|Home | About | Journals | Submit | Contact Us | Français|
The three closely related human Ras genes, Hras, Nras, and Kras, are all widely expressed, engage a common set of downstream effectors, and can each exhibit oncogenic activity. However, the vast majority of activating Ras mutations in human tumors involve Kras. Moreover, Kras mutations are most frequently seen in tumors of endodermally derived tissues (lung, pancreas, and colon), suggesting that activated Kras may affect an endodermal progenitor to initiate oncogenesis. Using a culture model of retinoic acid (RA)-induced stem cell differentiation to endoderm, we determined that while activated HrasV12 promotes differentiation and growth arrest in these endodermal progenitors, KrasV12 promotes their proliferation. Furthermore, KrasV12-expressing endodermal progenitors fail to differentiate upon RA treatment and continue to proliferate and maintain stem cell characteristics. NrasV12 neither promotes nor prevents differentiation. A structure-function analysis demonstrated that these distinct effects of the Ras isoforms involve their variable C-terminal domains, implicating compartmentalized signaling, and revealed a requirement for several established Ras effectors. These findings indicate that activated Ras isoforms exert profoundly different effects on endodermal progenitors and that mutant Kras may initiate tumorigenesis by expanding a susceptible stem/progenitor cell population. These results potentially explain the high frequency of Kras mutations in tumors of endodermal origin.
Somatic activating Ras mutations are detected in about 15 to 20% of all human malignancies (1, 10), highlighting the importance of Ras GTPase-mediated signaling pathways in oncogenesis (10, 21, 23, 31). These mutations, which give rise to a protein that is defective for GTP hydrolysis and, therefore, remains constitutively active in a GTP-bound form, have been detected in each of the three closely related human Ras genes—Hras, Nras, and Kras. However, the vast majority of mutations detected in human cancers arise in the Kras gene. This does not appear to reflect clear differences in the biochemical properties of the three Ras proteins, as they each exhibit very similar GTP binding and hydrolysis properties and their interactions with effector targets in vitro are largely indistinguishable (2, 10, 13, 21, 73, 78). In addition, mutationally activated forms of each Ras isoform exhibit transforming activity in cell culture models, such as NIH 3T3 fibroblasts. Notably, Hras is reported to be a considerably more potent oncogene in such models (23, 54, 73, 89) and has been used for the vast majority of cell-based studies of Ras-induced oncogenic transformation thus far. It has also been suggested that tissue-specific expression differences among the Ras genes might explain their association with tumors in distinct tissues, although these three Ras isoforms are all widely expressed (73, 78). Differences in the carboxyl termini of the Ras proteins, which determine their modification by lipids and consequent subcellular membrane distribution, are likely to contribute to their distinct oncogenic potential in vivo (4, 31, 61, 73).
The fact that activating Kras mutations are detected with highest frequency in tumors of endodermally derived tissues, such as those of lung (35%), pancreas (95%), and colon (30%) (1, 11, 21, 31, 88), raises the possibility that mutationally activated Kras exerts its oncogenic activity at the level of an endodermal precursor or stem cell. Indeed, a recent study revealed that activated Kras can expand a bronchoalveolar stem cell population in culture and that this may explain the role of Kras in promoting lung adenocarcinomas in a mouse model (48).
One of the most thoroughly studied in vitro models of stem cell differentiation along the endodermal lineage is the F9 mouse embryonal carcinoma stem cell model (82). Cultured F9 cells express well-established stem cell markers, including Oct3/Oct4, Nanog, SOX2, and SSEA-1, and they undergo self-renewal in vitro (12, 16, 65, 92). When F9 cells are treated with retinoic acid (RA) for 7 to 10 days, the expression of these stem cell markers is lost, the cells stop proliferating, they undergo a morphological transformation, and they begin to express genes associated with early endoderm differentiation, including GATA4 (14, 32). The physiologic relevance of this model is well supported by substantial evidence, indicating a requirement for RA signaling in the differentiation of endodermal tissues during normal vertebrate development (27, 55, 57, 59).
