The Loss of Nuclear PTEN Increases Tumorigenesis in a Preclinical Mouse Model for Hepatocellular Carcinoma

Summary

The PTEN gene is highly mutated in many cancers, including hepatocellular carcinoma. The PTEN protein is located at different subcellular regions—PTEN at the plasma membrane suppresses PI3-kinase signaling in cell growth, whereas PTEN in the nucleus maintains genome integrity. Here, using nuclear PTEN-deficient mice, we analyzed the role of PTEN in the nucleus in hepatocellular carcinoma that is induced by carcinogen and oxidative stress-producing hepatotoxin. Upon oxidative stress, PTEN was accumulated in the nucleus of the liver, and this accumulation promoted repair of DNA damage in wild-type mice. In contrast, nuclear PTEN-deficient mice had increased DNA damage and accelerated hepatocellular carcinoma formation. Both basal and oxidative stress-induced localization of PTEN in the nucleus require ubiquitination of lysine 13 in PTEN. Taken together, these data suggest the critical role of nuclear PTEN in the protection from DNA damage and tumorigenesis in vivo.

Graphical Abstract

Highlights

• •The nuclear localization of PTEN is regulated by ubiquitination of lysine 13
• •Nuclear PTEN suppresses DNA damage in vivo
• •Nuclear PTEN loss causes tumorigenesis in a mouse model for hepatocellular carcinoma

Cell Biology; Cancer

Introduction

Phosphatase and tensin homolog deleted on chromosome ten (PTEN) is one of the most critical tumor suppressors, which functions at different subcellular locations, including the plasma membrane and nucleus (Baker, 2007; Gil et al., 2007; Hopkins and Parsons, 2014; Kreis et al., 2014; Leslie et al., 2016; Planchon et al., 2008). At the plasma membrane, PTEN counteracts PI3 kinase signaling by dephosphorylating the potent second messenger PIP3 to PIP2. The loss of PTEN in cancer cells results in over-activation of AKT and mTOR signaling, leading to excessive stimulation of cell growth and inhibition of cell death (Chalhoub and Baker, 2009; Endersby and Baker, 2008; Fruman et al., 2017; Iijima and Devreotes, 2002; Iijima et al., 2002; Song et al., 2012). In the nucleus, PTEN functions in DNA repair, genome stability, and cell cycle control through associations with Rad51 and p53 (Baker, 2007; Bassi et al., 2013; Chung et al., 2006; Gil et al., 2007; Kreis et al., 2014; Leslie et al., 2016; Planchon et al., 2008; Shen et al., 2007; Song et al., 2011). PTEN in the nucleus also produces resistance to ionizing radiation therapy (Ma et al., 2019). Most PTEN functions in the nucleus are independent of its lipid phosphatase activity (Baker, 2007; Bassi et al., 2013; Chung et al., 2006; Gil et al., 2007; Kreis et al., 2014; Leslie et al., 2016; Planchon et al., 2008; Shen et al., 2007; Song et al., 2011). Therefore, the loss of PTEN produces multiple effects on the tumorigenesis at different subcellular locations.

The subcellular localization of PTEN is highly regulated (Kreis et al., 2014; Leslie et al., 2016). It has been shown that the recruitment of PTEN to the plasma membrane is regulated by phosphorylation of a cluster of serine and threonine residues at the C-terminal region (Das et al., 2003; Nguyen et al., 2014a, 2014b, 2015a, 2015b; Rahdar et al., 2009; Yang et al., 2017). The phosphorylation controls the conformation of PTEN via intramolecular interactions between the C terminus and the membrane-binding interface. Phosphorylation closes the PTEN conformation by promoting interactions, whereas dephosphorylation opens the PTEN conformation. Also, the C-terminal phosphorylation regulates the localization of PTEN in the nucleus (Nguyen et al., 2014a, 2015b). When the conformation of PTEN is opened, PTEN can accumulate in the nucleus, likely through exposing a nuclear localization signal. It is currently unknown what types of physiological and pathological cues control these conformations of PTEN.

The nuclear localization of PTEN changes in response to various stress. For example, oxidative stress with hydrogen peroxide (H₂O₂) induces the accumulation of PTEN in the nucleus (Chang et al., 2008; Choi et al., 2013; Hou et al., 2019). Genotoxic stress with ionizing radiation also changes the level of PTEN in the nucleus (Bassi et al., 2013; Ma et al., 2019). The ubiquitination of PTEN has been suggested to be important for the nuclear localization of PTEN at the basal condition; however, it is not well understood whether stress-induced nuclear localization of PTEN is regulated by ubiquitination (Nguyen et al., 2015b; Song et al., 2008; Trotman et al., 2007). In addition to the regulation of the localization of PTEN in the nucleus, the physiological role of nuclear PTEN in vivo remains uncertain. Although a study has reported that increased degradation of nuclear PTEN promotes cancer in a xenograft model (Ge et al., 2020), the in vivo effect of direct manipulation of PTEN localization on tumorigenesis remains to be determined.

