Cyclin D1-induced senescencein vivooccurs over a time frame of several weeks
: We used the Irbp-Cyclin D1
mouse in which the expression of Cyclin D1 in the pineal gland causes excessive proliferation that is limited by senescence. The net result is a hyperplastic but senescent pineal gland that does not progress into an invasive tumor unless either p53 or the Cdk4-inhibitor p18Ink4c
is lost [13
]. We examined the temporal evolution of senescence and the contribution of the p53 and Rb tumor suppressor pathways to cell cycle exit in vivo
. Histological studies of Cyclin D1-expressing pineal glands at various ages (post-natal day (P)10, P24, P35, and P49) showed that enhanced proliferation, measured by Ki67 immunostaining, was apparent at P10 but after this point it decreased such that essentially all cells had exited the cell cycle by P35 [Figure A, left panel, Quantitation in Figure B]. Cessation of proliferation preceded the formation of senescence-associated heterochromatin foci (SAHF); in fact, there was an unexpectedly long, two-week delay from P35 to P49 before SAHF were apparent [Figure A, middle and right panels]. Because SAHF have been observed in only a few mouse models of senescence [14
], but not in other murine cells like MEFs [17
], and because constitutive centromeric heterochromatin may show intense DAPI staining mimicking SAHF in normal murine cells [18
], we verified that these foci were indeed only seen in Cyclin D1-expressing cells and not the wild-type counterparts at P49 Additional file 1
: Figure S1A]. We could not detect positive staining for senescence-associated beta galactosidase (SABG), another marker of senescence [20
]. However, we believe this must be a technical issue (detecting the enzyme activity necessitates freshly frozen tissue), because when the Cyclin D1-expressing pineal cells were grown in vitro
, they showed features of senescence that included positive staining for SABG [see below, and Additional file 1
: Figure S1B]. In addition, we evaluated the pineal cells for other known markers of senescence, such as Dec1, DcR2, and p15Ink4b [21
]. We found that, concomitant with cell cycle exit and SAHF formation, all three markers of senescence were increased at P49 [Figure C]. We conclude that Cyclin D1-induced senescence in the pineal cells occurs by P49, and that cell cycle exit in this setting occurs many days prior to expression of bona fide
markers of senescence.
Figure 1 Cyclin D1-induced senescence occurs over several weeks. A) Ki67 staining of Irbp-Cyclin D1 pineal sections at the indicated ages (left); immunoflourescence staining for Histone 3 trimethylated at Lysine 9 (H3K9me3) in pineal sections at the indicated (more ...)
p53 activation occurs prior to cell cycle exit, and is mediated by a ROS-induced DNA damage response: To identify when p53 was engaged in tumor suppression, we evaluated cell lysates of pineal glands collected at different postnatal ages. We found that at P10, p53 protein expression was increased, as was its phosphorylation at Ser15/20 (a site phosphorylated by the DNA damage response) [Figure A]. The p53 target, p21Cip1, was also induced at that time-point [Figure A]. However, p53 activation did not persist once the cells had exited the cell cycle nor was it detectable at P49 when the cells displayed SAHF [Figure A].
Figure 2 Cyclin D1-induced cell cycle exit is associated with early p53 activation and a DNA damage response. A) Western blotting for the indicated proteins in Irbp-Cyclin D1 pineal glands at the indicated ages. HA: HA epitope on the Cyclin D1 transgene. B) Representative (more ...)
We hypothesized that Cyclin D1 expression may be inducing p53 through activation of the DNA damage response (DDR), as reported for Ras-induced senescence (reviewed in [22
]). Indeed, we found nuclear accumulation of phosphorylated histone H2AX (pH2AX) at P10, concomitant with p53 activation [Figure B, C]. In addition, Chk1 was phosphorylated concomitantly with phosphorylation of p53, further indicating that the DDR pathway was active in the transgenic pineal gland at this time, but not at later time-points [Figure D]. These findings reveal that deregulated Cyclin D1 enhances the DDR pathway and activates p53 while cells are proliferating, but ongoing DDR and active p53 are not needed after cells have undergone senescence.
