Protein Kinase C Delta (PKCδ) Affects Proliferation of Insulin-Secreting Cells by Promoting Nuclear Extrusion of the Cell Cycle Inhibitor p21Cip1/WAF1

Background High fat diet-induced hyperglycemia and palmitate-stimulated apoptosis was prevented by specific inhibition of protein kinase C delta (PKCδ) in β-cells. To understand the role of PKCδ in more detail the impact of changes in PKCδ activity on proliferation and survival of insulin-secreting cells was analyzed under stress-free conditions. Methodology and Principal Findings Using genetic and pharmacological approaches, the effect of reduced and increased PKCδ activity on proliferation, apoptosis and cell cycle regulation of insulin secreting cells was examined. Proteins were analyzed by Western blotting and by confocal laser scanning microscopy. Increased expression of wild type PKCδ (PKCδWT) significantly stimulated proliferation of INS-1E cells with concomitant reduced expression and cytosolic retraction of the cell cycle inhibitor p21Cip1/WAF1. This nuclear extrusion was mediated by PKCδ-dependent phosphorylation of p21Cip1/WAF1 at Ser146. In kinase dead PKCδ (PKCδKN) overexpressing cells and after inhibition of endogenous PKCδ activity by rottlerin or RNA interference phosphorylation of p21Cip1/WAF1 was reduced, which favored its nuclear accumulation and apoptotic cell death of INS-1E cells. Human and mouse islet cells express p21Cip1/WAF1 with strong nuclear accumulation, while in islet cells of PKCδWT transgenic mice the inhibitor resides cytosolic. Conclusions and Significance These observations disclose PKCδ as negative regulator of p21Cip1/WAF1, which facilitates proliferation of insulin secreting cells under stress-free conditions and suggest that additional stress-induced changes push PKCδ into its known pro-apoptotic role.


Introduction
Sufficient b-cell mass is required for adequate insulin secretion. Consequently, an elevated demand of insulin is controlled by increased proliferation of pancreatic endocrine cells while insufficient insulin secretion and the development of type-2 diabetes have been associated with b-cell death [1]. A variety of molecular changes are involved in b-cell failure including reduced insulin/IGF-1 receptor signaling, endoplasmic reticulum stress and mitochondrial dysfunction [2][3][4][5][6][7][8][9][10]. These changes are triggered by obesity-linked factors, such as oxidative stress, saturated free fatty acids, cytokines and interleukins. Previous observations from our and other groups suggested that protein kinase C delta (PKCd) plays a decisive role in b-cell failure induced by cytokines and free fatty acids [11][12][13][14][15]. Thus, mice with targeted overexpression of a kinase-negative PKCd (PKCdKN) mutant in b-cells are protected against high fat diet-induced glucose intolerance and show increased survival of islet b-cells [14]. Conversely, we have previously shown that exposure of b-cells to high concentrations of palmitate promotes PKCd-mediated nuclear accumulation of FOXO1, a pro-apoptotic transcription factor activated under stress conditions [14]. Furthermore, PKCd has been found to mediate iNOS mRNA stabilization induced by IL-1b, whereas ablation of PKCd protected mice against streptozotozin-induced hyperglycemia [11,12]. Thus, under certain stress conditions, PKCd promotes signaling pathways leading to apoptotic b-cell death.
