|Home | About | Journals | Submit | Contact Us | Français|
The tumor suppressor protein p27Kip1 plays a pivotal role in the control of cell growth and metastasis formation.
Several studies pointed to different roles for p27Kip1 in the control of Ras induced transformation, although no explanation has been provided to elucidate these differences. We recently demonstrated that p27kip1 regulates H-Ras activity via its interaction with stathmin.
Here, using in vitro and in vivo models, we show that p27kip1 is an important regulator of Ras induced transformation. In H-RasV12 transformed cells, p27kip1 suppressed cell proliferation and tumor growth via two distinct mechanisms: 1) inhibition of CDK activity and 2) impairment of MT-destabilizing activity of stathmin. Conversely, in K-Ras4BV12 transformed cells, p27kip1 acted mainly in a CDK-dependent but stathmin-independent manner.
Using human cancer-derived cell lines and primary breast and sarcoma samples, we confirmed in human models what we observed in mice.
Overall, we highlight a pathway, conserved from mouse to human, important in the regulation of H-Ras oncogenic activity that could have therapeutic and diagnostic implication in patients that may benefit from anti-H-Ras therapies.
The tumor suppressor protein p27kip1 (hereafter p27) was originally identified as a cyclin/CDK inhibitor, in particular of the CDK2-containing complexes[1, 2]. Subsequent studies demonstrated that it is also implicated in the regulation of several other biological activities, such as differentiation, apoptosis, motility and autophagy [1, 2].
Formal demonstration that p27 is a fundamental negative regulator of cell cycle progression with tumor suppressor properties primarily arose from the characterization of p27 knock-out mice[3-5]. Interestingly, most of the phenotypes of p27 null mice are reverted by the concomitant knock-out of the microtubules (MT)-destabilizing stathmin , demonstrating that p27/stathmin interaction plays a role in the control not only of cell motility [7-11] but also of cell proliferation[6, 12]. Mechanistically, we recently showed that p27 controls H-Ras-driven proliferation acting on its intracellular recycling and mono- bi-ubiquitination .
A large body of literature exists focusing on the cooperation between Ras and p27 during tumor onset and progression. It is widely accepted that p27 expression impacts on Ras-driven tumor progression. Interesting differences have been noted when p27 knock-out mice have been challenged with tumorigenic models relying on H-Ras versus K-Ras mutations. In tumors driven by K-Ras mutations, such as the urethane-induced [13, 14] or K-RasG12D-induced  lung cancers, p27 acts as a haploinsufficient tumor suppressor gene and the loss of one allele is sufficient to induce the maximum oncogenic cooperation. Conversely, in H-Ras-driven tumorigenesis, such as the MMTV-H-RasV12-induced breast and salivary gland cancers  or the DMBA/TPA skin carcinogenesis model , p27 acts as a classical tumor suppressor gene and loss of one p27 allele does not result in enhanced tumor growth. In these settings, many of the H-RasV12/p27 null tumors displayed features characteristic of highly aggressive tumors . Accordingly, it has been postulated that p27 controls cell H-RasV12-driven transformation via not only the inhibition of the cyclin/CDK/RB pathway but also via RB-independent pathways .
We have recently observed that p27 participates to the regulation of H-Ras activity by modulating its recycling and ubiquitination . Since H-Ras and K-Ras are differently regulated by recycling and ubiquitination [19, 20], it is expectable that p27 could differentially act in H-Ras versus K-Ras-driven tumor progression. Here, we address this hypothesis using different in vitro and in vivo model systems.
We previously observed that p27 null cells and mice displayed higher Ras activity, due to different recycling and decreased mono-bi-ubiquitination . To test if these differences may have any role in cell transformation, we generated and characterized H-RasV12-transformed WT and p27KO 3T3 fibroblasts [7, 21]. Endogenous p27 expression was maintained in H-RasV12 overexpressing-p27WT 3T3 cell clones (Figure (Figure1A).1A). WT and p27KO H-Ras cells were both less sensitive to serum deprivation than not transformed cells (Figure (Figure1B1B and Supplementary Figure S1A). Yet, p27KO H-RasV12 cells displayed increased S phase population when compared to WT H-RasV12 cells, in exponential growth (Figure (Figure1B)1B) and after contact inhibition (Figure (Figure1C).1C). Furthermore, they entered into the cell cycle sooner than WT H-RasV12 cells (Supplementary Figure S1A and B). Accordingly, in exponentially growing conditions p27KO H-RasV12 cells had a significantly lower doubling time (Supplementary Figure S1C), and proliferated significantly more than the WT counterpart (Figure (Figure1D1D and and1E1E).
