Nature Medicine
8, 1153 - 1160 (2002)
Published online: 16 September 2002; Corrected online: 23 September 2002 | doi:10.1038/nm761
PKB/Akt phosphorylates p27, impairs nuclear import of p27 and opposes p27-mediated G1 arrestJiyong Liang1, Judit Zubovitz2, Teresa Petrocelli1, Rouslan Kotchetkov1, Michael K. Connor1, Kathy Han1, Jin-Hwa Lee1, Sandra Ciarallo1, Charles Catzavelos3, Richard Beniston1, Edmee Franssen4
& Joyce M. Slingerland1, 51 Molecular and Cell Biology, Sunnybrook and Women's College Health Sciences Centre, St. Mary's Hospital, McGill University, Montreal, Quebec 2 Department of Pathology, Sunnybrook and Women's College Health Sciences Centre, St. Mary's Hospital, McGill University, Montreal, Quebec 3 Department of Pathology, St. Mary's Hospital, McGill University, Montreal, Quebec 4 Division of Clinical Epidemiology and Biostatistics, Toronto Sunnybrook Regional Cancer Centre, Toronto, Ontario, Canada 5 Medical Oncology, Toronto Sunnybrook Regional Cancer Centre, Toronto, Ontario, Canada 6 T.P. and R.K. contributed equally to this study.
Correspondence should be addressed to Joyce M. Slingerland jslingerland@med.miami.eduMechanisms linking mitogenic and growth inhibitory cytokine signaling and the cell cycle have not been fully elucidated in either cancer or in normal cells. Here we show that activation of protein kinase B (PKB)/Akt, contributes to resistance to antiproliferative signals and breast cancer progression in part by impairing the nuclear import and action of p27. Akt transfection caused cytoplasmic p27 accumulation and resistance to cytokine-mediated G1 arrest. The nuclear localization signal of p27 contains an Akt consensus site at threonine 157, and p27 phosphorylation by Akt impaired its nuclear import in vitro. Akt phosphorylated wild-type p27 but not p27T157A. In cells transfected with constitutively active AktT308DS473D (PKBDD), p27WT mislocalized to the cytoplasm, but p27T157A was nuclear. In cells with activated Akt, p27WT failed to cause G1 arrest, while the antiproliferative effect of p27T157A was not impaired. Cytoplasmic p27 was seen in 41% (52 of 128) of primary human breast cancers in conjunction with Akt activation and was correlated with a poor patient prognosis. Thus, we show a novel mechanism whereby Akt impairs p27 function that is associated with an aggressive phenotype in human breast cancer.NOTE: In the version of the article initially published online, the abstract contained one extraneous sentence. This error has been corrected in the HTML and PDF versions. The abstract will appear correctly in the forthcoming print issue.Cell-cycle deregulation is a hallmark of cancer. Loss of cytokine-mediated G1 arrest may confer an advantage during malignant progression. Resistance to the antiproliferative effects of transforming growth factor- (TGF-) often occurs despite intact TGF- signaling and such cells may manifest resistance to multiple inhibitory cytokines, suggesting underlying alterations in cell-cycle controls1,
2.
Cyclin-dependent kinases (cdks) are regulated by cyclin binding, phosphorylation and by two families of cdk inhibitors3. G1 progression is governed by D-type and E-type cyclin-cdk complexes. The inhibitors of cdk4 (INK4) family includes p15INK4B and the kinase inhibitor protein (KIP) family comprises p21Cip1, p27Kip1 and p57Kip2 (ref. 4). In addition to inhibition of cyclin E-cdk2, p21 and p27 also facilitate assembly and activation of cyclin D-cdks in early G1 (refs. 5,6).
p27Kip1 was first identified as a mediator of TGF-1-induced G1 arrest7,
8,
9. TGF- induces expression of the gene encoding p15INK4B. p15INK4B binds and inhibits cdk4 facilitating dissociation of p27 and cyclin D1 from cyclin D1−cdk4−p27 complexes and p27 binds and inhibits cyclin E-cdk2, leading to G1 arrest10,
11,
12. p27 also mediates G1 arrest induced by IL-6 (ref. 13).* p27 is a nuclear protein whose frequent deregulation in human cancers may confer resistance to antiproliferative signals. In cMyc or MAPK activated cancer-derived lines, cyclin D1−cdk4/6 complexes sequester p27, and cyclin E-cdk2-inhibition is impaired14,
15,
16. cMyc inhibits p15 induction by TGF- (ref. 17) and may also induce a factor that inactivates p2718. In up to 50% of human cancers, reduced p27 protein is associated with a poor prognosis19. In some tumors, p27 is mislocalized to the cytoplasm19,
20, however; the mechanism and significance of this has not been elucidated.
In human cancers, constitutive activation of phosphoinositol 3' kinase (PI3K) and its effector* PKB/Akt arise through oncogenic receptor tyrosine kinase activation, Ras activation, mutational loss of PTEN, or through activating mutation of the PI3K effector, protein kinase B (PKB)/Akt (hereafter termed Akt) itself21,
22. Akt can increase cyclin D1 levels23 and downregulate p27 by increasing p27 proteolysis24 or repressing p27 expression through Akt phosphorylation of a forkhead transcription factor25. However, in most cancers, reduced p27 does not result from transcriptional silencing19.
