Tumor suppressor p16^INK4a controls oncogenic K-Ras function in human pancreatic cancer cells

Abstract

Pancreatic cancer is characterized by oncogenic activation of K-Ras and inactivation of the cell cycle inhibitor p16^INK4a. We previously demonstrated that reintroduction of p16^INK4a reversed anoikis resistance and clonogenicity of human pancreatic cancer cells, properties commonly attributed to the transforming potential of oncogenic K-Ras. Therefore, we aimed to determine the role of Ras after p16^INK4a re-expression. Here, we show that restitution of p16^INK4a in pancreatic cancer cell lines elicits a profound suppression of K-Ras activity. A more detailed analysis in p16^INK4a reconstituted Capan-1 cells indicated selective reduction of both K-Ras activity and protein stability. Re-expression of K-Ras in p16^INK4a restituted Capan-1 cells reversed the anoikis-sensitive phenotype and increased colony formation, indicating that K-Ras suppression was required for p16^INK4a-mediated reversion of the transformed phenotype. Inducible expression of p16^INK4a in DanG cells confirmed inhibition of K-Ras activity as well as an increase in anoikis susceptibility. Thus, our results delineate a novel functional interaction with defined biological consequences for the two most frequent alterations observed in pancreatic cancer. (Cancer Sci 2012; 103: 169–175)

Pancreatic cancer is characterized by a set of genetic alterations,^(1,2) including oncogenic activation of K-Ras and functional inactivation of tumor suppressors p16^INK4a(p16), p53 and Smad4.^(3,4) Of these, activation of K-Ras appears to represent the first and most frequent lesion, followed by functional inactivation of tumor suppressor p16.^(3,4) A very recent overview of molecular aberrations in the progression to pancreatic cancer is given by Scarlett et al.⁽⁵⁾ p16 selectively binds and thereby inhibits cyclin-dependent kinases (CDK) 4 and 6 that phosphorylate the retinoblastoma protein (Rb) to promote cell cycle progression.⁽⁶⁾ Apart from this well characterized function, p16 has been implicated in processes such as apoptosis, cell senescence and spreading, angiogenesis and tumor invasion.⁽⁷⁾ In this context, we have demonstrated that re-expression of p16 in pancreatic cancer cells restored anoikis,⁽⁸⁾ a cellular apoptosis program activated in epithelial cells upon loss of contact with the extracellular matrix. Furthermore, restitution of p16 abrogated clonogenicity and tumorigenicity in nude mice,⁽⁸⁾ features that are intimately linked to the transforming potential of oncogenic Ras.^(9,10)

Ras family molecules are small guanosine triphosphatases (GTPases) that cycle between their active GTP-bound and their inactive GDP-bound states.^(11,12)Ras genes are mutated in various types of human cancer, encoding proteins that are locked in an active conformation and inappropriately activate effector pathways.^(11,12) Multiple Ras homologues with non-redundant functions are known to exist, that is, H-Ras, N-Ras and K-Ras, although isoform-specific signaling is poorly understood.^(11,13) Accumulating evidence indicates distinct localization to membrane microdomains, distinct interaction partners and specific trafficking.⁽¹⁴⁾ Isoform-specific differences extend to oncogenic action and specific Ras oncoproteins are linked to specific tumor entities.^(11,15)

The incidence of K-ras mutations in pancreatic adenocarcinomas exceeds 90%⁽⁴⁾ and has been termed a “virtual rite of passage” for this malignancy.⁽³⁾ However, specific effector pathways of K-Ras in pancreatic cancer cells that could account for this singular constellation remain to be identified. Genetic mouse models of pancreatic cancer confirmed an initiator function of oncogenic K-Ras in ductal pancreatic cancer, but also indicate strong cooperation with p16 mutation.⁽¹⁶⁾ A lot of effort has been made in attempts to understand how K-Ras and p16 interact in the context of pancreatic ductal cell transformation. Earlier concepts proposed Ras induced p16-mediated senescence to create selection pressure against p16 and subsequent loss of the tumor suppressor.^(17–19) However, oncogenic Ras does not necessarily induce senescence in cells with intact p16, but might promote proliferation despite the presence of the tumor suppressor^(20–23) (for review see Hezel et al.⁽²⁴⁾). This latter scenario would also fit the molecular pathology of early pancreatic intraepithelial neoplasia precursor lesions, which feature oncogenic K-Ras but still retain functional p16.⁽²⁵⁾