We have recently created a mouse model in which the level of nuclear PTEN is decreased (Igarashi et al., 2018). In the nuclear PTEN-deficient mice, the size of both neurons and brains is decreased, demonstrating the importance of nuclear PTEN in the central nervous system (Igarashi et al., 2018). This brain phenotype is in sharp contrast to increased brain size in mice lacking the total PTEN owing to enhanced PI3-kinase signaling (Backman et al., 2001; Endersby and Baker, 2008; Knobbe et al., 2008; Kwon et al., 2006). Furthermore, unlike the total loss of PTEN, the deficiency of nuclear PTEN did not show spontaneous tumorigenesis (Igarashi et al., 2018). In this report, we analyzed this mouse model to study the role of nuclear PTEN in stress-induced tumorigenesis in vivo. We found that nuclear PTEN-deficient mice have increased DNA damage and enhanced formation of hepatocellular carcinoma when the mice are exposed to a carcinogen and hepatotoxin. These data suggest that nuclear PTEN plays an important role in tumorigenesis in vivo potentially through preventing DNA damage.

Results

Lysine 13 Is Essential for the Nuclear Accumulation of PTEN

To investigate the mechanism by which PTEN is accumulated in the nucleus in response to oxidative stress, we treated human prostate cancer cell line DU145 and primary mouse hepatocytes with H₂O₂. Immunofluorescence microscopy using anti-PTEN antibodies showed that PTEN is accumulated after the H₂O₂ treatment in both cell types (Figures 1A and 1B). This nuclear accumulation of PTEN was reversible: when ectopically expressed in DU145 cells, PTEN-GFP was first accumulated in the nucleus upon H₂O₂ treatment, similar to endogenous PTEN, and was then re-distributed in the cytosol and nucleus at 6 h after washout of H₂O₂ (Figures 1C and 1D). Similar to H₂O₂, increasing levels of nitric oxide, another reactive oxygen species, with spermine NONOate (SPNO) also induced the nuclear accumulation of PTEN-GFP (Figures 1E and 1F).

Figure 1.

Oxidative Stress Induces the Accumulation of PTEN in the Nucleus

(A and B) DU145 cells (A) and mouse primary hepatocytes (B) were treated with 0.5 mM H₂O₂ for 1 h. Immunofluorescence microscopy using anti-PTEN antibodies was performed. Nuclear DNA was stained with DAPI. The intensity of the PTEN signal in the nucleus was quantified relative to that in the cytosol. Bars are average ± SD (n = 52–84 cells).

(C and D) DU145 cells expressing GFP-PTEN were treated with 0.5 mM H₂O₂ for 1 h. After washes, cells were incubated in the culture medium without H₂O₂ for 6 h. Live cell imaging was performed using confocal microscopy. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (D). Bars are average ± SD (n = 27–74 cells).

(E and F) DU145 cells expressing PTEN-GFP were treated with 400 μM SPNO for the indicated amounts of time and observed by confocal microscopy. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (F). Bars are average ± SD (n = 40–52 cells).

(G and H) HCT116 cells expressing GFP fused to phosphorylation defective PTEN mutants were treated with 0.5 mM H₂O₂ for 1 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (H). Bars are average ± SD (n = 25–55 cells).

(I and J) DU145 and HCT116 cells were treated with 0.5 mM H₂O₂ for 1 h and analyzed by western blotting. The PTEN and phosphorylated PTEN band intensities were quantified relative to those of GAPDH and PTEN, respectively, in (J). Bars are average ± SD (n = 4–5). Statistical analysis was performed using Student's t test in (A, B, and J) and one-way ANOVA with post hoc Tukey in (D, F, and H): ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

Phosphorylation of the C-terminal region of PTEN modulates the conformation of PTEN and regulates its localization to the nucleus and the plasma membrane (Nguyen et al., 2014a, 2015b). Consistent with these previous reports, alanine mutations of the phosphorylation sites (S380, T382, T383, and S385) accumulated PTEN_{S380A,T382A,T383A,S385A} in the nucleus even in the absence of H₂O₂ in human colon cancer cell line HCT116 (Figures 1G and 1H). H₂O₂ did not further increase the nuclear localization of PTEN_{S380A,T382A,T383A,S385A}. In contrast, other phospho-defective PTEN mutants, PTEN_T319A,T321A and PTEN_T366A,S370A, behaved similarly to wild-type (WT) PTEN-GFP (Figures 1G and 1H). Western blotting showed that phosphorylation of S380, T382, T383, and S385 was not changed in response to H₂O₂ treatment (Figures 1I and 1J). Therefore, the phosphorylation of the C terminus of PTEN controls its basal nuclear localization and is not regulated by oxidative stress.

