JOURNAL OF CELLULAR PHYSIOLOGY 210:740–749 (2007)

Oncogenic KRAS Provides a Uniquely Powerful and Variable Oncogenic Contribution Among RAS Family Members in the Colonic Epithelium JEFFREY W. KELLER,1 JEFFREY L. FRANKLIN,1 RAMONA GRAVES-DEAL,1 DAVID B. FRIEDMAN,2,3 CORBIN W. WHITWELL,2,3 AND ROBERT J. COFFEY1,4* 1 Department of Cell and Developmental Biology, Vanderbilt University Medical Center, Nashville, Tennessee 2 Mass Spectrometry Research Center, Vanderbilt University Medical Center, Nashville, Tennessee 3 Department of Biochemistry, Vanderbilt University Medical Center, Nashville, Tennessee 4 Department of Veterans Affairs Medical Center, Nashville, Tennessee Activating mutations of the RAS family of small GTPases are among the most common genetic events in human tumorigenesis. Constitutive activation of the three canonical family members, KRAS, NRAS, and HRAS segregate strongly by tissue type. Of these, KRAS mutations predominate in human tumors, including those arising from the colon and lung. We sought to compare the oncogenic contributions of different RAS isoforms in a comparable genetic setting and to explore downstream molecular changes that may explain the apparent differential oncogenic effects of the various RAS family members. We utilized colorectal cancer cell lines characterized by oncogenic KRAS in parallel with isogenically derived lines in which the mutant allele has been disrupted. We additionally attempted to reconstitute the isogenic derivatives with oncogenic forms of other RAS family members and analyze them in parallel. Pairwise analysis of HCT 116 and DLD-1 cell lines as well as their isogenic derivatives reveals distinct K-RASG13D signatures despite the genetic similarities of these cell lines. In DLD-1, for example, oncogenic K-RAS enhances the motility of these cells by downregulation of Rap1 activity, yet is not associated with increased ERK1/2 phosphorylation. In HCT 116, however, ERK1/ 2 phosphorylation is elevated relative to the isogenic derivative, but Rap1 activity is unchanged. K-RAS is uniquely oncogenic in the colonic epithelium, though the molecular aspects of its oncogenic contribution are not necessarily conserved across cell lines. We therefore conclude that the oncogenic contribution of K-RAS is a function of its multifaceted functionality and is highly contextdependent. J. Cell. Physiol. 210: 740–749, 2007. ß 2006 Wiley-Liss, Inc.

Activating mutations of the members of the RAS family of small GTPases are among the most common genetic events in human tumorigenesis and characterize approximately 30% of all human cancers (Bos, 1989). Constitutive RAS signaling contributes to the transformed phenotype by promoting proliferation, enhanced motility, survival and loss of anchorage dependence. Activating mutations of RAS family members do not occur with equal frequency across tumor types, but rather segregate strongly by tissue of origin, with one isoform predominating in any particular RAS-driven tumor. For example, mutations to KRAS are the most frequent RAS family mutations and occur in 30% of nonsmall cell lung cancers, 50% of colorectal adenocarcinomas and 90% of pancreatic adenocarcinomas. Fifteen percent of melanomas and 30% of acute myelogenous leukemias display activating NRAS mutations. Activating mutations of HRAS are far less common but contribute to squamous cell carcinomas, bladder carcinomas, and to renal cancers (Downward, 2003). Isoform-specific functions may explain the observed segregation by tumor type. Numerous recent studies have refuted the long-held belief that RAS isoforms are functionally interchangeable in vivo (Yan et al., 1997a,b, 1998; RodriguezViciana et al., 2004). Despite their extensive sequence identity, differences in primary sequence and posttranslational modifications of the C-terminal 24 amino acids provide the cell biological basis for functional differences in the members of this protein family. Because of the multiplicity of RAS function, discerning specific pathß 2006 WILEY-LISS, INC.

ways and isolating their contributions to tumorigenesis has proved daunting. This complexity is further magnified by the differential requirements for RAS signaling at different stages of tumor development (Chin et al., 1999; Lim and Counter, 2005). The frequency of KRAS mutations across a broad range of human tumors suggests the potency of the oncogenic contribution of the constitutively active form of this protein. In an attempt to better understand the basis for the prevalence of the KRAS oncogene in colorectal cancer, we employed isogenic pairs of colorectal cancer cell lines that owe much of their transformed phenotype to the presence of an activating mutation (G13D) of KRAS (Shirasawa et al., 1993). Using homologous recombination, the mutant allele has

This article includes Supplementary Material available from the authors upon request or via the Internet at http://www.interscience.wiley.com/jpages/0021-9541/suppmat. Contract grant sponsor: NCI SPORE; Contract grant number: 95103; Contract grant sponsor: NIH; Contract grant numbers: CA46413 and CA95103. *Correspondence to: Robert J. Coffey, Suite 4140 MRBIII, Vanderbilt University, 465 21st Avenue South, Nashville, Tennessee 37232. E-mail: [email protected] Received 20 May 2006; Accepted 28 August 2006 DOI: 10.1002/jcp.20898

741

VARIABLE ONCOGENIC SIGNATURE OF K-RAS

been disrupted to generate isogenic derivatives which undergo dramatic reversion of the transformed phenotype by conventional assays of tumorigenicity. To understand the K-RAS-mediated downstream signaling events that may explain this, we have attempted to recapitulate the transformed phenotype of these cells by overexpressing constitutively active forms of either N-RAS or H-RAS in the isogenic derivatives in order to gauge their relative oncogenic contributions. We utilized molecular and functional readouts to compare the oncogenic contributions of constitutively active RAS isoforms in a genetically comparable background. We identify K-RAS-mediated changes that partially explain the dramatic transformed phenotype observed in HCT 116 and DLD-1 colorectal cancer cell lines and present evidence that the molecular signatures of K-RAS-driven transformation are not conserved across cell lines derived from tumors with a common origin. We conclude that individual tumors may select for different aspects of K-RAS signaling in a context-dependent manner. MATERIALS AND METHODS