Previous studies have revealed that the expression of mutationally activated Hras (HrasV12) in F9 cells promotes their differentiation to early endoderm in the absence of RA (86, 95). To examine a potentially distinct consequence of activating the other Ras isoforms in endodermal precursors, we have compared the effects of the three different activated Ras genes in this system and identified a dramatic difference in their activities. In striking contrast with the case for HrasV12, expressing KrasV12 promotes an expansion of the F9 population and prevents cell differentiation in response to RA. NrasV12 is essentially inert in this system. This striking difference in biological activity among these closely related Ras proteins reflects differences in their C termini and may account for the high frequency of activating mutant Kras alleles detected in tumors of endodermally derived tissues.
Primary antibodies used include those raised against Nanog (Bethyl); SSEA1 and SSEA3 (gifts from D. Solter); Oct3/Oct4, pan-ras, and GM130 (Transduction Laboratories); Sox2 and actin (Sigma); pan-ras, Kras, and Hras (Calbiochem); and Raf-1, GATA4, and α-tubulin (Santa Cruz).
F9 cells (a gift from S. Strickland) and PCC4 cells were maintained on gelatin-coated tissue culture dishes in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum (Gibco). To induce endodermal differentiation, F9 or PCC4 cells were seeded sparsely and incubated with 0.2 to 1.0 μM all-trans RA (Sigma). Cells were maintained in the dark, and the medium was replaced every 48 h. F9 or PCC4 cells were transfected using Lipofectamine 2000 or nucleoporated using a Nucleofector II device (Amaxa). For stable transfections, cells were selected for their resistance to G418 (Gibco). For RA-resistant, F9-derived cells, cells were first selected in the appropriate antibiotic for 7 to 10 days after transfection and then incubated with antibiotic plus RA for an additional 7 to 10 days. Dishes were fixed with methanol and stained with Giemsa for documentation. Resistant colonies were either cloned using cloning cylinders or pooled as polyclonal populations and maintained in the presence of the selecting antibiotic, plus or minus RA. The expression vectors used were pEYFP-C1-Kras12V, pEYFP-C1-Hras61L, pEGFP-C3-Nras12D, pEYFP-C1-Hras12V-KrasTail, pEYFP-C1-Kras12V-HrasTail, pcRaf1-HrasTail, and pcRaf1-KrasTail, which were described previously (19, 20); pWP1-HA-KrasG12V, pWP1-HA-KrasG12V-E37G, pWP1-HA-KrasG12V-T35S, pWP1-HA-KrasG12V-Y40C, 2xmyc-NH2-tagged HrasG12V, 3XHA-NH2-tagged KrasG12V, and 3XHA-NH2-tagged NrasG12V in pcDNA3.1 (Invitrogen), which were obtained from the Guthrie Foundation cDNA Resource Center; and pcDNA3.1 with 3XHA as vector control. pMAXGFP (Amaxa) was used to evaluate the efficiency of transfection. The metabolic inhibitors LY294002 and UO126 were used at 10 μM, with solvent dimethyl sulfoxide (DMSO) used as control.
Growth curves were established by plating cells in triplicate or quadruplicate parallel dishes. Cells were counted daily with a hemocytometer to yield cell numbers. Alternatively, relative cell numbers were determined at daily intervals by using SYTO60, a red fluorescent nucleic acid stain (Molecular Probes). Triplicate or quadruplicate dishes were processed for SYTO60 staining, scanned, and quantified with the Li-Cor Odyssey system. Growth curves were generated using Microsoft Excel, with error bars indicating standards of deviation, and P values were generated using one-way analysis of variance. Experiments were performed two to four times.
Cell pellets were collected by removing the medium and washed twice with ice-cold phosphate-buffered saline (PBS), and cells were scraped into a microfuge tube and pelleted at 2,000 × g for 5 min at 4°C. The supernatants were aspirated, and the pellets were then maintained at −20°C. For immunoblotting, pellets were thawed on ice and lysed with Laemmli sample buffer without glycerol or reducing agent, with protease (Sigma) and phosphatase (Calbiochem) inhibitor cocktails, and boiled for 10 min. The supernatants were quantified using detergent-compatible DC protein assay (Bio-Rad) and a VMax kinetic microplate reader (Molecular Devices), with bovine serum albumin as a standard. Concentrations were adjusted with lysis buffer. Laemmli sample buffer (5× concentrated) was then added to the samples. Equal protein concentrations were loaded onto sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred to nitrocellulose filters, which were blocked overnight at 4°C. The blots were incubated for 1 h at room temperature with the indicated primary antibodies. Secondary antibodies, conjugated to horseradish peroxidase (Sigma or Cell Signaling Technology), were also incubated for 1 h at room temperature. Signal visualization was achieved using chemiluminescence (Pierce) and BioMax MR film (Kodak). Films were scanned using ScanMaker 8700 (Microtek).
