Subcellular localization, modification and protein complex formation of the

cdk-inhibitor p16 in Rb-functional and Rb-inactivated tumor cells

Kristina Nilsson and G€oran Landberg*

Division of Pathology, Department of Laboratory Medicine, Lund University, Malm€o University Hospital, S-205 02,Malm€o, Sweden

The cdk-inhibitor p16 is a tumor suppressor gene that is inacti-vated in many forms of cancer. Despite numerous studies, theexact mechanism of regulation of p16 has not been clarified, al-though the status of retinoblastoma (Rb) seems to be one importantfactor that influences the p16 expression. The specificity and valid-ity of cytoplasmic localization of p16 observed in some tumors hasfurther been questioned. Here, by subcellular fractionation of Rb-functional and Rb-inactivated cell lines, we show that p16 indeed isexpressed in the cytoplasm as well as in the nucleus. Post transla-tional modifications of p16 in different subcellular compartments aswell as its capacity to form protein complexes were further de-lineated. Two dimensional gel electrophoresis showed that twoforms of p16 appeared in the cytoplasm, while only one form wasdetected in the nucleus. Samples of basal cell carcinoma and squa-mous cell carcinoma of the skin with either functional or non-func-tional Rb also exhibited at least two forms of p16. In addition, cyto-plasmic p16 bound cyclin dependent kinase (cdk)4/6, potentiallyindicating that p16 could have a function in the cytoplasm.' 2005 Wiley-Liss, Inc.

Key words: p16; retinoblastoma gene product; two-dimensional gelelectrophoresis

Progression of a cell through the cell cycle is tightly regulatedby several control mechanisms involving positive and negativeregulators of the cell cycle. Uncontrolled regulation of these con-trol mechanisms leads to dysfunctional cell cycle control, uncon-trolled cell division and eventually tumor formation.1 Central inthe late G1 phase checkpoint is the retinoblastoma (Rb) pathway.In G1 phase, Rb binds to transcription factors such as E2Fs, pre-venting genes necessary for S-phase entry to be transcribed. Uponmitogen stimulation, there is an upregulation of cyclin D whichbinds to cyclin dependent kinases (cdks) 4/6 and the complex initi-ates the phosphorylation of Rb, which becomes inactive. Thisleads to release of E2Fs and transcription of genes that are neededfor the cell to continue into S-phase.2 The cdk-inhibitor p16, withthe major function to block the cell cycle, is a tumor suppressorgene that is inactivated in many cancer forms.3 When p16 in theG1-phase of the cell cycle binds to cdk4/6, the active cyclin D-cdk4/6 complex cannot be formed and phosphorylation of Rb isthereby prevented causing inhibition of the cell cycle.4,5

Small homozygous deletions, point mutations and methylationof the promoter are mechanisms that inactivate p16.3 When inacti-vated, there is, in some cases, a loss of p16 protein expression intumor cells as observed by immunohistochemistry.6 In contrast,p16 is overexpressed in some tumors, commonly explained as sec-ondary to Rb-inactivation.4,7 The exact mechanism inducing p16in Rb-inactivated tumors has not been elucidated, but both directtranscriptional effects and protein accumulation as a result ofmany population doublings have been proposed.8

In tumors, many of the p16 mutations map to the cdk-bindingregions of the protein, emphasizing the importance of cdk-inhibi-tion of p16 in tumor suppression.9 Mutations that do not affect thecdk-binding parts of p16 have also been observed in tumors butthe functional consequences for these mutations have not beenfully clarified. In a previous study, it was discovered that theregions affected by these mutations harbor phosphorylation sitesof p16 and phosphorylated p16 was shown to be associated withcdk4.10 In contrast, two other studies did not observe any phos-phorylation of p16.11,12 Further, the subcellular localization of p16

has been debated. It is generally believed that p16 exerts its cdk-inhibitory function in the nucleus. However, to our knowledge,there is no direct proof for this, although some studies report amainly nuclear localization of p16 in normal cells.13,14 In contrast,many studies have reported a cytoplasmic localization of p16 intumor cells, but the specificity of this expression has been ques-tioned.15 Moreover, in studies where the normal localization ofp16 is considered to be evenly throughout the cell, some variantsof mutated p16 have been shown to be mislocalized to either thenucleus or the cytoplasm.16–19

In this study, we aimed to investigate the subcellular localiza-tion and post translational modifications of p16 in Rb-functionaland Rb-inactivated cells as well as complex formation of p16. Toverify if p16 is expressed in the cytoplasm, subcellular fractions ofcells were analyzed for p16 expression. Post translational modifica-tions of p16 were determined by two-dimensional (2D) gel electro-phoresis and the presence of protein complex by gel filtration andimmunoprecipitation. The results showed that p16 can be expressedin the cytoplasm, and both nuclear and cytoplasmic p16 formedcomplexes with cdks. Nuclear p16 appeared predominantly in oneform, while cytoplasmic p16 displayed two forms. We further vali-dated the p16 expression and modifications using primary tumors ofthe skin showing similar findings as for the cell lines.

