CI Suppresses mTOR and Causes Reversible Arrest.

First, we plated normal retinal pigment epithelium (RPE) cells at regular density (RD) and HD. After 6 d, cells plated at RD reached subconfluency (Fig. 1A, RD), whereas the cells plated at HD became “packed,” i.e., were completely contact-inhibited (Fig. 1A, HD). Contact inhibited cells are typically characterized by positive β-gal staining (22⇓⇓⇓⇓⇓⇓⇓⇓–31). β-Gal staining is not only a hallmark of senescence but also of CI and serum starvation because all three conditions are associated with lysosomal activation (22⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓–33). Importantly, pS6, a marker of the mTOR activity, was undetectable in contact-inhibited cells (Fig. 1B). Contact-inhibited cells retained replicative potential (RP), meaning that the cells restarted proliferation after splitting of the culture (Fig. 1C). This indicated that CI was a reversible condition. In agreement, contact-inhibited cells were arrested in G1 phase as shown by flow cytometry (Fig. 1D). After splitting of the culture, cells reentered the cell cycle, so that 60% of cells were in S-phase after 24 h. Cells progressed to normal cell cycle distribution by 48 h (Fig. 1D). Similarly, human WI38t fibroblasts retained RP during CI, which was associated with decreased S6 phosphorylation (Fig. 1 E–G).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Characterization of reversible CI of normal cells. (A) β-Gal staining. RPE cells were plated at 1 million (HD) and 100,000 (RD) per well in six-well plates. After 6 d in culture, cells were packed in wells plated at 1 million (HD) and were confluent (with some spaces in monolayers) in wells plated at 100,000 (RD). (B) Immunoblot analysis. RPE cells were plated at RD and HD (A) and lysed after 6 d in culture. (C) RP: RPE cells were plated at RD and HD (A) and, after 6 d in culture, cells were split and replated at LD to regrow. After 8 d, cells were counted. (D) Cell cycle distribution: RPE cells were plated at HD and analyzed after 6 h (6 h initial condition) and 6 d in culture by flow cytometry. In parallel, a 6-d culture was replated at LD and cell cycle distribution was analyzed after 24 h and 48 h in culture. (E) β-gal staining. WI38t cells were plated at HD and RD in six-well plates. After 6 d in culture, cells were packed in wells plated at HD and were confluent (with some spaces in monolayers) in wells plated at RD. (F) Immunoblot analysis: WI38t cells were plated at RD and HD (A) and lysed after 6 d in culture. (G) RP: WI38t cells were plated at RD and HD (E) and, after 6 d in culture, cells were split and replated at LD and allowed to regrow. After 7 d, cells were counted.

p27 Induction Is Associated with mTOR Inhibition.

We next investigated the relationship between cell density, p27 induction, and phosphorylation status of S6. RPE cells were plated at increasing cell densities and incubated for 2 d. Levels of p27, a marker of CI (34⇓⇓⇓⇓⇓–40), were progressively increased with increasing cell density. pS6 became undetectable at the higher cell densities (Fig. 2A), i.e., when plated at 2 × 10⁵ per well in a 12-well plate and 4 × 10⁵ per well in a 24-well plate.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Inhibition of the mTOR pathway in CI associated with p27 induction. (A) Immunoblot analysis: RPE cells were plated at a range of densities (shown from highest to lowest) and lysed after 2 d in culture. (B and C) Immunoblot analysis: RPE cells (B) and WI38t cells (C) were plated at HD and lysed on the days indicated. (D) Immunoblot analysis: IEC18 (rat intestinal epithelial) cells were plated at HD and RD. After 4 d, cells were lysed. In a parallel set, culture media were changed to fresh medium for 1 h before lysis (Med Δ). (E) RPE and IEC18 cells were plated at HD. After 2 d in culture, one set of cells was lysed and a second set was replated at LD into fresh medium (marked as “F”) or CM collected from the respective HD culture. At 24 h after replating, cells were lysed for immunoblotting. (F) Immunofluorescence: RPE cells were grown in colonies and stained for p-S6. Cells were photographed under light and fluorescence microscopes.