In contrast to the well studied regulation of p16 tumor suppressor function by K-Ras,^(26,27) little is known on the converse interaction, that is, how p16 might affect the K-Ras oncoprotein. Here, we used human pancreatic cancer cell lines to address the impact of p16 re-expression on oncogenic K-Ras function. We found that p16 restricted the activity of the constitutive active K-Ras oncoprotein present in these cell lines and thereby counteracted cellular features that define the transformed phenotype.

Materials and Methods

Antibodies.  The following antibodies were used: K-Ras clone F234 and H-Ras clone F235 from Santa Cruz Biotechnology (Santa Cruz, CA, USA); N-Ras clone F155-227 and Pan-Ras clone RAS 10 from Oncogene Research Products (Bad Soden, Germany); p16^INK4a clone DCS-50.1/A7 from NeoMarkers (Fremont, CA, USA) or clone G175-405 from BD Pharmingen (Heidelberg, Germany); and Rho-A and integrin linked kinase (ILK) from Upstate Biotechnologies (Hamburg, Germany).

Cell culture.  Capan-1 and SW480 cells were obtained from the American Type Culture Collection (Rockville, ML, USA) and DanG cells were from DKFZ (Heidelberg, Germany) and cultured as recommended. Capan-1/p16 clones with expression of full-length human p16^INK4a from pRC_CMV have been previously described.⁽⁸⁾

Determination of anoikis and anchorage-independent growth assay.  To determine anoikis, 2 × 10⁵ cells were cultured on plates coated with poly 2-hydroxyethyl methacrylate (PolyHEMA) (Sigma, Deisenhofen, Germany) for 24 h or as adherent control cultures. Apoptotic cells were then identified from cell cycle analyses as previously described.⁽⁸⁾ The difference in apoptosis rates between suspended and adherent cultures is given as the anoikis rate. Colony formation in agar suspension was evaluated as previously described.⁽⁸⁾

Transfection of oligonucleotides and p16 or K-ras c-DNA constructs.  The phosphorothioate oligonucleotides used were: human K-rasV12 antisense-oligonucleotide AS-GTT (5′- CTA CGC CAA CAG CTC CA -3′), sense-oligonucleotide S-GTT (5′- TGG AGC TGT TGG CGT AG -3′), and random-oligonucleotide RANDOM (5′- GAT GCG GTT GTC CAC GA -3′), which have been previously described.⁽²⁸⁾ Capan-1 cells, 1–2 × 10⁵, were transfected with 5 μM oligonucleotides and 9.6 μL Enhancer/12 μL Effectene/mL as recommended by the manufacturer (Qiagen, Hilden, Germany) and cultured in medium supplemented with oligonucleotides.

Human K-ras cDNA was amplified from Capan-1 mRNA using 5′- CGG GAT CCG AGA GAG GCC TGC TGA A -3′ and 5′- AAG CGG CCG CTA CTG GCA CTT CGA GGA -3′ as primers and subcloned into pIREShyg (Clontech Laboratories, Palo Alto, CA, USA). Transfection of Capan-1/p16 cells with pIREShyg/K-ras2B and SW480 cells with pRC_CMV/p16 was performed using Effectene as recommended by the manufacturer. Control cells were transfected with the corresponding plasmid without insert. Cells were selected with 100 μg/mL hygromycin (Capan1) or 1.0 mg/mL G418 (SW480). Individual Capan-1/K-ras clones were isolated and expanded.

Inducible expression of human full-length p16 in DanG cells was achieved exactly as described before for MiaPaCa-2 cells.⁽²⁹⁾ Stable clones of DanG-TREx containing pcDNA4/TO-p16 were called DanG-TREx-p16.