In agreement with our previous observation [7, 8, 21, 22], WT versus p27KO-transformed cells displayed different ability to move, both in three-dimensional (3D)- (Supplementary Figure S1D) and in two-dimensional (2D)-extracellular matrix contexts (Supplementary Figure S1E), but not when 2D-migration was performed on plastic dish (Supplementary Figure S1F).
Subcutaneous injection of 1×106 WT H-RasV12 cells in nude mice (n = 5) were not sufficient to give rise to tumors, while the same number of p27KO H-RasV12 cells determined the appearance of tumor masses within 10-12 days (data not shown). Using 2×106 cells, all mice injected with WT H-RasV12 cells developed slow growing tumors within 15-17 days, but p27KO H-RasV12 cells formed fast growing tumors within 5-10 days (Figure (Figure1F)1F) as evidenced by differences in the explanted tumor masses (p < 0.0001, n = 5/clone), with no appreciable difference among the clones utilized (Figure (Figure1G1G).
In line with our previous results showing that p27/stathmin interaction modulates H-Ras activity  we showed that p27WT but not the p271-170 mutant (unable to bind stathmin) [7-9, 12, 21] reduced ERK1/2 phosphorylation when reintroduced in p27KO cells (Supplementary Figure 2A). In vitro experiments using growth curves (Figure (Figure2A)2A) and soft agar assays (Figure (Figure2B)2B) and in vivo experiments (Figure (Figure2C)2C) using subcutaneous injections, showed that p27KO H-RasV12 fibroblasts re-expressing p27WT or p271-170 were barely (in vitro) or not (in vivo) affected in their growth by either p27WT or p271-170. However, p27WT, but not p271-170, reduced the ability of p27KO H-RasV12 transformed fibroblasts to move in 3D-Matrigel (Figure (Figure2D)2D) and to intravasate, extravasate and settle at distant sites in nude mice, as demonstrated by the presence of cells expressing the H-RasV12 transgene in the blood and in the lungs of mice bearing subcutaneous tumors (5 mice/cell clone) (Figure (Figure2E2E and and2F2F).
To test whether stathmin was involved, at least in part, in determining the different in vitro and in vivo growth observed in WT versus p27KO H-RasV12 transformed fibroblasts, we used 3T3 fibroblasts derived from WT, p27KO and also double knock-out (DKO) for both, p27 and stathmin C57BL/6 embryos [12, 22]. Growth curve experiments confirmed that transformed p27KO cells grew much faster than WT cells and showed that DKO H-RasV12 cells displayed an intermediate growth rate (Figure (Figure3A)3A) and western blot analyses demonstrated higher ERK phosphorylation in p27KO H-RasV12 than in WT and DKO fibroblasts, both in basal conditions (Supplemental Figure S2A) and following serum (FBS) or Epidermal Growth Factor (EGF) stimulation (Figure (Figure3B3B).
Upstream activators of ERK include the H-Ras-related small GTPase RhoA that is also regulated by p27 [22, 23]. RhoA signals to ROCK1 to control both cell proliferation and migration [24, 25]. To distinguish the effects of p27 on H-Ras and RhoA we pharmacologically inhibited H-Ras (FTI-276) or ROCK1 (Y27632). FTI-276 treatment abrogated the differences in ERK phosphorylation between the different genotypes, while Y27632 treatment did not (Figure (Figure3C).3C). When used in growth curve experiments, both inhibitors reduced the proliferation of H-RasV12 transformed cells of all genotypes (Figure (Figure3D3D and and3E).3E). However, only FTI-276 inhibitor abrogated the differences in cell proliferation between p27KO and DKO cells (Figure (Figure3D).3D). Similar results were observed in soft agar assay experiments (Supplementary Figure S2B).