For immunostaining, cells were grown on 12-mm glass coverslips and coated with gelatin or Cell-Tak (Becton Dickinson). The cells were fixed with either 2% paraformaldehyde or −20°C methanol and washed with PBS. Ammonium chloride in PBS was used to diminish autofluorescence. Where necessary, cells were permeabilized with 0.2% Triton X-100 in PBS for 10 min on ice. Blocking was performed with PBS containing 2% serum for 30 min at room temperature. Primary antibodies and secondary antibodies, conjugated to Texas Red-X or Oregon Green (Molecular Probes), were diluted in blocking buffer and incubated with cells for 30 min at room temperature in a humidified chamber. After each incubation, cells were washed three times with PBS. Counterstaining with Hoechst 33258 (Molecular Probes) was performed at the second wash. Coverslips were then mounted on slides using Airvol, visualized using a Zeiss Axioplan2 immunofluorescence microscope, and imaged using an AxioCamMR digital camera and AxioVision 4.5 software. Phase-contrast microscopy was conducted using a Nikon Diaphot inverted microscope and an Olympus SP-350 digital camera.
To assess the subcellular localization of Ras in F9 cells, undifferentiated F9 cells or F9 cells treated with 200 nM RA for 5 days to induce differentiation were plated at 3 × 105 cells per plate into 35-mm dishes containing a 15-mm glass coverslip-covered cutout (MatTek). Cells were transfected 24 h later with either pEYFP-C1-Kras12V, pEYFP-C1-Hras61L, pEYFP-C1-Hras12V-KrasTail, pEYFP-C1-Kras12V-HrasTail, or pEYFP-C1-GalT (galactosyl transferase, a marker for the Golgi) by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Twenty-four hours posttransfection, cells were imaged with a Zeiss 510 inverted laser scanning confocal microscope. A minimum of five 0.45-μm z slices were acquired for each cell, and representative images were chosen to display plasma membranes and/or endomembranes. All images shown are representative of more than 90% of cells examined.
Activating Kras mutations are detected most frequently in tumors of endodermally derived tissues, such as those of lung (35%), pancreas (95%), and colon (30%) (1, 2, 9, 10, 88), suggesting that mutant Kras may initially exert its oncogenic activity in an endodermal progenitor or stem cell. To test the hypothesis that activated Kras can influence the properties of endodermal stem/progenitor cells in a manner distinct from activated Hras or Nras, we exploited the F9 mouse embryonal carcinoma model of stem cell differentiation to endoderm (82). These cells exhibit several stem cell properties, including the ability to undergo self-renewal, the capacity to differentiate along distinct lineages, and the expression of stem cell markers, including Oct3/Oct4, Nanog, SOX2, and SSEA1 (Fig. (Fig.1A).1A). In response to continuous treatment with RA, the cells undergo morphological transformation associated with reduced proliferative capacity (Fig. (Fig.1B),1B), they lose their stem cell markers, and they begin to express endodermal markers (GATA4 and SSEA3) (Fig. 1C and D and data not shown). Thus, this cell culture model recapitulates aspects of stem cell differentiation to early endoderm during vertebrate development, which also requires RA (27, 55, 57, 59).