Here we show that Akt causes resistance to cytokine-mediated G1 arrest. p27 phosphorylation by Akt impairs its nuclear import and leads to cytoplasmic p27 accumulation. In human breast cancers, cytoplasmic mislocalization of p27 is associated with Akt activation, loss of differentiation and poor patient outcome.
Activation of Akt in lines resistant to G1 arrest by TGF-
Ras has been shown to confer TGF- resistance. While investigating mechanisms of TGF- resistance, we found that two TGF- resistant lines showed activation of the Ras effector, PI3K-Akt pathway. TGF- causes G1 arrest of normal 184 human mammary epithelial cells (HMECs) and of the early stage melanoma line, WM35, but not the resistant 184A1L5R or advanced melanoma-derived line, WM239 (Fig. 1a and b). Resistant lines showed increased Akt activation (Akt-P) (Fig. 1c and d). Total Akt was similar in sensitive and resistant HMECs. Akt levels were somewhat higher in WM239 than WM35. PTEN loss contributed to Akt activation in WM239 (Fig. 1d).
 | | |
The increased PI3K-Akt activity in 184A1L5R and WM239 was linked to their TGF- resistance, as partial PI3K inhibition restored TGF- responsiveness. In 184A1L5R, low concentrations of the PI3K inhibitor, LY294002 (10−12.5  M) modestly reduced the S phase fraction (% S) but allowed continued proliferation, with a profile similar to that of 184 cells (Fig. 2a). Although TGF- alone had little effect, 10 M LY294002 together with TGF- caused G1 arrest of 184A1L5R (Fig. 2a). In TGF--resistant WM239, LY294002 at a concentration that did not independently inhibit proliferation led to G1 arrest when combined with TGF- (data not shown).
| | Figure 2. PI3K inhibition restores TGF- response in resistant cells and Akt transfection mediates TGF- resistance. | | |  | a and b, Asynchronous cells were treated with LY294002 alone ( ) or with TGF- ( ) for 24 h. a, Effects on the % S phase. b, Effects in 184A1L5R of 10 M LY294002 with or without TGF- on Akt-P levels, Akt kinase activities, cyclin E1-association with p27 and cyclin E-cdk2 activities. Lane 4 shows the antibody only control for the left and middle panels. Total Akt, p27 and -actin levels are shown in the right panel. c−e, The indicated cells were transformed with PKBDD, PTEN or empty vector (E). c, Ectopic expression of PKBDD was shown by western blot (Akt) and Akt activity confirmed by blotting for phosphorylation of the Akt substrate, GSK3- (GSK3--P) or by Akt kinase assays (Akt kinase). Levels of p27 and cyclin D1 are shown. d and e, Flow cytometric analysis of parental and transfected cells treated with () or without () TGF- (d) and without () or with () IL-6 (e) for 48 h.
NOTE: In the version of the article initially published online, Fig. 2b, in the center panel, bottom blot, lanes 2 and 3 were transposed. This has been corrected in the HTML and PDF versions, and will appear correctly in the forthcoming print issue.
 Full Figure and legend (67K) |
| |
Because inhibition of cyclin E-cdk2 by p27 contributes to G1 arrest by TGF-, cyclin E1 complexes were assayed in 184AIL5R. A low dose of LY294002 (10 M) inhibited Akt activity (Fig. 2b). LY294002 did not change p27 protein levels in HMECs (Fig. 2b), but modestly increased cyclin E1−bound p27 and partly inhibited cyclin E1-cdk2 activity (Fig. 2b). TGF- and LY294002 together increased further cyclin E1−bound p27 and inhibited cyclin E1-cdk2 causing G1 arrest. Thus, attenuation of P13K activity restored cyclin E1-cdk2 inhibition by p27 and G1 arrest by TGF-.
Akt inhibits responsiveness to antiproliferative cytokines
184 and WM35 cells were transfected with a constitutively active AktT308DS473D (PKBDD)26,
27 or empty vector (E). Transfectants showed increased Akt protein and activity (Fig. 2c). PKBDD-transfected lines were resistant to G1 arrest by TGF- (shown for WM35PKBDD) (Fig. 2d). Moreover, PKBDD transfection also conferred resistance to G1 arrest by IL-6 (Fig. 2e). In WM239, transfection of PTEN restored TGF- sensitivity (Fig. 2d).*
Akt causes cytoplasmic mislocalization of p27
Because Akt impairs the nuclear localization of some of its substrates28, we assayed if Akt activation affected p27 localization. Cellular p27 was predominantly nuclear in 184 HMEC. 184A1L5R and 184PKBDD cells showed both nuclear and cytoplasmic p27 (Fig. 3a). Increased cytoplasmic p27 in WM239 and WM35PKBDD, and in 184A1L5R and 184PKBDD compared with WM35 and 184, respectively, was confirmed by immunoblotting of fractionated cell lysates (Fig. 3b).
Transfected fluorescent-tagged wild-type p27 (YFPp27WT) was exclusively nuclear in most 184 and WM35 cells (Fig. 3c and Supplementary Table A online). YFPp27WT-transfected 184A1L5R and WM239 showed increased cytoplasmic p27. LY294002 restored the predominantly nuclear localization of YFPp27WT in TGF--resistant 184A1L5R and WM239 cells (Fig. 3c). In 184PKBDD and WM35PKBDD, transfected YFPp27WT showed increased cytoplasmic localization, similar to that in 184A1L5R and WM239.