These results, along with the data collected on normal fibroblasts , suggested that p27/stathmin interaction through the regulation of H-Ras/ERK activity partially controls H-RasV12 driven transformation. If this was really the case, then no difference in cell transformation should be observed between p27KO and DKO fibroblasts transformed with K-Ras4BV12 oncogene. H-Ras and K-Ras4B (alternative splicing of exone 4 of K-Ras gene) are highly homolog proteins (83% aminoacid identity in the first 165 aminoacids) with an hyper variable C-terminus (24 aa) which comprises the membrane targeting sequence (Figure (Figure4A)4A) . The different hyper variable region results in several unique features, such as their sub-cellular localization and their functional regulation. H-Ras and N-Ras, but not K-Ras4B, require recycling to be fully activated [19, 20] and H-Ras and N-Ras, but not K-Ras4B, are inhibited by mono-bi-ubiquitination . In line with these notions, we observed that K-Ras4BV12 was not mono-bi-ubiquitinated either in the presence or absence of p27 and stathmin (Figure (Figure4B4B).
K-Ras4BV12 p27KO and DKO transformed cells displayed similar levels of ERK1/2 phosphorylation (Figure (Figure5A)5A) and proliferated fairly at the same extent but significantly faster than WT cells (Figure (Figure5B).5B). No substantial difference in the transformation efficiency was detectable by soft agar assay between p27KO and DKO K-Ras4BV12 cells (Figure (Figure5C).5C). We next analyzed, by pull down assay, the levels of GTP-bound active Ras proteins in WT, p27KO and DKO cells transformed with H-RasV12 or with K-Ras4BV12, in basal conditions and after stimulation with serum. As expected all transformed cells displayed constitutively active Ras, whose activity only slightly increased upon serum stimulation (Figure (Figure5D).5D). However, basal and serum-stimulated H-RasV12 (but not K-Ras4BV12) activities were significantly higher in p27KO when compared to WT and DKO transformed cells (Figure (Figure5D),5D), supporting the possibility that regulation of H-Ras ubiquitination by p27 and stathmin participated in the control of its activity.
In all experiments reported above, we consistently observed that both H-RasV12 and K-RasV12 transformed immortalized 3T3 WT cells with very low ability, especially when used below passage 45 (Figures (Figures1,1, ,2,2, ,3,3, ,4,4, ,5).5). This observation pointed that the p27/CDK/RB pathway plays a fundamental role as gatekeeper from Ras-induced cell transformation. Our results also showed that this function cannot be fully rescued by the reintroduction of human p27WT in p27KO transformed cells. To exclude species-specific effects due to the use of human p27 (h-p27WT) in rescue-experiments, we re-performed some of the same assays in p27 null H-RasV12 transformed cells, reintroducing the mouse p27, either wild type (m-p27WT) or mutated in the binding to Cyclin/CDKs complexes [8, 12, 27] (m-p27CK-) (Supplementary Figure S3A and B). In line with the results obtained with the human protein, m-p27WT and mp27CK- proteins only slightly affected the ability of H-RasV12 transformed cells to grow in culture or in soft agar (Supplementary Figure S3C and D) but significantly reduced cell motility (Supplementary Figure S3E).
To avoid the possible bias due to clonal selection and in the generation of 3T3 cells we next concomitantly transduced with and SV40 Large TAg (LgTAg) and with H-RasV12 (Figure (Figure6A)6A) or K-Ras4BV12 (Figure (Figure6B)6B) primary MEF of the different genotypes. Since SV40 LgTAg oncogene simultaneously inactivates p53 and RB, we anticipated that in this model the relevance of CDK-inhibition by p27 would be less pronounced.
In this model p27KO transformed with LgTAg and H-RasV12 grew at higher extent than the correspondent WT and DKO cells both in culture and in anchorage independent manner (Figure 6C, 6E and and6G).6G). Correspondent cell clones transformed with K-Ras4BV12 proliferated all at similar level, in the presence or absence of p27 (Figure 6D, 6F and and6H6H).
