Temporally distinct roles for tumor suppressor pathways in cell cycle arrest and cellular senescence in Cyclin D1-driven tumor

Background Cellular senescence represents a tumor suppressive response to a variety of aberrant and oncogenic insults. We have previously described a transgenic mouse model of Cyclin D1-driven senescence in pineal cells that opposes tumor progression. We now attempted to define the molecular mechanisms leading to p53 activation in this model, and to identify effectors of Cyclin D1-induced senescence. Results Senescence evolved over a period of weeks, with initial hyperproliferation followed by cell cycle arrest due to ROS production leading to activation of a DNA damage response and the p53 pathway. Interestingly, cell cycle exit was associated with repression of the Cyclin-dependent kinase Cdk2. This was followed days later by formation of heterochromatin foci correlating with RB protein hypophosphorylation. In the absence of the Cdk4-inhibitor p18Ink4c, cell cycle exit was delayed but most cells eventually showed a senescent phenotype. However, tumors later arose from this premalignant, largely senescent lesion. We found that the p53 pathway was intact in tumors arising in a p18Ink4c-/- background, indicating that the two genes represent distinct tumor suppressor pathways. Upon tumor progression, both p18Ink4c-/- and p53-/- tumors showed increased Cdk2 expression. Inhibition of Cdk2 in cultured pre-tumorigenic and tumor cells of both backgrounds resulted in decreased proliferation and evidence of senescence. Conclusion Our findings indicate that the p53 and the RB pathways play temporally distinct roles in senescence induction in Cyclin D1-expressing cells, and that Cdk2 inhibition plays a role in tumor suppression, and may be a useful therapeutic target.


Introduction
Cellular senescence is a well-established tumor suppressor mechanism, activated in response to oncogenic signals, DNA damage, and telomere attrition among other protumorigenic insults (reviewed in [1,2]). Mechanistic insight into oncogene-induced senescence has emerged over the past few years. While it is clear that both the p53 and the Rb tumor suppressors are involved [3][4][5][6], their relative importance seems to vary depending on the activating insult and cellular context [3,[7][8][9][10]. Understanding the relative contributions of p53 and Rb to the induction and maintenance of senescence may have important implications, especially with development of targeted therapeutic agents.
Cyclin-dependent kinase 2 (Cdk2) was recently implicated in the senescence process, as Cdk2 loss was found to enhance senescence in Myc-induced tumors [11]. In addition, it was shown that Cdk2-dependent phosphorylation of Myc was necessary to bypass Ras-induced senescence [12]. This suggests that Cdk2 may act in senescence independently of its role in RB phosphorylation and cell cycle exit.
Here, we used a transgenic mouse model of premalignant Cyclin D1-driven pineal gland hyperplasia, to define the molecular mechanisms leading to p53 activation in response to Cyclin D1, and to identify effectors of Cyclin D1-induced senescence. The results shed light on the pattern of evolution of the senescence response in a premalignant lesion in vivo, and the differences in the contribution of the two major tumor suppressor pathways, p53 and RB. In addition, our findings suggest that Cdk2 inhibition may be a useful therapeutic approach, irrespective of the underlying genetic insult that led to senescence evasion.