Duplex polymerase chain reaction.  cDNA was obtained by reverse transcription from 2 μg total RNA. K-ras 2A/2B and GAPDH cDNA were then amplified simultaneously with the following primers: 5′- GAG AGA GGC CTG CTG AA -3′and 5′- TAC TGG CAC TTC GAG GA -3′ for K-ras; and 5′- ACC ACA GTC CAT GCC ATC AC -3′and 5′- TCC ACC ACC CTG TTG CTG TA -3′for GAPDH. Amplification conditions for 33 cycles were: 1 min at 95°C, 1 min at 58°C, 3 min at 72°C.

Nuclear run-off.  Nuclear run-off transcription was performed as previously described.⁽⁸⁾ In brief, nuclei from 10⁸ cells were isolated. Transcribed RNA products were labeled through [α-³²P] uridine triphosphate (Amersham Biosciences, Freiburg, Germany) incorporation and purified. Unincorporated nucleotides were removed using Microcons YM-10 (Millipore, Billerica, MA, USA). Run-off probes were adjusted to equal radioactivity and hybridized at 42°C for 48 h to cDNA for K-ras, heat shock protein 27 (hsp27), β-actin and GAPDH (5 μg/slot). Membranes were washed twice for 10 min in 40 mM NaPO₄/1% SDS buffer (pH 7.2) at 42°C, dried and exposed to X-ray films.

Western blot analysis.  Cells were lysed in 62.5 mM Tris/HCl (pH 6.8), 6 M urea, 10% glycerol, 2% SDS, 5%β-mercaptoethanol and 0.001% bromophenolblue, sonicated and incubated at 65°C for 15 min. Aliquots were separated by SDS-PAGE, blotted onto nitrocellulose and subjected to immunodetection according to the protocol provided by Upstate Biotechnologies with antibodies diluted 1:100 (H-Ras, K-Ras, N-Ras, Rho-A), 1:400 (Pan-Ras) or 1:1000 (ILK, p16). Bands were visualized by enhanced chemiluminescence.

Ras activation assay.  A total of 5 × 10⁶ cells were lysed in 500 μL magnesium lysis buffer (10 mM MgCl₂/1% NP-40) (MLB), sheared and centrifuged for 10 min at maximum speed and 4°C. Equal amounts of protein (0.8–1 mg in Capan-1 cells, 0.5 mg in DanG cells) were used to pull down Ras with 10 μg of Raf-1 Ras binding domain (RBD) agarose (Upstate Biotechnologies) according to the manufacturer’s protocol. GTP-bound Ras was then detected using western blotting of the precipitates.

Pulse-chase experiments.  Approximately 3–10 × 10⁶ adherent cells were starved with DMEM 5% FBS without methionine/cysteine for 30 min before application of 100 μCi/mL EasyTag EXPRE³⁵S³⁵S Protein Labeling Mix (NEN, Zaventem, Belgium). After 14 h, cells were chased with medium containing 2 mM methionine/cysteine. Cells were lysed in MLB (see Ras activation assay). K-Ras was precipitated from 500–600 μg aliquots using 1.5 μg of K-Ras antibody.

Statistical analysis.  Statistical analysis was performed using anova (Prism, GraphPad Software Inc., La Jolla, CA, USA). Differences were considered significant at P < 0.05.

Results

p16-reconstituted pancreatic cancer cells lack K-Ras activity and undergo anoikis.  As previously reported,⁽⁸⁾ reconstitution of Capan-1 cells with p16 restored anoikis when cells were subjected to suspension culture (Fig. 1A). We examined Ras activity under these experimental conditions using the Ras binding domain of Raf to selectively precipitate GTP-bound Ras (Fig. 1B, upper panel). Using a K-Ras-specific antibody for Ras-immunodetection, two important observations were made: (i) compared with adherent cultures, suspension cultures revealed a substantial increase in K-Ras activity; and (ii) restitution of p16 abolished K-Ras activity under both conditions. As Capan-1 cells harbor a GGT→GTT mutation in codon 12,⁽³⁰⁾ these observations reflect a major regulation in the activity of the K-RasV12 oncoprotein.