Similarly, tumors from LgTAg/H-RasV12 transformed p27KO MEF had significantly higher volume (Figure (Figure7A)7A) and number of Ki67 positive cells (Figure (Figure7B7B and and7C),7C), when compared to both WT and DKO counterparts. Conversely, LgTAg/K-Ras4BV12-transformed MEFs of all three genotypes displayed similar tumor volume (Figure (Figure7A)7A) and Ki67 expression (Figure (Figure7B7B and and7D).7D). In explanted tumors, the expression of the Ras-ERK downstream targets Egr-1, c-Fos and Jun-B was higher in p27KO cells only in the presence of H-RasV12 (Figure (Figure7E7E).
To evaluate the relevance of our findings in human cancer, we chose the soft tissue sarcomas (STS) as human counterpart of transformed fibroblasts. Following proliferative stimuli, MES-SA cells (established from primary STS) displayed high levels of p27 coupled with low levels of ERK1/2 activation and EGR1 expression, while HS-913T cells (established from metastatic STS) displayed low/null p27 levels, high ERK1/2 activation and EGR1 expression (Figure (Figure8A8A and and8B).8B). Using HT1080 cells, (a STS cell line harboring the N-RasQ61K mutation and a well-known model for Ras-driven transformation ) we next shown that Ras activity was reduced by one third by the overexpression of p27WT or p27CK- (Figure (Figure8C).8C). Since N-Ras shares the same recycling- and ubiquitin-mediated regulation of H-Ras [19, 20], our result supported the hypothesis p27 could control H/N-Ras activity in human STS. Accordingly, in a panel of already characterized human sarcoma specimens , low cytoplasmic p27/stathmin ratio was significantly associated with high levels of ERK1/2 phosphorylation (Figure (Figure8D)8D) and, when sarcoma specimens were segregated in primary vs metastatic tumors, we observed that metastatic tumors specimens displayed a significant lower p27/stathmin ratio coupled with a significantly higher ERK activation (Figure (Figure8E8E).
Finally, using BRAF-mutated colorectal carcinoma cells (i.e. Colo-201 and Colo-205 in Supplementary Figure S4A and B) and HER2-overexpressing mammary carcinoma cells (i.e. SK-BR-3 and MDA-MB-453 in Supplementary Figure S4C), we consistently detected an inverse correlation between p27/stathmin cytoplasmic ratio and the activation of ERK1/2 and the expression of EGR-1 (Supplementary Figure S4A-C). Importantly, knock-down of p27 in SK-BR-3 cells increased ERK1/2 phosphorylation (Figure S4D) and low cytoplasmic p27/stathmin ratio was significantly associated with a higher level of ERK1/2 phosphorylation in primary breast carcinomas (Supplementary Table S1 and Supplementary Figure S4E).
The results presented in this manuscript highlight several interesting features of the tumor suppressor gene p27 in the control of cell Ras-induced cell transformation and metastasis formation.
The most salient observation regards the susceptibility of WT or p27-null immortalized fibroblasts to the transforming activity of H-RasV12 and K-RasV12. The fact that p27-null fibroblasts are more prone to transformation and that reintroduction of p27 expression is not able to fully revert their phenotype reinforces the concept that control of CDK activity by p27 represents a significant barrier against cell transformation. The combined use of Large T-antigen and H-/K-RasV12 (Figures (Figures55 and and6)6) further confirms the relevance of p27 in controlling CDK activity during cell transformation. Yet, once cells are transformed by Ras, re-expression of p27 limited cell motility but failed to properly control in vitro proliferation and in vivo growth. Since we were concerned by the possibility that accumulation of concomitant mutations and/or genetic alterations could somehow affect the phenotype of p27KO H-RasV12 cells, we repeated our experiments using 2 additional models (i.e. C57Bl6 3T3 and primary MEFs transformation), overall confirming the observation made using the Sv129 3T3 cells (Figures (Figures11 and and2).2). It is interesting to note that in the case of v-Src transformed cells, reintroduction of the degradation resistant p27T187A mutant, completely reverted the phenotypes of p27-null cells , suggesting that the pathways activated by RasV12 and v-Src differently impact on the tumor suppressor roles of p27. Yet, one limitation of this study resides in the difficulty to discern whether one, or more of the multiple functions of p27, in the control of CDKs activities, MT-stability, actin reorganization, gene transcription and mitotic division [1, 2] could render p27 null cells more prone to transformation independently from the oncogene used.