Next we determined the time course of p27 induction and S6 dephosphorylation in several cell lines (Fig. 2 B–D). At the time of plating (day 0), the Akt/mTOR pathway was activated in RPE cells (Fig. 2B). By day 2, p27, a marker of CI, was induced. At that time, S6 became dephosphorylated and phosphorylation of Akt S473 was decreased. Noteworthy, p53 was not changed (Fig. 2B). Similar results were obtained in human fibroblast WI38t cells, in which p27 was strongly induced by day 2 and phospho-AKT and pS6 became undetectable at that time (Fig. 2C). p53 was not changed. Thus, we observed that the appearance of p27 coincided with dephosphorylation of S6 and AKT. Also, in normal human bladder cells (NBCs), pS6 was decreased at the cell density when p27 was induced (Fig. S1A). We also determined that the Akt/mTOR pathway was inhibited in contact inhibited rat intestinal epithelial IEC18 cells (Fig. 2D).

The change of the medium did not activate the Akt/mTOR pathway in IEC18 cells, suggesting that CI rather than medium conditioning/exhaustion was responsible for mTOR deactivation in this cell line. We also investigated the effect of conditioned medium (CM) from high-density cultures on S6 phosphorylation in the cells plated at RD. We found that CM, even collected from 2–3-wk-old contact-inhibited cultures only marginally (not completely) decreased pS6 level in cells from RD culture (Fig. S1B).

We next used a very sensitive method to detect the effect of CM on the activity of mTOR. Contact-inhibited cells (with undetectable pS6) were replated at regular cell density into fresh medium or into their own CM (Fig. 2E). We found that both p-S6 kinase (T389) and p-S6 were reactivated even when cells were split into CM (Fig. 2E). However, this reactivation was higher in fresh medium than in CM, because mTOR is regulated by numerous medium factors (nutrients, hormones, and mitogens, which may be at higher levels in fresh medium).

Finally, we determined the pattern of S6 dephosphorylation in RPE cells growing as separate dense colonies. As shown in Fig. 2F and Fig. S2, the cells were preferentially negative for pS6 in the middle of the colonies, whereas, at the edges, all cells were pS6-positive. This supports the conclusion that it is the CI rather than conditioning of the medium that is predominantly responsible for mTOR deactivation in normal cells from confluent cultures.

We next investigated whether stimulation of mTOR would convert CI into cellular senescence. To this end, we used lentivirus expressing TSC2 shRNA (shTSC2), which, as we previously described, decreased the levels of TSC2, thus activating mTOR (41, 42). Here we infected RPE and IEC18 cells and then plated them at high cell density to cause CI. After 4 d, the cells were replated at low cell density to determine if cells become senescent. In control, contact inhibited RPE and IEC18 cells (infected with empty vector) regrew after splitting (Fig. 3A). shTSC2-infected cells did not resume proliferation after replating at low density (LD; Fig. 3A), shTSC2-infected contact-inhibited cells acquired large-cell morphology when replated at LD (Fig. 3B). They retained β-gal positivity, which is typical for CI and senescence, even though they were no longer contact-inhibited (solitary cells). Control cells in contrast become β-gal–negative. Also, some shTSC2-infected cells displayed senescence-associated heterochromatin foci (Fig. S3), an additional marker of senescence.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Effect of shTSC2 on senescence of contact-inhibited cells. RPE (A) and IEC18 (B) cells were infected with lentivirus pLKO vector or pLKO-shTSC2 and then were plated at HD. After 4 d, cells were replated at LD and allowed to grow. RP was determined by fold increase in cell number after replating: RPE cells were counted after 8 d of regrowth, and IEC18 cells were counted after 3 d of regrowth. Replicate sets of replated RPE and IEC18 cells were stained for β-gal after 3 d of regrowth. Percentage of senescent cells was determined by counting β-gal–positive cells in three separate fields. Data presented as mean ± SD. (Scale bar: 100 µm.)

High Cell Density Prevents p21-Induced Senescence.