1% NP-40, 10 mM MgCl2, 1 mM EDTA, 2% glycerol, proteinase inhibitors (Roche) and phosphatase inhibitor cocktails I and II (Sigma, St. Louis, Missouri). Lysates were vortexed 5 sec, allowed to rest on ice 5 min and centrifuged at 13,400 rpm at 48C in a benchtop centrifuge for 15 min. Supernatants were then transferred to fresh tubes, and protein concentration determined by BCA assay. Samples were resolved by SDS– PAGE, transferred to Immobilon, blocked for 1 h with 5% nonfat milk in TBST (0.1% Tween-20), then incubated with primary antibody overnight. RAS specific antibodies were from Santa Cruz, California: H-RAS (sc-520); N-RAS (sc-31); and K-RAS (sc-30) were all used at 1:200; pan-RAS was from Sigma (RAS clone 10) and was used at 1:1,000. Antibodies against MEK (Cell Signaling, Beverly, Massachusetts 9124), pMEK (Cell Signaling 9127), ERK1/2 (Cell Signaling 9102), and pERK1/2 (Cell Signaling 9101) were used at 1:1,000. Monoclonal Rap1 antibody from BD Transduction Laboratories (610195) was used at 1:500. Blots were detected with appropriate secondary antibody conjugated to HRP and detected with enhanced chemiluminescence (Amersham, Uppsala, Sweden). Rap1 Activation kit (#17–321) was purchased from Upstate (Lake Placid, New York) and used per manufacturer’s instruction. The EZ-DetectTM RAS Activation Kit was purchased from Pierce (Rockford, IL).

Molecular biology

Retroviral vector LZRS-MS-EGFP (a kind gift of G. Nolan, Stanford University) was used for cloning H-RAS, N-RAS, and K-RAS4B (Ras cDNAs were the kind gift of M. Philips, New York University). G12V mutations of RAS cDNAs as well as of Rap1 were generated using site-directed PCR-based mutagenesis. Rap1 clone was purchased from ATCC. All constructs were bidirectionally sequenced to ensure accuracy. Rap1 sequences were cloned into pBABE Puro (blunt, HincII). siRNA sequences were generated with the use of pSICOLIGOMAKER software (http://web.mit.edu/ccr/labs/jacks). shRAP(2) target sequence: GAAGAACTGTTGCCTAATT; shRAP(3) target sequence: GGATGCATTTCAAATGTTA. Sequences were BLASTed to confirm target specificity. Hairpins were cloned into linearized pSICOR (XhoI/HpaI) and sequenced. Cell culture conditions, retroviral, and lentiviral transduction

Retroviral vectors were transfected using Fugene 6 (Roche, Indianapolis, IN) into Phoenix 293 cells. In all cases, these cells were selected in puromycin (1 mg/ml) for 24 h to generate packaging lines. Recovered packaging lines were passaged and grown to a density of 50–70% in a 100 mm dish. A minimal volume of media was added, and the packaging cells were placed at 328C to generate virus. After 12–24 h, media was harvested and filtered through a 0.22 mm syringe filter. Retroviral-containing media was supplemented with polybrene (5 mg/ml) and transferred to subconfluent, actively proliferating target cells. Successfully transduced target cells were sorted by flow cytometry (VA Flow Cytometry Core Facility, Nashville, TN). For lentiviral transduction of siRNA constructs, 293T cells were plated in a 100 mm dish and grown to 50–70% confluence. Cells were transfected overnight with the lentiviral vector pSICOR (5 mg) as well as the D8.2 packaging vector (2.5 mg) and the pMD26 packaging vector (2.5 mg) using Fugene 6 according to the manufacturer’s instructions. Media was changed and replaced with minimal volume. After 24 h, media was harvested and filtered through a 0.22 mm syringe filter. Filtered viral supernatant was transferred to a plate of target cells, and transduction continued for 24 h. Selection was begun the following day (puromycin 7.5 mg/ ml) and refreshed daily for 4–6 days until a nontransfected control plate was completely killed. Cells were maintained in DMEM, and selection was periodically repeated. Specificity of knockdown was verified by Western blot. MG132 and lactacystin were purchased from Calbiochem, San Diego, California. Immunoblotting and small G protein activation assays

For RAS Western blots, all cells were washed 3
with icecold PBS and lysed in 25 mM HEPES (pH 7.5), 150 mM NaCl, Journal of Cellular Physiology DOI 10.1002/jcp

Wound healing assays

Cells were grown to confluence in conditions described above. The monolayer was then ‘‘wounded’’ by making a single scratch with a standard pipette tip. The width of this wound was measured and recorded. Migration rates of the cells on the wound borders were measured on a daily basis, and rates of closure were calculated as the width at any given point expressed as a percentage of the initial wound. Rates of closure were monitored until the opposing borders met. 2D DIGE analysis

TNE buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 2 mM EDTA, pH 8.0, 1 mM Na3VO4, 1 mM leupeptin, 2 mM pepstatin, and 0.1 mM aprotinin) was added to each plate (150 mm) and cells scraped and collected in an Eppendorf tube. The suspended solution was then sonicated at 48C (3
10 pulses, power level 4), after which NP-40 (1% v/v) and PMSF (1 mM) were added to the lysate solution. Samples were then vortexed, allowed to sit on ice for 10 min and centrifuged at 10,000g for 10 min at 48C. Supernatants were removed and concentrations determined in triplicate by bicinchoninic acid (BCA) protein assay. DTT was then added to a final concentration of 2 mM, and samples were visualized on a 10% SDS–PAGE gel to confirm concentrations. Cy-dye labeling, 2D gel electrophoresis and imaging, DIGE analysis, in-gel digestion, and mass spectrometry were performed essentially as described previously (Friedman et al., 2004).

