p27 and Leukemia: Cell Cycle and Beyond
Cyclin dependent kinase inhibitors B (CDKN1B) or p27 belongs to the cip/kip family of CDKI (Sherr and Roberts, 1999). The functional importance of p27 is demonstrated by the presence of homologous proteins across the eukaryotic kingdom including yeast (Sic1) and Arabidopsis thaliana (AtKRP6 and AtKRP7) (Barberis et al., 2005; Guerinier et al., 2013). p27 shares significant homology with its other family members (p21 and p57), specifically in the amino terminal domain (Polyak et al., 1994b). The protein was first identified as an inhibitor of CDK2 containing complexes in G1 arrested lung epithelial cells under contact inhibition or when treated with transformation growth factor beta (TGFbeta) (Polyak et al., 1994a). Subsequently, the gene encoding p27 was cloned and also identified in a yeast tri-hybrid screen as a cyclin D–CDK4 interacting protein (Toyoshima and Hunter, 1994; Polyak et al., 1994b). Since then p27 has not only emerged as a prime regulator of cell cycle progression but has also been implicated in numerous malignancies including leukemia.
p27 is a potent inhibitor of various cyclin–CDK complexes. The crystal structures of human p27 inhibitory domain bound to cyclinA–CDK2 complex revealed the physical association of the conserved LFG sequence motif of p27 with conserved cyclin-box residues of cyclin A and further interactions of p27 with N terminal beta sheet residues of CDK2. The resulting complex induced changes in the catalytic cleft of CDK2 thereby occluding ATP binding and inhibiting its catalytic activity (Russo et al., 1996). In vitro studies have demonstrated that p27 acts as a stoichiometric inhibitor of all G1 cyclin–CDK complexes (Cheng et al., 1998). However, in vivo studies have revealed that p27 preferentially inhibits CDK2 complexes (Blain et al., 1997). Interestingly, p27 has been found to promote assembly of cyclin D–CDK4/6 complexes (LaBaer et al., 1997). Such an interaction facilitates sequestration of p27 in cyclin D–CDK4 complexes allowing the downstream activation of cyclinE– CDK2 complexes (Perez-Roger et al., 1999). Moreover, interaction of p27 with cyclin D–CDK4 complex also inhibits the activity of CDK4 (Kato et al., 1994). However, crucial differences were observed in terms of the levels of p27 required to inhibit the activities of CDK2 containing complexes vis-à-vis CDK4 containing complexes. Whereas CDK2 complexes bound to p27 were inactive, those bound to CDK4 still retained kinase activity. Additional analysis showed that high concentrations of p27 yield a stoichiometry of greater than
1:1 of p27 and the cyclin–CDK4 complexes thus inactivating cyclin–CDK4 (Blain et al., 1997). The importance of the interaction between p27 and CDKs can be assuaged from the wide array of experimental drugs targeting CDKs (such as flavopiridol) that have been shown to have profound anti- tumor or anti-leukemic activity (Shapiro, 2006; Malumbres et al., 2008). Together, the available data confirm that p27 is a versatile molecule that can exist in both cyclin bound inhibitory and non-inhibitory conformations (Vlach et al., 1997). Recently, it was shown that p27 can physically interact with minichromosome maintenance 7 (MCM7) a protein involved in the recognition of DNA replication origin and having DNA helicase activity (Nallamshetty et al., 2005).
p27 also modulates cell survival and apoptosis; however, the effects appear to be very much context dependent. Various studies have shown that p27 negatively regulates apoptosis by down regulating the activity of CDK2 in endothelial and fibroblast cells (Levkau et al., 1998; Hiromura et al., 1999).