A second interesting finding is the different potential displayed by p27 in restraining H-RasV12 versus K-RasV12-4B induced transformation. Only in H-RasV12 transformed cells p27 controls not only CDKs activation but also Ras activity/localization via stathmin. Consequently, p27-null cells had a further advantage in cell growth and invasion that is likely dependent by the lack of feedback control of Ras exerted by p27 when located into the cytoplasm. It is interesting to note that a role for cytoplasmic p27 in the inhibition of Ras activity, via GRB2 binding, was previously reported . Yet, it was not specified whether H- or K-Ras activity was tested. p27 expression was also reported to be necessary to mediate the inhibition of H-Ras-induced transformation induced either by STAT1 or by dominant negative Rho, likely in a RB-independent or partially dependent manner [18, 30]. More recent evidences demonstrated that in MEFs and in urinary bladder HT1197 cells, carrying a N-Ras Q61R mutation, p27 inhibits cell motility likely by reducing ERK activation , independently supporting our findings. Overall these data point to p27 as a relevant regulator of Ras activity in a cytoplasmic and CDK-independent manner.
Our work presents the limitation due to the use of constituvely active Ras mutant vectors. It has to be considered that the K-Ras4B gene used in this work is an alternatively spliced version of the K-Ras gene that could represent only a minor portion of the total K-Ras transcribed, as we observed in a model of skin carcinogenesis . Thus, our model could not fully recapitulate the activation of endogenous K-Ras that is transcribed as K-Ras4A and K-Ras4B, with the former still subjected to recycling to be fully activated [19, 20].
The role(s) played by p27 when located in the cytoplasm are at the center of an interesting scientific debate. It has been considered either a cellular escamotage to inactivate nuclear p27 [1, 32] or a more complex way to control other signaling pathways and processes [12, 23, 29], such as cell death and autophagy [33, 34] or cell motility [7, 23]. Most of these activities have been attributed to the C-terminal portion of p27, containing several important regulatory elements. It is interesting to note that, at least in MEN syndrome , breast cancer [36, 37], intestine neuroendocrine tumors  and prostate cancer , the CDKN1B gene (encoding for p27) is frequently mutated in the C-terminal portion of the protein [36-39]. One of the mutations, repeatedly identified in human breast cancers, is the E171®Stop [36, 37] that results in a truncated protein (p271-170) that we have characterized here and in previous publications [7-9, 12, 21].
Our data show that p27 and stathmin regulate in concert H-Ras but not K-Ras activity. This evidence could explain the contradictory tumor suppressor roles ascribed to p27 in different mouse models of cancer. In the presence of K-Ras mutations, p27 acts invariably as a haploinsufficient tumor suppressor gene [13-15]. In the presence of H-Ras, the complete loss of p27 is necessary to favor tumor growth in mice [16, 17], suggesting that the presence of one p27 allele is still able to restrain cell proliferation. According to this model, the fact that Ras-MAPK pathway activation results in phosphorylation of S10 and cytoplasmic delocalization of p27, likely represents a feedback control loop of particular importance to prevent unwanted cell cycle entry when the mitogenic extracellular stimuli are below the required threshold.
High levels of stathmin expression have often been linked to the acquisition of a metastatic phenotype . The use of knock-out mice and cells allowed us to exclude a primary role of stathmin in the onset of several type of primary tumors in mice , suggesting that the function here described for p27/stathmin interaction can be unmasked in normal mice, cells and tissues only in the context of p27 absence. This concept may be of particular relevance in human tumors where p27 levels and localization are finely regulated and may also explain the contradictory results reported so far for stathmin in the control of cell growth, when gain-of-function [42, 43] or knock-out [44, 45] models were considered.