RESULTS K-RAS4BG12V cannot be highly overexpressed in DKS-8 or HKE-3 (KRASG13D/þ) cells though H-RASG12V and N-RASG12V are maintained

Constitutive activation of RAS can yield numerous different outcomes depending on cell context. In an untransformed environment, constitutive Ras activity promotes cell cycle arrest, triggering feedbacks within the cell that under normal circumstances mitigate the potentially dangerous effects of unregulated signaling through a number of pathways (Serrano et al., 1997). The same event, occurring in a cell in which these mechanisms have been impaired, however, contributes significantly to transformation, enabling focus formation, altering morphology and motility and increasing proliferation (Ruley, 1983; Hirakawa and Ruley, 1988). We hypothesized that context is similarly important in the derivation of isogenic cell lines by means of targeted disruption of an oncogenic allele. A cell that has adapted to, or even become dependent on, constitutive signaling

742

KELLER ET AL.

events, likely undergoes dramatic changes following removal of those events (Chin et al., 1999; Jain et al., 2002). To explore the relationship of DLD-1 and the HCT 116 cell lines to their isogenic derivatives, we transduced these derivatives with a retrovirus carrying constitutively active K-RAS4BG12V (and a separate GFP marker under transcriptional control of an IRES element) in an effort to recapitulate the transformed phenotype of the parent cell lines. Following transduction, we sorted the pool of positively expressing cells by flow cytometry for GFP intensity as an indicator of expression levels into three distinct expression levels (high, medium, and low) (Liu et al., 2000). We were able to generate three distinct and equivalent pools of RASoverexpressing target cells for both HKE-3 and DKS-8 (high, medium, and low) as measured by GFP expression. In order to verify the correlation between differential expression levels of GFP and K-RAS4B, we immunoblotted lysates made from these cells for GFP and K-RAS4B and measured RAS activity by the use of the RBD fragment (Fig. 1, part A). While we were able to verify three distinct levels of expression for GFP, the immunoblotting results for K-RAS4B did not correlate with those for GFP (Fig. 1, part A). The three distinct levels of GFP observed in the high, medium, and lowexpressing pools of K-RAS imply the presence of transcripts, but the lack of corresponding K-RAS protein suggests some posttranscriptional downregulation. This observation corroborates previous findings (Habets et al., 2001). In contrast, similar transduction with retrovirus carrying H-RASG12V or N-RASG12V followed by the same sorting strategy produced detectable overexpression of active N-RAS and H-RAS (Fig. 1, part C). We further investigated the reduced levels of K-RAS4B protein levels by treating these cells with inhibitors of proteasomal degradation, MG132, and lactacystin, followed by immunoblot. Interestingly, proteasomal inhibition resulted in elevated levels of K-RAS4B protein, though not complete restoration as compared with protein levels seen in the H-RASG12V and N-RASG12V overexpressors (Fig. 1, part B). This result is perhaps not surprising given that ubiquitination that targets the protein for proteasomal degradation frequently occurs on internal lysine residues, and that the distinguishing motif of K-RAS4B hypervariable domain is a polylysine motif (Glickman and Ciechanover, 2002). Preferential degradation of K-RAS4BG12V, but not H-RASG12V or N-RASG12V, suggests also the comparatively deleterious and potent effects of constitutive K-RAS4B signaling. Further, this observation supports the possibility that these isogenic derivatives have undergone changes to adapt to the loss of oncogenic K-RAS4B signaling that cannot simply be reverted by reconstitution. Neither overexpression of N-RASG12V nor H-RASG12V confer the tumorigenic properties of constitutively active endogenous K-RAS

Disruption of the oncogenic KRAS allele compromises several tumorigenic properties of these cells, including enhanced proliferation as well as anchorageindependent growth as measured by soft agar assays (Shirasawa et al., 1993). To compare the oncogenic contribution of endogenous K-RASG13D with overexpressed H-RASG12V and N-RASG12V, we measured the anchorage-independent growth properties and growth rates of the isogenically derived DKS-8 cells overexpresJournal of Cellular Physiology DOI 10.1002/jcp

Fig. 1. Overexpression of oncogenic N-RAS, H-RAS, and K-RAS4B mutants in DKS-8 cells. A: High levels of overexpression of K-RAS4BG12V are not achievable in DKS-8 cells. Three distinct expression levels (high, medium, and low) were generated by sorting for GFP expression. K-RAS4B protein levels shown by immunoblot, however, did not correspond to GFP expression, suggesting dowregulation. B: Treatment of these cells with proteasome inhibitors lactacystin (10 mM) or MG132 (50 mM) for 4 h resulted in partial restoration of protein levels as measured by immunoblot. C: DKS-8 cell lines maintain different levels of overexpression of H-RASG12V and N-RASG12V as shown by immunoblot. These mutant proteins are functional, as shown by their detection with the Raf binding domain Ras activity assay (RBD).

sing H-RASG12V or N-RASG12V relative to the parental DLD-1 cell lines (Fig. 2). In both these assays, overexpression of H-RASG12V proved to more closely recapitulate the oncogenic contribution of endogenous K-RASG13D than did N-RASG12V. Although both represented an increase over the DKS-8 cells expressing vector alone, neither approached the levels of the parental line (Fig. 2, part A). Quantification of soft agar assays reveals that the respective oncogenic contributions were dose-dependent, and therefore attributable to RAS overexpression

VARIABLE ONCOGENIC SIGNATURE OF K-RAS

743

Fig. 2. Overexpression of oncogenic H-RAS or N-RAS only partially recapitulates the tumorigenic properties of the endogenous oncogenic K-RAS. A: The DKS-8 isogenic derivatives (KRASþ/) overexpressing H-RASG12V or N-RASG12V do not recapitulate the ability to grow in soft agar seen in the DLD-1 (KRASG13D/þ) parent cell line. B: The

oncogenic contribution of H-RASG12V and N-RASG12V was dosedependent as shown by quantitative measurements of all colonies larger than 50 mm in diameter. C: H-RASG12V and N-RASG12V only partially restored the growth rate of the DLD-1 (KRASG13D/þ) parent line. Results are representative of three independent experiments.