The expression of mutationally activated Hras (HrasV12) in F9 cells is sufficient to promote their differentiation to primitive endoderm in the absence of RA (86, 95). To identify a potentially distinct activity of the other Ras isoforms in F9 differentiation, we established expression vectors for HrasV12, NrasV12, and KrasV12 (Fig. (Fig.2).2). We initially confirmed the functionality of each of these three proteins by demonstrating their focus-forming activities in transfected NIH 3T3 fibroblasts (data not shown), and we then compared their activities in the F9 model system. As was previously reported, we found that HrasV12 promotes the differentiation of F9 cells in the absence of RA. Within 2 days of transfection, HrasV12-transfected cells lose expression of the stem cell markers Oct3/Oct4 (Fig. (Fig.2A2A and and3A)3A) and SSEA1 (data not shown) and exhibit altered morphologies (Fig. (Fig.3A).3A). By 7 days posttransfection, they cease proliferating and assume epithelial morphologies (Fig. (Fig.3B3B and data not shown) and they cannot be propagated further. In contrast to HrasV12, F9 cells transfected with NrasV12 or KrasV12 are morphologically indistinguishable from vector-transfected F9 cells (Fig. (Fig.2A2A and and3B)3B) and they continue to express stem cell markers and can be readily passaged (Fig. (Fig.2A2A and data not shown). We observed similar properties among these Ras isoforms following transfection with a range of plasmid concentrations or with different vector backbones and fusion tags, indicating that the observed biological differences do not reflect expression level differences (data not shown).
We next examined the consequences of sustained expression of the activated Ras isoforms in F9 cells. Two weeks posttransfection, vector-transfected cell colonies can be readily selected, whereas HrasV12-transfected cells fail to yield detectable colonies (Fig. (Fig.3C),3C), reflecting HrasV12's ability to promote differentiation and growth arrest (Fig. 3A and B). However, KrasV12- and NrasV12-transfected F9 cells remain undifferentiated and can be readily propagated (Fig. 3B and C and data not shown). Upon RA treatment, vector-transfected and NrasV12-transfected cells differentiate (Fig. (Fig.3D)3D) and cannot be propagated beyond two passages (Fig. (Fig.4E4E and data not shown). RA-treated, HrasV12-transduced cells, which are already undergoing HrasV12-induced differentiation, senesce and eventually die (Fig. (Fig.3D3D).
In striking contrast to the case for HrasV12- or NrasV12-transfected F9 cells, KrasV12-transfected cells appear to remain undifferentiated upon RA treatment (Fig. (Fig.3D3D and and4A).4A). Stable KrasV12-expressing cells were expanded for further characterization. They exhibit increased proliferative potential and a reduced serum requirement relative to parental F9 cells (Fig. 4B to D) and can be passaged indefinitely in the presence of RA without obvious consequence (beyond passage 25) (Fig. (Fig.4E).4E). Moreover, unlike RA-treated F9 cells, KrasV12-expressing cells cultured long term in RA retain stem cell markers (Fig. 4F to H). Notably, these cells remain RA responsive, as indicated by the fact that RA treatment promotes the loss of SSEA1 (Fig. (Fig.4G).4G). We determined that multiple, clonally derived KrasV12-transfected cell lines expressing widely varying levels of KrasV12 protein (Fig. (Fig.4F4F and data not shown) exhibit indistinguishable phenotypes with respect to resistance to differentiation, further confirming that the distinct activities of the Ras isoforms do not reflect expression differences. Thus, KrasV12 is uniquely able to promote the expansion of a population of undifferentiated endodermal stem/progenitor cells.
To determine whether the distinct activities of the activated Ras isoforms are limited to a particular stem/progenitor lineage, we performed analogous studies of PCC4 cells, a mouse embryonic carcinoma stem cell line that responds to RA by differentiating into mesenchyme but not endoderm (Fig. (Fig.5A).5A). The expressions of the various activated Ras isoforms fail to induce PCC4 differentiation, and they each yield G418-resistant cells that can be readily propagated (Fig. 5B to D and data not shown). Moreover, RA-treated HrasV12-, KrasV12-, or NrasV12-expressing PCC4 cells remain indistinguishable from vector-transfected cells in that they all differentiate into mesenchymal cells that cannot be serially passaged (Fig. (Fig.5E5E and data not shown). Together, these findings suggest that the lineage-specific context of the progenitor/stem cell determines its response to the various activated Ras proteins and that stem cells partially committed to an endodermal fate (e.g., F9 cells) exhibit very different biological responses to the three mutant Ras isoforms. These findings also suggest that mutationally activated Kras can uniquely expand endodermal stem/progenitor cells.