Simultaneously, p27 has also been reported to induce apoptosis especially in solid tumours (Katayose et al., 1997; Fujieda et al., 1999; Eguchi et al., 2004). A more detailed analysis of the various functions of p27 may be found in the reviews (Borriello et al., 2007; Besson et al., 2008).
p27 and its interactors
p27 like its other cip/kip family members is an intrinsically unstructured or disordered protein. Before binding to cyclin– CDK complexes, p27 is largely disordered and devoid of stable tertiary structure. Upon binding to complexes such as cyclin A– CDK2 it acquires nascent secondary structure (Lacy et al.,2004). This intrinsically disordered state of free p27 not only explains its wide array of protein-protein interactions but also makes it perceptible as to how various protein interactions and phosphorylation events may modulate its cyclin–CDK inhibitory activity (Otieno and Kriwacki, 2012). An example of such a regulation is obtained from thermodynamic studies of the interaction between p27 with either cyclin D–CDK4 or cyclin A–CDK2 complexes. p27 was found to fold incompletely when bound to cyclin D–CDK4 thus exposing Y88 residue of p27 for further modifications ensuing cell cycle entry (Ou et al., 2011). As expected from the intrinsically disordered structure of p27, the post translational modifications of the protein are key determinants of its binding specificity. Phosphorylation on Tyr-74, -88 and -89 by Src, Lyn, Abl or JAK2 kinases impairs the ability of p27 to inhibit the activity of CDK (Chu et al., 2007a; Grimmler et al., 2007; James et al., 2008; Jakel et al., 2011). Importantly, oncogenic tyrosine kinases such as Bcr-Abl and JAK2V617F are known to phosphorylate p27 on Y88 leading to its degradation and cellular proliferation (Grimmler et al., 2007; Jakel et al., 2011). In addition, various other interacting partners including Pin1, CUL4A, tuberin, Ca2þ/CaM-dependent kinase 1 (CaMKI), Grb2, valosin-containing protein (VCP) have been found to be associated with p27 thereby regulating its function and/or degradation (Sugiyama et al., 2001; Rosner et al., 2007; Miranda-Carboni et al., 2008; Zhou et al., 2009; Mallampalli et al., 2013; Yu et al., 2013).
Sub-cellular localization of p27
Another important aspect of p27 biology that is controlled by phosphorylation of some of its key residues pertains to its localization (Table 1). p27 shuttles between the nuclear and cytoplasmic compartments. Phosphorylation of three specific residues (S10, T157, and T198) was found to contribute to its cytoplasmic retention and localization. Phosphorylation on S10 represents the most abundant phosphorylation of p27 (Ishida et al., 2000). The modification not just greatly increases the stability of the protein, but also contributes to its nuclear export through interaction with the CRM1/Ran-GTP complex (Rodier et al., 2001; Connor et al., 2003; Besson et al., 2006). Various cellular kinases have been implicated in this modification including human kinase interacting stathmin (hKIS), Mirk/dyrk1B, Akt1, and Erk (Boehm et al., 2002; Deng et al., 2004; Nacusi and Sheaff, 2006; Borriello et al., 2007). In fact, hKIS has been known to promote proliferation of leukemic cell lines through phosphorylation of p27 on ser10 (Nakamura et al., 2008). On the other hand phosphorylation of p27 on T157 and T198 by Akt and its eventual sequestration in the cytoplasm provides yet another means to incorporate external cues into induction of cellular proliferation.
Importantly, phosphorylation of T157 within the nuclear localization signal (NLS) of p27 prevents its nuclear import (Liang et al., 2002; Shin et al., 2002; Viglietto et al., 2002). Recently, TRIP6 an important adaptor protein regulating cell motility has been shown to facilitate Akt mediated recognition and phosphorylation of p27 on T157 (Lin et al., 2013). Again, phosphorylation at T198 by Akt or p90 ribosomal protein S6 kinase sequesters p27 in the cytoplasm in a 14-3-3 binding dependent manner (Fujita et al., 2002, 2003). 14-3-3 has been known to bind to phosphorylated T157 thereby preventing interactions with importin alpha (Sekimoto et al., 2004). In mice, phosphorylation of p27 on S183 by IKKalpha was found to contribute to the nuclear export of p27. Incidentally, IKKalpha induced nuclear export of p27 was found to contribute to the metastatic potential of Erb2 induced breast cancer cells in mice (Zhang et al., 2013b).