It has been suggested that lysine residues are important for the nuclear localization of PTEN (Bassi et al., 2013; Nguyen et al., 2015b; Song et al., 2008; Trotman et al., 2007). Using PTEN-GFP, we asked which lysines are required for PTEN to be accumulated in the nucleus in response to oxidative stress. We first mutated lysines that have been proposed to be subject to post-translational modifications, including K6, K13, K66, K80, K254, K289, and five lysines located in the CBR3 loop (K260, K263, K266, K267, and K269) to arginine (Kreis et al., 2014; Leslie et al., 2016). Substitution of these residues did not change the localization of PTEN in the presence or absence of H₂O₂, except for K13 in HCT116 cells (Figures 2A and 2B). PTEN_K13R-GFP was excluded from the nucleus before H₂O₂ exposure and failed to accumulate in the nucleus after H₂O₂ exposure. Similar results were obtained when we tested PTEN-GFP and PTEN_K13R-GFP in human hepatocarcinoma cell line HepG2 (Figures 2C and 2D). These data suggest that K13 is a key lysine that controls the nuclear localization of PTEN. Although PTEN has been shown to dimerize (Papa et al., 2014), we found no effects of endogenous PTEN (regardless of presence or absence of PTEN) on the distribution of WT PTEN-GFP or PTEN_K13R-GFP in response to H₂O₂ in three cell lines, which carry two copies (HCT116 cells), one copy (DU145 cells), or no copy (human prostate cancer cell line PC3) of PTEN (Figure S1).

Figure 2.

Lysine 13 is Critical for the Nuclear Localization of PTEN

(A and B) HCT116 cells expressing GFP fused to PTEN lysine mutants were treated with 0.5 mM H₂O₂ for 1 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (B). Bars are average ± SD (n = 23–63 cells).

(C and D) HepG2 cells expressing PTEN-GFP or PTEN_K13R-GFP were treated with 0.5 mM H₂O₂ for 1 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (D). Bars are average ± SD (n = 58–83 cells). Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (B and D): ∗∗∗p < 0.001.

See also Figure S1 and Table S1.

To further test whether all other lysines regulate PTEN's nuclear localization, we replaced all 34 lysines to arginines in PTEN, creating PTEN_K0 (Figures 3A and 3B). PTEN_K0-GFP was greatly excluded from the nucleus in the presence or absence of H₂O₂, similar to PTEN_K13R-GFP (Figures 3C and 3D). We then put back K13 in PTEN_K0 and created PTEN_K0K13. PTEN_K0K13-GFP restored both basal nuclear localization and H₂O₂-induced nuclear accumulation (Figures 3C and 3D). These data show that K13 is the major lysine residue that controls the nuclear localization of PTEN in response to oxidative stress.

Figure 3.

Lysine 13 is Sufficient for the Nuclear Localization of PTEN

(A) Lysine residues in PTEN. PTEN contains 34 lysines.

(B) Schematic diagrams of PTEN constructs. In the K0 mutant, all 34 lysines were changed to arginine. In the K0K13 mutant, all 34 lysines were changed to arginine expect for lysine 13.

(C and D) HEK293T cells expressing the indicated PTEN-GFP mutants were treated with 0.5 mM H₂O₂ for 1 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (D). Bars are average ± SD (n = 57–88 cells). Student's t test: ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

To ask if K13 undergoes ubiquitination, we expressed HA-ubiquitin along with WT PTEN-GFP, PTEN_K0-GFP, or PTEN_K0K13-GFP in HEK293T cells and immunoprecipitated using GFP-Trap beads. Western blotting showed that WT PTEN-GFP is at least mono-, bi-, and tri-ubiquitinated under the basal condition, and these ubiquitinations are dramatically decreased in PTEN_K0-GFP (Figures 4A and 4B). The majority of ubiquitination was restored in PTEN_K0K13-GFP (Figures 4A and 4B). Also, we found that mono-, bi-, and tri-ubiquitinations of WT PTEN-GFP are increased in response to H₂O₂ treatment (Figures 4C and 4D). In contrast, PTEN_K13R-GFP showed decreases in H₂O₂ induced bi- and tri-ubiquitinations (Figures 4C and 4D). These data suggest that K13 is the major site of ubiquitination in PTEN under both basal and oxidative stress conditions.

Figure 4.

The Ubiquitination of Lysine 13 Controls the Nuclear Localization of PTEN in Response to H₂O₂

(A and B) Whole-cell lysates of HEK293T cells expressing HA-ubiquitin along with the indicated PTEN-GFP constructs were incubated with GFP-Trap. The pellet fractions were analyzed by western blotting using antibodies to GFP and HA. Mono-, bi-, and tri-ubiquitination of PTEN-, PTEN_K0-, and PTEN_K0K13-GFP were quantified relative to that of PTEN-, PTEN_K0-, and PTEN_K0K13-GFP in (B). Bars are average ± SD (n = 4).

(C and D) HEK293T cells carrying HA-ubiquitin along with GFP-PTEN or PTEN_K13R-GFP were treated with 0.5 mM H₂O₂ for 1 h. PTEN-GFP and PTEN_K13R-GFP were pulled down using GFP-Trap beads. The pellet fractions were analyzed by western blotting using antibodies to GFP and HA. Ubiquitination was quantified in (D). Bars are average ± SD (n = 4–8).

(E) K13 was replaced by ubiquitin in PTEN_K13-Ubi.