Very few studies have investigated the role of PKCd for normal b-cell function, in particular under stress-free conditions. Surprisingly, mice with increased transgenic expression of PKCd in b-cells develop and age normally under chow diet and maintain normal glucose tolerance (unpublished observations). As a matter of fact, although PKCd can serve as a pro-apoptotic signal, depending on the cellular context, it can also elicit anti-apoptotic and survival signals in a variety of cell systems [16][17][18]. These proliferative effects might involve a direct interference of PKCd with cell cycle regulation [19,20]. Intriguingly, proliferation of differentiated bcells is a rare event although proteins which are important for cell cycle progression are expressed [21]. In adult mice less than 0.4% of b-cells stain positive for BrdU, in cultured human islet preparations only 0.3% of the cells proliferate [21][22][23]. Proliferation is tightly controlled by the sequential expression and activation of cell cycle regulators, such as cyclins and cyclin-dependent kinases (CDKs). The mitogenic activity of cyclin-CDK complexes is limited through binding of transiently expressed cell cycle inhibitors [24]. Inhibitors of the Cip/Kip family, p21 Cip1/WAF1 , p27 kip1 and p57 Kip2 are ubiquitously expressed proteins that slow down proliferation and cell cycle progression at G1/S or G2/M phase transitions [25]. While p57 Kip2 regulates cell cycling mainly during development, p21 Cip1/WAF1 and p27 kip1 accumulate in mitogen-starved cells and mediate cell cycle arrest upon DNA damage [26][27][28]. In accordance with a minor role of p21 Cip1/WAF1 during development, mice deficient of p21 Cip1/WAF1 show normal growth and differentiation of the endocrine pancreas [22]. In contrast, mice that specifically overexpress p21 Cip1/WAF1 in b-cells have impaired b-cell replication and develop age-related hyperglycemia due to increased apoptosis [29].
The activity of p21 Cip1/WAF1 is regulated further by its subcellular distribution which is controlled by phosphorylation of p21 Cip1/WAF1 at residues located in the C-terminal domain in proximity to the nuclear localization sequence [30]. PKB/Aktmediated phosphorylations at Ser146 and at Thr145 sequester p21 Cip1/WAF1 into the cytosol [31]. In vitro phosphorylation assays have further shown that PKCd can phosphorylate directly p21 Cip1/WAF1 at Ser146, which triggers its cytosolic accumulation and influences the stabilization of p21 Cip1/WAF1 [20].
In the present study, we examined the role PKCd plays in proliferation and survival of insulin-secreting cells. Our results suggest that PKCd phosphorylates the cell cycle inhibitor p21 Cip1/WAF1 at Ser146, which favors its nuclear extrusion and supports proliferation under stress-free conditions. However, under stress conditions such as free fatty acids PKCd turns into a pro-apoptotic kinase.

PKCd affects proliferation and apoptosis of insulin-secreting cells
The first observation that PKCd may influence cell growth was made with INS-1E cells which were transfected with either an active PKCd (PKCdWT) or an inactive, kinase dead PKCd (PKCdKN, Fig. 1A). Surprisingly, PKCdWT transfected cells displayed 2.4 times more nuclei stained positive for the proliferation marker Ki67 when compared to untransfected control cells (Fig. 1B). The phosphorylation of PKCd at Ser643 and Thr505 was increased proportionally to the protein amount in PKCdWT INS-1E cells under standard culture conditions, which is indicative for an active PKCd (Fig. 1A). In PKCdKN INS-1E cells phosphorylation of PKCd at Thr505, a phosphorylation site of PDK1, is also increased proportionally to the protein amount, while phosphorylation at the autophosphorylation site Ser643 is reduced (Fig. 1A) [32][33][34]. It is noteworthy that the PKCdKN mutant remains inactive regardless of the degree of phosphorylation. These observations suggest that PKCd supports proliferation of INS-1E cells.
PKCd controls cytosolic-nuclear trafficking of p21 Cip1/WAF1 in INS-1E cells and primary mouse islet cells The analysis of proteins which regulate proliferation revealed that PKCdWT cells expressed a significantly lower amount of the cell cycle inhibitor p21 Cip1/WAF1 than control or PKCdKN INS-1E cells, while no change in expression of p27 kip1 was apparent ( Fig. 2A, B). Notably, due to the shorter length of rodent p21 Cip1/WAF1 compared to the human orthologue, the protein band of p21 Cip1/WAF1 displayed an apparent molecular weight lower than 21 kDa. This band was specific for p21 Cip1/WAF1 , as a protein of the same size was detected in bleomycin-treated wild type mouse embryonic fibroblasts (MEFs) but not in p21 Cip1/WAF1 -deficient MEFs (Fig. S1A). Moreover, the specificity of two antibodies used in this study was confirmed by immunocytochemistry, which revealed a p21 Cip1/WAF1 -specific staining in bleomycin-treated MEFs that was completely absent in untreated cells or in p21 Cip1/WAF1 -deficient cells exposed to the DNA damaging agent (Fig. S1B). The reduced expression of p21 Cip1/WAF1 in PKCdWT INS-1E cells may be responsible for accelerated proliferation.