We next investigated whether mTOR inhibition in high cell density can prevent cellular senescence caused by senescence-inducing agents. As described previously, induction of ectopic p21 causes cellular senescence in HT-p21 cells (5, 43⇓–45). In these cells, addition of isopropyl-thio-galactosidase (IPTG) induces p21, causing cell cycle arrest and irreversible senescence, meaning that the cells cannot restart proliferation when IPTG is removed. Here HT-p21 cells were plated at increasing cell densities and treated with IPTG for 3 d (Fig. 4A). After 3 d, cells were trypsinized and replated at very LD: 1,000 viable cells per well (six-well plate). After 8 d in culture formed colonies were stained and counted (Fig. 4A). A number of colonies (which corresponded to a number of nonsenescent cells, i.e., replication-competent cells) increased when cells were treated with IPTG at HD. A number of nonsenescent cells (colonies) increased from 4.7% to 25.8%, when the plating cell density was increased from 50,000 to 1,000,000 cells per well (Fig. 4A). Therefore, IPTG did not cause senescence in dense cultures. However, IPTG still induces p21 in dense cultures. So, the conversion from cell cycle arrest to cellular senescence (geroconversion) was suppressed in HD.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Effects of high-density and CM on p21-induced senescence in HT-p21 cells. (A) Effect of cell density on senescence in HT-p21 cells. RP: HT-p21 cells were treated with IPTG at different densities (as indicated). After 3 d, cells were trypsinized, and 1,000 viable cells were replated in fresh medium without IPTG. Colonies were stained after 8 d. Number of colonies is presented as percentage of 1,000 seeded cells ± SD. (B) β-Gal staining: CM was collected from HT-p21 cells growing for 3 d at LD (CM-LD, 10,000 per well in a six-well plate) or HD (CM-HD, 1,000,000 per well in a six-well plate). Then, HT-p21 cells were plated at LD and treated with IPTG in CM-HD or CM-LD. After a 3-d treatment, cells were stained for β-gal. Percentage of senescent cells was determined by counting β-gal–positive cells in three separate fields. Data presented as mean ± SD. (Scale bar: 100 µm.) (C) Immunoblot analysis: HT-p21 cells in RD were treated with IPTG, when indicated, in fresh medium or CM-HD (CM-HD collected from HD culture of HT-p21 cells after 1 d and 2 d in culture) and lysed 24 h later. I, IPTG 50 µg/mL; R, rapamycin 500 nM.

We next investigated whether CM from HD cultures can inhibit IPTG-induced senescence in HT-p21 cells. HT-p21 cells were placed in CM collected from LD and HD cultures (CM-LD and CM-HD, respectively) and treated with IPTG. CM-HD, but not CM-LD, prevented senescent morphology of HT-p21 cells (Fig. 4B). Consistently, CM-HD inhibited pS6 in freshly plated cells at RD (Fig. 4C), whereas it did not prevent induction of p21 (Fig. 4C). Neither rapamycin nor CM prevented p21 induction and p21-dependent cell cycle arrest (41, 42, 45⇓–47). mTOR is not needed for arrest. However, it is needed for geroconversion from cell cycle arrest to irreversible senescent phenotype (12). Rapamycin and CM inhibited mTOR and thus inhibited geroconversion (i.e., conversion from reversible arrest to senescence). Therefore, inhibition of mTOR does not prevent p21-mediated arrest but it prevents the final step: geroconversion. Although it does not affect p21 expression, inhibition of mTOR prevents the senescent program.

Given that inhibition of mTOR by any means suppresses geroconversion of IPTG-treated HT-p21 cells, this inhibition could account for senescence-inhibiting properties of CM-HD. We confirmed this finding in two additional cancer cell lines. Like in HT-p21 cells, p27 was not detected in high cell density cultures of MCF7 and RKO cells, yet the Akt/mTOR pathway was inhibited (Fig. S1 C and D). The change of the media reactivated the pathway (Fig. S1 C and D). Thus, in cancer cells lacking CI, mTOR is still inhibited at high cell density as a result of conditioning of medium.