Material and methods

Cell culturing

BT-549 (American Type Culture Collection (ATCC), Rockville,MD) was grown in Dulbecco’s Modified Eagle’s Medium(DMEM) with 4.5 g/l glucose supplemented to contain 10% FCS,10 mM Hepes and 5.2 g/l sodium bicarbonate. MDA-MB-468(ATCC) was grown in RPMI 1640 supplemented to contain 10%FCS, 1 mM sodium pyruvate and 1.5 g/l sodium bicarbonate. SCC-4 (ATCC) was grown in DMEM/HAM0S-F12 supplemented tocontain 10% FCS and 3 g/l sodium bicarbonate. The breast cancercell lines BT-549 and MDA-MB-468 are Rb-inactivated cell lineswith a partially deleted Rb gene 20 and SCC-4 is a squamous cellcarcinoma cell line with functional Rb.21 BT-549 and SCC-4 haveno p16 mutations.22,23 All cells were grown at 37�C in 5% CO2.

Paraffin embedding and immunohistochemical staining of cells

Cells were harvested, fixed in 4% paraformaldehyde for 30 minand stained with Mayer’s hematoxylin. The paraformaldehydewas removed, 70% ethanol was added and cells were incubatedover night. The cells were dehydrated using increasing concentra-tions of ethanol and finally xylene. After dehydration, the cellswere embedded in paraffin. Paraffin sections of 4 lm were de-par-

Grant sponsors: Swedish Cancer Society; Swedish Research Council;Gunnar, Arvid and Elisabeth Nilsson Cancer Foundation; Per-Eric and UllaSchyberg Foundation; Lund University Research Funds; Malm€o UniversityHospital Research and Cancer Funds; Swegene/Wallenberg ConsortiumNorth.*Correspondence to: Division of Pathology, Department of Laboratory

Medicine, Lund University, Malm€o University Hospital, S-205 02, Malm€o,Sweden. Fax:146-40-337063. E-mail: goran.landberg@med.lu.seReceived 14 January 2005; Accepted after revision 8 July 2005DOI 10.1002/ijc.21466Published online 13 September 2005 in Wiley InterScience (www.

interscience.wiley.com).

Int. J. Cancer: 118, 1120–1125 (2006)' 2005 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

affinized using xylene and rehydrated using descending concentra-tions of ethanol according to standard protocol. For anti-humanp16 antibody (1:200, BD Pharmingen, San Jose, CA), antigenretrieval was achieved by microwave heating in 10 mM citratebuffer, pH 6.0. Staining was performed in a DakoTechmate 500(Dako A/S, Glostrup, Denmark) according to the manufacturer’sinstructions.

Subcellular fractionation

For subcellular fractionation, cells were pelleted and suspendedin Nuclei buffer (10 mM Hepes, pH 7.9, 1.5 mM MgCl2, 10 mMKCl, 1% Triton X-100, 0.5 mM DTT), kept on ice for 10 min andcentrifuged at 10000g at 4�C for 10 min. The supernatants (cyto-plasmic fractions) were stored at 280�C and the pellets were sus-pended in High Salt Buffer (10 mM Hepes, pH 7.9, 400 mM NaCl,0.1 mM EDTA, 5% glycerol, 0.5 mM DTT), kept on ice for20 min and centrifuged at 10000g at 4�C for 10 min. The superna-tants (nuclear fractions) were stored at 280�C.

Western blotting

For western blotting, 25 lg of protein was separated on a 12%SDS-PAGE and transferred to Hybond ECL nitrocellulose mem-branes (Amersham Biosciences, Little Chalfont, UK). The mem-branes were incubated with primary anti-human p16 (1:500, BDPharmingen) or anti-human PARP (1:5,000, BIOMOL ResearchLaboratories, Plymouth Meeting, PA) antibodies followed by per-oxidase-conjugated anti-mouse (1:5,000, Amersham Biosciences)antibody. The proteins were detected using enhanced chemilumi-nescence detection system plus (ECL) reagent (Amersham Bio-sciences) according to the manufacturer’s instructions andexposed on ECL Hyper Film (Amersham Biosciences).