Even more interesting is that the analysis of the subcellular distribution of p21 Cip1/WAF1 using confocal laser scanning microscopy showed that PKCdWT cells expressed p21 Cip1/WAF1 almost exclusively in cytoplasm. In contrast, in PKCdKN cells and to a lesser extent in control INS-1E cells p21 Cip1/WAF1 nuclear accumulation of the inhibitor was apparent (Fig. 2C). The reduced nuclear localization of p21 Cip1/WAF1 in PKCdWT cells was confirmed by Western blotting of cytosolic and nuclear fractions (Fig. 2D). In the cytosolic fractions the relative amount of p21 Cip1/WAF1 was similar in control and PKCdWT cells. Furthermore, the cell cycle inhibitor remained cytosolic in PKCdWT cells also after synchronization of cells by serum removal (Fig. S2). Although an increased number of nuclei of PKCdKN INS-1E cells stained positive for p21 Cip1/WAF1 , the amount of p21 Cip1/WAF1 protein detected on Western blots was not increased relative to the amount of nuclear proteins (data not shown). These findings do also support the hypothesis that PKCdWT cells proliferate faster due to reduced p21 Cip1/WAF1 activity.
To transfer this finding to native b-cells, proliferation of islet cells was examined in pancreatic slices of WT and b-cell specific PKCdWT transgenic mice. Even after high fat feeding, Ki67 staining was not detectable neither in WT nor in PKCdWT b-cells, which suggests that proliferation remained low (data not shown). However, p21 Cip1/WAF1 immunoreactivity was found in nuclei of cultured mouse and human islet cells ( Fig. 3A and 3B). Similarly, mice islet cells with targeted expression of PKCdWT in b-cells showed reduced nuclear accumulation of p21 Cip1/WAF1 , whereas a prominent nuclear staining was evident in islet cells from control mice and PKCdKN transgenic mice (Fig. 3A). These observations suggest that PKCd-dependent regulation of p21 Cip1/WAF1 might also influence proliferation of native b-cells.
PKCd-dependent phosphorylation of p21 Cip1/WAF1 at Ser146 regulates its subcellular distribution and function To investigate the molecular mechanism of PKCd-dependent subcellular distribution of p21 Cip1/WAF1 , the phosphorylation of p21 Cip1/WAF1 at two regulatory sites was examined next. In comparison to control INS-1E cells phoshorylation of p21 Cip1/WAF1 at Ser146 was significantly increased in PKCdWT cells, while it was reduced in PKCdKN cells (Fig. 4A). Phosphorylation of Thr145 was not detectable (Fig. 4B). To substantiate the effect of PKCd on p21 Cip1/WAF1 , its phosphorylation at Ser146 was examined in cells stimulated with the phorbol myristate acetate (PMA). When cells were starved overnight, phosphorylation at Ser146 declined (Fig. 4C, first and second line). PMA stimulated p21 Cip1/WAF1 phosphorylation in starved cells, an effect that was completely abolished by the PKCd inhibitor rottlerin. Inhibition of phosphorylation was accompanied by an increase in p21 Cip1/WAF1 protein (Fig. 4C). Moreover, the effects were specific for PKCd, as inhibition of protein kinase B or ERK1/2 neither inhibited phosphorylation at Ser146 nor promoted nuclear accumulation of p21 Cip1/WAF1 (Fig. S3). As a matter of fact nuclear staining of p21 Cip1/WAF1 was reduced after stimulation of PKCs with PMA and was increased after treatment of INS-1E cells with rottlerin ( Fig. 5). Similar but less pronounced results were obtained with PKCdWT INS-1E cells (data not shown). Although rottlerin is a sensitive inhibitor of PKCd, it also affects other kinases such as CaM kinase III [35]. Therefore, we used the more specific siRNAapproach to reduce PKCd expression and activity.
