The cyclin D1 expression pattern is not altered by signaling inhibitors
If the PI3K/AKT/GSK3 pathway stabilizes cyclin D1 levels specifically during G1 and G2 phases as suggested above, inhibitors of this pathway would produce a reduction in cyclin D1 expression during these cell cycle phases to the low levels seen during S phase. Thus, inhibition of these signaling pathways would be expected to result in low, uniform expression of cyclin D1 throughout the cell cycle. PI3K was inhibited by LY294002, while the kinase mTOR was inhibited by rapamycin in actively cycling human diploid fibroblast (MRC5) cultures. After 2 hrs treatment, including a terminal pulse with BrdU, the culture was fixed and stained with fluorescent antibodies against both cyclin D1 and BrdU, while DNA was stained with DAPI. Individual images of each fluorochrome were collected with a sensitive CCD camera, and subjected to image analysis to accurately quantitate the level of each fluorochrome in each cell (see ). The results were displayed by plotting the cyclin D1 level of each cell against its DNA level, with BrdU positive cells noted (Fig. 1A). Treatment with LY294002 dramatically reduced cyclin D1 levels, yet the overall cell cycle pattern of expression was maintained, with higher levels observed during G1 and G2 phases as in untreated cultures. Rapamycin had a similar but weaker effect on the cyclin D1 level (Fig. 1A). Neither inhibitor produced the uniform cyclin D1 expression through the cell cycle we had predicted.
We next inhibited AKT activity by microinjecting a plasmid expressing a dominant inhibitory mutant. This plasmid, pCDNA3-Akt-DN (K179 M, a kind gift of Dr. Nissim Hay) generates a kinase-dead AKT protein that can be detected within the cells by antibody staining. Staining confirmed the high level expression of the mutant protein at 4 and 24 hrs following plasmid injection. This protein, however, did not induce any obvious alteration in cyclin D1 expression after 8 hrs (Fig. 2A). The average expression levels of cyclin D1 in injected cells at 8 hrs (not shown) and 24 hrs (Fig. 2B) confirmed that the there was no change in the cyclin D1 profiles in cells expressing an excess of inhibitory AKT protein.
It is possible that proliferative signaling molecules other than those inhibited above might be responsible for S phase cyclin D1 suppression. To test this possibility proliferative signaling in general was disrupted by serum withdrawal for 4 hrs. In this case, the level of Thr-286 phosphorylation of cyclin D1 was directly determined with a phosphorylation site-specific antibody (the generous gift of Michelle D. Garrett and David R. Mason), and compared to total cyclin D1. To avoid rapid degradation of phosphorylated cyclin D1 , the cells were treated with the proteasomal inhibitor MG132 for 2 hrs prior to analysis. Serum withdrawal for 4 hrs slightly reduced both total and Thr-286-phosphorylated cyclin D1. Significantly, however, the ratio of phospho/total cyclin D1 was not changed by serum deprivation in any cell cycle phase (Fig. 1B). This indicates that the overall rate of cyclin D1 Thr-286 phosphorylation is not altered in any cell cycle phase by disruption of proliferative signaling following serum removal. Taken together, the above results fail to support the notion that alterations in proliferative signaling through the cell cycle are responsible for the cell cycle specific expression pattern of cyclin D1.
The activity of critical signaling molecules remains constant through the cell cycle
To extend the above observations, we designed experiments to directly analyze the activity of individual proliferative signaling molecules through the cell cycle, to determine if there are alterations in proliferative signaling activity sufficient to account for the proposed variations in GSK3 activity. We first analyzed the activity of PI3K in individual cells of an asynchronous NIH3T3 culture using a plasmid expressing the PH domain of AKT linked to GFP . T. Balla and associates have shown that the resulting protein strongly binds phosphatidylinositol-3-phosphate, the lipid product of PI3K activity, resulting in its association with the plasma membrane upon activation of PI3K . To directly visualize the redistribution of fluorescence at the time of PI3K stimulation, the AKT-PH-GFP plasmid was microinjected into quiescent NIH3T3 cells. The intracellular distribution of fluorescence was determined by confocal analysis 15 hrs later without serum addition, or 20 min following addition of 10% serum to the culture. Without serum addition, the fluorescence was uniformly distributed throughout the cytoplasm and nucleus (Fig. 3B). Upon serum stimulation, however, the fluorescence became associated with the plasma membrane (Fig. 3A), including particularly high concentrations associated with tiny projections on the plasma membrane. Thus, the loss of fluorescence from the nucleus and cytoplasm, coupled with its redistribution to plasma membrane structures was diagnostic for PI3K activity [see Additional file, movies 1, 2, 3].