CI Suppresses Senescent Program.

Cancer cells do not undergo CI. To imitate the conditions of CI, a very small number of HT-p21 cells was trapped among contact-inhibited RPE cells. (RPE cells exert strong CI without conditioning the medium.) The same small number of HT-p21 cells were plated without RPE cells, as a regular culture condition (Fig. 5). HT-p21 cells express GFP and therefore could be distinguished from RPE cells under fluorescence microscopy. We treated cells with IPTG for 3 d.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

CI of cancer HT-p21 cells in confluent cultures of RPE cells prevents IPTG-induced senescence. (A) HT-p21 cells express GFP and can be distinguished from RPE cells under fluorescence microscopy. HT-p21-9 cells were plated at LD alone (regular culture) or together with HD RPE cells (number of plated HT-p21 cells was ∼5% of the number of RPE cells, i.e., 2,500:50,000; CI, culture) and treated with IPTG. After 3 d, cells were fixed and stained for pS6. Cells were photographed under a fluorescence microscope. (B–D) A small number of HT-p21 cells were plated together with a high number of RPE cells as in A and treated with IPTG. After 3 d, cells were trypsinized and replated: One thousand HT-p21 cells were plated per well in six-well plates. (B) On day 1, green (HT-p21) cells were photographed. (C) RP: after 7 d, green cells were counted. Data are presented as fold increase in number of HT-p21 (green) relative to initially plated numbers. (D) After 7 d, cells were photographed under light and fluorescence microscopes. Overlaid images display green colonies of HT-p21 cells on the top of nonfluorescent monolayer of RPE cells. (E–G) HT-p21 cells were plated and treated with IPTG in regular culture (alone) or together with RPE cells (CI culture) as in B–D. After 3 d, treatment cells were sorted for GFP by using flow cytometry, and green cells were plated at 500 per well in 12-well plates in drug-free medium. Colonies were allowed to grow for 6 d and stained with crystal violet. (E) Cells were photographed under light and fluorescence microscopes on day 2 after plating. (F) Cells were photographed 6 d after replating. (G) RP: colonies of HT-p21 cells treated with IPTG in regular culture or in the presence of HD RPE cells (CI culture), then sorted for GFP and grown in drug-free medium for 12 d.

As shown in Fig. 5 A and B, HT-p21 cells plated alone (regular culture) acquired senescent morphology. In contrast, HT-p21 cells trapped among RPE cells remained small in size. In regular culture of HT-p21 cells (green nucleus), cells remained “red” (pS6-positive cytoplasm). When HT-p21 cells were trapped in the RPE monolayer (Fig. 5A and Fig. S4), all “green” (HT-p21 cells) became pS6-negative (Fig. 5A and Fig. S4).

To measure RP of such cells, IPTG was washed out and cells were replated at LD and allowed to regrow. HT-p21 cells treated with IPTG inside of RPE monolayer retained proliferative potential whereas HT-p21 cells from standard culture did not (Fig. 5C). We conclude that HT-p21 cells inhibited by RPE monolayer did not become senescent. In agreement, they formed large colonies upon replating, whereas HT-p21 cells from regular culture did not (Fig. 5D). To minimize the effect of contaminating RPE cells on colony formation, we next sorted HT-p21 cells from mixed cultures by using GFP marker and replated them at LD in drug-free medium. After attachment, HT-p21 cells from regular culture displayed senescent morphology, whereas the cells from CI culture did not (Fig. 5E). In agreement, CI-culture cells proliferated upon replating and formed colonies, in contrast to HT-p21 cells from regular culture (Fig. 5 F and G).

We also investigated whether CI suppresses senescence caused by PD0332991, an inhibitor of CDK4/6, in normal RPE cells (46). PD0332991 caused morphological senescence in LD cultures, whereby cells were p-S6 positive (Fig. S5). These senescent cells did not restart proliferation when PD0332991 was washed out (Fig. S5). In CI culture, the cells were p-S6 negative (Fig. S5). Despite the treatment with PD0332991, contact inhibited cells did not become senescent and resumed proliferation after replating.