Gelfiltration

Whole cell protein extracts were prepared by lysing cells inlysis buffer (150 mM NaCl, 50 mM Tris, pH 7.6, 0.1% NP-40)supplemented with the protease inhibitor cocktail Complete Mini(Roche, Mannheim, Germany) and PMSF. After incubation on icefor 20 min, the lysates were centrifuged at 14000g at 4�C for10 min and the supernatants were stored at 280�C. Gelfiltrationwas performed using a Bio-Gel A-0.5 m column and a BioLogicLP System (Bio-Rad Laboratories, Hercules, CA). Protein samples(3–5 mg) were loaded onto the column together with Gel FiltrationStandard (Bio-Rad Laboratories) and separated in TBS (10 mMTris-HCl, 150 mM NaCl, pH 7.5) at a flow rate of 0.25 ml/min.After 30 min, collection was started and each fraction was col-lected for 2 min. A total of 72 fractions were collected, and a sub-set was used for Western blotting, as shown in figure 3a. Thirtymicroliters of each fraction was separated on an SDS-PAGE, andWestern blotting with anti-human p16 or anti-human cdk4(1:1,000, Chemicon International, Temecula, CA) antibodies wasperformed as described earlier. For immunoprecipitation, 4 frac-tions were pooled and concentrated using a centricon YM-10(Millipore, Billerica, MA). Samples were diluted in lysis bufferand immunoprecipitation done as described later.

Immunoprecipitation and immunodepletion

Immunoprecipitation was carried out with 500 lg of proteinand 3 lg of anti-human p16 (Alexis Biochemicals, Lausen, Swit-zerland) or anti-human cdk4 (Santa Cruz Biotechnology, SantaCruz, CA) antibodies. For immunoprecipitation of cdk6, 7.5 ll ofanti-human cdk6 antibody (BD Pharmingen) was used. As nega-tive controls, 3 lg of anti-human Lamin B (Santa Cruz Biotech-nology) or anti-mouse IgG (Amersham Biosciences) antibodieswere used. After incubation with protein G sepharose (AmershamBiosciences) at 4�C for 2 hr, the samples were washed 3 timeswith lysis buffer and boiled in 53 sample buffer for 10 min. Sepa-ration of the samples on SDS-PAGE and Western blotting withanti-human p16, anti-human cdk4 and anti-human cdk6 (1:10,000,BD Pharmingen) antibodies was done as described earlier. Immu-

noprecipitations that were to be analyzed by 2D gel electrophore-sis were carried out as described earlier but with 1,000 lg of pro-tein. Samples were diluted in 220 ll of rehydration buffer (seelater), vortexed for 20 min and then loaded onto ReadyStripimmobilized pH gradient (IPG) Strip, pH 4–7 (Bio-Rad Laborato-ries) and treated as described later. For immunodepletion, 25 ll ofthe supernatants from the immunoprecipitations were analyzed.

Two dimensional gel electrophoresis

For 2D gel electrophoresis, cells were suspended in ReadyPrepSequential Extraction Kit Reagent 3 (Bio–Rad Laboratories), vor-texed for 5 min and centrifuged at 19000g for 10 min at room tem-perature. The supernatants were stored at 280�C. 150–200 lg ofprotein was diluted in rehydration buffer (8.3 M Urea, 2% CHAPS,32 mM DTT, 0.2% Biolytes 3/10, 0.001% bromophenol blue),loaded onto a ReadyStrip IPG Strip, pH 4–7 (Bio-Rad Laborato-ries), and isoelectric focusing was carried out using a PROTEANIEF cell (Bio-Rad Laboratories) according to the manufacturer’sinstructions. Before separating the focused samples by SDS-PAGE,the strips were equilibrated in equilibration buffer I (6 M Urea, 2%SDS, 0.375 M Tris-HCl, pH 8.8, 20% glycerol, 130 mM DTT) andequilibration buffer II (6 M Urea, 2% SDS, 0.375 M Tris-HCl,pH 8.8, 20% glycerol, 135 mM iodoacetamide). After separation,the proteins were transferred to Hybond ECL nitrocellulose mem-brane (Amersham Biosciences) and detection of p16 by Westernblotting was done as described earlier.

