in the presence and absence of 1 mM dithiothreitol (DTT) as indicated. (E) HCT116, LN229, and 293T cells were harvested in lysis buffer containing 1% NP-40–150 mM NaCl–20 mM Tris-HCl, pH 7.5. One milligram of protein lysate was incubated overnight with 1 μg of PTEN N19 antibody or 1 μg normal goat IgG and protein A/G agarose beads. Immunocomplexes were washed in lysis buffer and boiled in sample buffer, and Western blotting was performed.

Next, immunofluorescence was performed to determine whether PTEN and actin colocalize in human cells. A lentivirus that expresses green fluorescent protein (GFP) GFP-PTEN was generated and used to infect HCT116 PTEN^−/− cells. Infected cells were then fixed and stained with Alexa-conjugated phalloidin, which binds to and stains actin filaments. Cells were then imaged with fluorescence microscopy (Fig. 11). As previously reported, the majority of GFP-PTEN was diffusely present in the cytoplasm and the nucleus, with a minority present at the plasma membrane. While GFP alone did not colocalize with actin, GFP-PTEN and actin colocalized at the plasma membrane (yellow arrows). This colocalization was seen as a subtle but distinct overlap of GFP and phalloidin staining. These signals also overlapped with staining on the membrane-associated actin network. These data are consistent with the immunoprecipitation and Western blot data depicted in Fig. 10.

Fig. 11.

Colocalization of PTEN and F-actin. HCT116 PTEN^−/− cells infected with GFP lentivirus or GFP-PTEN lentivirus were fixed and stained as described in Materials and Methods. Alexa 594-phalloidin-labeled F-actin colocalized with GFP-PTEN but not GFP alone at the protrusions of the cells and in the magnified actin network (denoted by the yellow arrows). The lower panels of GFP-PTEN and GFP alone are magnified images derived from the squares in the upper panels. DIC, differential interference contrast.

To determine if the interaction between PTEN and actin was regulated by DNA damage, PTEN and actin colocalization was measured by immunofluorescence either in unirradiated cells or 30 h after irradiation with 6 Gy. DNA damage failed to enhance the degree of colocalization to any measurable extent. Similarly, the presence of tumor-derived mutations R11A, Y16C, F21A (each described in detail in Fig. 6), and G129E in the GFP-PTEN construct failed to affect the colocalization between PTEN and actin.

Pharmacological inhibition of actin depolymerization abrogates cell size checkpoint control in PTEN^+/+ cells.

We next considered the possibility that a defect in actin remodeling could be responsible for the absence of size checkpoint control in HCT116 PTEN^−/− cells. In this case, we would expect that pharmacological inhibition of actin remodeling in PTEN^+/+ cells would be phenotypically equivalent to deletion of PTEN. To test this, we measured the effect of cytochalasin D, a potent inhibitor of actin polymerization, on the cell size checkpoint in HCT116 PTEN^+/+ and PTEN^−/− cells. Cells were pretreated with 200 nM cytochalasin D, treated with 6 Gy IR, and then cultured for 3 days. Cell sizes were then measured. Pharmacological inhibition of actin polymerization abrogated cell size checkpoint control in PTEN^+/+ cells, recapitulating the phenotype of PTEN deletion (Fig. 12 A and B). Importantly, cytochalasin D had no effect on the size of PTEN^−/− cells, demonstrating that the effect of the drug on cell size checkpoint control was specific to PTEN^+/+ cells. However, depletion of gelsolin or EPLIN individually was insufficient to abrogate cell size checkpoint control. Taken together, these data indicate that the postirradiation cell size control defect in PTEN^−/− cells is caused by a generalized defect in the ability to normally regulate actin dynamics.

Fig. 12.

Pharmacological inhibition of actin remodeling leads to abrogation of cell size checkpoint control in PTEN^+/+ cells. (A) HCT116 PTEN^+/+ and PTEN^−/− cells were exposed to 200 nM cytochalasin D, treated with 6 Gy IR, cultured for either 3 or 6 days, and measured with a Multisizer III. (B) Quantification of cell size in irradiated PTEN^+/+ and PTEN^−/− cells treated with either 200 nM cytochalasin D or with vehicle alone.

DISCUSSION

The genetic and biochemical mechanisms that regulate cell size during cellular proliferation and cell cycle arrest remain mostly obscure. To date, most published work on cell size checkpoints has focused on the existence (or lack thereof) of a sensing mechanism in the G₁ phase of the eukaryotic cell cycle that halts the cell cycle until the cell has reached sufficient size and mass to support cell division (examples in references 6, 12, 13, 19, and 52). In the studies presented here, we have focused our attention on a related but different issue—the mechanism responsible for ensuring that human cells arrested in the G₁ or G₂ phases of the cell cycle simultaneously stop increasing in size. We focus specifically on the cell size checkpoint that is enacted during DNA damage-induced arrest.

In the work described in this paper and in a previous publication (34), we identified the PTEN tumor suppressor as a required effector of this cell size checkpoint. Cells in which PTEN has been deleted by human somatic cell gene targeting or in which PTEN is inactivated by naturally occurring tumor-derived mutations are unable to normally arrest their cell size during DNA damage-induced cell cycle arrest. Mutational analysis of PTEN revealed that the lipid phosphatase activity of PTEN is required for this PTEN-dependent cell size checkpoint, while the ability of PTEN to modulate Akt phosphorylation is dispensable for this checkpoint. This conclusion was subsequently confirmed with the use of Akt inhibitors. Endogenous PTEN was shown to interact at the membrane with an actin-remodeling complex that contains actin-remodeling proteins, such as gelsolin, a protein known to be regulated by PIP₂. Treatment of PTEN^+/+ cells with cytochalasin D, a potent inhibitor of actin remodeling, led to abrogation of the cell size checkpoint. Importantly, this inhibitor produced no effect on cell size control in otherwise isogenic PTEN^−/− cells. Taken together, these data indicate that direct control of actin remodeling but not control of Akt phosphorylation is required for PTEN-dependent cell size checkpoint control (Fig. 13).