Functional importance of cytoplasmic p27
Cytoplasmic localization of p27 has a direct effect on cell cycle regulation. The physical sequestration of p27 ensures that p27 cannot inhibit the activity of cyclin–CDK complexes. Despite the imminent significance of sequestration of p27 in the cytoplasm, it
was evident that cytoplasmic p27 may affect pathways other than those regulating cellular proliferation (Liang et al., 2002). Indeed, studies demonstrated that cytoplasmic p27 could modulate the actin cytoskeleton structure through Rho-GTPase (Nagahara et al., 1998; Besson et al., 2004). p27 was found to inhibit the interaction between RhoA and its activators the guanine nucleotide exchange factors (GEFs) thereby inhibiting the activity of the downstream ROCK (Besson et al., 2004). As a result, p27—/— cells showed increased migration defects along with increased numbers of stress fibres and focal adhesions (Besson et al., 2004). Additionally, p27S10 phosphorylation and the resulting cytosolic localization have been implicated in cell migration (McAllister et al., 2003). Therefore it was not surprising that many metastatic cancers including melanoma, papillary thyroid, hepatocellular carcinoma, and breast cancer cells show a predominance of cytoplasmic p27 (Li et al., 2006; Wu et al., 2006; Denicourt et al., 2007; Wang et al., 2008; Ahn et al., 2009).
Interestingly, p27 has been found to be phosphorylated by a neuron specific CDK5 (Zheng et al., 2010). The phosphorylation of p27 in post-mitotic neurons has been linked to stabilization of p27 and induction of neuronal migration by blocking RhoA signalling through its interaction with Neurogenin2 protein (Kawauchi et al., 2006; Nguyen et al., 2006). To add to the already growing complexity, RhoA has been found to promote the translation efficiency of p27 mRNA while in a separate study constitutively active RhoA (RhoAL63) has been found to induce cyclin E–CDK2 mediated degradation of p27 (Hu et al., 1999; Vidal et al., 2002). In recent times,cytoplasmic p27 has been proposed to play the role of a potential oncogene (Serres et al., 2011).
Regulation of the expression of p27
The expression of p27 within cells has been primarily found to be governed at the translational level (Hengst and Reed, 1996). In fact, a translation mediated up regulation of p27 without a corresponding increase in its mRNA expression has been found in the Bcr-Ablþ CD34þ cells indicating the importance of post- transcriptional and translational regulation (Jiang et al., 2000; Chu et al., 2010). However, transcriptional regulation as well as degradation of p27 also controls its expression (Pagano et al., 1995). Numerous transcription factors such as FKHR-L1, AFX, FOXO, Sp1, E2F1, and BRCA1 have been found to promote its transcription (Dijkers et al., 2000; Medema et al., 2000; Andres et al., 2001; Stahl et al., 2002; Williamson et al., 2002; Wang et al., 2005). Simultaneously, c-myc, Id3 and Ap1 have been found to repress the transcription of p27 (Yang et al., 2001; Garrett- Engele et al., 2007; Khattar and Kumar, 2010). Similarly, various mechanisms have been shown to regulate the translation efficiency of p27 mRNA including increased association with ribosome and cap independent translation (Millard et al., 1997; Miskimins et al., 2001; Kullmann et al., 2002; Gopfert et al., 2003; Bellodi et al., 2010). Degradation of p27 by the nuclear SCFskp2 ubiquitin ligase marks the entry of cells into the S phase. The phosphorylation of p27 on T187 by the Cyclin–CDK2 complexes marks them for the SCFskp2 mediated ubiquitin- proteasomal degradation (Carrano et al., 1999). KPC (Kip1 ubiquitin promoting complex), a cytoplasmic ubiquitin ligase complex has also been implicated in the degradation of p27 (Kamura et al., 2004). Details of the complex pathways mediating p27 degradation may be found in the review (Lu and Hunter, 2010; Starostina and Kipreos, 2012).
p27 and Leukemia
Human p27 is located on chromosome 12p13. This chromosomal location had been identified very early as a region of chromosomal translocations in lymphoid and myeloid malignancies (Sato et al., 1995). Subsequently, this chromosomal segment not only lead to the identification of TEL (ETV6) as a prime locus involved in haematopoietic malignancies but also increased the curiosity of the scientific community towards the possible roles of p27 in such disorders.