Tumor materials

Basal cell carcinomas and an invasive squamous cell carcinomaof the skin were frozen in liquid nitrogen after surgery. Sectionswere cut and stained with hematoxylin and the tumors were eithermicrodissected, selecting tumor areas with different growth prop-erties, as described before 24 or examined as a whole. Dissectedareas were mechanically disrupted in ReadyPrep SequentialExtraction Kit Reagent 3 (Bio-Rad Laboratories) using scissors,vortexed for 7 min and lysates were clarified by centrifugation at19000g for 10 min at room temperature. Lysates were diluted inrehydration buffer and 2D gel electrophoresis was performed asdescribed earlier. For immunohistochemical staining, paraffin sec-tions of 4 lm from two of the tumors were deparaffinized usingxylene and rehydrated using descending concentrations of ethanolaccording to standard protocol. For the antibodies anti-humanp16INK4a (1:200, BD Pharmingen) and anti-human phospho-Rb(1:150, Cell Signaling Technology, Beverly, MA), antigen re-trieval was achieved by microwave heating in 10 mM citrate buf-fer, pH 6.0. Staining was performed in a DakoTechmate 500(Dako) according to the manufacturer’s instructions.

Results

Cytoplasmic p16 staining has been observed in differenttumors, but also been argued and often classified as unspecificstaining. To better characterize the potential localization of p16 inthe cytoplasm, we made subcellular fractionation of cells from threedifferent cell lines showing either both nuclear and cytoplasmicstaining or mainly cytoplasmic staining of p16 by immuno-histochemistry. As illustrated in figure 1a, Western blot analysis ofthe fractions showed a band corresponding to p16 both in the cyto-plasmic and the nuclear fractions. The Western blot results furthermimicked the immunohistochemistry staining of p16 (Fig. 1b)confirming a presence of p16 both in the cytoplasm as well as inthe nucleus of some tumor cell lines. In accordance with that inthe literature,4,7 the total level of p16 was considerably higher inthe Rb-inactivated cell lines compared to the cell line with func-tional Rb (Fig. 1a and 1b).

In two previous studies, p16 appeared in two or more forms whenanalyzed by 2D gel electrophoresis.10,12 In this study, 2D gel elec-trophoresis of whole cell extracts of the three different cell linesconfirmed the presence of two major populations of p16 (Fig. 2a).

1121SUBCELLULAR LOCALIZATION OF p16 IN TUMOR CELLS

The more acidic form (closest to the anode, 1) represented,according to Gump et al., a phosphorylated form of p16, whereasthe basic form (closest to the cathode, 2) was unphosphorylatedp16.10 The relation between the two p16 forms varied slightlybetween the cell lines and in the Rb-inactivated cell line BT-549,which has the highest expression of p16, additional populationscould be observed besides the two distinct forms. When analyzingthe nuclear and cytoplasmic fractions of the cell lines BT-549 andMDA-MB-468 using 2D gel electrophoresis, the more acidic formwas observed both in the nucleus and in the cytoplasm, while thebasic form predominantly was observed in the cytoplasmic fraction(Fig. 2b). This experiment was not performed with the cell lineSCC-4, since this cell line expresses p16 only in the cytoplasm, asshown in figure 1a and 1b. In both, previous studies demonstratingat least two forms of p16, it has been shown that the more acidicforms of p16 bind to cdks.10,12 However, when analysing p16 andcdk4 immunoprecipitations (IPs) from BT-549 by 2D gel electro-phoresis, all forms of p16 bound to cdk4 (Fig. 2c).

The cdk-inhibitor p16 can form complexes with cdk4 and cdk6and we wanted to characterize the presence of potential differentp16 protein complexes in the various subcellular fractions. Proteincomplexes were separated according to native size by gel filtrationand in the subsequent Western blot of filtrated whole cell extractof BT-549, we observed that p16 was separated into two differentportions of respectively 37–50 kDa and 17 kDa (Fig. 3a), which isin line with another report.25 When separating the nuclear and thecytoplasmic fractions from the same cell line by gelfiltration, p16appeared in roughly the same fractions.

To characterize the protein content of the p16 complexes sepa-rated by gel filtration, we used the p16 containing fractions fromwhole cell extract for IP. p16-cdk4/6 complexes are about 45–55 kDa 25 and as is shown in figure 3b, the 37–50 kDa peak con-tained p16 bound to cdk4 while p16, as expected, was not boundto cdk4 in the 17 kDa peak. The binding of p16 to cdk6 was notanalyzed because of very low levels of cdk6 in the cell line usedfor the experiment.

According to the results from the gel filtration experiment andIP, both nuclear and cytoplasmic p16 seemed to be bound to cdk4.To further validate this, we performed p16 and cdk4 IP usingnuclear and cytoplasmic extracts of BT-549 and MDA-MB-468.The results clearly showed that p16 was bound to cdk4 in bothsubcellular compartments (Fig. 3c), although immunodepletionindicated that a portion of p16 was not (Fig. 3d). IP of whole cellextract from SCC-4 showed that p16 did not bind to cdk4 (data

not shown), but as is shown in figure 3c, p16 bound to cdk6 in thecytoplasm of SCC-4.