Hras and Kras are highly homologous, except for their carboxy-terminal tails, which determine the association with cellular membranes (4, 19, 31, 73, 81). Kras is largely restricted to the plasma membrane, while Hras localizes to plasma membrane and Golgi. In F9 cells, Kras is similarly restricted to plasma membrane, whereas Hras is detected both at plasma membrane and at Golgi (Fig. (Fig.6A).6A). To directly test a role for the Ras carboxyl terminus in the F9 model, we expressed an HrasV12 chimeric protein in which the carboxyl terminus was substituted with that of Kras (Fig. (Fig.6B,6B, lane HrasKTail). As expected, HrasKTail no longer localizes to Golgi (Fig. (Fig.6A6A and data not shown). Unlike HrasV12, but similarly to KrasV12, HrasKTail does not repress Oct3/Oct4 expression or induce differentiation (Fig. 6C to E). It is well tolerated in F9 cells, and we could readily establish stably transfected colonies that express stem cell markers (Fig. 6D and E and data not shown). Moreover, RA-treated HrasKTail-expressing cells, like KrasV12-expressing cells, fail to differentiate and they maintain stem cell-like properties, including the expression of Oct3/Oct4 and the ability to be passaged indefinitely (Fig. (Fig.6E6E and data not shown). In a complementary experiment, we examined the properties of a Ras chimera in which the Hras carboxyl terminus was fused to Kras (Fig. (Fig.6B,6B, lane KrasHTail). As expected, this chimeric protein demonstrated Golgi localization due to the Hras tail (Fig. (Fig.6A).6A). However, unexpectedly, it did not repress Oct3/Oct4 expression or induce F9 differentiation (Fig. 6C to E and data not shown). Thus, while the ability of Hras to induce differentiation seems dependent upon its ability to signal from the Golgi, the ability of Kras to maintain a stem cell phenotype is not strictly dependent on its subcellular localization. However, a critical role for sequences within the Kras C terminus is supported by the additional observation that the expression of an activated version of the less commonly detected Kras isoform, Kras-4A, a splice isoform which differs from the far more abundantly expressed Kras-4B isoform (used in all other studies here) only at the C terminus, fails to promote RA-resistant stem cell maintenance in F9 cells and instead causes apoptosis (data not shown).
The distinct and seemingly opposing phenotypes associated with HrasV12 and KrasV12 expression in F9 cells prompted us to compare their signaling outputs in the context of F9 differentiation. Ras interacts with multiple downstream effectors via a highly conserved, 9-amino-acid, amino-terminal domain (switch I), encompassing amino acid residues 32 to 40 (91). The three best-established Ras effectors implicated in its oncogenic properties include the Raf kinase, phosphatidylinositol 3-kinase (PI-3 kinase), and the Ral GTPase activator RalGDS (2, 10, 13, 21, 28, 56, 73, 78). Notably, the PI-3 kinase-AKT, Raf-MEK-extracellular signal-regulated kinase, and RalGDS-Ral GTPase pathways have previously been implicated in F9 differentiation by RA or HrasV12 (6, 86, 87) as well as in oncogenic Ras transformation (28, 91). Therefore, we examined the role of these Ras effectors in vector-transfected and KrasV12-expressing F9 cells. Pharmacologic inhibition of PI-3 kinase (with LY294002) (Fig. (Fig.7A)7A) does not affect the proliferation of vector-transfected F9 cells or their expression of Oct3/Oct4 (Fig. 7B and C, panel b, and 8B and T). However, PI-3 kinase is required for RA-induced expression of GATA4 and endoderm differentiation (Fig. (Fig.7C,7C, panel h, and 8K and W). Moreover, PI-3 kinase is not required for KrasV12-mediated stem cell maintenance (Fig. 8BB and TT), but is required for sustained proliferation (Fig. (Fig.7B,7B, Ce, and Ck) and to prevent the loss of stem cell maintenance by RA treatment (Fig. 8KK and WW). In contrast, LY294002-treated F9 cells transiently transfected with HrasV12 maintain Oct3/Oct4 expression and consequently fail to undergo differentiation and growth arrest (Fig. (Fig.7D).7D). Together, these results indicate a complex role for PI-3 kinase downstream of both activated Hras and activated Kras that involves distinct requirements in endodermal stem/progenitor cell maintenance, proliferation, and differentiation as well as resistance to the actions of RA.