Given the roles played by p27 during cell cycle progression and commitment to G1/S transition, it was widely expected that p27 may act as a tumor-suppressor. This was indeed found to be the case when homozygous null mice for p27 were created. These mice showed hyper-cellularity of organs along with pituitary tumours and intestinal adenomas. Interestingly, the sensitivity to p27 dosage demonstrated that p27 acts as a haplo-insufficient tumor suppressor (Fero et al., 1998). Although various studies have reported mutations within the p27 gene in
haematological malignancies, yet p27 deletions have remained fairly uncommon (Pietenpol et al., 1995; Stegmaier et al., 1996; Komuro et al., 1999; Markaki et al., 2006; Takeuchi et al., 2002).
p27 and Bcr-Abl oncogene
Expression studies of p27 in chronic myeloid leukemia (CML) patient samples showed absence of specific mutations that could be directly linked to the disease (Iolascon et al., 1998). CML is characterised by the presence of the oncogenic tyrosine kinase Bcr-Abl. Early reports have shown that conditional expression of Bcr-Abl in Ba/F3 cells caused a marked decrease in the expression of p27 (Jonuleit et al., 2000). Moreover the same study also showed that Bcr-Abl could similarly regulate the expression of p27 in human cell line M07.
Later studies also confirmed that conditional -expression/over- expression of Bcr-Abl did indeed cause a decrease in the expression of p27 in cell lines while inhibition of Bcr-Abl with ST1571 reversed this effect (Gesbert et al., 2000; Parada et al., 2001; Tomoda et al., 2005; Roy et al., 2013). At the same time results revealed the increased expression of p27 in CD34þ cells isolated from CML patients when compared to normal CD34þ cells. This study also showed that p27 majorly localised in the nucleus in normal CD34þ cells whereas >80% of p27 localised in the cytoplasm of CD34þ cells in CML (Jiang et al., 2000). Remarkably, similar to previous reports no mutations were seen in the nuclear localization signal (NLS) of p27 in the CML samples studied thus negating a change in NLS as a possible cause of the cytoplasmic retention of p27 in CML. Several other studies also demonstrated that transduction of Bcr-Abl into primary CD34þ cells derived from cord blood induced an increase in the expression of p27 (Zhao et al., 2001). Physiologically transplantation of mice with WT or Bcr-Abl transduced bone marrow cells resulted in increased numbers of leukemic stem cells along with significantly higher mortality in Bcr-Abl transduced bone marrow recipient mice (Zhang et al., 2012).
Many mechanisms have been put forth to explain the effect of Bcr-Abl on the expression of p27 (Sengupta and Banerjee, 2007). Bcr-Abl is known to activate multiple signalling pathways. This includes the PI3K/Akt pathway which is constitutively activated in Bcr-Ablþ cells. This was mediated through direct interaction of Akt with Bcr-Abl and was found to be essential for Bcr-Abl mediated in vivo leukemogenesis (Skorski et al., 1995, 1997; Atfi et al., 2005). Bcr-Abl mediated Akt signaling was found to affect p27 expression as well as cell
cycle kinetics in Bcr-Abl expressing cell lines (Gesbert et al., 2000). Subsequently, Akt was reported to phosphorylated p27 on T157 causing its cytoplasmic retention and thereby blocking cell cycle inhibitory functions (Liang et al., 2002; Shin et al., 2002; Viglietto et al., 2002). Additional studies have shown that Bcr-Abl induced Akt mediated phosphorylation of p27 on T157 was responsible for the cytoplasmic localization of p27 in CD34þ cells of CML patients (Chu et al., 2010). This study, along with previous reports, shows that Y177 of Bcr-Abl was necessary for cytoplasmic localization of p27 (Chu et al., 2007b). Another report indicated that Bcr-Abl signals through PI3K to induce expression of skp2 and thereby mediate degradation of p27 (Andreu et al., 2005). Additionally, FKHRL1 (forkhead transcription factor rhabdo-myosarcoma-like1, FOXO3a) a member of the forkhead family of transcription factors is phosphorylated by PI3K/Akt which thereby renders it inactive by sequestering it in the cytoplasm. In cells expressing Bcr-Abl, FKHRL1 is constitutively phosphorylated. Inhibition of Bcr-Abl leads to a decrease in the phosphorylation of FKHRL1 along with an increase in the expression of p27. Also, a dominant negative form of FKHRL1 enhanced the sensitivity of Bcr-Ablþ cell lines to ST1571 (Komatsu et al., 2003). Another report indicated that active FKHRL1 could overcome ST1571 resistance in CML cells (Kikuchi et al., 2007). Notably, forkhead family of transcription factors have been known to directly regulate the transcription of p27 (Medema et al., 2000).