Results
Cyclin D1-induced senescence in vivo occurs 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 p18 Ink4c 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 1A, left panel, Quantitation in Figure 1B]. 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 1A, middle and right panels]. Because SAHF have been observed in only a few mouse models of senescence [14][15][16], 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 [18,19], 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 D1expressing 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 1C]. 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. 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 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 ages (middle), with corresponding DAPI staining (right). B) Quantitation of Ki67-positive cells as percent of total pineal cells at indicated ages. C) Immunostaining for Dec1, DcR2, and p15Ink4b, in Irbp-Cyclin D1 pineal glands at the indicated ages. 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 2A]. The p53 target, p21 Cip1 , was also induced at that time-point [ Figure 2A]. 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 2A].
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 2B, 2C]. 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 2D]. 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][24][25]. 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 3A]. 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 3B, left panel], and to increased expression of two p53 pathway effector proteins, 14-3-3 and p21 [ Figure 3B, 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 3B], as well as evasion of senescence [ Figure 3C]. 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 3D], and there was also increased expression of the mitochondrial superoxide dismutase protein MnSOD [ Figure 3E], which is induced by ROS stress in vivo [26]. 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.
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 4A]. We found that Rb was phosphorylated at Cdk2dependent sites (Ser612) at P10, when cells were proliferating, but decreased after cells exited the cell cycle by P24 [ Figure 4A]. 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 4A]. Rb phosphorylation at Cdk4-dependent sites was reversed at P49 as SAHF formed [ Figure 4A].
To understand the mechanism of Rb activation, we investigated the expression of the Cdk-inhibitors p16Ink4a, 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; OL = overlay. B) Immunofluorescence staining for the respective proteins in cultured Cyclin D1-expressing pineal cells (top panel), which have been treated with N-Acetyl Cysteine (NAC) or vehicle. The middle panel shows the respective DAPI-stained nuclei, and the lower panel shows the overlay (OL). C) Staining for senescence-associated beta galactosidase (SABG) in explanted Cyclin D1-expressing cells that have been treated with either NAC or vehicle, as indicated. D) Representative immunofluorescence staining for 4-hydroxynonenal (4HNE) in 10-day old Irbp-Cyclin D1 (D1) and wild-type (WT) pineal sections, as indicated. E) Representative immunostaining for the mitochondrial superoxide dismutase MnSOD in wild-type (WT) and Irbp-Cyclin D1 (D1) pineal sections, as indicated.
p15Ink4b, p18Ink4c, and p27Kip1 [27][28][29][30]. There was increased expression of p18 Ink4c at all time points [ Figure 3B], an increase in p15 Ink4b expression at P49 [see Figure 1D], but no changes in expression of p16 Ink4a and p27 Kip1 (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 3B], but interestingly we found that Cdk2 expression was markedly reduced from P10 to P24 [ Figure 3B], 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.
p18 Ink4c loss delays p53-dependent cell cycle exit: Because of the observed increase in expression of p18 Ink4c , we used a genetic approach to evaluate its role in Cyclin D1-induced senescence. In contrast to what occurs without p53where cell proliferation only slightly decreases from P10 to P35 and then increases as invasive tumor progresses [ Figure 4C, top panel; and [13]], proliferation decreased from P10 through P49 without p18Ink4c [ Figure 4C, bottom panel], but exceeded that in the Irbp-Cyclin D1 cells [ Figure 4C, bottom panel, compare with Figure 1B].
Interestingly, loss of Cdk4-dependent Rb phosphorylation still occurred in the absence of p18Ink4c, again correlating with the appearance of SAHF [ Figure 4D]. 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 -/cells [13]. 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 p18 Ink4c slowed proliferation but was not essential for most Cyclin D1 expressing cells to cease proliferating and become senescent.
p53 and p18 Ink4c act 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 5A, left]; this border disappeared as the tumor infiltrated the whole pineal gland [ Figure 5A, right], and SAHF were lost in the emerging Irbp-Cyclin D1, p18Ink4c -/tumor [ Figure 5B]. Dual immunostaining for BrdU and SAHF clearly demonstrated that proliferating pinealocytes were distinct from those that displayed SAHF [ Figure 5C].
Cdk2 is induced in both p18Ink4c -/and p53 -/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 5F, compare with Figure 4D], in which neither cell cycle arrest nor senescence was observed [see Figure 4A]. 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 6A], and Cdk2 expression correlated with Ki67-positivity in emerging tumors [ Figure 6B]. Areas in the Irbp-Cyclin D1, p18Ink4c -/tumors that remained Ki67-negative displayed little Cdk2 [ Figure 6B and 6C].
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 pretumorigenic cells, but not in wild-type pineal cells [ Figure 7A, quantitation in Figure 7B]. 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).
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 [32,33]. Inhibition of Cdk4 decreased proliferation (measured by BrdU incorporation), though to a lesser extent than seen with Cdk2 inhibition [ Figure 7C]. However, unlike Cdk2 inhibition, it did not result in any detectable increase in SABG staining [ Figure 7D]. This demonstrates that Cdk2 inhibition was specifically relevant to induction of senescence in Cyclin D1-expressing pineal cells.