Pharmacologic MEK inhibition (Fig. (Fig.7A)7A) does not inhibit stem cell maintenance of vector-transfected F9 cells (Fig. (Fig.7B7B and Cc and 8C and U), but does block RA-induced differentiation to endoderm (Fig. (Fig.7B7B and compare panels Cg and Ci and 8, compare panels J and V with panels L and X, respectively), consistent with a requirement for the endogenous extracellular signal-regulated kinase pathway in RA-mediated F9 differentiation (99). Similarly, MEK inhibition blocks the ability of HrasV12 to induce F9 differentiation (87), suggesting that RA- and HrasV12-induced differentiation of endodermal progenitors involve similar signaling pathways. We determined that MEK inhibition does not affect Oct3/Oct4 expression (stem cell maintenance) of KrasV12-expressing F9 cells, even in the presence of RA (Fig. (Fig.8,8, compare CC and UU with FF and XX), indicating that the actions of KrasV12 in resisting differentiation are MEK independent. However, long-term culture (more than 1 week) of the KrasV12-expressing F9 cells in the presence of the MEK inhibitor leads to growth suppression and a reduction in the number of Oct3-expressing/Oct4-expressing cells (data not shown). These results indicate that like PI-3 kinase, MEK signaling downstream of activated Hras and Kras is utilized for distinct aspects of stem cell maintenance, expansion, and differentiation.
Previous studies demonstrated that F9 cells expressing a Raf kinase fusion protein that includes the Hras-derived carboxyl terminus undergo differentiation, indicating that Raf is a key effector of HrasV12-induced differentiation (87). To examine a potential role for compartmentalized Ras-Raf signaling in the differential function of activated Hras and Kras, we tested the effect of forced Raf localization to a subcellular region corresponding to that of Kras by generating a Raf fusion protein containing the carboxyl terminus of Kras (Raf.KTail) and expressing it in F9 cells (Fig. (Fig.7G).7G). Raf.KTail-expressing cells, but not Raf.HTail-expressing cells, could be established as stable lines that, like Krasv12- and Hras.KTail-expressing cells, are also refractory to RA-induced differentiation (Fig. (Fig.7E),7E), and they maintain Oct3/Oct4 expression (Fig. (Fig.7G).7G). Notably, F9 cells expressing the Raf.KTail protein, but not parental F9 cells, exhibit reduced proliferation when maintained in the presence of the MEK inhibitor U0126 (Fig. (Fig.7F),7F), implicating a MEK-dependent Kras function in stem/progenitor expansion. Taken together with data presented above indicating a MEK-independent role for KrasV12 in endodermal stem cell maintenance, these findings suggest a bifurcation within this pathway that distinguishes the activities of activated Hras and Kras and reveals both MEK-dependent and MEK-independent functions for KrasV12 in stem cell expansion and stem cell maintenance, respectively.