Successful targeting of the Ras-MAPK pathway represents a promising tool to treat aggressive human cancers, but optimal selection is needed to identify patients who would benefit from such therapies. Our findings suggest that evaluation of p27 and stathmin expression may contribute to this selection. This is particularly relevant for breast cancer, where Ras oncogenes are infrequently mutated but often hyperactive  and p27 mutations could be driver oncogenic events [36, 37].
Detailed description of the material and methods used is provided in supplementary material.
Sarcoma and breast cancer tissues were collected at CRO Aviano, Italy and stored in the Institutional Biobank, provided that the specific informed consent was obtained from the patient. Scientific use of biological materials was approved by the Internal Review Board (IRB) of CRO Aviano (# IRB-07/2015).
All animal experimentation were reviewed and approved by the CRO institutional Animal Care and Use Committee (OPBA), authorized by Italian Ministry of Health (# 616/2015-PR) and conducted according to that OPBA's guidelines.
Primary tumors were established by subcutaneous injection of 1×106 or 2×106 transformed cells into the flanks of female athymic nude mice (Harlan, 7-8 weeks of age). Tumor growth was monitored every other day for up to 26 days from injection.
Primary wild type (WT), p27 knock-out (p27KO) and p27/stathmin double KO (DKO) mouse embryo fibroblasts (MEF) were prepared from embryos at day 13.5, according to standard procedures [6, 7]. 3T3 fibroblasts were generated from primary MEFs, as described . MEF, 3T3 fibroblasts, 293T/17, HEK 293, MDA-MB-453 and SK-BR-3 human mammary adenocarcinoma cells and MES-SA, HS-913T, HT-1080 sarcoma cell lines, were all cultured in DMEM supplemented with 10% FBS (Sigma). Colo-201 and Colo-205 human colorectal adenocarcinoma cells were cultured in RPMI-1640 supplemented with 10% FBS (Sigma).
3T3 fibroblasts were transformed using H-RasV12 (gently provided by Dr. R. Baserga) and K-Ras4BV12 (from ADDGENE, donated by Dr. T. Jacks) cDNAs both cloned in pMSCV-hygro retroviral vector (Clontech). Primary MEFs were transformed using concomitantly SV40 Large TAg (provided by Dr. R. Maestro) and pMSCV-Hygro-H-RasV12 or pMSCV-Hygro-K-Ras4BV12. Human p27WT or p271-170 were described [7, 21, 22]; mouse p27WT and p27CK- cDNAs were provided by Dr. Bruno Amati.
Proliferation assays include: growth curve experiments, using the Trypan Blue exclusion test and the MTS assay (Promega); cell cycle distribution using flow cytometry; BrdU incorporation assay (Roche); soft agar assays; and tissues staining with Ki67.
Motility assays include were performed essentially as described [7, 21, 22] and include 3D-Matrigel™ evasion assay ; transwell-based migration assay using HTS Fluoroblok™ coated with 20μg/ml fibronectin (Sigma) and wound-healing assay.
A total of 17 sarcoma (leiomyosarcomas and fibromyosarcomas) and 37 breast tumor specimens were collected and diagnosed at Centro di Riferimento Oncologico (CRO) of Aviano (Italy), according to the World Health Organization (WHO) criteria. Sarcoma samples were described elsewhere  and derived from primary (n = 8) or metastatic samples (n = 9). Breast cancer specimens derived from locally advanced primary tumors, as better specified in Supplementary Table 1.
We thank Dr. R. Maestro, Dr. T. Jacks, Dr. B. Amati and Dr. R. Baserga for providing reagents and all members of the S.C.I.C.C. lab for helpful scientific discussion.
CONFLICTS OF INTEREST
The authors have declared that no conflict of interest exists.
This work was supported by Associazione Italiana Ricerca sul Cancro (AIRC) to GB (IG 12854) and by CRO Intramural research grant to GB. IS is a recipient of AIRC/FIRC Fellowship.
Author ContributionsB.B. and G.B. designed the research; I.P., L.F., S.B., I.S., F.C., M.C., S.D.A. and S.B. performed the experiments, developed the methodologies and analyzed the data; T.P. and V.C. provided pathological analyses of human and mouse tumor specimens; S.M. and M.S., provided reagents/tools; and B.B. and G.B. wrote the paper.