To verify if the observed forms of p16 and localization to differ-ent subcellular compartments was present not only in cell lines, weanalyzed a couple of microdissected in vivo tumors with 2D gelelectrophoresis. We used a basal cell carcinoma with a functionalRb-pathway determined by a phospho-specific Rb-antibody, show-ing nuclear staining, and a squamous cell carcinoma of the skinlacking nuclear staining and consequently having a nonfunctionalRb-pathway (Fig. 4). Interestingly, both the basal cell carcinomaand the squamous cell carcinoma contained both forms of p16 asillustrated in figure 4. p16 was also present in both the cytoplasmand the nucleus in both tumors as determined by immunohistochem-istry (Fig. 4). To investigate if a particular form of p16 is expressedin a certain tumor area or tumor type, we analyzed p16 in a subsetof basal cell carcinomas by 2D gel electrophoresis. Tumors that dis-played areas with different growth properties were microdissectedbefore analysis. As is illustrated in figure 5, the relation between thedifferent forms of p16 varied but the basic form of p16 seemed tobe most expressed in the analyzed primary tumors. Further, therewas no major difference between the growth properties of the basalcell carcinomas and the expression of the various p16 forms.

FIGURE 1 – p16 expression in cell lines. (a) Western blot showingthe p16 expression in nuclear (nucl) and cytoplasmic (cyt) fractions inRb-functional (SCC-4) and Rb-inactivated (BT-549 and MDA-MB-468) cell lines. PARP was used as a nuclear marker. (b) Immunohisto-chemical p16 staining showing the same subcellular localization ofp16 as the Western blot in (a).

FIGURE 2 – 2D gel electrophoresis followed by Western blot of p16in cell lines. (a) 2D gel electrophoresis of whole cell lysates fromthree different cell lines. Two major populations of p16 were observedin both Rb-functional and Rb-inactivated cell lines. (b) 2D gel electro-phoresis analysis of nuclear and cytoplasmic fractions of BT-549 andMDA-MB-468. The nuclear fraction of the cell lines contained mainlythe more acidic form (closest to the anode, 1) of p16, whereas thecytoplasmic fraction included both populations. (c) 2D gel electropho-resis analysis of p16 and cdk4 immunoprecipitations using whole cellextract from BT-549.

1122 NILSSON AND LANDBERG

Discussion

The presence of p16 in the cytoplasm in tumor cells has beenobserved and reported in many studies, but the validity and rele-vance of this localization of the cdk-inhibitor has been ques-tioned.15 Cytoplasmic p16 staining of tumor cells has further beenshown to be of prognostic value. In primary breast carcinoma, astrong cytoplasmic expression of p16 was associated with a highlymalignant phenotype,26 whereas in malignant melanoma, no asso-ciation between cytoplasmic p16 and patient outcome was found.27

In another study, high p16 expression both in the nucleus and inthe cytoplasm of mammary carcinoma cells was associated withunfavorable prognostic parameters.28 It is nevertheless importantto establish whether p16 can be localized to the cytoplasm in sometumor cells, so as to avoid studies potentially reporting cross reac-tivities for the p16-antibodies or even ignoring cytoplasmic stain-ing in sub groups of tumors. Here, we show by subcellular fractio-nations that p16 indeed is expressed in the cytoplasm in several celllines, regardless of the Rb-status of the cell line. This is in line withearlier indications, and we conclude that p16 can be expressed andlocalized to the cytoplasm in certain malignancies.

Other cell cycle inhibitors with a main effect in the nucleushave also been shown to be localized to the cytoplasm. The cdk-inhibitor p21 has separate functions in the cytoplasm compared tothe nucleus 29 and p53 can be inactivated by a cytoplasmic local-ization.30 In some breast cancer samples, p27 is inhibited from

cdk2 binding by a phosphorylation of the nuclear localization sig-nal region hindering nuclear import and causing a cytoplasmiclocalization of the protein.31 So far, no direct evidence for a func-tion of p16 in the cytoplasm has been reported. Possibly, this cyto-plasmic localization can be a means of inactivating p16 or it maybe secondary to mutations and inactivation of p16, as discussedlater. We have previously reported an upregulation of p16, fol-lowed by a proliferative decrease, in infiltrative parts of basal cellcarcinoma.24 These results support a functional Rb-pathwayin basal cell carcinomas and further links p16 and ceased prolifer-ation to infiltrative properties. In squamous cell carcinoma of theskin, we observed a similar phenomenon, with a clear p16 upregu-lation in infiltrative cells. In these tumors however, the Rb-path-way is disrupted either by an inactivation of Rb or by a delocali-zation of p16 to the cytoplasm.32 These data, where p16 isincreased in infiltrative cells regardless of its subcellular localiza-tion, indicate a possible function for cytoplasmic p16 in infiltrativebehaviors.