(F and G) HCT116 cells expressing PTEN-GFP, PTEN_K13R-GFP, or PTEN_K13-Ubi-GFP were treated with 0.5 mM H₂O₂ for 1 h. Live cell imaging was performed using confocal microscopy. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (G). Bars are average ± SD (n = 45–91 cells).

(H and I) HEK293T cells carrying the indicated PTEN constructs were treated with 15 μM PR-619, a broad-range reversible and cell-permeable inhibitor of the deubiquitylating enzyme, for 2 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol (I). Bars are average ± SD (n = 33–134 cells). Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (B, D, G, and I): ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

To determine whether artificial fusion of ubiquitin to PTEN can replace the role of K13 for the nuclear localization, we placed ubiquitin between residue 12 and 14 (lysine 13 was removed) in PTEN-GFP and expressed PTEN_K13-Ubi-GFP in HEK293T cells (Figure 4E). PTEN_K13-Ubi-GFP restored both the basal nuclear localization and the nuclear accumulation in response to H₂O₂, similar to WT PTEN-GFP (Figures 4F and 4G). Therefore, the ubiquitination of K13 is important for the nuclear localization of PTEN.

If the ubiquitination of PTEN is critical for its nuclear localization, inhibiting de-ubiquitination might promote this localization. Indeed, when treated with a broad inhibitor to de-ubiquitination enzymes, PR-619, PTEN-GFP was accumulated in the nucleus without H₂O₂ (Figures 4H and 4I). This PR-619-induced nuclear accumulation depends on K13 since both PTEN_K13R-GFP and PTEN_K0-GFP did not accumulate in the nucleus in the presence of PR-619 (Figures 4H and 4I). Furthermore, PTEN_K0K13-GFP restored the PR-619-induced nuclear accumulation (Figures 4H and 4I). These data further support the importance of the ubiquitination of K13 in the nuclear localization of PTEN.

Reactive oxygen species create DNA damage in the nucleus. To test whether DNA damage induces nuclear accumulation of PTEN, we treated cells with a topoisomerase inhibitor, camptothecin. We found that camptothecin induces the nuclear accumulation of WT PTEN-GFP in HEK293T (Figures 5A and 5B), HepG2 (Figures 5C and 5D), HCT116 (Figure S1), DU145 (Figure S1), and PC3 (Figure S1) cells. In contrast, PTEN_K13R-GFP was unable to accumulate in the nucleus in the presence of camptothecin in all of these five cells (Figures 5A–5D and S1). Therefore, K13 is important for the nuclear accumulation of PTEN induced by DNA damage.

Figure 5.

Lysine 13 is Necessary for the Nuclear Localization of PTEN in Response to Camptothecin

(A and B) HEK293T cells expressing the indicated PTEN-GFP constructs were treated with 0, 10, 20, and 50 μM camptothecin (a topoisomerase inhibitor) for 6 h. An image of cells treated with 50 μM camptothecin is presented. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (B). Bars are average ± SD (n = 24–63 cells).

(C and D) HepG2 cells expressing PTEN-GFP or PTEN_K13R-GFP were treated with 0 (DMSO) or 50 μM camptothecin for 6 h. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (D). Bars are average ± SD (n = 61–79 cells). Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (B and D): ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

To test whether antioxidants can block the nuclear accumulation of PTEN in response to reactive oxygen species and DNA damage, we treated HCT116 cells expressing PTEN-GFP with either H₂O₂ or camptothecin in the presence of glutathione monoethyl ester (GSH-MEE). GSH-MEE significantly blocked the nuclear accumulation of PTEN induced by H₂O₂ (Figures 6A and 6B). In contrast, GSH-MEE was not able to block the nuclear recruitment of PTEN induced by camptothecin (Figures 6C and 6D). These data suggest that reactive oxygen species cause DNA damage and then promote nuclear PTEN localization.

Figure 6.

The Effect of an Antioxidant on the Nuclear Localization of PTEN

(A–D). After preincubation with 5 mM GSH-MEE for 16 h, HCT116 cells expressing PTEN-GFP were treated with 0.5 mM H₂O₂ for 1 h (A and B) or 50 μM camptothecin for 6 h (C and D). The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (B and D). Bars are average ± SD (n = 61–84 cells). Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (B and D): ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

Dynamics of Nuclear Import and Export of PTEN

Live cell imaging of PTEN-GFP showed that WT PTEN-GFP gradually accumulated in the nucleus after H₂O₂ treatment (Figures 7A and 7B). In contrast, PTEN_K13R-GFP remained in the cytosol after H₂O₂ treatment (Figures 7A and 7B). To determine how the K13R mutation affects the dynamics of PTEN's localization, we tagged a marker protein, mEos (Zhang et al., 2012), to PTEN and expressed PTEN-mEos in DU145 cells. mEos is a monomeric form of an Eos fluorescent protein with UV-inducible green to red color conversion (Zhang et al., 2012) When we photoconverted PTEN-mEos in the cytosol, a fraction of photoconverted PTEN-mEos was imported into the nucleus (Figures 7C, 7E, and 7G). Treatment with H₂O₂ did not affect the rate of the nuclear import of PTEN-mEos (Figures 7C, 7E, and 7G). In contrast, when we measured the rate of nuclear export by photoconverting PTEN-mEos in the nucleus, nuclear export was significantly decreased by H₂O₂ (Figures 7D, 7F, and 7G). These data suggest that the H₂O₂-induced nuclear accumulation of PTEN is achieved by decreasing the nuclear export rate without changing the nuclear import rate (Figure 7H).