Figure 1.  p16 abrogates K-RasV12 activity in Capan-1 pancreatic cancer cells and causes downregulation of K-Ras expression levels. (A, B) Capan-1 cells were cultured for 24 h either adherent (ad) or suspended on polyHEMA-coated plates (susp). Anoikis rates (A) and K-Ras activity and expression (B) were determined based on the pre-G1 fraction of cell cycle analyses, the precipitation of GTP-bound Ras and immunoblotting, respectively. (A) Representative DNA histograms of cell cycle analyses for mock and p16 clones (n = 4). Number insets give the percentage of apoptotic cells. (B) Determination of K-Ras in p16 reconstituted cells (p16) or controls (UT: untransfected, mock). Representative immunoblots performed on a Ras binding domain (RBD) pull-down assay (upper panel) and on whole cell lysates (50 μg/lane) (lower panel) are shown. The antibodies used had K-Ras or Pan-Ras specificity. (C) Analysis of Ras activity in Capan-1 −/+ p16 using K-Ras (middle panel) or Pan-Ras (lower panel) antibodies. Upper row illustrates p16 expression status in these clones. (D) Representative immunoblots illustrate the expression of K-, N- or H-Ras or RhoA. Because the signal obtained for H-Ras was barely detectable, Capan-1 cells with expression of H-RasN17 were used as a positive control (outer right lane).

p16 selectively inhibits K-Ras expression.  To explore the mechanism underlying K-RasV12 inhibition, Ras expression was determined (Fig. 1B, lower panel). Immunoblotting was performed using an antibody that recognizes the K-, H- and N-Ras isoforms, with K-Ras represented by the upper band. In control cells (untransfected [UT] and mock transfected cells), this upper band was slightly induced by suspension culture conditions. However, irrespective of the culture condition applied, K-Ras protein abundance was profoundly and selectively reduced in cells with reconstitution of p16, suggesting that K-Ras downregulation accounted for the loss of K-Ras activity.

To verify the loss of K-RasV12 expression and activity in p16 reconstituted cells, a panel of three independent p16 clones as well as mock transfected controls were assessed for Ras activity (Fig. 1C). Using either the K- or the Pan-Ras specific antibody, we established that all three independent Capan-1/p16 clones were deficient in K-Ras activity. Furthermore, the signals obtained using either antibody were virtually congruent, suggesting that K-RasV12 accounts for most, if not all, of the Ras activity in randomly cycling Capan-1 cells. Subsequently, K-Ras expression was examined in the same panel of clones (Fig. 1D). Compared with control cells, K-Ras expression was severely reduced in Capan-1/p16 cells. In contrast, N-Ras and H-Ras content did not seem to be altered due to p16 expression, although H-Ras expression was barely detectable. Also, expression of a small GTPase from the Rho family, Rho-A was unchanged (Fig. 1D, lower panel). Thus, p16 restitution profoundly and selectively inhibited K-RasV12 activity on the basis of reduced K-RasV12 expression.

K-Ras inhibition in p16-reconstituted cells is due to decreased K-Ras protein stability.  To further define the molecular mechanism by which p16 downregulates K-Ras expression, we determined the K-ras transcription rate in nuclear run-off assays (Fig. 2A). However, levels of K-ras de novo mRNA transcripts were comparable in p16-deficient controls and p16 restituted cells. Furthermore, duplex PCR analyses of K-ras steady-state mRNA levels (Fig. 2B) revealed no p16-induced significant differences for either the K-ras A or predominant K-ras B splice variant. GAPDH (Fig. 2B) or major histocompatibility complex (not shown) were used as an internal amplification standard. Taken together, these experiments suggest that inhibition of K-RasV12 was not due to reduced transcription rates or modified mRNA stability, but rather occurred at the protein level. Therefore, K-RasV12 protein stability was determined by pulse chase experiments (Fig. 2C). Following a 35[S]-methionine/-cysteine pulse, comparable amounts of radioactivity were recovered in K-RasV12 immunoprecipitates from p16 restituted and control cells, excluding major changes in K-RasV12 synthesis (compare lanes 2 and 6 in Fig. 2C). A distinct signal was obtained in Capan-1 control cells, which persisted until at least 12 h, although a slight increase in apparent molecular weight was noted during the chase period (compare lanes 2–4 in Fig. 2C). In contrast, ³⁵[S]-labeled K-RasV12 protein rapidly declined in Capan-1/p16 cells and was barely detectable at 6 h. Inhibition of protein synthesis using cycloheximide confirmed the shortened half-life and defined it at 4 h (Fig. S1). Thus, restitution of p16 had severely reduced K-RasV12 protein stability.