Furthermore, Jab1 a member of the COP9/signalosome complex has been found to induce nuclear export and degradation of p27 (Tomoda et al., 1999). In cell lines expressing Bcr-Abl as well as in CML samples, the Jab1 signalosome is up regulated through the MAPK and PI3K pathways. Thus, the pathway ensures p27 degradation in Bcr- Ablþ cells (Tomoda et al., 2005). A recent report also suggests that Bcr-Abl mediates direct transcriptional up regulation of Jab1 through co-operative interaction between beta-catenin and STAT1 at the Jab1 promoter (Yang et al., 2011).
Conspicuously, expression of skp2 is regulated by oncogenic tyrosine kinases such as Bcr-Abl, JAK2V617F, FLT3-ITD, and TEL-PDGFRbeta thus inducing degradation of p27. In addition, absence of skp2 was found to substantially decrease Bcr-Abl mediated leukemogeneis (Agarwal et al., 2008).
p27 and JAK2 mutation
Cytokine receptors are crucial to blood cell formation and a pivotal player in this signaling mechanism is the JAK/STAT pathway. Studies have indicated that JAK2 tyrosine kinase directly binds and phosphorylates p27 (Jakel et al., 2011).JAK2 could bind to p27 through its FERM and kinase domains and phosphorylated Y88 of p27. The report further showed that patients harbouring JAK2V617F showed substantial phosphorylation of p27 on Y88. Thus phosphorylated p27 was found to reside majorly in the cytoplasm. Both JAK2 and p27 are known to reside in the nucleus as well as in the cytoplasm.
As a result, an interaction between p27 and JAK2 in the nucleus is plausible and is likely to have functional ramifications. At the same time, these effects are also likely to be cell and context dependent. Therefore, JAK2-p27 interactions are likely to have implications in malignancies associated with constitutively active JAK2 kinase. JAK2 is one of the candidate oncogenic fusion proteins described in various leukemias (Lacronique et al., 1997; Peeters et al., 1997; Reiter et al., 2005). JAK2V617F mutation was identified in myeloproliferative neoplasms and has been found to be present in 95% of patients with Polycythemia Vera. Treatment of JAK2V617F expressing cell line HEL with a JAK2 inhibitor or siRNA targeting JAK2 was found to arrest cells in G1 phase of the cell cycle. A corresponding dose dependent increase in the expression of p27 was also observed (Walz et al., 2006). Furthermore in Ba/ F3 cells expressing JAK2V617F, p27 expression was found to be deregulated through modulation of the expression of skp2 (Furuhata et al., 2009). Additional evidence indicating the importance of known p27 interacting domain of JAK2 in the regulation of leukemogenesis was obtained from experiments with Ba/F3 cells expressing the Epo receptor. These cells were transformed readily by JAK2V617F but not when their FERM domain was mutated. The FERM domain mutant of JAK2V617F also failed to down regulate p27 (Wernig et al., 2008).
p27 and other leukemic onco-proteins
MLL is involved in chromosomal translocations with greater than 50 different genes in leukemia. These fusion genes have been found to cause haematopoietic cell transformation in vitro and leukemia in vivo (Lavau et al., 1997). MLL functions as a methyl transferase specifically on histone3 lysine4 (H3K4). Early studies had indicated that the amino terminus of MLL was involved in the up regulation of p27 and sufficient to induce growth arrest and monocytic differentiation of U937 cell lines (Caslini et al., 2000). p27 was also later identified as a direct target of MLL mediated transcriptional regulation (Milne et al., 2005; Xia et al., 2005). Recently, a report has shown that p27 is required to maintain quiescence in MLL-AF9 transformed cells (Zhang et al., 2013a). By transplanting MLL-AF9 transduced p27þ/þ or p27—/— hematopoietic stem and progenitor cells (HSPCs) into lethally irradiated congenic mice, the study found that deletion of p27 delayed MLL-AF9 leukemia development. Deletion of p27 was on the other hand found to significantly promote the proliferation of primitive CD117þ CD11blo MLL-AF9 transduced leukemic cells. Similarly, the PML protein was first characterised as a fusion protein with retinoic acid receptor alpha (RARalpha) in acute promyelocytic leukemia (APL). PML was demonstrated to contain growth suppressive properties and over expression of PML was found to down regulate the expression of p27 in HeLa cells (Mu et al., 1997). In addition, PML was found to induce apoptosis in cells. This was associated with localization of p27 and Bax in nuclear bodies (NB). NB associated p27 was found to be delocalised in APL cells and retargeted upon treatment with retinoic acid or arsenic (Quignon et al., 1998).