Figure 7.

Time-Lapse Analysis of PTEN Localization

(A and B) DU145 cells expressing PTEN-GFP or PTEN_K13R-GFP were treated with 0.5 mM H₂O₂ for 1 h and analyzed by time-lapse confocal microscopy. The intensity of GFP in the nucleus was quantified relative to that in the cytosol in (B). Bars are average ± SD (n = 20 cells for PTEN-GFP and 22 cells for PTEN_K13R-GFP).

(C–G) DU145 cells were transfected with plasmids carrying PTEN- or PTEN_K13R-mEOS. mEos in the two regions of the cytosol in (C) or in the region of the nucleus in (D) was photoconverted using a 405-nm laser on laser scanning confocal microscope (indicated by asterisks). Images were obtained for both photoconverted and unconverted mEos signals in the presence or absence of 0.5 mM H₂O₂ for 30 min. Images before the photoconversion were also obtained (pre). The intensity of photoconverted mEos in the nucleus relative to that in the cytosol (in C and E, to measure the nuclear import of PTEN) and in the cytosol relative to that in the nucleus (in D and F, to measure the nuclear export of PTEN) was quantified. Bars are average ± SD (n = 10–12 cells). Rates of nuclear import and export are calculated in (G).

(H) Summary of the data. Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (G): ∗p < 0.05, ∗∗∗p < 0.001. Scale bars, 20 μm.

See also Table S1.

The rate of the nuclear import of PTEN_K13R-mEos was much lower than that of WT PTEN-mEos in the absence of H₂O₂ (Figures 7C, 7E, and 7G). This lower rate of import of PTEN_K13R-mEos was unchanged upon H₂O₂ (Figures 7C, 7E, and 7G). The rate of nuclear export of PTEN_K13R-mEos was also lower than that of WT PTEN-mEos and did not change in response to H₂O₂ (Figures 7D, 7F, and 7G). Therefore, defects in the nuclear accumulation of PTEN_K13R in the nucleus likely result from decreased rates of nuclear import, and oxidative stress does not change the export rate of PTEN_K13R. Therefore, the majority of PTEN_K13R remains in the cytosol regardless of oxidative stress (Figure 7H).

Oxidative Stress-Induced Nuclear Localization of PTEN Is Decreased in the Liver of Nuclear PTEN-Deficient Mice

To examine the nuclear localization of PTEN in vivo, we used a hepatotoxin, carbon tetrachloride (CCl₄), which produces oxidative stress in the liver (Recknagel et al., 1989; Sheweita et al., 2001). CCl₄ was intraperitoneally injected into WT mice and nuclear PTEN-deficient mice, in which the Pten gene had been edited to carry the K13R and D384V mutations using the CRISPR-Cas9 system (Figure 8A) (Igarashi et al., 2018). We have previously shown that the combination of K13R and mutations that block phosphorylation of the C-terminal tail, such as D384V, more effectively inhibits the localization of PTEN in the nucleus compared with the single K13R mutation (Igarashi et al., 2018; Nguyen et al., 2015b). Indeed, the K13R and D384V mutations specifically decreased the level of PTEN in the nucleus without affecting the level of PTEN in the cytosol (Figures 8B and 8C). Livers were subjected to immunofluorescence microscopy with anti-PTEN antibodies at 0, 2, 4, and 7 days after the CCl₄ injection. In WT mice, without CCl₄ injection, PTEN is located both in the cytosol and nucleus (Figures 8D and 8E). Upon CCl₄ injection, PTEN was accumulated in the nucleus with a peak at day 2 and gradually returned to the normal distribution at day 7 (Figure 8E). In contrast, PTEN in nuclear PTEN-deficient mice showed a decreased level of nuclear localization before CCl₄ injection and remained unchanged in the nucleus after the CCl₄ injection (Figures 8D and 8E). Therefore, PTEN is accumulated in the nucleus in vivo in response to oxidative stress, similar to cultured cells in vitro, and nuclear accumulation is blocked in nuclear PTEN-deficient mice.

Figure 8.

Oxidative Stress Induces the Nuclear Accumulation of PTEN in the Liver

(A) Nuclear PTEN-deficient mice carry K13R and D384V mutations. WT and nuclear PTEN-deficient mice were subjected to intraperitoneal injection of 20% CCl₄/mineral oil (10 mL/kg body weight) at 2–3 months of age.

(B) Liver sections from WT, nuclear PTEN-deficient mice, and heterozygous PTEN knockout mice were placed on the same slides and subjected to immunofluorescence microscopy with (left panel) or without (right panel) the primary antibody.