We next assessed whether reactive oxygen species (ROS) may be contributing to Cyclin D1-induced senescence in pineal cells, based on their role in Ras-induced senescence in cultured fibroblasts [23
]. Using DCF-DA assay in explanted pineal cells, we found that ROS were indeed induced in response to Cyclin D1 expression, but not in wild-type pineal cells grown in the same conditions [Figure A]. We investigated whether ROS were responsible for activation of the DNA damage response in this setting. Indeed, Cyclin D1-induced ROS resulted in DNA damage foci marked by pH2AX, as well as the ROS-induced incorporation of oxidized dNTP, 8-Oxo-dGTP, into DNA Additional file 1
: Figure S1D]. It also led to phosphorylation of Chk1, a component of the DNA damage response pathway [Figure , left panel], and to increased expression of two p53 pathway effector proteins, 14-3-3 and p21 [Figure B, middle and right panels]. Treating cells with the ROS scavenger N-Acetyl Cysteine (NAC) resulted in abrogation of DNA damage Additional file 1
: Figure S1D], abrogation of DDR activation, and absence of p53 pathway activation [Figure B], as well as evasion of senescence [Figure C]. By staining for 4-hydroxy-nonenal (HNE), a marker of lipid oxidation, we found evidence of oxidative stress in pineal sections of Irbp-Cyclin D1
mice, but not wild-type mice [Figure D], and there was also increased expression of the mitochondrial superoxide dismutase protein MnSOD [Figure E], which is induced by ROS stress in vivo
]. Thus, from the above in vitro
and in vivo
evidence, we conclude that Cyclin D1 expression results in accumulation of ROS, which in turn leads to activation of the DDR and the p53 pathway, resulting in induction of senescence.
Figure 3 Cyclin D1 expression leads to ROS accumulation, which results in a DNA damage response and senescence induction. A) DCFDA assay in cultured pineal cells that are either wild-type (WT) or express Cyclin D1 (Cyclin D1). PC=phase contrast; (more ...)
RB pathway activation occurs after cell cycle exit and is associated with SAHF: To evaluate when the Rb pathway was engaged, we used western blotting to investigate the Cdk-dependent phosphorylation of Rb in pineal cell lysates. We investigated the status of Rb phosphorylation at Cdk4-dependent sites such as Ser790, and at Cdk2-dependent sites such as Ser612 [Figure A]. We found that Rb was phosphorylated at Cdk2-dependent sites (Ser612) at P10, when cells were proliferating, but decreased after cells exited the cell cycle by P24 [Figure A]. In contrast, Rb was phosphorylated at Cdk4-dependent sites (Ser790) from P10 through P35, even though most cells had ceased to proliferate by P24 [Figure A]. Rb phosphorylation at Cdk4-dependent sites was reversed at P49 as SAHF formed [Figure A].
Figure 4 p18Ink4c loss delays Cyclin D1-induced senescence. A) and B) Western blotting for the indicated proteins in wild-type (WT) and Irbp-Cyclin D1 (D1) pineal glands at the indicated ages. Hsc70 is a loading control. C) Quantitative analysis of Ki67-positive (more ...)
To understand the mechanism of Rb activation, we investigated the expression of the Cdk-inhibitors p16Ink4a, p15Ink4b, p18Ink4c, and p27Kip1 [27
]. There was increased expression of p18Ink4c
at all time points [Figure B], an increase in p15Ink4b
expression at P49 [see Figure D, but no changes in expression of p16Ink4a
(not shown). We also evaluated the expression of Cdk4 and Cdk2, especially since Cdk2 inhibition was recently found to be important for Myc-induced senescence [11
]. We observed a modest decrease in Cdk4 expression from P10 through P49 [Figure B], but interestingly we found that Cdk2 expression was markedly reduced from P10 to P24 [Figure B], coincident with the timing of cell proliferation arrest and loss of Rb phosphorylation at Cdk2-specific sites. We conclude that Cdk2 repression correlates most closely with the initial proliferation arrest; and that diminished Cdk4-dependent Rb phosphorylation occurs at a later time-point and correlates with formation of SAHF.
p18Ink4closs delays p53-dependent cell cycle exit:
Because of the observed increase in expression of p18Ink4c
, we used a genetic approach to evaluate its role in Cyclin D1-induced senescence. In contrast to what occurs without p53 – where cell proliferation only slightly decreases from P10 to P35 and then increases as invasive tumor progresses [Figure C, top panel; and [13
]], proliferation decreased from P10 through P49 without p18Ink4c
[Figure C, bottom panel], but exceeded that in the Irbp-Cyclin D1
cells [Figure C, bottom panel, compare with Figure B].