(Fig. 2, part B). The consequences of constitutive K-RAS activity that makes its overexpression difficult in DKS-8 cells may also constitute the profound oncogenic contribution in the DLD-1 cell line.

with the transformed phenotype; and (2) that K-RAS may differentially utilize effector pathways differently across cell lines of similar origin (Fig. 3). Overexpression of H-RASG12V or N-RASG12V resulted in elevated levels of MEK and ERK1/2 phosphorylation, yet as shown previously, these different RAS isoforms promote the transformed phenotype to much lesser degrees than the endogenous K-RAS (Fig. 2). We additionally examined changes in two other canonical RAS effector pathways, RalGDS activation and activation of the PI(3)K/Akt pathway as measured by Ser473 phosphorylation of Akt. We detected no differences in either RalGDS activity or Akt phosphorlyation between HCT 116, DLD-1 cell lines and their respective isogenic derivatives (data not shown).

Oncogenic KRAS promotes MEK and ERK1/2 phosphorylation in HCT 116, but not DLD-1 cell lines

In light of the profound reduction in classical tumorigenic measures observed in HCT 116 and DLD1 cells upon disruption of the mutant KRAS allele, we sought to explore the molecular differences between the parent cell lines, the isogenic derivatives and the isogenic derivatives overexpressing H-RASG12V or NRASG12V to identify the pathways that K-RAS may differentially utilize to so aggressively promote the transformed phenotype. The best-characterized RAS effector pathway is the Raf/MEK/ERK serine/threonine kinase cascade. We measured phosphorylation levels of MEK and ERK1/2 across these cell lines to assess the correlation between ERK1/2 phosphorylation, RAS activity, and tumorigenicity. HCT 116 (KRASG13D/þ) showed elevated levels of MEK and ERK1/2 phosphorylation relative to HKE-3 (KRASþ/) as expected, though phosphorylation levels were comparable between DLD-1 (KRASG13D/þ) and DKS-8 (KRASþ/), suggesting that (1) K-RAS-mediated activation of this pathway does not necessarily correlate Journal of Cellular Physiology DOI 10.1002/jcp

Oncogenic K-RAS downregulates Rap1 activity to promote cell motility in DLD-1 cells but not HCT 116 cells

In further analysis of possible signaling changes that may be unique to constitutive K-RAS signaling, we analyzed the activation state of another effector pathway, Rap1. We utilized a RalGDS Rap1 binding domain fusion protein to precipitate Rap1 in its GTP-bound state. We observed downregulation of active Rap1 as detected by this assay in DLD-1 though not in DKS-8 cells expressing either control vector, H-RASG12V or N-RASG12V. Total Rap1 levels were consistent across cell

744

KELLER ET AL.

Fig. 3. Oncogenic K-RAS differentially activates the RAF/MEK/ ERK1/2 pathway in HCT 116 and DLD-1 cell lines relative to their isogenic derivatives. Immunoblot for phospho- and total MEK and ERK1/2 reveal constitutive K-RAS activity correlates differently with the activation status of this pathway as shown by phosphorylation. Elevated phosphorylation of this pathway is not a conserved feature of

K-RAS-mediated transformation or maintenance of the transformed state, and similarly, is not sufficient for the transformed phenotype, given that oncogenic H-RAS and N-RAS engage this effector pathway effectively yet do not restore the transformed phenotype of the parent lines. A: Total and phospho-MEK. B: Total and phospho-ERK1/2. Results representative of three independent blots.

lines (Fig. 4). Consistent with our findings above that endogenous oncogenic K-RAS engages effector pathways differently across cell lines, downregulation of Rap1 activity was not observed in HCT 116 and derivative cell lines. Of the pathways tested, downregulation of Rap1 is the only one in which we observe an effect mediated by endogenous K-RASG13D but not by overexpressed NRASG12V or H-RASG12V. The K-RAS-specific nature of this downregulation led us to hypothesize that Rap1 signaling is an important mediator of the tumorigenic behavior of these cells. To investigate this possibility, we generated cell lines expressing constitutively active Rap1 mutants (G12V) or siRNA sequences which specifically knockdown levels of Rap1 (Fig. 5, part A). Rap1 mediates cell/substrate (Bos et al., 2003) and cell/cell adhesion (Knox and Brown, 2002; Price et al.,

2004; Zhang et al., 2005). Given these functional roles, we compared the motility of DLD-1 cells expressing either wild-type Rap1, constitutively activated Rap1 or cells expressing siRNA constructs specific for Rap1 in wound healing assays. We observed significantly different rates of wound closure in these various cell lines as well as changes in the morphology of the migratory edge of the monolayer. DLD-1 cells overexpressing wild-type Rap1, and more dramatically, expressing constitutively active Rap1 (G12V), migrated more slowly than control cells in this assay (Fig. 5, parts B and C). Consistent with this, cells with decreased total levels of Rap1 exhibited dramatically increased motility relative to control (Fig. 5, parts B and C). These results were not due to differences in proliferation rates, as neither Rap1 overexpressors nor knockdown cell lines display significantly altered growth rates relative to control (data not shown). We also examined the localization of E-cadherin in these various cell lines to assess the contribution of dysregulated cell/cell adhesion to altered motility. Constitutively active Rap mutants display stronger lateral E-cadherin staining than do those cells in which Rap1 expression has been knocked down, consistent with enhanced cell/cell adhesion and reduced motility (Fig. 5, part D).