This assay was then applied to actively cycling NIH3T3 cells, where we found that the fluorescence of all cells remained evenly distributed throughout the cytoplasm and nucleus. Specific association of fluorescence with the plasma membrane was not observed in any of these cells (Fig. 3C). To demonstrate that cycling cells did retain the ability to produce a high level of PI3K activity, the PH-AKT-GFP plasmid was injected into proliferating NIH3T3 cells, after which serum was removed from the culture for 5–12 hrs, and then added back. Within 20 minutes of the addition of serum back to these cultures the fluorescence became associated with the plasma membrane, characteristic of PI3K activation described above for quiescent cells (Fig. 3D–G).
These results present an interesting model of proliferative signaling in cycling cells. The high levels of PI3K activity observed following serum addition to quiescent cultures most likely represents the response to a change in growth condition rather than a normal consequence of cell cycle progression. While PI3K activity is present in and required for the proliferation of actively cycling cells, the levels required are apparently much lower than observed upon serum stimulation. Importantly, there was no evidence of alterations in PI3K activity during S phase or any other cell cycle period in asynchronous cultures, reducing the likelihood that such an alteration might be responsible for the elevation in GSK3 activity, and the corresponding decline in cyclin D1 levels during S phase.
AKT is activated uniformly through S and G2 phases
The above study of individual cells in an actively cycling culture was next confirmed by biochemical analyses in synchronized cell populations. A single treatment of NIH3T3 cells with 2 mM thymidine has been shown to result in their synchronization in S phase. Following removal of thymidine, treated cells remain in S phase for 4 hrs, and enter G2 phase approximately 5 hrs later [24,27]. Critically, we have confirmed that the cell cycle expression profile of cyclin D1 in such synchronized cultures behaves as in asynchronous cultures, with low levels in S phase cells, followed by increasing levels upon entry into G2 phase (see Fig. 4D). Cultures prior to or at various times following release from thymidine blockage were tested by western analysis for the activating phosphorylation of AKT (Fig. 4A,C), while cyclin D1 expression levels are presented for comparison (Fig. 4B,D). No evidence of alterations in the activation level of AKT was apparent at any time following release from thymidine blockage, as the cells passed from S to G2 phase. As a control, the levels of the activating phosphorylation on AKT were seen to increase dramatically within 15 and 90 min following treatment of quiescent NIH3T3 cells with serum. Moreover, in the cultures which showed no difference in AKT activation levels, the levels of cyclin D1 protein were seen to increase during passage from S to G2 phase (as previously demonstrated ). We conclude that alteration in the activity of AKT is not responsible for suppression of cyclin D1 levels during S phase.
GSK3 activity is constant in synchronized S and G2 phase cells
From the above results it is clear that the proliferative signaling pathway upstream of GSK3 does not vary through the cell cycle. However, it is possible that a pathway other than PI3K/AKT is responsible for regulation of GSK3 activity . Therefore, antibody staining as well as biochemical analyses were performed to directly study variation of GSK3 activity through the cell cycle. NIH3T3 cells were synchronized in S phase by thymidine treatment and released as described above. Most of the cells progressed from S to G2 phase 5 hours after the release. The cells were collected at the indicated times, and the GSK3 phosphorylation was analyzed by western analysis with an antibody specific to the inhibitory phosphorylation of GSK3β on position 9 (Fig. 5A, upper panel). Total GSK3 was also determined and the phospho-GSK3/total GSK3 ratio presented (Fig. 5A, lower panel). We next directly determined the activity of GSK3 by synchronizing cells as described above, and immunoprecipitating GSK3 at the indicated times following release. The immunoprecipitated protein was then assayed with a synthetic peptide substrate whose phosphorylation was analyzed by PAGE analysis. Due to the low level of activity of GSK3 in NIH3T3 cells, this assay was repeated 10 times to obtain consistent results (Fig. 5B). There was some increase in GSK3 activity immediately upon release from thymidine blockage, but no evidence of any alteration upon passage from S to G2 phase. For comparison, there was a reduction in activity upon addition of serum to quiescent cultures, while the levels of GSK3 activity declined only gradually following serum removal from actively cycling cultures [see Additional file 4]. As a control, the inhibition of GSK3 activity by 50 mM LiCl is demonstrated (Fig. 5B). It is clear from these results than neither the upstream signaling pathway leading to GSK3 control, nor the activity of GSK3 itself varies during passage from S to G2 phase in synchronized cultures, or in actively cycling cells. This places doubt on the role of proliferative signaling or GSK3 activity in suppressing cyclin D1 levels during S phase.