Notably, p53 activation measured by phosphorylation at Ser15/20, and expression of the p53-target p21Cip1, persisted until P24 in Irbp-Cyclin D1, p18Ink4c -/- cells [Figure D], correlating with the prolonged cellular proliferation. Further, Cdk2 expression persisted until P24 in Irbp-Cyclin D1, p18Ink4c -/- cells [Figure ]. These findings indicate that p18Ink4c loss delayed but did not prevent p53-dependent events leading to cell cycle exit.
Interestingly, loss of Cdk4-dependent Rb phosphorylation still occurred in the absence of p18Ink4c
, again correlating with the appearance of SAHF [Figure D]. In fact, the majority of Irbp-Cyclin D1, p18Ink4c -/-
cells displayed SAHF by P49 Additional file 2
: Figure S2A], whereas SAHF never formed in Irbp-Cyclin D1, p53 -/-
]. In addition to SAHF, the senescence markers Dec1 and DcR2 were also expressed in Irbp-Cyclin D1, p18Ink4c -/-
cells at P49 Additional file 1
: Figure S2B]. Findings were similar in vitro
using pineal cells explanted from P10 animals and cultivated for 10-20
days: Explanted Irbp-Cyclin D1
cells showed evidence of senescence, including loss of proliferation (measured by BrdU incorporation), and positive staining for SABG, by 10
days in culture Additional file 1
: Figure S1B, 1C], while the Irbp-Cyclin D1, p53 -/-
cells continued to proliferate and did not senesce Additional file 1
: Figure S1B, bottom]. In contrast, the Irbp-Cyclin D1, p18Ink4c -/-
cells did show evidence of senescence, but it was delayed until close to 20
days in culture Additional file 2
: Figure S2C]. We conclude that p18Ink4c
slowed proliferation but was not essential for most Cyclin D1 expressing cells to cease proliferating and become senescent.
p53 and p18Ink4cact independently in suppressing Cyclin D1-driven tumors:
The persistence of a small number of proliferating cells by P49, in Irbp-Cyclin D1, p18Ink4c -/-
mice, was important because it led to pineoblastoma by 7-10
months of age in all mice examined (n
15). Examining mice at 3-5
months of age, we observed an obvious border between the malignant and pre-malignant parts of the pineal gland [Figure A, left]; this border disappeared as the tumor infiltrated the whole pineal gland [Figure A, right], and SAHF were lost in the emerging Irbp-Cyclin D1, p18Ink4c -/-
tumor [Figure B]. Dual immunostaining for BrdU and SAHF clearly demonstrated that proliferating pinealocytes were distinct from those that displayed SAHF [Figure C].
Figure 5 p18Ink4c loss promotes emergence of tumor from a premalignant senescent lesion. A) Representative pineal sections of Irbp-Cyclin D1, p18Ink4c-/- mice at the indicated ages, with Ki67 immunostaining. Arrowheads mark examples of Ki67-positive cells. (more ...)
We considered whether the prolonged proliferation in the absence of p18Ink4c might have derailed a p53-dependent arrest in the malignant tumors. Western blotting showed that Irbp-Cyclin D1, p18Ink4c -/- tumors still expressed the p53 protein [Figure D], and sequencing of p53 exons 5-8 did not reveal mutations in genomic DNA from nine different Irbp-Cyclin D1, p18Ink4c -/- pineal tumors (data not shown). Further, using primary cultures of pineal tumor cells, we found that both gamma irradiation and treatment with etoposide resulted in increased p53 phosphorylation and in p53-dependent increases in p21Cip1 and 14-3-3 in Irbp-Cyclin D1, p18Ink4c -/- but not Irbp-Cyclin D1, p53 -/- tumor cells [Figure E]. These findings confirmed that p53 remained intact in Irbp-Cyclin D1, p18Ink4c -/- tumor cells. In contrast, there was decreased p18Ink4c expression in Irbp-Cyclin D1, p53-/- tumors, suggesting that p18Ink4c may act as a tumor suppressor, even in a p53-null setting [Figure D, F]. However, preliminary results show no enhanced tumor susceptibility in Irbp-Cyclin D1, p53-/-, p18Ink4c -/- (double knock-out) animals (data not shown).