Figure 2.  K-RasV12 suppression results from accelerated degradation of K-RasV12. (A) Nuclear run-off analysis of K-ras transcription rates in Capan-1 control and Capan-1/p16 cells. A representative autoradiograph of three independent experiments is shown. (B) Duplex PCR reactions with amplification of full-length alternative K-ras splice variants K-ras2A (897 bp) and K-ras2B (773 bp) as well as GAPDH as an internal standard. A representative out of two amplification reactions derived from independent RNA preparations is shown. UT, untransfected Capan-1 cells. (C) Pulse-chase ^[35]S-labeling and subsequent immunoprecipitation of K-Ras to determine the effects of p16 on K-RasV12 protein turnover in Capan-1 cells. Capan-1 cells and a representative p16 clone were analyzed. Precipitates were separated on a 13.5% SDS-PAGE. An autoradiograph representative of three independent experiments is shown.

K-Ras suppression is required for the anoikis-sensitive phenotype and the loss of clonogenicity observed in Capan-1/p16 cells.  Next, we aimed to characterize the functional relevance of p16-dependent suppression of K-RasV12 activity in Capan-1 cells and pursued two independent approaches. The first approach addressed the role of K-RasV12 for the anoikis-resistant phenotype of Capan-1 cells using K-RasV12-specific antisense oligonucleotides to successfully suppress K-RasV12 expression compared with cells that had received sense or random oligonucleotides (Fig. 3A). In the activity assay, antisense oligonucleotides caused a significant decrease in K-Ras activity compared with the sense oligonucleotides (Fig. S2). When the oligonucleotide-transfected cultures were then subjected to anoikis conditions, cells with K-RasV12 inhibition revealed significantly elevated anoikis rates, consistent with an anoikis-protective function of K-RasV12 in Capan-1 cells (Fig. 3B).

Figure 3.  K-RasV12 inhibition in Capan-1 cells restores anoikis susceptibility. Capan-1 cells were treated with 5 μM of control (random, K-rasV12 sense) or K-rasV12 antisense oligonucleotides for 5 days, and Ras expression (A) as well as anoikis rates (B) were determined. Data in the graph represent the mean ± SEM from four independent experiments (*P < 0.05).

The second approach directly addressed the relevance of K-RasV12 in the context of p16-dependent anoikis induction and suppression of clonogenicity. Accordingly, Capan-1/p16 cells received oncogenic K-ras2B V12 in a second transfection. This procedure created clones with moderate to high expression of K-Ras2B V12 (referred to as K-RasV12 from now on; Fig. 4A, upper panel, upper row) and successfully restored K-RasV12 activity to levels of Capan-1 control cells or above (Fig. 4A, upper panel, lower row). Additional immunoblots confirmed that clones with restored K-Ras activity retained p16 expression and did not exhibit compensatory regulation of the N-Ras isoform (Fig. 4A, lower panel). Thus, having established their properties, we then determined anoikis in Capan-1/p16-K-rasV12 clones and control cells (Fig. 4B). In Capan-1 or vector-control cultures, few cells underwent anoikis. As expected, Capan-1/p16 cells revealed substantial anoikis rates, which remained unchanged if these cells received an empty vector in the second transfection. In contrast, anoikis rates returned to the level of anoikis-resistant Capan-1 parental cells, if Capan-1/p16 cells received oncogenic K-rasV12. Similar results were observed when clonogenicity in soft agar was examined (Fig. 4C). In these colony-formation assays, parental Capan-1 cells generated a high number of clones consistent with their transformed phenotype. In contrast, Capan-1/p16 cells produced significantly less colonies. Again, transfer of a mock construct in the second transfection had no influence on colony formation. However, if K-RasV12 was re-expressed, clonogenicity returned to Capan-1 levels.