Fig. 4. Oncogenic K-RAS selectively suppresses Rap1 activity in DLD-1 but not HCT 116 cells. Rap1 activity was gauged by use of recombinant RalGDS Rap1 binding domain. This assay demonstrated that constitutive K-RAS signaling suppresses Rap1 activity in the DLD-1 parent line, but not in HCT 116 relative to their respective isogenic derivatives. Neither constitutive H-RAS or N-RAS activity mediate this effect. Results representative of three independent blots.

Journal of Cellular Physiology DOI 10.1002/jcp

2D DIGE analysis of cells expressing activated RAS isoforms reveals commonalities in classes of proteomic changes and differences in intensity according to isoform

Given our findings that the same endogenous mutation to KRAS can engage common effector pathways differently even in cell lines of common tissue origin, we sought to assess more comprehensively the changes

VARIABLE ONCOGENIC SIGNATURE OF K-RAS

745

Fig. 5. K-RAS-mediated suppression of Rap1 activity facilitates motility in DLD-1 cells. A: Generation of stable cell lines overexpressing wild-type Rap1 or the constitutively active Rap1 mutant (G12V); siRNA sequences were identified against Rap1 and lentivirally transduced to create stable cell lines. B and C: Rap1 activity correlates with motility and cell morphology in wound healing assays. Constitutive Rap1 activity retarded the rate of wound closure in a monolayer of DLD-1 cells (i) in contrast to control (ii) or DLD-1 cells in which Rap1 had been knocked down (iii). In the absence of Rap1 in these lines, cells migrated more rapidly than the control population.

Insets show differential cell morphology associated with Rap1 activity: constitutive Rap1 activity produces a tightly packed healing edge relative to the less densely packed cell of the Rap1 knockdown line in which cells appear to be migrating into the wound. Images taken approximately 48 h postwounding. D: Rap1 activity and retarded migration correlate with redistribution of E-cadherin to the cell membrane, suggesting intact adherens junctions. Rap1 mutant and knockdown lines proliferate at a rate comparable to that of the parent line such that differential proliferation does not explain the differences observed here (data not shown).

resulting from KRAS mutation. We utilized 2D DIGE technology to compare the proteomic profiles of the DLD-1 (KRASG13D/þ) parent line plus vector against DKS-8 (KRASþ/) plus vector, H-RASG12V or N-RASG12V (the DLD-1 ‘‘suite’’). We predicted that the changes that result from constitutive RAS activation might fall into two classes: those that are common to all RAS isoforms but that may differ in intensity and those that are isoform-specific. Through this analysis, we hoped to eliminate proteomic changes common to all RAS isoforms and thus to identify proteomic changes unique to K-RAS that may underlie the unique oncogenicity of constitutive K-RAS signaling. The experimental design and methodology for these studies is illustrated in Supplementary Table 1 and is

based on a strategy we have previously reported (Friedman et al., 2004). We conducted this analysis in parallel on the DLD-1 and HCT 116 suites of cell lines. Changes in the abundance of individual proteins between the lines based on DLD-1 and those based on HCT 116 showed little conservation; levels of only one protein, cathepsin D, showed significant conserved changes across cell lines (Fig. 6). This change was observed at the protein level as well as at the mRNA level as detected by TaqMan analysis (Fig. 6). Changes to broader classes were conserved, however. For example, endogenous oncogenic K-RAS showed downregulation of several protein components of the ubiquitin/ proteasome machinery. Similarly, proteins with a known role in survival and apoptosis regulation also

Journal of Cellular Physiology DOI 10.1002/jcp

746

KELLER ET AL.

Fig. 6. Cathepsin D is the only conserved proteomic change in a comprehensive 2D-DIGE analysis of HCT 116 and DLD-1 and their respective cell lines. The aspartyl protease cathepsin D is strongly downregulated at the protein (A) and mRNA (B) levels in the parent lines (KRASG13D/þ) relative to their isogenic derivatives (KRASþ/). Downregulation of this protein is K-RAS specific; while constitutive H-RAS and N-RAS signaling also promote its downregulation, they do so to a lesser degree. Part A shows the average fold decreases with associated Student’s t-test P-values reported below the axis.

were downregulated more strongly by endogenous oncogenic K-RAS signaling than by H-RASG12V or NRASG12V. We present a grouping of the significant proteomic changes by known function in Supplemental Table 2. One trend that emerged in the analysis of changes mediated by the various RAS family members was that for any given change (upregulation or downregulation of a given protein), oncogenic K-RAS produced a more significant effect, followed by H-RASG12V, then N-RASG12V (K > H > N), relative to the isogenic control. Thirty-four percent of changes identified in the DLD-1 suite fit this pattern as did 51% of the changes identified in the HCT 116 suite. Of all changes, endogenous K-RAS resulted in the most dramatic effect of the RAS isoforms (K > H or N) in 61% of changes in the HCT 116 suite and 52% of changes in the DLD-1 suite. Our complete data set is available as Supplementary Data (Supplementary Table 3). DISCUSSION

Oncogenic mutation of RAS genes is among the most common genetic events in human tumorigenesis, but Journal of Cellular Physiology DOI 10.1002/jcp

mutations of the three canonical RAS family members segregate strongly by tissue and tumor type, suggesting specificity of function of the individual isoforms in different intracellular environments. We report here results of an extensive analysis of the specific contribution of endogenous oncogenic K-RAS in human colorectal cancer cells that confirm not only functional differences between family members but also functional differences of an individual family member in different cellular contexts. The DLD-1 and HCT 116 cell lines have been extensively used in conjunction with their isogenic derivatives as a means of dissecting the mechanisms of transformation utilized by K-RAS (Shirasawa et al., 1993). Our observations in these systems reveal some of the complexity of this approach. Oncogenic changes accumulated during transformation are not simply additive, discrete units, but rather function in combination to achieve a cumulative effect. For example, our finding that H-RASG12V or N-RASG12V overexpression is maintained in DKS-8 and HKE-3 cells while