In contrast to effects of LY294002, treatment of WM239 cells with the MEK inhibitor, UO126, did not result in redistribution of p27 from cytoplasm to nucleus. Moreover, in two independent cell lines, inducible overexpression of constitutively activated MEK (MEKEE) and MAPK activation did not cause cytoplasmic mislocalization of p27 (Fig. 3d and f). Thus, in this culture model, MAPK activation is neither necessary nor sufficient for cytoplasmic mislocalization of p27. As both Akt and MAPK activation can increase cyclin D1 levels, we tested the effect of cyclin D1 transfection on p27 localization. Cyclin D1 overexpression did not mediate cytoplasmic localization of p27 (Fig. 3e and g).
Akt binds and phosphorylates cellular p27Kip1
A minimal consensus motif has been defined for Akt (ref. 29). p27 contains a putative Akt consensus sequence between amino acids 152 and 157 (RKRPAT). Immunoprecipitated cellular Akt could phosphorylate recombinant p27 directly in vitro as well as the known Akt substrate, histone H2B. p27T157A, generated by replacing the T157 with alanine, was much less efficiently phosphorylated by Akt (Fig. 4a). Thus, T157 is identified as a putative Akt phosphorylation site in p27. The inability of Akt to phosphorylate p27T157A is not due to loss of Akt binding as both endogenous p27 and ectopically expressed p27WT and p27T157A coprecipitated with activated cellular Akt-P (Fig. 4b).
| | Figure 4. p27 is a substrate of Akt and binds Akt in vivo. | | |  | a, Akt kinase assays using either histone H2B, or recombinant wild-type p27 (p27WT) as substrates. Control reactions using normal rabbit IgG (NR-IgG) are shown (right lanes). Akt phosphorylated p27WT but not p27T157A in in vitro kinase assays. Input p27 is immunoblotted in the top right panel. b, Ser473-phosphorylated Akt immunoprecipitates were resolved and immunoblotted to show Akt-P and associated endogenous cellular p27 (left panel) or transfected YFPp27WT and YFPp27T157A (right panel). c, Cellular p27 shows reactivity with antibody specific for the phosphorylated Akt consensus motif. While p27 levels are increased by LY294002, reactivity with the phospho-Akt substrate antibody (P-Akt substrate) is diminished. Lower graph: , without LY294002; , with LY294002. d, WM239 cells were transfected with YFPp27WT or YFPp27T157A and then treated with or without LY294002 for 20 hours. The top band shows Akt-P. YFPp27 was immunoblotted with antibodies to P-Akt substrate, anti-p27T157-P and total p27.
NOTE: In the version of the article initially published online, Fig. 4a, in the left panel, "Akt" was incorrectly placed. It should have been aligned with the unlabelled positivity indicators. This error has been corrected in the HTML and PDF version. The figure will appear correctly in the forthcoming print issue.
Full Figure and legend (58K) |
| |
To demonstrate that phosphorylation of p27 by Akt can occur in cells, p27 immunoprecipitates from asynchronous and LY294002-treated WM239 and MCF-7 cells were immunoblotted with a phospho-specific antibody against the Akt phosphorylation consensus motif (P-Akt substrate antibody). Immunoreactivity with this antibody demonstrated cellular p27 phosphorylation at the Akt consensus motif in asynchronous cells. This was inhibited by LY294002 (Fig. 4c).
In normal 184 HMECs, Akt-P is minimal in G0 and increases rapidly when cells enter early G1 (data not shown). Using an antibody specific for T157-phosphorylated p27 generated by Viglietto et al., we showed that cellular T157 phosphorylated p27 was minimal in G0 and the relative amount of T157 phosphorylated p27 over total p27 rose by 1.8-fold within 4 hours of exit from quiescence (data not shown). Densitometric analysis of p27T157-P and total p27 blots showed that the relative amount of T157 phosphorylated over total p27 was two-fold higher in proliferating WM239 than in WM35 (data not shown). Thus, activation of Akt in vivo is associated with increased T157 phosphorylation of p27 in both normal HMECs and in tumor-derived cells.
To provide further evidence for p27 phosphorylation by Akt in vivo, WM239 cells were transfected with either YFPp27WT or YFPp27T157A and treated with 20 M LY294002. LY294002 markedly reduced reactivity of YFPp27WT with the antibody to the phosphorylated Akt consensus motif (P-Akt substrate) and the p27T157-P antibody. p27T157A showed minimal reactivity with the antibody to phosphorylated Akt substrate and none to anti-p27T157-P (Fig. 4d).
p27T157A is nuclear in cells with constitutive Akt
T157 is located within the nuclear localization signal (NLS) of p2730,
31. Phosphorylation of the NLS regulates nuclear localization of many proteins32. To assay whether the potential for T157 phosphorylation might influence p27 localization, WM35PKBDD cells were transfected with YFPp27WT, YFPp27T157A and YFPp27T157D (Fig. 5a and Supplementary Table B online). While approximately 30% of YFPp27WT and YFPp27T157D expressing cells showed both nuclear and cytoplasmic p27, YFPp27T157A was nuclear.
| | Figure 5. T157 phosphorylation impairs nuclear import of p27. | | |  | a, p27 localization following transfection of the indicated p27 alleles into WM35 cells expressing PKBDD (WM35PKBDD). b, WM239 cells were transiently transfected with YFPp27 alleles and the cell-cycle profile of YFP-positive cells analyzed by flow cytometry. , untransfected;  , wild-type; , T157A-transfected. c, p27 nuclear import was assessed by the addition of His-tagged p27 (His-p27) to digitonin permeabilized cells for 60 minutes (60) and nuclear, (N), and supernatant, (S), fractions were immunoblotted for p27. Input His-p27WT is shown in the left-most lane (t=0 minutes). Pre-treatment of His-p27WT with cellular Akt (+Akt) impaired p27 import. Import reactions for His-p27T157A (T157A) and His-p27T157D (T157D) are shown. Import of T157A was abolished by addition of wheat germ agglutinin (+ WGA).