Figure 4.  K-RasV12 re-expression reverts anoikis susceptibility and colony formation of p16 restituted Capan-1 cells. K-RasV12 activity was restored in Capan-1/p16 cells using a K-rasV12 expression construct. (A) Characterization of Ras expression and activity in Capan-1/p16 cell clones that were stably transfected with pIREShyg-K-ras2B (p16/K-ras). Controls were parental Capan-1 cells, pRC_CMV vector controls (Mock), Capan-1/pRC_CMV-p16 (p16) and Capan-1/pRC_CMV-p16/pIREShyg (p16/Mock_ras). Immunoblots for K-Ras (upper panel), N-Ras and p16 (lower panel) as well as a K-Ras activity assay (upper panel) are shown. (B) Clones were subjected to suspension culture and anoikis rates were determined. Data in the graph represent the mean ± SEM from three independent experiments (*P < 0.05). (C) The same set of clones was subjected to soft agar assays and colony numbers were obtained after 12 days (mean ± SEM, n = 4, *P < 0.05).

Taken together, restoration of K-RasV12 activity was capable of fully reversing p16-induced restoration of anoikis as well as p16-mediated suppression of clonogenicity in Capan-1 cells, indicating that K-RasV12 suppression was required for these events. We propose direct interaction of p16 with K-Ras as a possible mechanism for the changes in K-Ras activity/stability and consecutively in the tumorigenic character (Fig. S3), but we are not sure if this is the mechanism under physiological conditions. Constitutively active K-Ras is mostly membrane associated and needs a stringent buffer to be solved from the membrane, while p16 co-immunoprecipitates need a buffer of very low stringency so that our trial with the low-stringent buffer co-precipitated a low amount of K-Ras only.

p16-dependent suppression of Ras can be reproduced in other pancreatic and colon carcinoma cell lines.  To exclude a cell line-specific phenomenon for the observed interaction between p16 and K-RasV12, we studied DanG pancreatic and SW480 colon cancer cells, which also harbor oncogenic K-RasV12 (data not shown). Following stable transfection, p16 restituted cells revealed reduced K-RasV12 expression (Fig. 5A) when compared with their respective controls. In contrast, expression of ILK (Fig. 5A) and expression of proliferating cell nuclear antigen (data not shown) remained unaltered, excluding unspecific downregulation or unequal sample loading. However, G1 cell cycle arrest or apoptosis induction in these stably transfected cell lines precluded further experiments. We therefore took advantage of a tetracycline inducible expression system and generated DanG-TREx-p16 cells, which express p16 within 24 h of the addition of 1 μg/mL doxycycline (dox) to the culture medium (not shown). Dox treatment resulted in reduced K-Ras activity at 4 days of treatment in two independent DanG-TREx-p16 clones and also significantly increased the anoikis fraction (2.2 ± 0.45-fold control, P < 0.05). One of the clones is shown exemplary in Figure 5(B). Following persistent induction of p16 for 8 days, we noted a reduction of K-Ras protein in whole cell lysates of DanG cells (Fig. 5B, right panel). The amount of K-Ras was reduced to 50% after 10 days (data not shown). Finally, induction of p16 in MiaPaCa-2-TREx-p16, which harbor a different oncogenic K-Ras mutation (GGT to TGT mutation in Codon 12 of the K-ras gene, Cosmic data bank of the Sanger Institute, Cambridge, UK) also produced a reduction of K-Ras activity (Fig. S4). Thus, the regulatory mechanism that links p16 and K-Ras applies to a broader spectrum of tumor cells with oncogenic K-Ras.