Discussion
Emerging evidence supports the concept that cellular senescence represents a likely mechanism through which oncogenic transformation is suppressed. Senescence has been observed in pre-malignant lesions in mouse and in man, but not in fully transformed counterparts of these lesions (reviewed in [21,34]). Because of the time lag in progression of premalignant lesions and the incomplete penetrance, it has been assumed that accumulation of as yet poorly-defined genetic or epigenetic changes likely contribute to the emergence of a tumor from a premalignant, apparently senescent lesion.
Our work provides new insight into cellular and molecular events that occur as oncogene-expressing cells arrest, become senescent, and finally emerge or escape from the senescent state as a malignant tumor. This is the first in vivo description of temporal morphologic and molecular events accompanying the evolution of an oncogene-driven senescent state, showing a previously unrecognized temporal sequence where cell cycle exit preceded formation of heterochromatin foci by several weeks.
Two tumor suppressor genes, p53 and p18Ink4c, played distinct roles during this process. P53 activation occurred concomitantly with an active DNA damage response, and was essential to drive cell cycle exit, temporally associated with Cdk2 repression and loss of Cdk2-dependent phosphorylation of the retinoblastoma protein Rb. Days later, reversal of Cdk4-dependent phosphorylation of the Rb protein correlated with the emergence of morphological and biochemical changes of oncogene-induced senescence. At that point, though, there was no evidence of p53 pathway activation. This is the first direct in vivo evidence for distinct temporal roles for these two tumor suppressors in the senescence process.
The early and transient activation of the p53 pathway suggested that p53 was integral for the initial cell cycle exit but not directly involved in formation of SAHF. Other models have also shown conflicting and contextdependent evidence for the role of p53 in the formation of SAHF [35][36][37]. In contrast, Rb activation was delayed and stable. Rb seemed to be important in both cell cycle exit as well as formation of SAHF: compromise of the Rb pathway through loss of p18Ink4c (the major known function of which is inhibition of Cdk4/6 to indirectly activate Rb [38][39][40]) led to a delay in initial cell cycle exit, and eventually to complete penetrance of tumor progression within the senescent-like lesion. Taken together, these findings implicate Rb, rather than p53, as the key protein needed to foster the emergence and maintenance of SAHF, thought to be responsible for repression of cell-cycle genes [3,35]. Involvement of Rb in the formation of SAHF has been shown in other settings: human RB was shown to directly co-localize with SAHF [35,36], and inactivation of the p16INK4a/RB pathway impairs formation of RasG12V-induced SAHF in human fibroblasts [35]. In our model, p53-driven cell cycle exit correlated with hypo-phosphorylation of Rb at Cdk2dependent sites, while formation of SAHF correlated with hypo-phosphorylation at Cdk4-dependent sites. This suggests that hypo-phosphorylation of these specific residues may be involved in SAHF formation. It will be interesting to evaluate whether mutated forms of Rb that cannot be phosphorylated at these particular Cdk4 sites can more robustly foster the appearance of SAHF.
Our results suggest that p53 and p18Ink4c represent separate tumor suppressor pathways in Cyclin D1-driven pineoblastoma. While tumors progressed within 3-4 months in p53 -/animals, they appeared much later, after 7-10 months in p18Ink4c -/mice. Also, the p53 pathway was intact in p18Ink4c -/tumors, further proving that the two pathways of tumor suppression are distinct. Tumor suppression required functional p53 and p18Ink4c, as neither was sufficient to prevent tumor progression alone. While the cell cycle exit after P10 was clearly p53-dependent, absence of p18Ink4c delayed the cell cycle exit but did not prevent it in the majority of cells, which went on to express other markers of senescence. However, few cells continued to proliferate, resulting in tumorigenesis. It thus appears that, while p53 loss resulted in abrogation of cell cycle exit altogether, loss of p18 Ink4c decreased the threshold for bypass of the p53dependent cell cycle exit in a subset of cells.
In our model, p53-dependent cell cycle arrest was associated with marked Cdk2 repression, while Cdk2 levels were maintained in Irbp-Cyclin D1, p53 -/cells which never exited the cell cycle [see Figure 5F]. While a role for Cdk2 repression in facilitating senescence has been shown in an earlier report [11], ours is the first description of Cdk2 repression occurring in a temporal association with p53-dependent cell cycle exit. This indicates that Cdk2 repression might be a novel p53-dependent mechanism to foster cell cycle exit, especially since no similar changes were seen in the related cell cycle regulator Cdk1. However, additional work will be needed to investigate whether Cdk2 repression is a direct p53dependent effect, and whether it is sufficient to induce cell cycle exit and induction of senescence. Moreover, since Cdk2 and other Cdks are also regulated post-transcriptionally by phosphorylation and by their binding to CDK-inhibitors, future work should focus on elucidating these molecular aspects for complete mechanistic understanding of the role of Cdk2 and other Cdks in inducing senescence. Future studies utilizing the established in vitro model, as well as genetically engineered mouse models, should be able to specifically dissect the role of Cdk2 in tumor progression, and upstream and downstream mechanisms leading to its repression and to cell cycle exit.
Finally, although tumors arising in a p53 -/setting were molecularly different from those arising in a p18Ink4c -/setting, Cdk2 levels were high in both (compared to senescent cells) and both cell types responded to Cdk2 inhibition. While Cdk4 inhibition also decreased cell proliferation, only Cdk2 inhibition resulted in features of senescence in treated cells. These conclusions are based on the published specificity of the inhibitors used at the corresponding concentrations. Ideally, we would have preferred to document inhibitor-induced pRb specific phosphorylations and specific kinase activity in each experiment, however the low number of primary cells used in these experiments was prohibitive.
Keeping the above limitations in mind, these findings provide a rationale for exploring the use of pharmacological Cdk inhibition, specifically Cdk2, to induce senescence in tumor cells, irrespective of whether the p53 pathway is compromised. Such an approach to therapy may be especially useful in tumors where the primary insult lies with deregulated Cyclin D1 expression, as in the reported model.