Full Figure and legend (43K) |
| |
WM239 cells, whose Akt is constitutively activated, were transiently transfected with either YFPp27WT or YFPp27T157A. Flow cytometric analysis of YFP positive cells at 20 hours post-transfection revealed that the cell-cycle inhibitory function of p27WT was significantly impaired, while p27T157A retained G1 inhibitory function in WM239 (Fig. 5b). Equal expression of YFPp27WT and YFPp27T157A was demonstrated (Fig. 4d, lower). In contrast, YFPp27WT and YFPp27T157A both caused G1 arrest in 184 HMECs and WM35 cells lacking constitutive Akt activation (data not shown). Thus p27 phosphorylation by Akt impairs its G1 inhibitory function.
Phosphorylation by Akt impairs nuclear import of p27
p27 nuclear import was assessed by the incubation of His-tagged p27 (His-p27) with digitonin permeabilized MCF-7 cells. Nuclei and supernatant fractions were then immunoblotted for His-p27. Prior reaction of recombinant His-p27WT with cellular Akt kinase impaired nuclear import of p27. His-p27T157A showed a faster rate of import than His-p27WT, and His-p27T157D protein showed essentially no nuclear import above that of negative controls (reactions carried out at 4 °C, without ATP or in the presence of wheat germ agglutinin) (Fig. 5c). These data suggest that cytoplasmic p27 in Akt activated cells results from impaired p27 nuclear import.
Cytoplasmic p27 and activated Akt in human breast cancers
p27 protein levels and localization were reviewed in 128 primary breast cancers previously stained for p27 by immunohistochemistry33. Levels of p27 were scored as high (>50%) or low ( 50%) tumor nuclei staining as previously described33. While normal breast ductal epithelium and lymphocytes and a majority of tumors (Fig. 6a) showed exclusively nuclear p27, 42% (52/128) of the breast cancers showed either nuclear and cytoplasmic or predominantly cytoplasmic p27 (Fig. 6b and c). Of cancers with reduced p27 levels, 44% (31/70) showed cytoplasmic p27, while 36% (21/58) of tumors with high p27 levels showed cytoplasmic p27. p27 protein levels and cytoplasmic p27 mislocalization were not statistically correlated (chi square analysis, P = 0.3542).
| | Figure 6. Cytoplasmic mislocalization of p27 in primary breast cancers is associated with Akt activation and poor patient outcome. | | |  | a−c, Primary human breast cancers were immunostained for p27 (left panels). The same tumors were stained for phosphorylated Akt (right panels) and counterstained with hematoxylin. a, An intermediate grade tumor with exclusively nuclear p27 staining in >50% of tumor nuclei (N only; p27>50% nuclei positive) shows no immunoreactivity for phosphorylated Akt. b and c, Moderately and poorly differentiated cancers showing intense (b) and reduced (c) staining of p27 in both nuclei and cytoplasm (N+C, p27 >50% nuclei positive and N+C, p27 50% nuclei positive, respectively). d, Kaplan−Meier overall survival curves stratified by nuclear only (green line) versus nuclear and cytoplasmic (purple line) p27 (P = 0.49). e, Kaplan−Meier curves show that for each level of p27 staining (p27 >50% or 50% tumor nuclei positive), the detection of cytoplasmic p27 was associated with a significantly reduced overall survival compared with that of the nuclear only group (P = 0.05). f, The best prognostic subgroup of breast cancers shows strong, exclusively nuclear p27 (p27 >50%; green line), whereas cancers with both reduced levels and cytoplasmic mislocalization of p27 (p27 50%; red line) have the lowest survival (P = 0.02).
Full Figure and legend (68K) |
| |
Cytoplasmic p27 was not correlated with menopausal, nodal or ER/PR status. Tumors with high levels of exclusively nuclear p27 (N only; p27 >50% nuclei positive) (Fig. 6a) were almost all well differentiated or of low grade, whereas tumors with high levels of p27 but cytoplasmic localization (N+C; p27 >50%) (Fig. 6b) were more poorly differentiated (P < 0.001 on  2 analysis) (Supplementary Table C). The Kaplan−Meier curve in Fig. 6d shows the influence of p27 localization on patient survival. Fig. 6e shows that for each level of p27 staining (high, >50% nuclei positive or low, 50% nuclei positive), when p27 is seen in the cytoplasm (N+C) rather than in the nucleus alone (N), patient survival is worse. These data were statistically significant for overall survival (P = 0.05, Wilcoxon test) (Fig. 6e) and for disease-free survival (P = 0.003, data not shown). Patients with breast cancers with high levels of exclusively nuclear p27 (N only; p27 >50%) had the best outcome, whereas the worst survival was seen in those with reduced p27 levels and detectable cytoplasmic p27 (N+C; 50%, P = 0.02, Log-rank test) (Fig. 6f).