VARIABLE ONCOGENIC SIGNATURE OF K-RAS

reexpressed K-RAS4B is preferentially degraded suggests important differences in the oncogenic capacities of the various Ras family members and also reveals significant differences between these parent lines and their isogenic derivatives. The difficulty of restoring expression of oncogenic K-RAS we demonstrate is corroborated by earlier reports, and together, these findings support the conclusion that addition or removal of an oncogene may have myriad effects on cell physiology. This finding is consistent with previous reports that restoration of oncogenic signaling following its inactivation cannot reconstitute the transformed phenotype (Habets et al., 2001; Jain et al., 2002). Sudden disruption of oncogenic signaling, as happens following the disruption of the mutant KRAS allele in these parent cells, likely triggers a range of events (including, in some instances, cell death) that precipitate tumor regression and allow surviving cells to redefine their equilibrium (Chin et al., 1999; Felsher and Bishop, 1999; Pelengaris et al., 1999). For these reasons, we must carefully consider whether any change observed between the parent and isogenic lines can be attributed directly to constitutive K-RAS signaling. These cell lines represent a uniquely useful tool for studying the contribution of endogenous oncogenic RAS to the transformed state of a disease-relevant cell population as they obviate the need for overexpression and avoid the pitfalls associated with that strategy (Shields et al., 2000). We also note that although we use a single isogenically derived clonal population (HKE-3 and DKS-8) from each of these parent lines, these isogenic derivatives are representative of multiple clones characterized from each of the lines as shown previously (Shirasawa et al., 1993). The effects of constitutive K-RAS signaling that may promote the transformed phenotype are not necessarily conserved, even across cell lines derived from tumors of common origin. An excellent example of such a nonconserved effect is the suppression of Rap1 activity in the presence of constitutive K-RAS signaling in DLD-1 cells relative to HCT 116. DLD-1 cells show significantly reduced levels of active Rap1 compared to HCT 116. The isogenically derived DKS-8 cell line exhibits higher levels of Rap1 relative to DLD-1, suggesting a suppressive effect by K-RAS activity. The analogous HCT 116 derivative shows no such change. Rap1 is a known antagonist of K-RAS-driven transformation, and suppression of its activity is similarly observed in a mouse model of lung tumorigenesis that relies on oncogenic K-RAS (Kitayama et al., 1989; Sweet-Cordero et al., 2005). Overexpression of constitutively active Rap1 mutants, as well as RNA interference studies, demonstrate that Rap1 suppression results in enhanced cell motility and dysregulation of adherens junctions, both commonly observed changes in invasive carcinomas. The inconsistency of this suppression between the HCT 116 and DLD-1 cell lines suggests that Rap1 suppression may represent a selective advantage in the development of some, but not all, colon tumors. We also note that the presence of oncogenic K-RAS does not correlate strictly with ERK1/2 phosphorylation. RAS activation of the RAF/MEK/ERK serine/ threonine kinase cascade is the best-understood RAS effector pathway and is considered an important mechanism of RAS-driven transformation. Underscoring the significance of this pathway is the presence of activating BRAF mutations in approximately 12% of colorectal cancers in nonoverlapping fashion with KRAS mutations, suggesting a possibly epistatic relationship between these oncogenes (Davies et al., 2002). While Journal of Cellular Physiology DOI 10.1002/jcp

747

HCT 116 cells display elevated levels of pERK1/2 relative to HKE-3 cells as expected, DLD-1 cells display minimal activation of this pathway as measured by ERK1/2 and MEK phosphorylation relative to their isogenic derivatives. Moreover, overexpression of HRASG12V or N-RASG12V in either the HKE-3 or DKS-8 cell lines drives phosphorylation of ERK1/2, yet these cells are weakly tumorigenic by soft agar assay (Fig. 2) and do not recapitulate the morphology or proliferation rates of the parent lines (data not shown). Despite comparable levels of total ERK1/2 and MEK, different levels of phosphorylation in HCT 116 relative to DLD-1 suggest that activation of this canonical pathway does not appear to be an absolute requirement of the transformed state. These results are consistent with xenograft studies that suggest that activation of this pathway may be more important in establishing rather than maintaining tumors (Lim and Counter, 2005). Our results provide additional evidence that regulation of this pathway in some instances becomes insensitive to constitutive K-RAS signaling in fully transformed cells. We also observe that activation of this pathway may result from constitutive signaling of any RAS isoform, and, therefore, that its activation is likely not sufficient to promote the transformed phenotype. 2D DIGE analysis of DLD-1 and HCT 116 and their respective derivatives suggests K-RAS-mediated proteomic changes are conserved across classes of proteins, rather than individual proteins themselves. In only one instance, the downregulation of cathepsin D levels in the presence of oncogenic K-RAS, do we see a strongly conserved individual change across DLD-1 and HCT 116 cell lines. More typically, we observe a pattern of downregulated expression of any given protein most strongly in the case of endogenous oncogenic K-RAS, less strongly by H-RASG12V and least strongly by NRASG12V. The hierarchical effects of the different RAS family members on the up- or downregulation of a given protein are conserved in a significant percentage of cases in this data set. Far more rare are instances of proteins that change in the setting of constitutive activity of one isoform but not the others. Further, the hierarchy of RAS-mediated effects we observe here, (K > H>N) or (K > H or N), is the same as that seen in the functional reconstitution assays: endogenous oncogenic K-RAS conferred the most transformed phenotype by soft agar growth assays and proliferation measurements, expression of H-RASG12V or N-RASG12V only partially recapitulated these behaviors. These data imply that different RAS isoforms may not mediate their oncogenic effects so much through functions that are unique to individual isoforms as through their relative potencies in serving common functions. We note that this would not preclude different Ras isoforms’ coordinated interaction in particular pathways. Cathepsin D, the single conserved change across data sets, is also one of the most significantly downregulated proteins observed. Cathepsin D in some settings mediates early apoptotic pathways, specifically the activation of Bax and the efflux of apoptosis-inducing factor from the intermembrane space of mitochondria (Bidere et al., 2003). Examination of 59 colorectal tumors revealed altered cathepsin D expression in the majority of samples. Interestingly, while approximately half of samples showed downregulation relative to normal tissue, one-third of samples showed increased expression (Iacobuzio-Donahue et al., 2004). It is notable that the percentage of samples that show