Figure 5.  p16 inhibits K-Ras in Dan-G pancreatic and SW-480 colonic cancer cells. (A) Immunoblots for p16 and K-Ras in stably transfected individual Dan-G clones (left) and in a p16 transfected population of SW-480 cells (right) are shown. Blots were stripped and reprobed with an antibody against integrin-linked kinase (ILK) (lower row). (B) Dan-G cells with doxycycline (dox) inducible expression of p16 were stimulated for 4 or 8 days with 1 μg/mL dox and whole cell lysates were analyzed using immunoblotting for activity and expression of K-Ras as well as for expression of p16. Blots were stripped and reprobed with an antibody against beta-actin or ILK to confirm equal loading.

Discussion

Coincident mutation in K-ras and loss of p16 are a hallmark of human pancreatic cancer, suggesting that both alterations might coordinately promote a distinct tumorigenic pathway. Our current results establish a novel functional link between these two most common genetic alterations, in that p16 attenuates the activity of K-RasV12. Disruption or restitution of this functional interaction corresponded with gain or loss of a fully transformed phenotype, underscoring its crucial importance for the tumor biology of human pancreatic cancer cells.

Capan-1 cells permit high level re-expression of p16, possibly due to the genetic background of the cell line harboring TP53 and SMAD4 mutations besides CDKN2A deletion and the activating mutation of K-ras (Kalthoff et al.⁽³⁰⁾ and Cosmic data bank of the Sanger Institute) and to the loss of pRb, which occurred after p16 re-expression in Capan-1 cells.⁽⁸⁾ Otherwise, cells would have been lost after a few days or passages as mentioned for the SW-480 used. K-Ras activity in Capan-1 p16 transfectants was virtually abrogated despite the activating mutation. Concurrent with the loss of activity, we noted a profound reduction of K-Ras protein. Thus, p16-mediated regulation of K-Ras activity differs from the well studied mechanisms that control Ras activity via GDP-GTP exchange⁽¹²⁾ and, in contrast to the latter, is functional in oncogenic Ras variants. Our subsequent inducible strategy of p16 re-expression in pancreatic cell lines not only confirmed K-Ras regulation but also revealed that reduction of K-Ras activity preceded the reduction of cellular K-Ras content, suggesting a dual or two-step mechanism of regulation. A delay in full expression of p16 could be the reason.⁽³¹⁾ This implies a (dys-)balance of p16 and K-Ras levels, which could determine the fate of the cells. The lectin galectin-3 was described as an interaction partner capable of enhancing K-Ras. GTP nanoclustering and signaling.^(32–35) We found reduced galectin-3 levels in Capan-1/p16 cells (data not shown) that could be responsible for the decrease in K-Ras activity, but which cannot explain the decrease in stability. Decrease in activity as well as in stability could be due to direct interaction of p16 with K-Ras, which was difficult to show, but would be intelligible. P16 consists of just four ankyrin repeats and is known to mediate numerous protein–protein interactions.⁽³⁶⁾

The altered protein stability was selectively observed for the K-RasV12 homologue and did not affect the N-Ras protein. New data discovered isoform-specific differences⁽³⁷⁾ that could explain K-Ras characteristics in the future. Another study identified the multifunctional phosphoproteins Nucleophosmin and Nucleolin stabilizing specifically the K-Ras isoform or its oncogenic counterpart at the plasma membrane.⁽³⁸⁾ Until recently, little information was available on the stability of fully processed Ras proteins and data are in part derived from overexpression studies. For c-H-Ras a half-life in the range of 9–20 h was described,^(13,39,40) and an extended half-life from 20 to 42 h in the constitutive active v-H-Ras counterpart.⁽⁴⁰⁾ c-K-Ras is reportedly less stable,⁽¹³⁾ and we observed a half-life between 6 and 12 h for constitutively active K-RasV12 in Capan-1 cells. Because a ubiquitin-mediated degradation pathway is established for the H-Ras and N-Ras but not the K-Ras homologue,⁽⁴¹⁾ the exact mechanism that destabilized K-Ras still remains to be identified. This is in contrast to the findings of Kim et al.⁽³⁹⁾ who detected a reduced cellular content of K-Ras in co-transfection experiments by overexpression of the F-box protein β-transducin repeat-containing protein (β-Trcp), which mediates ubiquitylation. We assume that K-Ras could be phosphorylated in Capan-1 cells while being degraded and perhaps also in the p16 transfectants, which showed a little shift just after the S³⁵ pulse. Protein kinase C-mediated phosphorylation of K-Ras was already shown in 1987⁽⁴²⁾ and led to translocation of K-Ras from the plasma membrane to the outer membrane of mitochondria and the induction of apoptosis.⁽⁴³⁾