53 breast cancers were stained with phospho-Akt antibody (Fig. 6a−c, right). Normal quiescent breast acini, lymphocytes and stroma showed no Akt-P staining. None of 23 tumors with exclusively nuclear p27 showed Akt activation, whereas 28/30 tumors with cytoplasmic p27 showed phospho-Akt reactivity. Detection of cytoplasmic p27 in human breast tumors was highly statistically significantly associated with Akt activation (2, P < 0.001).
Discussion
This study suggests that constitutive activation of the PI3K-Akt pathway mediates TGF- resistance. TGF-- and IL-6-sensitive cells were rendered resistant by ectopic Akt activation. Akt activity was increased in 184A1L5R and WM239 cells. Moreover, PTEN transfection into WM239 cells, and inhibition of PI3K by LY294002 in TGF--resistant cells restored p27 binding and inhibition of cyclin E-cdk2 and G1 arrest by TGF-.
Whereas PI3K signaling can inhibit p27 transcription or accelerate p27 degradation in different cell types, we show that Akt can bind p27 and phosphorylate T157. Akt phosphorylated p27WT but not p27T157A in vitro. Moreover, the reactivity of cellular and transfected p27 with phospho-Akt substrate and p27T157-P antibodies was strongly reduced by Akt inhibition following LY294002 treatment. The p27T157A showed minimal reactivity with the antibody to phosphorylated Akt substrate and none with anti-p27T157-P. Cellular p27 phosphorylation at T157 increased in normal HMECs in association with Akt activation as cells moved from G0 into G1. Moreover, T157 phosphorylation of p27 was two-fold higher in PTEN-deficient WM239 than in WM35 cells, indicating that T157 of p27 is a putative Akt site in vivo.
T157 lies within the nuclear localization signal of p27 (NLS, aa 153−166)30,
31. Our data indicate that an Akt-dependent pathway regulates p27 localization. Although the T157A mutation may facilitate p27 import irrespective of PKB, phosphorylation at this site appears to inhibit nuclear import of p27. Cells with activated Akt showed cytoplasmic p27. LY294002 treatment or expression of p27T157A in cells with activated Akt restored nuclear p27 localization. Moreover, AktDD transfection led to cytoplasmic mislocalization of p27. p27 phosphorylation by Akt impaired nuclear p27 import in vitro. Phosphorylation within or near the NLS has been shown to inhibit nuclear import of other proteins32. Thus, as for other Akt substrates p21, and some forkhead transcription factors (refs. 28,34), in cancers the constitutive p27 phosphorylation by Akt may cause a relative cytoplasmic sequestration of p27. This would limit the p27 available to bind to and inhibit cyclin E-cdk2, compromising the arrest response to antiproliferative factors such as TGF- and IL-6. In contrast, in normal HMECs, T157 phosphorylation of p27 accompanies the periodic activation of Akt and may regulate normal p27 function in early G1.* It is noteworthy that the PKB consensus sequence in p27 is imperfect and shows some species variation. While PKB may phosphorylate p27 at this site in humans, the possibility that other kinases phosphorylate T157 cannot be excluded.
PKBDD transfection increased cyclin D1 levels in the WM35PKBDD but not in 184PKBDD, but p27 was mislocalized to cytoplasm in both cell types. PKBDD mediated cytoplasmic p27 mislocalization does not result from increased cyclin D1, since cyclin D1 overexpression did not cause cytoplasmic sequestration of p27 in our assays.
Whereas many reports have shown the importance of accelerated p27 proteolysis causing reduced p27 in human cancer19, only one study showed an effect of cytoplasmic p27 on outcome in esophageal cancer20. Patient data in which p27 loss and p27 localization are analyzed together with respect to disease outcome have not been published to date. Here, we demonstrate that Akt-mediated phosphorylation of p27 in its NLS impairs its nuclear import in vitro and show that cytoplasmic p27 localization is linked to Akt activation in human breast cancer in vivo and is associated with reduced patient survival.
Three independent reports (including this study) demonstrate cytoplasmic p27 in up to 40% of primary human breast cancers in association with activated Akt (Viglietto et al. and Shin et al. in this issue). Serine 473-phosphorylated Akt was not detected in normal breast epithelium and Akt-P staining showed greater intensity in invasive than in non-invasive tumor areas within individual breast cancers (unpublished data). Tumors with uniquely nuclear p27 localization showed no Akt activation.
Although Akt can inhibit p27 gene expression by targeting the forkhead transcription factors25 and loss of PTEN may lead to accelerated p27 proteolysis24, this seems to be cell-type dependent and may require additional changes in signaling pathways. LY294002 increased p27 protein in the melanoma lines but not in normal HMECs. In HMECs and in human breast cancers, Akt activation was not always associated with p27 loss. There was no statistical correlation between cytoplasmic p27 and reduced p27 protein in primary breast cancers. Approximately the same proportion of tumors with high and low p27 scores showed cytoplasmic p27. One possible implication of this is that the processes leading to cytoplasmic p27 mislocalization and accelerated p27 proteolysis may arise independently in vivo and each may contribute to tumor progression.