748

KELLER ET AL.

cathepsin D downregulation roughly corresponds to the frequency of KRAS mutation in these lesions (Vogelstein et al., 1988), though overlap between KRAS mutation and cathepsin D downregulation is not known. Oncogenic mutation of KRAS occurs in approximately half of colorectal tumors and provides a significant oncogenic contribution (Fearon and Vogelstein, 1990; Shirasawa et al., 1993). This leaves approximately half of colon tumors that arise without the oncogenic contribution of constitutive K-RAS signaling, thus implying multiple means of transformation. It is perhaps not surprising that a specific RAS-driven pathway or set of effectors is therefore not conserved across tumor-derived cell lines. Given the extraordinary selection pressures that exist within a heterogeneous tumor cell population, individual cells likely utilize selected aspects of constitutive RAS signaling depending on context and changing selective pressures. It has been demonstrated that the requirements for oncogenic RAS change over the course of tumor initiation and maintenance which may well explain the differences we observe here (Lim and Counter, 2005). Parallel studies of these cell lines demonstrate that oncogenic K-RAS may not produce a fixed constellation of molecular changes but rather promote transformation by a dynamic combination of events which are not necessarily conserved even in cell lines derived from tumors of common origin and that display similar genetic background (Supplemental Table 4) (Rodrigues et al., 1990; Shirasawa et al., 1993; Markowitz et al., 1995; Ilyas et al., 1997; Lengauer et al., 1997; Morin et al., 1997; Cahill et al., 1998; Fodde et al., 2001; Gayet et al., 2001). The lack of conservation of K-RASmediated changes across tumor lines argues for a more nuanced and dynamic model of RAS-driven transformation, one in which cells may select for certain K-RASmediated effects that confer survival, proliferative, or motility advantages depending on the rapidly changing context of a tumor. Constitutive K-RAS activity may simply engage a greater range of effector pathways than other RAS isoforms or activate commonly utilized pathways more vigorously as the results of our proteomic studies suggest. Either case provides a conceivable selective advantage. The fact that the KRAS gene codes for two splice variants (4A and 4B), which likely display significant differences in membrane affinity and therefore possibly effector access, may explain this phenomenon (Voice et al., 1999). The results of these studies serve as a powerful reminder of the protean nature of any molecular or genetic change and of the importance of context in understanding the contribution of any oncogene. The divergence we observe in the effects of a common oncogenic change across cell lines underscores the qualities of RAS signaling that make it both such a difficult target of study and such a potent and dangerous oncogene. ACKNOWLEDGMENTS

This work has been supported by the NCI SPORE Grant in GI Cancer CA 95103 and National Institutes of Health Grant CA46413. We also acknowledge support for Proteomics through the Vanderbilt Academic Venture Capital Fund. LITERATURE CITED Bidere N, Lorenzo HK, Carmona S, Laforge M, Harper F, Dumont C, Senik A. 2003. Cathepsin D triggers Bax activation, resulting in selective apoptosisinducing factor (AIF) relocation in T lymphocytes entering the early commitment phase to apoptosis. J Biol Chem 278:31401–31411.