Fig. 13.

Model of PTEN-dependent cell size checkpoint. After DNA damage, cells undergo p53/p21-dependent arrests in the G₁ and G₂ phases of the cell cycle. DNA-damaged cells also undergo a simultaneous arrest in cell size, which requires PTEN and its ability to modulate the activity of an actin remodeling complex.

It was surprising to us that the PTEN-dependent size phenotype described herein was Akt independent, since there are numerous reports in the literature of Akt being a central player in cell size control. In D. melanogaster, activation of Akt leads to increased cell and organ growth, and regulation of Akt appears to be required for the effects of PTEN on cell and organ size (41, 55). Akt has also been shown to promote cell and organ growth in mice (33, 48), though the presence of multiple Akt homologs has complicated testing its epistasis with PTEN. We do not understand the molecular basis of the discrepancies between these types of published studies and the data presented herein. Possible explanations include (i) mechanistic differences between cell size control during organismal development and DNA damage-induced cell cycle arrest, (ii) mechanistic differences in cell size control between humans, mice, and flies, and/or (iii) the possibility that PTEN and Akt function in parallel pathways to control cell size.

Currently, PTEN is the only known major regulator of the DNA damage-induced cell size checkpoint. It is worth noting, however, that a variety of genes, including the LK6, S6K, TSC1, and TSC2 genes and myc, have been shown to regulate cell size during proliferation (examples in references 1, 17, and 29). The fact that many of these genes are cancer related raises the important question whether the abrogation of cell size checkpoint control is fundamental to neoplastic transformation in a fashion analogous to that of abrogation of the G₁ and G₂ checkpoints. Clearly, several cytopathological findings that present in PTEN-deficient cancers are likely due to defective PTEN-dependent cell size checkpoint control. The presence of giant cells in tumors and the existence of tumor types that are composed exclusively of enlarged cells are two such cytopathological presentations.

Despite these findings, whether abrogation of cell size checkpoint control actually drives neoplasia is not clear. Since Akt is thought to be a key effector of PTEN-dependent tumor suppression but is clearly dispensable for cell size checkpoint control in the systems studied here, the cell size checkpoint may not be related to driving neoplasia. However, recent studies have called into question whether Akt is actually a required effector of PI3K pathway-driven oncogenesis (54). Furthermore, emerging data suggest that Akt inhibitors may be of limited clinical utility in tumors driven by mutations in PTEN. Thus, the extent to which Akt is a required effector of PTEN tumor suppression is not clear at this time.

How might abrogation of cell size checkpoint control actually drive neoplasia? We hypothesize that the explanation may be related to the eukaryotic cell checkpoint that halts cell division at the G₁ stage of the cell cycle until cells have reached sufficient size to separate their biomass into two daughter cells. Whereas in normal-sized cells, this checkpoint is vigilant in preventing cell division and proliferation, in oversized PTEN-deficient cells, this checkpoint may permit cells to enter the cell cycle, contributing to enhanced proliferation and neoplasia. This hypothesis, however, remains experimentally untested.

In addition to demonstrating that Akt is dispensable for cell size checkpoint control, we identified actin remodeling as a critical PTEN-regulated process that is involved in regulating cell size control. These findings are consistent with the early work of Goberdhan et al., who demonstrated that in D. melanogaster, PTEN affects cytoskeletal organization in multiple cell types (18). Here we have identified a physical interaction between PTEN and an actin-remodeling complex that includes β-actin, γ-actin, and several actin-remodeling proteins, including gelsolin and EPLIN. This finding raises another unresolved question: which of these proteins interacts directly with PTEN? We speculate that PTEN interacts directly with actin and indirectly with the remodeling proteins, since actin appears to be the most abundant protein (other than PTEN itself) in PTEN immunoprecipitates (Fig. 9A). In addition, PTEN contains a domain with homology to tensin, a known actin-interacting protein. A definitive answer to this question will require the ability to recapitulate the interactions with purified components, and these efforts are ongoing in our laboratory.

This newly identified interaction between PTEN and the actin-remodeling complex is reminiscent of the recent work of van Diepen et al., who demonstrated that PTEN interacts with myosin V in neurons (53). These researchers further showed that this interaction is critical for the ability of PTEN to control the size of these neurons. While we did not specifically identify myosin V as a PTEN-interacting protein in our study, we speculate that this omission is due to cell type-specific differences in the expression pattern of the myosin V gene. Determination of whether myosin V is part of a larger actin-containing complex in the neurons used in this study will be interesting. Furthermore, our work is also reminiscent of other recent studies that demonstrated that PTEN colocalizes with actin and myosin during chemotaxis in Dictyostelium (42). Our studies suggest that this reported colocalization may result from direct (or indirect) physical interaction. In addition, Goranov et al. have suggested that direct regulation of actin remodeling may be an important biochemical mechanism for eukaryotic cell size control (19, 20).

In summary, we have identified and evaluated a PTEN-dependent cell size checkpoint in human cancer cells. Current work is focusing on better understanding the structural nature of the identified interaction between PTEN and the actin-remodeling complex and evaluating how and why abrogation of PTEN-dependent cell size checkpoint control either directly or indirectly drives neoplasia.