For all levels of p27 staining, the presence of p27 in the tumor-cell cytoplasm was associated with reduced differentiation and lower disease-free survival (P = 0.003)* and overall survival (P = 0.05). These data have implications for the clinical application of p27 as a prognostic factor. Taking into account the presence or absence of cytoplasmic p27 may add to the prognostic significance of reduced p27 levels. Larger studies will be needed to confirm this.
Although one report suggested that p21 phosphorylation by Akt leads to its cytoplasmic sequestration34, two other groups confirmed that p21 is a Akt substrate, but did not observe cytoplasmic mislocalization of p21 upon Akt activation35,
36. p21 is expressed more sporadically and at lower levels than p27 in human breast cancers and previous studies have shown conflicting results regarding the prognostic potential of p21 (ref. 37). Although mislocalization of both p21 and p27 could theoretically cooperate to promote tumor progression and a comprehensive study of p21 and p27 levels and localization would be of value, p27 may have greater potential clinical utility in cancer prognosis.
Recent reports suggest that MAPK activation accelerates p27 proteolysis38,
39. This study, together with reports from Viglietto and Arteaga, suggest that constitutive PI3K-Akt activation contributes to oncogenesis through inhibition of nuclear p27 import and hence its cdk inhibitory function. Cytoplasmic mislocalization of p27 worsens the prognosis associated with reduced p27 levels in breast cancer supporting the relevance of these mechanisms to human tumorigenesis. Although some tumors show evidence of both accelerated p27 proteolysis and cytoplasmic localization, others show only one or the other. In human cancers, mutational activation of Ras and loss of the tumor suppressor PTEN are not infrequent21,
22. Overexpression of receptor tyrosine kinases (RTKs), such as Her2, can also activate PI3K-Akt in human breast and other cancers40. As both PI3K and MAPK are downstream of RTK-Ras, it will be of interest to determine what additional pathways direct RTK signaling to mediate either p27 proteolysis or cytoplasmic mislocalization in some breast cancers, while in others both coexist.
Akt phosphorylates proteins involved in signal transduction, apoptosis and gene expression40, and increasing evidence suggests that this pathway contributes importantly to cell-cycle regulation. Although this pathway affects multiple cell-cycle effectors, including cyclin D and p21, the present study, together with those of Arteaga and Viglietto, define a novel mechanism linking Akt activation with impaired nuclear p27 import and p27 deregulation in human cancer.
Methods
Cell culture.
Finite life span HMEC strain 184 (ref. 41) and immortalized 184AIL5R (ref. 12), MCF-7 cells42 and WM35 and WM239 lines43 were cultured as described. Cells were treated with 10 ng/ml TGF-.
Plasmids, site-directed mutagenesis and transfection.
The retroviral pBABE vector or pBABE constructs carrying the constitutively activated, hemagglutinin (HA)-tagged PKBT308D/S473D (PKBDD)26,
27 or wild-type PTEN were transfected as described26. Human wild-type p27 cDNA (p27WT) was inserted into the pIND vector. T157 of p27 was replaced with alanine (p27T157A) or aspartic acid (p27T157D) by site-directed mutagenesis using a QuickChange site-directed mutagenesis kit (Strategene, Loyola, California). The WT and mutant p27 cDNAs were then subcloned into pEYFP-C1 vector (Clontech, Palo Alto, California) encoding an N-terminal yellow-green variant of the Aequorea victoria green fluorescent protein (YFP). The YFP-p27 vectors were transfected using LipofectAMINE/PLUS reagents (Invitrogen, Carlsbad, California) according to the manufacturer's protocol. Muristerone A (MA) inducible MEKEE expression was achieved in MCF-7 cells using the 2-plasmid system from Invitrogen. In the absence of MA, cells showed basal MAPK activity. MA increased MAPK-P in 2 different MEKEE inducible lines. pRC-CMV-cyclin D1 (provided by P. Hinds) was transfected into MCF-7 cells to generate cyclin D1−overexpressing cell lines.
Intracellular localization of p27.
HMECs were EGF-depleted for 48 h and endogenous p27 detected by immunocytochemistry as described33. Nuclear cytoplasmic fractionation was carried out by digitonin-permeabilization as described44. Nuclei were isolated by centrifugation and the supernatant containing the cytosolic fraction collected. Equal cell volumes of nuclear and cytoplasmic lysates were assayed for p27 by immunoblotting. The nuclear protein, RCC1, was blotted as a fractionation control. YFP-tagged p27 expression was detected by direct fluorescence microscopy of transfected cells. Photographs were taken at  400 magnification using an Empix digital camera and 'CoolSnap' (Photomajics, Pittsfield, Massachusetts) software.
Nuclear import assays.
Import assays were carried out as described44. MCF-7 cells were digitonin permeabilized and isolated nuclei incubated with cytosolic proteins (4 g/l), an ATP-regenerating system and recombinant his-tagged WT, T157A or T157D p27 for 60 min at 21 °C. Nuclear and supernatant fractions were then separated by centrifugation and immunoblotted for His-p27. Where indicated, His-p27WT was reacted in vitro with cellular Akt before import assays. Pre-incubation with wheat germ agglutinin (200 g/ml) abolished active p27 import.
Flow cytometry.