Gump et al. has shown that p16 can be phosphorylated.10 Fourdifferent phosphorylation sites were identified on exogenouslyexpressed p16, and phosphorylation of serine 152 was found onendogenous p16 and was further linked to cdk4 binding properties.Phosphorylation of p16 was nevertheless not necessary for theinteraction between p16 and cdk4.10 In contrast, two other studiescould not identify p16 as a phospho protein,11,12 but one reportedthe acidic form of p16 as responsible for cdk-binding.12 In line with

FIGURE 3 – Complex formation of p16. (a) Gelfiltration of whole cell lysate of BT-549 followed by p16 and cdk4 Western blot analyses ofthe different fractions. p16 was separated into two major portions corresponding to 37-50 kDa respectively 17 kDa. Approximate elution peaksof gel filtration standards are indicated in the figure. (b) Immunoprecipitation (IP) of p16 followed by Western blot (WB) of p16 and cdk4 indifferent fractions from the gelfiltration experiment, indicating that the 37-50 kDa complex of p16 contained cdk4. (c) Immunoprecipitation ofp16 and cdk4 in the nuclear (nucl) and cytoplasmic (cyt) fractions of BT-549 and MDA-MB-468 as well as of p16 and cdk6 in the cytoplasmicfraction of SCC-4. Lamin B and IgG were used as negative controls. (d) Immunodepletion of cdk4 in the nuclear and cytoplasmic fractions ofBT-549 and MDA-MB-468.

1123SUBCELLULAR LOCALIZATION OF p16 IN TUMOR CELLS

previous reports, we observed two separate p16 populations by 2Dgel electrophoresis of several cell lines. To investigate the role thestatus of Rb has on the p16 expression, we included Rb-functionalas well as Rb-inactivated cell lines, but no obvious differences werefound by 2D gel electrophoresis. When analyzing p16 in the differ-ent subcellular compartments, cytoplasmic p16 was present in twoforms while nuclear p16 seemed to be predominantly representedby one form. These findings indicate one possible mechanismthat may influence the subcellular localization of p16. In contrastto previous studies, we found that all forms of p16 bound to cdk4.This discrepancy may be due to different experimental set ups andcell lines analyzed. The results from the 2D gel electrophoresisand IP experiments suggested that the nuclear as well as the cyto-plasmic p16 could bind to cdk4. By gel filtration, we next con-firmed that nuclear and cytoplasmic p16 interacted with other pro-teins and by immunoprecipitation, we showed that p16 in bothsubcellular compartments bound to cdk4/6. The different popula-tions of p16 observed by 2D gel electrophoresis were not onlyobserved in cell lines but also in two different types of skin tumorsunderscoring the generality of this post translational modificationof an important tumor suppressor. Whether the variation ofthe p16 forms observed between different tumors or areas withinthe same tumor is of significance or not needs to be further investi-gated.

The binding to cdk4/6 in the cytoplasm may be an additionalmechanism for p16 to inactivate the kinase and inhibiting it fromforming complexes with cyclin D. One possibility is that p16binds to cdk4/6 in the nucleus and the complex is then transportedto the cytoplasm. This will add another dimension to the p16inhibitory effect of cdk4/6 that might be of importance in theunderstanding of how cdk-inhibitors can block the cell cycle.Cdk4/6 might also have additional roles in the cytoplasm, whichcould be affected by p16. Cdk6 has been implicated in functionsin the cytoplasm33 and an undefined substrate for cdk4 and 6 thatis more prevalent in the cytoplasm than in the nucleus has beenfound.34 The localization of p16 in the cytoplasm might further bea mechanism to inactivate p16 by preventing p16 inhibitory activ-ities in the nucleus despite the cdk4/6 binding in the cytoplasm, asshown earlier. Future studies have to reveal if this is the case, butsimilar findings have been observed for other important tumorsuppressor genes.30,31

Another reason for the cytoplasmic localization of p16 may bethat the gene is mutated and the defect protein is localized to thecytoplasm. Some studies have also shown that various p16mutants are expressed predominantly in the cytoplasm16–19 and itis most likely that a fraction of the tumors with cytoplasmic p16can be explained by p16 mutations. There are nevertheless celllines with wild type p16,22 where p16 is expressed in both the

FIGURE 4 – Analysis of p16 by 2D gel electrophoresis and Western blot of samples of squamous cell carcinoma of the skin and basal cell car-cinoma. Protein extracts from microdissected areas of a squamous cell carcinoma resection, including (a) normal tissue, (b) invasive parts and(c) cancer in situ as well as extract from (d) a basal cell carcinoma were separated by 2D gel electrophoresis and Western blotted using p16 anti-bodies. The right panels show immunohistochemical staining of p16 (e and g) and phosphorylated Rb (f and h) in respective tumor.