(C) The intensity of PTEN signals in the cytosol and nucleus was determined after the subtraction of the nonspecific background signal based on the intensity in samples without the primary antibody. Bars are average ± SD (n = 3–4 mice).

(D and E) (D) At 2 days after CCl₄ injection, frozen liver sections were analyzed by immunofluorescence microscopy using anti-PTEN antibodies. The intensity of PTEN signals in the nucleus was quantified relative to that in the cytosol at the indicated days after CCl₄ injection in (E). Bars are average ± SD (n = 101–202 cells).

(F and G) (F) At 2 days after CCl₄ injection, frozen liver sections were analyzed by immunofluorescence microscopy using anti-phospho-H2AX (serine139) antibodies along with DAPI staining. The intensity of phospho-H2AX signals in the nucleus was quantified at the indicated days after CCl₄ injection in (G). Bars are average ± SD (n = 3–6 livers).

(H and I) Livers were isolated from WT and nuclear PTEN-deficient mice at the indicated days after CCl₄ injection and analyzed by western blotting with antibodies to phospho-H2AX and GAPDH. The intensity of phospho-H2AX signals was quantified relative to that of GAPDH in (I). Bars are average ± SD (n = 4 livers).

(J) Serum ALT levels were measured at the indicated days after CCl₄ injection (n = 4–9 mice). Statistical analysis was performed using one-way ANOVA with post hoc Tukey in (C), (E), (G), (I), and (J): ∗∗p < 0.01, ∗∗∗p < 0.001. Scale bars, 20 μm.

DNA Damage Is Increased in Nuclear PTEN-Deficient Livers in Response to Oxidative Stress

To examine the effect of nuclear PTEN deficiency on DNA damage in the liver, liver sections were analyzed by immunofluorescence microscopy with antibodies to phosphorylated histone 2AX (phospho-H2AX), which is a well-established marker for DNA damage (Bassi et al., 2013), after injection of CCl₄. We measured the signal intensity of phospho-H2AX in each nucleus in the liver. Without CCl₄ injection, we barely detected phospho-H2AX in both WT and nuclear PTEN-deficient mice (Figures 8F and 8G). After CCl₄ injection, the signal intensity of phospho-H2AX in the nucleus was increased with a peak at day 2 and then decreased at day 4 in WT mice (Figures 8F and 8G). At day 2, nuclear PTEN-deficient mice showed a significantly higher signal intensity of phospho-H2AX, compared with WT mice (Figures 8F and 8G). Supporting these immunofluorescence data, western blotting of livers also revealed an increased level of phospho-H2AX in nuclear PTEN-deficient mice at day 2 (Figures 8H and 8I). As control experiments, we measured serum levels of alanine aminotransferase as a readout for liver damage and found similar activities in WT and nuclear PTEN-deficient mice before and after CCl₄ injection (Figure 8J). Therefore, equivalent damages were produced in the liver of these mice by CCl₄ injection. These data suggest that nuclear PTEN decreases DNA damage in response to oxidative stress in vivo.

When we analyzed oncogenic signaling pathways by western blotting of livers, we found similar levels of phosphorylation of ERK, AKT, S6, and GSK-3α/β in WT and PTEN-deficient mice (Figure 9). We also found normal insulin-stimulated AKT phosphorylation in the liver and glucose uptake in nuclear PTEN-deficient mice (Figure 10). Thus, nuclear PTEN-deficient mice show normal signaling involving AKT and ERK in the liver.

Figure 9.

Phosphorylation of AKT, ERK, GSK-3, and Ribosomal Protein S6 Is Not Affected in the Liver of Nuclear PTEN-Deficient Mice

(A and B) Western blotting of livers dissected from WT and nuclear PTEN-deficient mice at the indicated days after CCl₄ injection using the indicated antibodies. Band intensity was quantified in (B). Bars are average ± SD (n = 4–9 livers). Statistical analysis was performed using one-way ANOVA with post hoc Tukey.

Figure 10.

Nuclear PTEN-Deficient Mice Normally Respond to Insulin

(A) WT and nuclear PTEN-deficient mice were subjected to intraperitoneal injection of insulin (5 U/kg body weight). At 10 min after injection, blood glucose levels were measured. Bars are average ± SD (n = 3–11 mice).

(B and C) Western blotting of livers using antibodies to AKT, phospho-AKT (S473). Band intensity was quantified in (C). Bars are average ± SD (n = 3 livers). Statistical analysis was performed using one-way ANOVA with post hoc Tukey: ∗∗p < 0.01, ∗∗∗p < 0.001.

Increased Hepatocellular Carcinoma in Nuclear PTEN-Deficient Mice in Response to Oxidative Stress

To test whether nuclear PTEN suppresses tumorigenesis in vivo, we used a preclinical mouse model of hepatocellular carcinoma induced by a single injection of a carcinogen, N-nitrosodiethylamine (DEN), followed by repeated injection of the hepatotoxin CCl₄ (Heindryckx et al., 2009; Uehara et al., 2014). We intraperitoneally injected DEN at 2 weeks of age and subsequently injected CCl₄ twice per week from 6 to 14 weeks of age (Figure 11A). We found similar body and liver weights in WT and nuclear PTEN-deficient mice with or without DEN and CCl₄ treatments (Figures 11B and 11C).