Journal of Cellular Physiology DOI 10.1002/jcp

Bos JL. 1989. ras oncogenes in human cancer: A review. Cancer Res 49:4682– 4689. Bos JL, de Bruyn K, Enserink J, Kuiperij B, Rangarajan S, Rehmann H, Riedl J, de Rooij J, van Mansfeld F, Zwartkruis F. 2003. The role of Rap1 in integrinmediated cell adhesion. Biochem Soc Trans 31:83–86. Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Markowitz SD, Kinzler KW, Vogelstein B. 1998. Mutations of mitotic checkpoint genes in human cancers. Nature 392:300–303. Chin L, Tam A, Pomerantz J, Wong M, Holash J, Bardeesy N, Shen Q, O’Hagan R, Pantginis J, Zhou H, Horner JW II, Cordon-Cardo C, Yancopoulos GD, DePinho RA. 1999. Essential role for oncogenic Ras in tumour maintenance. Nature 400:468–472. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, Chenevix-Trench G, Riggins GJ, Bigner DD, Palmieri G, Cossu A, Flanagan A, Nicholson A, Ho JW, Leung SY, Yuen ST, Weber BL, Seigler HF, Darrow TL, Paterson H, Marais R, Marshall CJ, Wooster R, Stratton MR, Futreal PA. 2002. Mutations of the BRAF gene in human cancer. Nature 417:949–954. Downward J. 2003. Targeting RAS signalling pathways in cancer therapy. Nat Rev Cancer 3:11–22. Fearon ER, Vogelstein B. 1990. A genetic model for colorectal tumorigenesis. Cell 61:759–767. Felsher DW, Bishop JM. 1999. Reversible tumorigenesis by MYC in hematopoietic lineages. Mol Cell 4:199–207. Fodde R, Kuipers J, Rosenberg C, Smits R, Kielman M, Gaspar C, van Es JH, Breukel C, Wiegant J, Giles RH, Clevers H. 2001. Mutations in the APC tumour suppressor gene cause chromosomal instability. Nat Cell Biol 3:433– 438. Friedman DB, Hill S, Keller JW, Merchant NB, Levy SE, Coffey RJ, Caprioli RM. 2004. Proteome analysis of human colon cancer by two-dimensional difference gel electrophoresis and mass spectrometry. Proteomics 4:793–811. Gayet J, Zhou XP, Duval A, Rolland S, Hoang JM, Cottu P, Hamelin R. 2001. Extensive characterization of genetic alterations in a series of human colorectal cancer cell lines. Oncogene 20:5025–5032. Glickman MH, Ciechanover A. 2002. The ubiquitin-proteasome proteolytic pathway: Destruction for the sake of construction. Physiol Rev 82:373–428. Habets GG, Knepper M, Sumortin J, Choi YJ, Sasazuki T, Shirasawa S, Bollag G. 2001. cDNA array analyses of K-ras-induced gene transcription. Methods Enzymol 332:245–260. Hirakawa T, Ruley HE. 1988. Rescue of cells from ras oncogene-induced growth arrest by a second, complementing, oncogene. Proc Natl Acad Sci USA 85:1519–1523. Iacobuzio-Donahue C, Shuja S, Cai J, Peng P, Willett J, Murnane MJ. 2004. Cathepsin D protein levels in colorectal tumors: Divergent expression patterns suggest complex regulation and function. Int J Oncol 24:473–485. Ilyas M, Tomlinson IP, Rowan A, Pignatelli M, Bodmer WF. 1997. Beta-catenin mutations in cell lines established from human colorectal cancers. Proc Natl Acad Sci USA 94:10330–10334. Jain M, Arvanitis C, Chu K, Dewey W, Leonhardt E, Trinh M, Sundberg CD, Bishop JM, Felsher DW. 2002. Sustained loss of a neoplastic phenotype by brief inactivation of MYC. Science 297:102–104. Kitayama H, Sugimoto Y, Matsuzaki T, Ikawa Y, Noda M. 1989. A ras-related gene with transformation suppressor activity. Cell 56:77–84. Knox AL, Brown NH. 2002. Rap1 GTPase regulation of adherens junction positioning and cell adhesion. Science 295:1285–1288. Lengauer C, Kinzler KW, Vogelstein B. 1997. Genetic instability in colorectal cancers. Nature 386:623–627. Lim KH, Counter CM. 2005. Reduction in the requirement of oncogenic Ras signaling to activation of PI3K/AKT pathway during tumor maintenance. Cancer Cell 8:381–392. Liu X, Constantinescu SN, Sun Y, Bogan JS, Hirsch D, Weinberg RA, Lodish HF. 2000. Generation of mammalian cells stably expressing multiple genes at predetermined levels. Anal Biochem 280:20–28. Markowitz S, Wang J, Myeroff L, Parsons R, Sun L, Lutterbaugh J, Fan RS, Zborowska E, Kinzler KW, Vogelstein B, et al. 1995. Inactivation of the type II TGF-beta receptor in colon cancer cells with microsatellite instability. Science 268:1336–1338. Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, Vogelstein B, Kinzler KW. 1997. Activation of beta-catenin-Tcf signaling in colon cancer by mutations in beta-catenin or APC. Science 275:1787–1790. Pelengaris S, Littlewood T, Khan M, Elia G, Evan G. 1999. Reversible activation of c-Myc in skin: Induction of a complex neoplastic phenotype by a single oncogenic lesion. Mol Cell 3:565–577. Price LS, Hajdo-Milasinovic A, Zhao J, Zwartkruis FJ, Collard JG, Bos JL. 2004. Rap1 regulates E-cadherin-mediated cell-cell adhesion. J Biol Chem 279:35127–35132. Rodrigues NR, Rowan A, Smith ME, Kerr IB, Bodmer WF, Gannon JV, Lane DP. 1990. p53 mutations in colorectal cancer. Proc Natl Acad Sci USA 87:7555– 7559. Rodriguez-Viciana P, Sabatier C, McCormick F. 2004. Signaling specificity by Ras family GTPases is determined by the full spectrum of effectors they regulate. Mol Cell Biol 24:4943–4954. Ruley HE. 1983. Adenovirus early region 1A enables viral and cellular transforming genes to transform primary cells in culture. Nature 304:602–606. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. 1997. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 88:593–602. Shields JM, Pruitt K, McFall A, Shaub A, Der CJ. 2000. Understanding Ras: ‘it ain’t over’ ‘til it’s over’. Trends Cell Biol 10:147–154. Shirasawa S, Furuse M, Yokoyama N, Sasazuki T. 1993. Altered growth of human colon cancer cell lines disrupted at activated Ki-ras. Science 260: 85–88. Sweet-Cordero A, Mukherjee S, Subramanian A, You H, Roix JJ, Ladd-Acosta C, Mesirov J, Golub TR, Jacks T. 2005. An oncogenic KRAS2 expression signature identified by cross-species gene-expression analysis. Nat Genet 37:48–55.

VARIABLE ONCOGENIC SIGNATURE OF K-RAS Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC, Leppert M, Nakamura Y, White R, Smits AM, Bos JL. 1988. Genetic alterations during colorectal-tumor development. N Engl J Med 319:525–532. Voice JK, Klemke RL, Le A, Jackson JH. 1999. Four human ras homologs differ in their abilities to activate Raf-1, induce transformation, and stimulate cell motility. J Biol Chem 274:17164–17170. Yan Z, Chen M, Perucho M, Friedman E. 1997a. Oncogenic Ki-ras but not oncogenic Ha-ras blocks integrin beta1-chain maturation in colon epithelial cells. J Biol Chem 272:30928–30936.

Journal of Cellular Physiology DOI 10.1002/jcp

749

Yan Z, Deng X, Chen M, Xu Y, Ahram M, Sloane BF, Friedman E. 1997b. Oncogenic c-Ki-ras but not oncogenic c-Ha-ras up-regulates CEA expression and disrupts basolateral polarity in colon epithelial cells. J Biol Chem 272: 27902–27907. Yan J, Roy S, Apolloni A, Lane A, Hancock JF. 1998. Ras isoforms vary in their ability to activate Raf-1 and phosphoinositide 3-kinase. J Biol Chem 273: 24052–24056. Zhang Z, Rehmann H, Price LS, Riedl J, Bos JL. 2005. AF6 negatively regulates Rap1-induced cell adhesion. J Biol Chem 280:33200–33205.