In cells transfected with siRNA against PKCd a significant reduction of PKCd expression was accompanied by a reduced phosphorylation at Ser146 and a concomitant increase in the protein amount of p21 Cip1/WAF1 (Fig. 6A), an effect not found in cells treated with control siRNA. Consistent with the results obtained with PKCdKN cells, the knockdown of PKCd by siRNA resulted in an increased nuclear translocation of p21 Cip1/WAF1 (Fig. 6B).
These data strongly suggest that p21 Cip1/WAF1 is a substrate of PKCd in insulin-secreting cells. Phosphorylation of p21 Cip1/WAF1 by PKCd results in its nuclear extrusion and thereby may support proliferation.

Functional consequences of reduced PKCd expression in insulin-secreting cells under stress-free conditions
As PKCd supports fatty acid induced apoptosis, the effect of changes in PKCd expression on cell death was examined in more detail. Surprisingly, inhibition of endogenous PKCd with siRNA or with PKCdKN mutant almost doubled the incidence of apoptotic cell death under non-stress conditions, as revealed by TUNEL staining (Fig. 6C and Fig. 6D). In accordance, isolated islet cells from transgenic mice expressing PKCdKN in b-cells displayed increased TUNEL staining when compared to control mice (Fig. 6E). In contrast, overexpression of PKCdWT in INS-1E cells and mice b-cells did not stimulate apoptosis under control culture conditions ( Fig. 6D and 6E). These observations suggest that PKCd per se is not pro-apoptotic but rather promotes proliferation.
Finally, the impact of changes in PKCd expression on cell cycle was examined. When cells were stained for the G2/M marker phospho-Ser10 histone H3, the same amount of control and PKCdWT nuclei (7%), but significant more PKCdKN nuclei (17%) stained positive for phospho-Ser10 histone H3 (Fig. S4). Cell cycle analysis by flow cytometry of propidium iodide-stained nuclei revealed two distinct DNA peaks (Fig. S5A). While the major peak of control cells (58%) and significant more PKCdWT INS-1E cells (70%) showed similar DNA staining which represents G1 (2n chromosomes), a minor part of the cells resided in G2 (4n). The first DNA peak of PKCdKN cells was visible at 4n (80%) and the second peak at 8n (20%), which probably represent cells with increased DNA content at G1 and G2, respectively. That the DNA peaks correlate to 2n, 4n and 8n was confirmed by treatment of the cells with colchicine (0.5 mM for 2 d) which arrests cell cycle at G2/M transition (Fig. S5B). When DNA from freshly isolated mouse islet cells was examined, more than 95% of WT and PKCdKN cells stained for 2n and less than 2% of the cells for G2 (Fig. S6). The prominent peak (2n) of WT and PKCdKN islet cells suggests that mouse islet cells are arrested in G0/G1 and that PKCdKN expression did not affect the arrest. This study deciphers a direct link between PKCd and the cell cycle inhibitor p21 Cip1/WAF1 which may influence b-cell proliferation. The mechanism which drives PKCd from a proliferative into a pro-apoptotic role under stress conditions remains to be elucidated.