Figure 11.

Increased Hepatocarcinoma in Nuclear PTEN-Deficient Mice

(A) Experimental design for DEN/CCl₄-induced liver cancer in mice. WT and nuclear PTEN-deficient mice were subjected to intraperitoneal injection of DEN at 2 weeks of age. Mice were further treated with CCl₄ twice per week from 6 to 14 weeks of age. Mice were analyzed at 15 weeks.

(B and C) Weights of the body (B) and liver (C) (n = 4 male and 4 female mice for control, 7 male and 7 female mice for DEN/CCl₄-treated).

(D) Images of the front and back sides of livers that are isolated from male mice. Scale bar, 1 cm.

(E and F) The number of tumors (E) and the size of the largest tumors (F) observed on the surface of the livers were quantified. Bars are average ± SD (n = 4 male and 4 female mice for control, 7 male and 7 female mice for DEN/CCl₄-treated).

(G–M) Liver sections were subjected to H&E staining (G), Sirius red staining (H), Ki67 immunostaining (J), and active caspase-3 immunostaining (L). For active caspase-3, liver sections at 2 days after CCl₄ injection were also analyzed (CCl₄). T, tumors; N, non-tumor regions. Scale bars, 200 μm. (I) Fibrotic regions were quantified based on Sirius red staining. Bars are average ± SD (n = 6–13 livers). (K) Ki67-positive cells were quantified. Bars are average ± SD (n = 5–9 livers). (M) Active caspase-3-positive cells were quantified. Bars are average ± SD (n = 3–5 livers). Statistical analysis was performed using Student's t test in (B), (C), (E), (F), (I), (K) and (M): ∗p < 0.05, ∗∗∗p < 0.001.

In male WT mice, the DEN and CCl₄ treatments induced hepatocellular carcinoma at 15 weeks (Figures 11D–11G), as previously reported (Heindryckx et al., 2009; Uehara et al., 2014). We found increases in both the number of hepatocellular carcinomas and the size of the largest tumors in nuclear PTEN-deficient mice (Figures 11E and 11F). It has been shown that female mice are protected from DEN-induced hepatocarcinogenesis owing to estrogen-mediated inhibition of interleukin-6 (Naugler et al., 2007). Indeed, we found a much higher level of hepatocellular carcinoma in male WT mice compared with female WT mice, in which hepatocellular carcinoma was almost undetectable (Figure 11E). Female nuclear PTEN-deficient mice modestly increased hepatocellular carcinoma compared with female WT mice, but this difference did not reach statistical significance (Figures 11E and 11F). Without the treatment with DEN and CCl₄, no hepatocellular carcinoma was observed in WT and nuclear PTEN-deficient mice regardless of their sex (Figures 11D and 11E).

DEN and CCl₄ treatments induce fibrosis, as in human patients with hepatocellular carcinoma (Heindryckx et al., 2009; Uehara et al., 2014). We found similar levels of fibrosis in male WT and nuclear PTEN-deficient mice using Sirius red stain of liver sections (Figures 11H and 11I). Therefore, liver injuries produced by DEN and CCl₄ treatments were similar in WT and nuclear PTEN-deficient mice. In addition, Ki67 immunostaining showed similar percentages of Ki67-positive cells in regions of hepatocellular carcinoma in male WT and nuclear PTEN-deficient mice (Figures 11J and 11K). Furthermore, immunostaining with antibodies to active caspase-3 revealed similar levels of apoptosis in response to CCl₄ in WT and nuclear PTEN-deficient mice (Figures 11L and 11M). No detectable apoptosis was observed in normal regions and tumors in WT and nuclear PTEN-deficient mice at the basal level (Figures 11L and 11M). These data suggest that hepatocellular carcinoma grows at similar rates without gross apoptosis in WT and nuclear PTEN-deficient mice once they are formed and that the loss of nuclear PTEN increases the frequency of the onset of tumorigenesis. Thus, nuclear PTEN functions as a tumor suppresser and protects against oxidative stress-induced hepatocellular carcinoma in vivo.