Cells were pulse-labeled with 10 M bromodeoxyuridine (BrdU), stained with anti-BrdU-conjugated FITC (Becton Dickinson, Mountain View, California) and propidium iodide and cell-cycle data acquired as described12.
Recombinant protein, affinity purification and kinase assays.
The p27WT, p27T157A and p27T157D cDNAs were subcloned into pET28a. Recombinant His-p27 was purified on nickel-agarose beads. An antibody against serine 473 (Ser473) phosphorylated Akt (ref. 26) was used to assay Akt activation by immunoblotting. For assays of Akt activity, Akt was immunoprecipitated from 1 mg cell lysates and reacted with GSK-3- as substrate using an Akt kinase kit (New England Biolabs, Beverly, Massachusetts). Akt kinase assays were also carried out using either 5 g histone-H2B or 10 g recombinant His-p27 as substrates as described45. Cyclin E1-dependent kinase activity was assayed and quantified as described9.
Antibodies and immunoblotting.
Antibodies to Akt, GSK-3-, phospho-Akt, phospho-GSK3- (Thr21) were obtained from New England Biolabs; to p27 (C-19) and cdk2 from Santa Cruz Biotechnology (San Cruz, California); to p27 (DCS72) from Neomarkers (Fremont, California); to p27 from Transduction Labs (Lexington, Kentucky); to -actin from Sigma (Oakville, Ontario). Cyclin E1 antibodies (mAbs E12 and E172) were obtained from E. Harlow. Antiphospho-Akt-substrate antibody was from New England Biolabs. The antibody specific for T157-phosphorylated p27 was generated and provided by G. Viglietto. Cells were lysed in ice-cold NP-40 lysis buffer and immunoblotted as described12.
Patient population and statistical methods.
The population studied was a group of 128 patients who underwent surgery for non-metastatic primary breast cancer between 1986 and 1992 at the Sunnybrook Health Sciences Center. This study was approved by the Research Ethics Board of the hospital. Kaplan-Meier survival and disease free survival curves were generated using nuclear p27 score and nuclear versus nuclear and cytoplasmic p27 as strata. Log-rank, Wilcoxon and -2log (LR) were used to assess significance. The association between discrete variables was tested using the 2 test.
Immunohistochemistry.
Paraffin sections of tumor blocks were stained for p27 as described33 using monoclonal p27 antibody (Transduction Labs, Lexington, Kentucky) diluted 1:1000 (0.25 g/ml) in PBS or for phospho-Akt using the phospho-Akt (Ser473) antibody (NEB) diluted 1:200. Sections were counterstained with hematoxylin. The degree and localization of p27 staining was scored independently by two pathologists (C.C. and J.Z.) and J.M.S. as described33. Tumors showing both nuclear and cytoplasmic or exclusively cytoplasmic p27 in at least 35% of cells were scored 'N+C'. Tumors with exclusively nuclear p27 were scored as nuclear only (N only). Phospho-Akt was scored by J.Z. and J.M.S.
Note: Supplementary information is available on the Nature Medicine website.
Received 17 April 2002; Accepted 14 August 2002; Published online: 16 September 2002.
REFERENCES
- Donovan, J. & Slingerland, J. Transforming growth factor- and breast cancer: Cell cycle arrest by transforming growth factor- and its disruption in cancer. Breast Cancer Res. 2, 116–124 (2000). | Article | PubMed | ISI | ChemPort |
- Kretzschmar, M. Transforming growth factor- and breast cancer: Transforming growth factor-/SMAD signaling defects and cancer. Breast Cancer Res. 2, 107–115 (2000). | Article | PubMed | ISI | ChemPort |
- Sherr, C.J. G1 phase progression: Cycling on cue. Cell 79, 551–555 (1994). | Article | PubMed | ISI | ChemPort |
- Sherr, C.J. & Roberts, J.M. CDK inhibitors: Positive and negative regulators of G1-phase progression. Genes Dev. 13, 1501–1512 (1999). | PubMed | ISI | ChemPort |
- LaBaer, J. et al. New functional activities for the p21 family of CDK inhibitors. Genes Dev. 11, 847–862 (1997). | PubMed | ISI | ChemPort |
- Cheng, M. et al. The p21(Cip1) and p27(Kip1) CDK 'inhibitors' are essential activators of cyclin D-dependent kinases in murine fibroblasts. EMBO J. 18, 1571–1583 (1999). | Article | PubMed | ISI | ChemPort |
- Koff, A., Ohtsuki, M., Polyak, K., Roberts, J.M. & Massague, J. Negative regulation of G1 in mammalian cells: Inhibition of cyclin E-dependent kinase by TGF-. Science 260, 536–539 (1993). | PubMed | ISI | ChemPort |
- Polyak, K. et al. p27Kip1, a cyclin-Cdk inhibitor, links transforming growth factor- and contact inhibition to cell cycle arrest. Genes Dev. 8, 9–22 (1994). | PubMed | ISI | ChemPort |