FIGURE 5 – Analysis of p16 by 2D gel electrophoresis and Westernblot of five basal cell carcinomas (BCC 1-BCC 5). Three of the tumorswere microdissected, and areas with different growth properties wereselected.

1124 NILSSON AND LANDBERG

nucleus and the cytoplasm, as shown here. At least one study hasfurther shown that cytoplasmic p16 observed in tumor-derived celllines was not linked to p16 gene mutations.26

In summary, by subcellular fractionation of several cell lines, wehave shown that cytoplasmic p16 observed by immunohistochemis-try in cell lines, as well as in tumors, indeed is specific. Our resultsalso suggest that post translational modifications of p16 may influ-ence the subcellular localization of the protein; although, this needsfurther investigation. By gelfiltration and immunoprecipitation, we

further show that p16 in both the cytoplasm and nucleus is presentin a larger protein complex and binds to cdk4/6. The results are inline with our previous observation of nuclear and cytoplasmic p16upregulation in infiltrative cells at the edge of tumors and suggestthat p16 could have a function in the cytoplasm in tumor cells.

Acknowledgement

We thank Elise Nilsson for excellent technical assistance.

References

1. Sherr CJ. Cancer cell cycles. Science 1996;274:1672–7.2. Reed SI. Control of the G1/S transition. Cancer Surv 1997;29:7–23.3. Rocco JW, Sidransky D. p16(MTS-1/CDKN2/INK4a) in cancer pro-

gression. Exp Cell Res 2001;264:42–55.4. Serrano M, Hannon GJ, Beach D. A new regulatory motif in cell-

cycle control causing specific inhibition of cyclin D/CDK4. Nature1993;366:704–7.

5. Serrano M. The tumor suppressor protein p16INK4a. Exp Cell Res1997;237:7–13.

6. Hirabayashi H, Fujii Y, Sakaguchi M, Tanaka H, Yoon HE, KomotoY, Inoue M, Miyoshi S, Matsuda H. p16INK4, pRB, p53 and cyclinD1 expression and hypermethylation of CDKN2 gene in thymomaand thymic carcinoma. Int J Cancer 1997;73:639–44.

7. Ruas M, Peters G. The p16INK4a/CDKN2A tumor suppressor and itsrelatives. Biochim Biophys Acta 1998;1378:F115–F177.

8. Hara E, Smith R, Parry D, Tahara H, Stone S, Peters G. Regulation ofp16CDKN2 expression and its implications for cell immortalizationand senescence. Mol Cell Biol 1996;16:859–67.

9. Russo AA, Tong L, Lee JO, Jeffrey PD, Pavletich NP. Structural basisfor inhibition of the cyclin-dependent kinase Cdk6 by the tumour sup-pressor p16INK4a. Nature 1998;395:237–4.

10. Gump J, Stokoe D, McCormick F. Phosphorylation of p16INK4A cor-relates with Cdk4 association. J Biol Chem 2003;278:6619–22.

11. Thullberg M, Bartkova J, Khan S, Hansen K, Ronnstrand L, Lukas J,Strauss M, Bartek J. Distinct versus redundant properties amongmembers of the INK4 family of cyclin-dependent kinase inhibitors.FEBS Lett 2000;470:161–6.

12. Weebadda WK, Jackson TJ, Lin AW. Expression of p16INK4A var-iants in senescent human fibroblasts independent of protein phosphor-ylation. J Cell Biochem 2005;94:1135–47.

13. Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss M, PetersG, Bartek J. Retinoblastoma-protein-dependent cell-cycle inhibitionby the tumour suppressor p16. Nature 1995;375:503–6.

14. Bartkova J, Lukas J, Guldberg P, Alsner J, Kirkin AF, Zeuthen J,Bartek J. The p16-cyclin D/Cdk4-pRb pathway as a functional unitfrequently altered in melanoma pathogenesis. Cancer Res 1996;56:5475–83.

15. Geradts J, Hruban RH, Schutte M, Kern SE, Maynard R. Immunohisto-chemical p16INK4a analysis of archival tumors with deletion, hyper-methylation, or mutation of the CDKN2/MTS1 gene. A comparisonof four commercial antibodies. Appl Immunohistochem Mol Morphol2000;8:71–9.

16. Walker GJ, Gabrielli BG, Castellano M, Hayward NK. Functionalreassessment of P16 variants using a transfection-based assay. IntJ Cancer 1999;82:305–12.