Emerging roles of cytoplasmic p27 in leukemia
Although p27 has been classically considered as a tumor suppressor, recent evidences indicate that it may have oncogenic property as well. The cytoplasmic localized p27 was observed to associate with poor prognosis in AML. In this study, cytoplasmic localization was found to correlate with constitutive phosphorylation of Akt on ser473. Further, expression of a constitutively active form of Akt induced an increase in the cytoplasmic localization of p27 (Min et al., 2004). In CML, a similar increase in the cytoplasmic localization of p27 was observed which correlated with the expression of Bcr-Abl (Jiang et al., 2000). Bcr-Abl Y177 was found to be crucial in regulating the altered localization of p27 through its effects on Akt activation (Chu et al., 2010). Akt mediated phosphorylation of p27 on T157 was found to induce cytoplasmic retention of p27. CD34þ cells isolated from CML samples showed increased phosphorylation of p27 on T157 compared to normal control cells (Chu et al., 2010). Moreover, phosphorylation of p27 by JAK2 and JAK2V617F ensues a predominant cytoplasmic localization of p27 (Jakel et al., 2011). These reports indicate that cytoplasmic p27 has may have important consequences in the progression of leukemia. This has been validated by a recent report from our laboratory that has shown that cytoplasmic localization of p27 increases during the progression of CML from chronic to blast phase (Roy et al., 2013).
Although cytoplasmic localization of p27 has been a prominent feature in leukemia, the functional consequences of such an altered localization are yet to be unraveled. Study from our laboratory reports the interaction of p27kip1 and RhoA in CD34þ cells of CML which was found to directly lead to decreased expression of RhoA-GTP. In fact, this study presents yet another mode of regulation of RhoA-GTP levels by Bcr-Abl. Previously, Bcr-Abl has been known to activate RhoA (Daubon et al., 2008). The DH domain of p210Bcr-Abl directly regulates the activity of RhoA (Sahay et al., 2008). Alternatively, p210Bcr-Abl transfected Ba/F3 cells showed decreased phosphorylation of RhoGDI and hence increased RhoA activation (Unwin et al., 2005). p27 may also interact with other cytoskeletal proteins such as Rac to mediate cell migration (McAllister et al., 2003). Thus, the association of p27 with cytoplasmic partners has important functional consequences in disease pathogenesis.Although a number of interactors of p27 are known, the spectrum of cytoplasmic interactors of p27 is yet to be fully identified.
Conclusion
p27 was long considered only as a cell cycle regulator and hence its function in leukemia and other malignancies has classically been linked to its tumour suppressive properties. However, by the turn of the last decade, p27 has emerged as both a tumour suppressor and a potential oncogene. The ability of p27 to modulate functions as diverse as cell proliferation and motility ensures that even moderate changes in its expression/ localization have far reaching consequences. In addition, the regulation of p27 via several mitogenic signalling kinases diversifies its effects. Although mutations in p27 are rare in leukemia, its aberrant expression especially in the cytosol has emerged as a crucial determinant of leukemic progression. An understanding of the diverse interactors of p27 especially in the cytosol would provide new insights into the field of leukemia prognosis and therapeutics. Furthermore, it would be interesting to analyze the effects of cytoplasmic p27 on hematopoietic stem cell homing and localization. Future studies could be directed to understand whether expression of cytoplasmic p27 contributes towards maintenance of leukemic stem cells and chemo-resistance in leukemic cells. In conclusion, p27 is being increasingly appreciated as a potential oncogene. Indeed, whether and how such an intrinsically unstructured protein contributes towards leukemic progression SEL120-34A may be investigated in future.