- Slingerland, J.M. et al. A novel inhibitor of cyclin-Cdk activity detected in transforming growth factor -arrested epithelial cells. Mol. Cell. Biol. 14, 3683–3694 (1994). | PubMed | ISI | ChemPort |
- Hannon, G.J. & Beach, D. p15INK4B is a potential effector of TGF--induced cell cycle arrest. Nature 371, 257–261 (1994). | Article | PubMed | ISI | ChemPort |
- Reynisdottir, I., Polyak, K., Iavarone, A. & Massague, J. Kip/Cip and Ink4 Cdk inhibitors cooperate to induce cell cycle arrest in response to TGF-. Genes Dev. 9, 1831–1845 (1995). | PubMed | ISI | ChemPort |
- Sandhu, C. et al. Transforming growth factor stabilizes p15INK4B protein, increases p15INK4B -cdk4 complexes and inhibits cyclin D1/cdk4 association in human mammary epithelial cells. Mol. Cell Biol. 17, 2458–2467 (1997). | PubMed | ISI | ChemPort |
- Florenes, V.A. et al. Interleukin-6 dependent induction of the cyclin dependent kinase inhibitor p21WAF1/C1P1 is lost during progression of human malignant melanoma. Oncogene 18, 1023–1032 (1999). | Article | PubMed | ISI | ChemPort |
- Bouchard, C. et al. Direct induction of cyclin D2 by Myc contributes to cell cycle progression and sequestration of p27. EMBO J. 18, 5321–5333 (1999). | Article | PubMed | ISI | ChemPort |
- Cheng, M., Sexl, V., Sherr, C.J. & Roussel, M.F. Assembly of cyclin D-dependent kinase and titration of p27Kip1 regulated by mitogen-activated protein kinase kinase (MEK1). Proc. Natl. Acad. Sci. USA 95, 1091–1096 (1998). | Article | PubMed | ChemPort |
- Perez-Roger, I., Kim, S.H., Griffiths, B., Sweing, A. & Land, H. Cyclins D1 and D2 mediate Myc-induced proliferation via sequestration of p27Kip1 and p21Cip1. EMBO J. 18, 5310–5320 (1999). | Article | PubMed | ISI | ChemPort |
- Warner, B.J., Blain, S.W., Seoane, J. & Massague, J. Myc downregulation by transforming growth factor required for activation of the p15INK4B G1 arrest pathway. Mol. Cell Biol. 19, 5913–5922 (1999). | PubMed | ISI | ChemPort |
- Vlach, J., Hennecke, S., Alevizopoulos, K., Conti, D. & Amati, B. Growth arrest by the cyclin-dependent kinase inhibitor p27Kip1 is abrogated by c-Myc. EMBO J. 15, 6595–6604 (1996). | PubMed | ISI | ChemPort |
- Slingerland, J. & Pagano, M. Regulation of the cdk inhibitor p27 and its deregulation in cancer. J. Cell Physiol. 183, 10–17 (2000). | Article | PubMed | ISI | ChemPort |
- Singh, S.P. et al. Loss or altered subcellular localization of p27 in Barrett's associated adenocarcinoma. Cancer Res. 58, 1730–1735 (1998). | PubMed | ISI | ChemPort |
- Bos, J.L. Ras oncogenes in human cancer: A review. Cancer Res. 49, 4682–4689 (1989). | PubMed | ISI | ChemPort |
- Di Cristofano, A. & Pandolfi, P.P. The multiple roles of PTEN in tumor suppression. Cell 100, 387–390 (2000). | Article | PubMed | ISI | ChemPort |
- Diehl, J.A., Cheng, M., Roussel, M.F. & Sherr, C.J. Glycogen synthase kinase-3 regulates cyclin D1 proteolysis and subcellular localization. Genes Dev. 12, 3499–3511 (1998). | PubMed | ISI | ChemPort |
- Sun, H. et al. PTEN modulates cell cycle progression and cell survival by regulating phosphatidylinositol 3,4,5,trisphosphate and Akt/protein kinase B signaling pathway. Proc. Natl. Acad. Sci. USA 96, 6169–6204 (1999).
- Medema, R.H., Kops, G.J., Bos, J.L. & Burgering, B.M. AFX-like Forkhead transcription factors mediate cell-cycle regulation by Ras and PKB through p27kip1. Nature 404, 782–787 (2000). | Article | PubMed | ISI | ChemPort |
- Alessi, D.R. et al. Mechanism of activation of protein kinase B by insulin and IGF-1. EMBO J. 15, 6541–6551 (1996). | PubMed | ISI | ChemPort |
- Wang, Q. et al. Protein kinase B/Akt participates in GLUT4 translocation by insulin in L6 myoblasts. Mol. Cell Biol. 19, 4008–4018 (1999). | PubMed | ISI | ChemPort |
- Muslin, A.J. & Xing, H. 14-3-3 proteins: Regulation of subcellular localization by molecular interference. Cell Signal 12, 703–709 (2000). | Article | PubMed | ISI | ChemPort |
- Obata, T. et al. Peptide and protein library screening defines optimal substrate motifs for AKT/PKB. J. Biol Chem. 275, 36108–36115 (2000). | Article | PubMed | ISI | ChemPort |
- Reynisdottir, I. & Massague, J. The subcellular locations of p15(Ink4b) and p27(Kip1) coordinate their inhibitory interactions with cdk4 and cdk2. Genes Dev. 11, 492–503 (1997). | PubMed | ISI | ChemPort |
- Zeng, Y., Hirano, K., Hirano, M., Nishimura, J. & Kanaide, H. Minimal requirements for the nuclear localization of p27Kip1, a cyclin-dependent kinase inhibitor. Biochem. Biophys. Res. Commun. 274, 37–42 (2000). | Article | PubMed |
|