Cdk2 is induced in bothp18Ink4c -/-andp53 -/-tumors, and may be a suitable therapeutic target:
Several changes in Cdk2 expression suggested that it may represent a critical effector of Cyclin D1-driven tumorigenesis. In the Irbp-Cyclin D1,
and Irbp-Cyclin D1, p18Ink4c -/-
animals, Cdk2 was repressed as cells ceased to proliferate, and repression was markedly blunted in the Irbp-Cyclin D1, p53 -/-
pineal gland [Figure F, compare with Figure D], in which neither cell cycle arrest nor senescence was observed [see Figure A]. Repression of Cdk2 seemed specific because there was no repression of another closely related cell cycle protein, Cdk1 [Additional file 2
: Figure S2D]. Lastly, Cdk2 increased in tumors progressing from the largely senescent, Irbp-Cyclin D1, p18Ink4c -/-
pineal gland [Figure A], and Cdk2 expression correlated with Ki67-positivity in emerging tumors [Figure B]. Areas in the Irbp-Cyclin D1, p18Ink4c -/-
tumors that remained Ki67-negative displayed little Cdk2 [Figure B and C].
Figure 6 Cdk2 is a therapeutic target in tumors that have bypassed senescence. A) Western blotting for the indicated proteins in Irbp-Cyclin D1, p18Ink4c -/- (D1p18null) pineal lysates at the indicated ages, Irbp-Cyclin D1, p18Ink4c -/- (D1p18null) tumors that (more ...)
To address the role of Cdk2 repression as a possible therapeutic target, we treated explanted Irbp-Cyclin D1, p18Ink4c -/-
and Irbp-Cyclin D1, p53 -/-
pineal tumor cells with the Cdk2-inhibitor CVT313, at a concentration of 5
μM, known to specifically inhibit Cdk2 [11
]. CVT313 treatment decreased cell number (estimated by total cellular area) in Irbp-Cyclin D1, p18Ink4c -/-
and Irbp-Cyclin D1, p53 -/-
tumor cells in 8-well chamber slides [Figure D, upper panel]. Furthermore, CVT313-treated cells showed an increase in positive staining for SABG activity [Figure D, lower panel, and Figure E]. Importantly, treatment of pre-tumorigenic Irbp-Cyclin D1
pineal cells with CVT313 also decreased the apparent cell number [Figure E], while treatment of wild-type pineal cells did not seem to have a noticeable effect, either on cellularity or SABG positivity [Figure E].
We assessed whether Cdk2 inhibition by CVT313 was primarily affecting cellular proliferation by BrdU incorporation assay. Indeed, we found that CVT313 treatment decreased proliferation in oncogene-expressing and pre-tumorigenic cells, but not in wild-type pineal cells [Figure A, quantitation in Figure B]. There was no evidence of any increase in apoptotic cells after CVT313 treatment in either cell type, as measured by TUNEL staining (negative data not shown).
Figure 7 Cdk2 inhibition in tumor cells leads to features of senescence. A) BrdU-incorporation and immunofluorescence staining (red), and corresponding DAPI staining (blue), of cultured wild-type (WT), Irbp-Cyclin D1 (D1), and Irbp-Cyclin D1,p53-/- (D1p53-/-) (more ...)
To investigate whether the effects on senescence induction were specific to Cdk2 inhibition, we treated explanted Irbp-Cyclin D1, p53-/-
and Irbp-Cyclin D1, p18Ink4c-/-
cells with a specific Cdk4-inhibitor, NSC 625987 [32
]. Inhibition of Cdk4 decreased proliferation (measured by BrdU incorporation), though to a lesser extent than seen with Cdk2 inhibition [Figure C]. However, unlike Cdk2 inhibition, it did not result in any detectable increase in SABG staining [Figure D]. This demonstrates that Cdk2 inhibition was specifically relevant to induction of senescence in Cyclin D1-expressing pineal cells.