Inhibition of GSK3 activity has no effect upon cyclin D1 expression
Experiments were next designed to determine if GSK3 activity plays a role in cyclin D1 expression or phosphorylation in any cell cycle phase. GSK3 activity was inhibited by 25 mM LiCl, and the proteasomal inhibitor MG132 was added to allow the accumulation of phosphorylated cyclin D1. To determine if the inhibition of GSK3 would reduce the phosphorylation of cyclin D1, the effect of this treatment upon phospho-Thr-286 accumulation was determined by image analysis following staining with the phospho-Thr-286-specific antibody . LiCl influenced neither the total amount of cyclin D1 nor its level of phosphorylation following staining of human diploid fibroblast MRC5 cells. This is apparent from the cell cycle expression profiles (Fig. 6A), and from average levels determined from these profiles (Fig. 6B). To confirm the results with LiCl, we performed similar experiments on NIH3T3 cells with two other GSK3β inhibitors, sodium valproate  and GSK3 inhibitor II from Calbiochem. Neither of these treatments altered the ratio of total cyclin D1 to phosphorylated cyclin D1 (Fig. 6C).
This critical result was next confirmed by western analysis. NIH3T3 cultures were treated with medium containing 25 mM or 50 mM LiCl to inhibit GSK3 activity, or with medium containing equivalent amounts of NaCl as a control. MG132 was added to block degradation of phosphorylated proteins as above. After treatment for 4 hrs the levels of total and phosphorylated β-catenin were determined by western analysis with appropriate antibodies, while actin was analyzed as a loading control. β-catenin is a well characterized substrate of GSK3. Its phosphorylation would, therefore, serve as an indication of the activity of GSK3 within the treated cells. Since the lysate was not separated into nuclear and cytoplasmic fractions [31,32], little alteration in the total β-catenin levels were seen (Fig. 7A). On the other hand, high levels of phosphorylated β-catenin were observed in cells cultured in normal medium (Fig. 7 column 2). Phosphorylated β-catenin levels were dramatically reduced by 25 mM LiCl (Fig. 7 column 3) and almost completely eliminated by treatment with 50 mM LiCl (Fig. 7 column 5). Levels of the phosphorylated proteins were observed only when their degradation was blocked by treatment with the proteasome inhibitor, MG132 (Fig. 7 columns 3, 4). The results were confirmed by quantitating the ratio of phosphorylated to total β-catenin (Fig. 7C). On the other hand, in these same lysates neither concentration of LiCl had any effect upon the phosphorylation of cyclin D1 on Thr-286 (Fig. 7 column 3, 5). This conclusion was confirmed following quantitation of the western bands to determine the ratio of phosphorylated to total cyclin D1 levels (Fig. 7B). The fact that LiCl was able to block the phosphorylation of a well known GSK3 substrate, without influencing the phosphorylation of cyclin D1within the same cells, confirms the above conclusion that GSK3 is not a major kinase for cyclin D1 in these cells.
To strengthen this conclusion, GSK3 protein was ablated with a commercial siRNA mixture reported to suppress levels of both GSK3 α and GSK3β. To confirm its effectiveness, this siRNA was injected into NIH3T3 cells 12 hrs prior to a second injection of a plasmid expressing the GSK3 β gene (the gift of J. R. Woodgett). Following these injections the cells were fixed and stained with a fluorescent antibody stain against GSK3. While the plasmid expressed high levels of GSK3 when injected alone (Fig. 8A), prior injection with siRNA totally blocked expression of the high exogenous levels of GSK3 (Fig. 8B). Even though the endogenous levels of GSK3 are low, these endogenous levels were also suppressed by the injected siRNA (Fig. 8B). We conclude that the siRNA is able to efficiently ablate GSK3 expression. This siRNA was injected into NIH3T3 cells, which were left over night prior to treatment with MG132 and staining with the phospho-Thr-286 cyclin D1 antibody. The siRNA had no influence upon phosphorylation of cyclin D1 (Fig. 8C), or upon total cyclin D1 levels (not shown). We conclude that inhibition of GSK3 does not alter normal cell cycle regulated phosphorylation of cyclin D1.