Our results outline a new mechanism whereby tumor suppressor pathways achieve control of Ras oncoprotein activity via regulation of Ras turnover. Accordingly, a balance situation is conceivable where cells that harbor oncogenic K-rasV12 mutations lack transforming K-RasV12 activity. Thus, it could operate in situations where K-ras mutations are present without overt malignancy, for example, in premalignant conditions such as chronic pancreatitis or even in normal pancreatic ductal cells.⁽⁴⁴⁾ Functional inactivation or loss of p16 would then unleash the restrain imposed on K-RasV12 activity by its regulated degradation. In support of such a mechanism, loss of p16 is associated with progression from focal premalignant lesions to highly invasive and metastatic pancreatic cancer in mice that express a mutant K-Ras allele in the pancreas.⁽¹⁶⁾ Our results are in excellent agreement with emerging concepts of Ras-driven tumorigenesis, which have emphasized a dose-dependent outcome of Ras oncoprotein activation.^(45,46) Using an inducible approach of K-Ras oncogene expression in mouse mammary cells in vivo, Sarkisian et al.⁽⁴⁶⁾ reported that an increase in the activity of mutant K-Ras was required for progression from hyperproliferative lesions to invasive cancer. We suggest that inactivation of p16 also accounts for the increased Ras oncoprotein activity postulated.

The dramatic increase in K-Ras oncoprotein activity in Capan-1 control cells subjected to anoikis conditions represents an eloquent example of this emerging plasticity of K-Ras oncoprotein signaling. In conjunction with our functional studies it strongly suggests that activation of K-Ras signaling is required to protect Capan-1 cells from anoikis, whereas lack of K-Ras activity in p16 reconstituted adherent cells did not translate into differences in basal apoptosis or proliferation (Fig. 1 and Plath et al.⁽⁸⁾). K-Ras suppression fully accounted for the restitution of anoikis and the reduction of clonogenicity observed upon re-expression of p16 in Capan-1 cells.⁽⁸⁾ Thus, suppression of key features of the transformed phenotype by p16 was due to its capacity to regulate oncogenic K-Ras. Our functional link between K-RasV12 activation and anoikis as well as clonogenicity is in excellent agreement with the results obtained by selective knockdown of K-Ras in Capan-1 cells, which also abrogated clonogenicity.⁽⁹⁾ Furthermore, both studies concur that K-Ras-depleted Capan-1 cells maintain regular proliferation and cell cycle distribution.^(8,9)

In summary, the current study presents a novel functional interaction of p16 and K-Ras. Re-expression of the tumor suppressor in pancreatic cancer cells restricted Ras oncoprotein activity, which resulted in loss or reduction of their transforming potential. Given that pharmacological inhibition of oncogenic K-Ras signaling still represents an unresolved challenge, further unraveling of the pathway that links p16 to suppression of K-Ras activity might open alternative venues to disrupt K-Ras-driven transformation and intervene in pancreatic cancer.

Acknowledgments

The authors dedicate this study to Stefan Rosewicz who has died but considerably contributed to this work. A. R. was supported by scholarships from DFG and Sonnenfeld–Stiftung. N. O. was supported by a scholarship from DFG. K. M. D. was supported by a grant from Wilhelm–Sander Stiftung.

Disclosure Statement

The authors have no conflict of interest.