Discussion
The present study discloses the cell cycle inhibitor p21 Cip1/WAF1 as a target of PKCd in insulin-secreting cells. Phosphorylation of p21 Cip1/WAF1 at Ser146 by PKCd leads to its nuclear extrusion, thereby favoring cell proliferation and survival. The fact that p21 Cip1/WAF1 is a substrate of PKCd is consistent with a previous report [20] and supported by our observation that both RNA interference as well as a pharmacological inhibitor of PKCd suppressed phosphorylation of p21 Cip1/WAF1 , whereas the PKC activator PMA increased p21 Cip1/WAF1 phosphorylation. Furthermore, inhibition of PKCd activity by expression of a kinase-inactive PKCd mutant reduced phosphorylation and increased nuclear accumulation of p21 Cip1/WAF1 . In contrast, in PKCdWT-expressing cells p21 Cip1/WAF1 was phosphorylated at Ser146 and largely confined to the cytoplasm. Interestingly, inhibition of PKB and ERK1/2 did not diminish phosphorylation of the cell cycle inhibitor, indicating that PKCd is the major regulator of p21 Cip1/WAF1 in insulin-secreting cells. Thus, our data suggest that in proliferating insulin-secreting cells PKCd supports proliferation, at least in part, by reducing nuclear accumulation and stability of p21 Cip1/WAF1 (Fig. 7). In pancreatic slices of PKCdWT mice, proliferation was not detectable. Similarly, in p21 Cip1/WAF1 KO mice, proliferation of pancreatic islet cells was also not significantly increased (0.4% in WT and 0.6% in KO) [22]. These observations suggest that p21 Cip1/WAF1 does not induce proliferation. Indeed, cell cycle inhibitors rather influence the velocity of proliferation while induction of b-cells proliferation occurs only under special conditions such as in new born and pregnant and lactating animals or after 90% pancreatectomy.
PKCd has been found to be involved in a variety of cellular events. Although several reports indicate a pro-apoptotic role, PKCd was also shown to exert anti-apoptotic and proliferative effects in various cell types. Such opposing effects of PKCd may be cell type-or stimulus-specific or mediated by spatio-temporal differences of PKCd activation. There is evidence that, similar to p21 Cip1/WAF1 , different functional effects of PKCd are connected with the diverse compartmentalization of the enzyme. Upon stimulation with phorbol ester or fatty acids, PKCd redistributes between a cytosolic, a membrane-bound and a cytoskeletonassociated compartment in b-cells [36]. The pro-apoptotic effect of PKCd is linked to its nuclear accumulation [37,38]. Furthermore, cleavage of PKCd by caspase-3 releases a constitutively active fragment that promotes apoptosis [39]. Our data suggest that a substantial amount of PKCdWT in transgenic cells is phosphorylated at Ser643, an autophosphorylation site and, consequently, is stimulated under control culture conditions. This increased activity does not induce apoptotic cell death suggesting that additional factors generated under stress conditions are needed to turn PKCd into a pro-apoptotic kinase (Fig. 7). Reduced PKCd activity significantly augmented apoptosis consistently in PKCdKN INS-1E cells (up to 2-fold), in PKCdKN transgenic mouse b-cells (by 60%), as well as after down regulation of PKCd by RNA interference in control INS-1E cells (2-fold). Although not discussed in detail, in a study by Cantley et al. using PKCdKO mice, the rate of apoptosis was 80% higher in knockout cells than in control cells [11]. The physiological impact of this finding remains unclear, especially as islet size and insulin content were not reduced in PKCdKO mice when compared to wild type mice [40]. In agreement, mice that express PKCdKN exclusively in bcells show no reduction in islet size and insulin content [14,41].
In addition to cell cycle regulation, emerging evidence suggests that p21 Cip1/WAF1 exerts other functions in diverse cellular processes, including cell differentiation and survival. The impaired replication and increased apoptosis of b-cells of p21 Cip1/WAF1 transgenic mice mirror our observations obtained in insulin secreting cells with reduced PKCd activity and may thus result from prolonged nuclear accumulation of p21 Cip1/WAF1 [29]. Interestingly, similar to PKCdKO mice, these mice show improved recovery from streptozotocin-induced hyperglycemia, which has been attributed to an increased regeneration of insulinproducing cells [11,29]. In line, our previous study disclosed protection of mice with b-cell specific expression of PKCdKN against HFD-induced hyperglycemia. Apparent contradictory to this assumption are two studies which link stress-induced expression of p21 Cip1/WAF1 to reduced insulin mRNA and b-cell failure. In one study oxidative stress-induced expression or exogenous overexpression of p21 Cip1/WAF1 in rat islets suppressed insulin biogenesis [42]. In the second study using a mouse model with deficient DNA repair reduced b-cell proliferation and the onset of diabetes was accompanied by increased expression of p21 Cip1/WAF1 [43]. One difference which might explain these opposing results between the two latter studies and our cell models is the persistent expression of p21 Cip1/WAF1 , while in our cell systems endogenous p21 Cip1/WAF1 expression is transient.