Mouse Studies
Irbp-Cyclin D1 transgenic mice [41] were bred with p53 -/-(Jackson Laboratory, Maine), or p18Ink4c -/mice [42] and maintained in a mixed C57BL/6 × 129/Sv genetic background. PCR for targeted alleles was used to verify mouse genotypes as described [41,42]. Animals were euthanized at defined time points or when obviously ill in accordance with the American University of Beirut (AUB) Institutional Animal Care and Use Committee guidelines; all studies were approved by this committee.
For evaluation of p53 pathway, pineal tumors were excised, dispersed, and plated onto permanox chamber slides, and grown in culture in 10% FBS in DMEM for 72 hours, then treated with 6 Gy irradiation at 1 Gy/min, or 10 μM Etoposide. Cells were collected 24 hours after irradiation, or 6 hours after Etoposide treatment, and protein extraction and western blotting was performed as above.

DCFDA Assay
To measure intracellular ROS in vitro, cells were treated with a peroxide sensitive reagent CM-H2DCF-DA (CM-H2DCF-DA; Molecular Probes, Eugene, OR) at 10 μM for 20 min at 37°C and observed under a fluorescence microscope.

N-Acetyl Cysteine treatment
Explanted pineal cells were treated with N-Acetyl Cysteine (Sigma Aldrich) at a concentration of 5 mM. Media was renewed daily. Cells were treated for 10 days, and stained for SABG as described [20]. For DDR pathway analysis, cells were fixed and stained after 4 days.

CVT313 and NSC625987 treatment
Explanted cells were treated with CVT313 at 5 μM (Santa Cruz Biotechnology), NSC625987 at 1 μM (Tocris Bioscience, Ellisville, MO, USA), or DMSO vehicle; media was renewed every 3 days. Cells were fixed and stained for SABG after 7 days, and counterstained with eosin. For quantification of proportion of cells positive for SABG, 10 random fields were selected, and digital photomicrographs were analyzed using Adobe Photoshop CS4 software, by color selection and area analysis. For quantification of cellular accumulation, all the area of the well was photographed over 12 fields. Digital photomicrographs were analyzed using Adobe Photoshop CS4 software, by area selection tool. For BrdU incorporation assay, cells were treated with BrdU at a concentration of 33 μM for 2 hours, fixed with 50%methanol/50%acetone solution for 2 minutes, then processed as detailed above.
Additional file 2: Figure S2: A) Representative immunohistochemical staining for Ki67 (left) and immunoflourescence for H3K9me3 (right) and corresponding DAPI (middle) staining of Irbp-Cyclin D1, p18Ink4c-/pineal sections at the indicated ages. B) Immunostaining for Dec1 and DcR2, in Irbp-Cyclin D1, p18Ink4c-/pineal glands at the indicated ages. C) SABG staining of cultured Irbp-Cyclin D1, p18Ink4c -/pineal cells after 14 and 20 days in culture, as indicated. D) Western blotting for the indicated proteins in pineal glands of the indicated genotypes, at the indicated ages.