17. Rizos H, Darmanian AP, Holland EA, Mann GJ, Kefford RF. Muta-tions in the INK4a/ARF melanoma susceptibility locus functionallyimpair p14ARF. J Biol Chem 2001;276:41424–34.

18. Hashemi J, Lindstrom MS, Asker C, Platz A, Hansson J, Wiman KG.A melanoma-predisposing germline CDKN2A mutation with func-tional significance for both p16 and p14ARF. Cancer Lett 2002;180:211–21.

19. Ghiorzo P, Villaggio B, Sementa AR, Hansson J, Platz A, Nicolo G,Spina B, Canepa M, Palmer JM, Hayward NK, Bianchi-Scarra G.Expression and localization of mutant p16 proteins in melano-

cytic lesions from familial melanoma patients. Hum Pathol 2004;35:25–33.

20. Wang NP, To H, Lee WH, Lee EY. Tumor suppressor activity of RBand p53 genes in human breast carcinoma cells. Oncogene 1993;8:279–88.

21. Kim MS, Li SL, Bertolami CN, Cherrick HM, Park NH. State of p53,Rb and DCC tumor suppressor genes in human oral cancer cell lines.Anticancer Res 1993;13:1405–13.

22. Craig C, Kim M, Ohri E, Wersto R, Katayose D, Li Z, Choi YH,Mudahar B, Srivastava S, Seth P, Cowan K. Effects of adenovirus-mediated p16INK4A expression on cell cycle arrest are determinedby endogenous p16 and Rb status in human cancer cells. Oncogene1998;16:265–72.

23. Lee JK, Kim MJ, Hong SP, Hong SD. Inactivation patterns of p16/INK4A in oral squamous cell carcinomas. Exp Mol Med 2004;36:165–71.

24. Svensson S, Nilsson K, Ringberg A, Landberg G. Invade or prolifer-ate? Two contrasting events in malignant behavior governed byp16(INK4a) and an intact Rb pathway illustrated by a model systemof basal cell carcinoma. Cancer Res 2003;63:1737–42.

25. Ragione FD, Russo GL, Oliva A, Mercurio C, Mastropietro S, PietraVD, Zappia V. Biochemical characterization of p16INK4- and p18-containing complexes in human cell lines. J Biol Chem 1996;271:15942–49.

26. Emig R, Magener A, Ehemann V, Meyer A, Stilgenbauer F, Volk-mann M, Wallwiener D, Sinn HP. Aberrant cytoplasmic expression ofthe p16 protein in breast cancer is associated with accelerated tumourproliferation. Br J Cancer 1998;78:1661–8.

27. Straume O, Sviland L, Akslen LA. Loss of nuclear p16 proteinexpression correlates with increased tumor cell proliferation (Ki-67)and poor prognosis in patients with vertical growth phase melanoma.Clin Cancer Res 2000;6:1845–53.

28. Milde-Langosch K, Bamberger AM, Rieck G, Kelp B, Loning T.Overexpression of the p16 cell cycle inhibitor in breast cancer is asso-ciated with a more malignant phenotype. Breast Cancer Res Treat2001;67:61–70.

29. Asada M, Yamada T, Ichijo H, Delia D, Miyazono K, Fukumuro K,Mizutani S. Apoptosis inhibitory activity of cytoplasmic p21(Cip1/WAF1) in monocytic differentiation. Embo J 1999;18:1223–34.

30. Nikolaev AY, Li M, Puskas N, Qin J, Gu W. Parc: a cytoplasmicanchor for p53. Cell 2003;112:29–40.

31. Blain SW, Massague J. Breast cancer banishes p27 from nucleus. NatMed 2002;8:1076–8.

32. Nilsson K, Svensson S, Landberg G. Retinoblastoma protein functionand p16(INK4a) expression in actinic keratosis, squamous cell carci-noma in situ and invasive squamous cell carcinoma of the skin andlinks between p16(INK4a) expression and infiltrative behavior. ModPathol 2004;17:1464–74.

33. Fahraeus R, Lane DP. The p16(INK4a) tumour suppressor proteininhibits alphavbeta3 integrin-mediated cell spreading on vitronectinby blocking PKC-dependent localization of alphavbeta3 to focal con-tacts. Embo J 1999;18:2106–18.

34. Kwon TK, Buchholz MA, Gabrielson EW, Nordin AA. A novel cyto-plasmic substrate for cdk4 and cdk6 in normal and malignant epithe-lial derived cells. Oncogene 1995;11:2077–83.

1125SUBCELLULAR LOCALIZATION OF p16 IN TUMOR CELLS