Phosphorylation events presumably not only regulate the compartmentalization but also the protein stability of p21 Cip1/WAF , although controversial data have been reported on this issue. It was previously shown that PKB-dependent phosphorylation of p21 Cip1/WAF1 at both Thr145 and Ser146 increases the stability of the CDK inhibitor and enhances its anti-apoptotic activity [31,44,45]. However, in our experiments, inhibition of PKB by Akti-1/2 affected neither phosphorylation nor nuclear accumulation of p21 Cip1/WAF1 , suggesting a minor role of PKB in our cell system. Likewise, inhibition of ERK1/2 with PD98059 had no impact on the subcellular distribution and phosphorylation of p21 Cip1/WAF1 . Although we did not study the effect of PKCd on p21 Cip1/WAF1 stability in detail, PKCdWT cells revealed a significantly lower amount of p21 Cip1/WAF1 than control or PKCdKN cells, suggesting that Ser146 phosphorylation affects p21 Cip1/WAF1 stability. These data are consistent with reports demonstrating that phosphorylation of p21 Cip1/WAF1 at Ser146 by PKCd leads to destabilization of the CDK inhibitor [46].
Thus, our data indicate that p21 Cip1/WAF1 exerts a dual effect depending on its subcellular distribution (Fig. 7). When trapped in the cytosol due to phosphorylation, p21 Cip1/WAF1 might favor proliferation, a notion supported by the increased Ki67 staining in PKCdWT cells. This cytosolic localization of p21 Cip1/WAF1 is known to exert anti-apoptotic effects by CDK-dependent or independent mechanisms [47,48]. One mechanism of antiapoptotic action of p21 Cip1/WAF1 involves its direct binding to and inhibition of the pro-apoptotic kinases ASK1 or JNK [31,49,50].
In contrast, nuclear p21 Cip1/WAF1 inhibits cell cycle progression and might eventually lead to apoptosis. In addition to the binding to cyclin-CDK complexes, p21 Cip1/WAF1 interacts directly with the proliferating cell nuclear antigen (PCNA), and thereby inhibits PCNA-dependent DNA replication [51]. Improper and prolonged nuclear accumulation of p21 Cip1/WAF1 may explain the observations that mice overexpressing p21 Cip1/WAF1 specifically in b-cells develop age-related hyperglycemia under normal feeding [29].
Whether PKCd-dependent regulation of p21 Cip1/WAF1 affects b-cell function in humans needs further experimental evidence. The expression of p21 Cip1/WAF1 in human islets and the fact that PKCd reduces nuclear accumulation of p21 Cip1/WAF1 in primary mouse islet cells supports the view that the cell cycle inhibitor could play a regulatory role also in adult human b-cells under special proliferative conditions [23].
In conclusion, our study demonstrates that PKCd induces posttranslational modifications of p21 Cip1/WAF1 which, in turn, determine its subcellular distribution and function in INS-1E cells. This study reveals that PKCd is not per se a pro-apoptotic kinase and underlines the importance of understanding molecular mechanisms for the evaluation of therapeutic targets in the treatment of diabetes mellitus.

Ethics Statement
The use for scientific purposes of isolated human islets was approved by the local ethics committee (University of Tuebingen, Medical Faculty No. 533/2010BO2). All animal experiments were done in accordance with the accepted standard of human care of animals and approved by the local Animal Care and Use Committee (Notification from 12.01.10).

Cell cycle analysis by Nicoletti
After 2 d culture, cells were detached by trypsin and resuspended in Nicoletti buffer containing 0.1% sodium citrate, pH 7.4, 0.1% Triton X-100 and 50 mg/ml propidium iodide. DNA staining was analyzed by flow cytometry using the FL2-H channel.

Statistics
Data are expressed as means 6 SEM, p,0.05 (unpaired Student's t-test or 2-way ANOVA followed by Bonferroni post test where applicable) was considered significant.