Discussion

Hepatocellular carcinoma is associated with persistent liver damages due to viral infections, environmental chemicals, excess alcohol consumption, and metabolic syndrome, such as non-alcoholic steatohepatitis, and is manifested by liver fibrosis or cirrhosis (Erstad et al., 2019; Mishra et al., 2009; Yang et al., 2019). PTEN has been shown to be mutated or lost in up to 30% of hepatocellular carcinoma (Khalid et al., 2017; Parikh et al., 2010). Also, decreased levels of PTEN are found in more than 50% of hepatocellular carcinoma and could be used as a prognostic marker (Khalid et al., 2017; Parikh et al., 2010). Therefore, it is important to understand the function of PTEN in the pathogenesis of hepatocellular carcinoma. Because PTEN has been shown to function in multiple locations (Baker, 2007; Gil et al., 2007; Hopkins and Parsons, 2014; Kreis et al., 2014; Leslie et al., 2016; Planchon et al., 2008), deciphering its functional importance at distinct subcellular locations will be critical to revealing how the loss of PTEN contributes to pathogenesis. In this study, we investigated the role of PTEN in the nucleus in hepatocellular carcinoma using a preclinical platform, in which hepatocellular carcinoma is induced by the combination of carcinogen (DEN) and repeated injection of the oxidative stress-inducing hepatotoxin (CCl₄). This mouse model induces extensive fibrosis in the liver, similar to human patients (Heindryckx et al., 2009; Uehara et al., 2014). We found that the formation of hepatocellular carcinoma was greatly accelerated in nuclear PTEN-deficient mice with increased DNA damage. These data provide the first in vivo evidence that nuclear PTEN mitigates the tumorigenesis by controlling DNA damage.

Although PTEN negatively controls PI3-kinase/AKT signaling as a lipid phosphatase at the plasma membrane and, therefore, total loss of PTEN enhances this oncogenic signaling pathway, nuclear PTEN-deficient mice maintain normal signaling involving AKT, mTOR, and ERK. Previous studies have reported that the loss of total PTEN in the liver in hepatocyte-specific PTEN knockout mice results in increase liver sizes and abnormality in insulin signaling due to increased PI3-kinase/AKT signaling (Horie et al., 2004; Stiles et al., 2004). These hepatocyte-specific Pten knockout mice also showed spontaneous adenomas and hepatocellular carcinoma at around 1 year old. Altered liver size and physiology and tumorigenesis in this mouse model likely involve a combination of consequences that are caused by the loss of PTEN functions in all of its subcellular locations. In contrast, nuclear PTEN-deficient mice showed a normal liver size and normal insulin response. Therefore, our data suggest that nuclear PTEN is unlikely to control the growth of the liver and glucose metabolism under normal conditions. Since nuclear PTEN-deficient mice do not produce spontaneous tumors until at least 1.5 years of age (Igarashi et al., 2018), nuclear PTEN plays an important role in protecting the liver from tumorigenesis when animals are exposed to stress and chronic liver damage.

CCl₄ is metabolized to highly reactive trichloromethyl radical by cytochrome p450, mainly in the liver, leading to oxidative stress and liver injuries (Recknagel et al., 1989; Sheweita et al., 2001). Our data suggest that K13 is important for the nuclear localization of PTEN as well as its nuclear accumulation in the liver under oxidative stress induced by CCl₄. Consistent with these in vivo data, K13 is important for the nuclear accumulation of PTEN in cell culture systems in response to H₂O₂. Using live-cell imaging with PTEN fused to photoconvertible mEos, we found that K13 controls the nuclear import of PTEN. Also, although WT PTEN decreases its nuclear export under oxidative stress, PTEN_K13R was unable to decrease the nuclear export under oxidative stress. Therefore, K13 controls both the nuclear import rate at the basal condition and the modulation of the nuclear export rate in response to oxidative stress.

It has been shown that PTEN undergoes multiple post-translational modifications, such as SUMOylation, phosphorylation, ubiquitination, and acetylation (Kreis et al., 2014; Leslie et al., 2016). These modifications of PTEN regulate its subcellular localization, enzymatic activity, and protein stability. The multiple layers of regulation enable PTEN to function as an important tumor suppressor with multiple functions at different locations. In our experiments, K13 was critical for the dynamics of the nuclear localization and accumulation; however, we do not rule out the possibility that, under different physiological and pathological conditions, other lysines and residues could provide crucial regulatory mechanisms through different post-translational modifications. Supporting this notion, multiple ubiquitin E3 ligases, protein kinases, and other enzymes have been reported to modify PTEN in previous studies (Ge et al., 2020; Kreis et al., 2014; Leslie et al., 2016). These regulatory mechanisms could also cross talk and create a sophisticated regulatory network for PTEN. It would be of great interest to determine how different regulations affect each other in healthy cells and animals and how changes in the regulatory system drive the pathogenesis of cancers.

Limitation of the Study

First, we demonstrate that the nuclear localization of PTEN depends on its ubiquitination at lysine 13. However, we have not addressed what ubiquitin ligase mediates this post-translational modification. Second, we show that oxidative stress and DNA damage drive the accumulation of PTEN in the nucleus by decreasing its nuclear export. Yet it remains to be determined what machinery controls this export process. Third, although the lack of nuclear PTEN accelerates the formation of hepatocellular carcinoma that is induced by hepatotoxin and oxidative stress, it is unknown whether nuclear PTEN functions as a tumor suppressor in other types of cancers. We will address these questions in our future studies.

Resource Availability

Lead Contact

Miho Iijima (miijima@jhmi.edu).

Materials Availability

This study did not generate new reagents.

Data and Code Availability

All data are included in the published article and the Supplemental Information, and any additional information will be available from the lead contact upon request.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: October 23, 2020