Over-expression of activated GSK3 β does not suppress cyclin D1 levels
The influence of GSK3 upon cyclin D1 degradation in NIH3T3 and MRC5 cells was next directly analyzed by microinjection of a plasmid expressing a constitutively active GSK3 β protein (S9A: the gift of J. R. Woodgett) [33,34]. Injected cells were stained for GSK3 and cyclin D1 at 8 and 24 hrs following injection. GSK3 staining was strong in injected cells at both time points, and displayed a cytoplasmic localization in most of the cells; whereas the staining was close to background level in the uninjected cells, indicating that the exogenous GSK3 level is far above the endogenous level (Figure 9A). However, the exogenous, constitutively active GSK3β did not apparently alter cyclin D1 levels at 8 hrs (Fig. 9A) or 24 hrs (not shown) following injection.
The effect of injected GSK3β upon cyclin D1 expression was next analyzed quantitatively. At 8 hrs following injection of the GSK3β expression plasmid cells were fixed and stained with fluorescent antibodies against cyclin D1 and GSK3. The intensity of each stain in each cell was determined by quantitative image analysis, and the GSK3 levels were plotted vs. cyclin D1 levels (Fig. 9B). The basal cyclin D1 level was apparent from the analysis of uninjected cells. This level of cyclin D1 expression was altered little if at all in cells containing even high levels of exogenous GSK3β. We conclude that GSK3β has little influence over cyclin D1 levels in NIH3T3 cells 8 hrs following injection. Similar results were obtained at 24 hrs in NIH3T3 cells, and in MRC5 cells (not shown). Finally, the average expression level of phospho-Thr-286 cyclin D1 was determined in each cell cycle phase following injection of the GSK3β plasmid as above, and compared to uninjected cells. In no case was there any difference between injected and uninjected cells in the level of phospho-Thr-286 cyclin D1 (Fig. 9C).
GSK3 is potentially involved in G2 phase
The above experiments show no evidence for the involvement of GSK3 in cyclin D1 expression or phosphorylation in any cell cycle phase. It should be emphasized, however, that all of those studies were performed upon cells cultured with normal growth factors which might have induced proliferative signaling within the cells to neutralize all GSK3 activity, thereby masking its potential role in regulating cyclin D1 in the absence of serum. This possibility is supported by the fact that while cyclin D1 suppression following serum removal is primarily due to reduced mRNA stability, there is also evidence for a limited role of altered protein stability upon serum removal . To test the possibility that GSK3 might play a role in cyclin D1 regulation in the absence of serum, NIH3T3 cells were given a brief pulse of BrdU prior to serum removal for 11 hrs. Thus, cells in S phase at the time of serum removal would be labeled with BrdU. It is known that these cells would progress normally through mitosis, but with low cyclin D1 levels . To determine if GSK3 might play a role in maintaining cyclin D1 at low levels in these serum-deprived cells, they were treated with 12.5 or 25 mM LiCl to inhibit GSK3 activity at the time of serum removal. Our attention was focused upon the BrdU labeled cells to determine if their cyclin D1 levels remained low upon entry into G2 phase. The addition of 12.5 mM LiCl failed to alter cyclin D1 levels, but 25 mM LiCl treatment resulted in the elevation of cyclin D1 levels in some of the cells passing from S to G2 phase in the absence of serum (Fig. 10A). Three duplicate experiments were performed and compared to demonstrate a significant increase in G2 phase cyclin D1 levels in serum deprived cells cultured with 25 mM LiCl (Fig. 10B). We interpret this result as potential evidence for a limited role of GSK3 in regulating G2 phase levels of cyclin D1, particularly when proliferative signaling is limited. This conclusion, however, is complicated by the fact that 25 mM LiCl slowed passage of cells into mitosis, resulting in an extended G2 phase (see Fig. 10A). We do not know if this extended G2 length might contribute to the slight elevation of cyclin D1 levels induced by LiCl in this experiment.