To further analyze the roles of the major Ras effectors in the ability of KrasV12 to promote the expansion and maintenance of endodermal progenitor cells, F9 cells were transfected with various KrasV12 effector domain missense mutants that selectively activate RalGDS (E37G), Raf (T35S), or PI-3 kinase (Y40C) (91). The expression of KrasV12Y40C did not significantly affect F9 differentiation capacity, and transfected cells underwent growth arrest in the presence of RA (Fig. (Fig.9A).9A). In contrast, the KrasV12E37G mutant was completely competent for maintenance of Oct3/Oct4 expression (Fig. (Fig.9C)9C) and cell proliferation in the presence of RA (Fig. 9A and B). Notably, these same effector mutants behaved analogously in the context of HrasV12-induced myeloid differentiation (64). However, cotransfection of the Y40C with E37G Kras mutants diminished the ability of E37G to promote F9 proliferation, suggesting an antagonism with the RalGDS pathway, but was consistent with the inability of Y40C to resist RA when transfected alone (Fig. 9A and B). Interestingly, the KrasV12T35S mutant was able to maintain Oct3/Oct4 expression (Fig. (Fig.9C),9C), but transfected cells exhibited significantly reduced proliferative potential (Fig. 9A and B), consistent with the observations described above using the Raf.KTail chimera. While activation of PI-3 kinase by itself was not sufficient to maintain stem cell properties, the proliferative response to KrasV12T35S expression could be enhanced by cotransfection with the Y40C mutant (Fig. 9A to C), further confirming an important role for PI-3 kinase signaling. Cotransfection of T35C was also able to further increase the proliferative increase associated with E37G expression (Fig. 9A and B), suggesting a cooperative effect of activating these two pathways, which have previously been shown to cooperate in tumor progression (90). Taken together, the studies with pharmacologic inhibitors and Kras effector mutants reveal distinct but important requirements for each of these major Ras effectors in the ability of KrasV12 to promote stem cell expansion and resistance to differentiation.
Mutational activation of Ras is one of the most common oncogenic events detected in human cancers. Although accumulating evidence suggests that Ras activation is an important event in the initiation of tumors of the lung, pancreas, and colon (3, 11, 47, 76, 88) and many of the downstream effectors of Ras have now been identified, its precise role in cancer initiation and progression is still unclear. Moreover, the observation that the vast majority of oncogenic Ras alleles arise in the Kras isoform, as opposed to the very closely related Hras and Nras isoforms, has yet to be explained mechanistically. Here, we provide evidence from cell culture studies that mutationally activated Kras, but not Hras or Nras, can promote the expansion of an endodermal stem/progenitor cell population and prevent differentiation. These observations can potentially explain the relatively high frequency of Kras mutations in human cancer, and they reveal a possible role for Kras activation as an initiating event in tumorigenesis. Moreover, our findings are consistent with recent observations in mouse studies in which it was demonstrated that mutational activation of Kras, but not Nras, promotes expansion of the stem cell compartment of the mouse colon (K. Haigis and T. Jacks, MIT, personal communication), thereby supporting the physiological relevance of these cell culture studies.
The F9 cell culture model faithfully recapitulates the differentiation of a stem/progenitor cell to early endoderm. The fact that RA treatment can specifically drive this process in culture and that RA signaling is required for the formation of early endoderm in vivo also supports the physiological relevance of the model. Previous studies have demonstrated that mutationally activated Hras is sufficient to promote the differentiation of F9 cells to primitive endoderm in the absence of RA in transfection studies (86, 87, 95). Our findings are consistent with those reports, and this observation can potentially explain why activating Hras mutations are not seen in tumors of endodermal origin, such as pancreatic, lung, and colorectal cancers. Notably, activated Hras has also been reported to promote the differentiation of several other lineages, including adipocytes (8) and neurons (5, 69). However, a role for activated Kras in these settings has been largely unexplored. Our observation that Hras, Nras, and Kras exert very different functions in the context of stem/progenitor cell differentiation suggests that it may be informative to compare the activities of the various Ras isoforms more broadly in the context of differentiation.
The observed differences in differentiation phenotypes associated with distinct Ras isoforms may be related to previous reports implicating tissue-dependent contexts in susceptibility to the transforming activity of particular oncogenes. Thus, lineage-specific factors associated with differentiation programs may contribute to the susceptibility of different tissue types to tumorigenic conversion by various oncogenes and to the resultant transformed phenotype (39, 72). For example, such differences may underlie the observations that activated Nras is associated with myeloid malignancies (66), germ cell tumors (34), congenital melanocytic nevi, and cutaneous melanomas derived from neural crest, but not mucosal melanomas, which are not derived from neural crest (7, 93). In addition, Hras activation is associated with tumorigenesis of the bladder (45, 79, 100) and salivary gland (97, 98), tissues that arise from an ontogenetic transitional zone, a region where endoderm and ectoderm meet. Notably, the expression of activated Hras, Nras, or Kras in PCC4 cells, which can give rise to mesenchymal (but not endodermal) cells in culture, does not result in an observable phenotype. Together, these findings suggest that the distinct tumor types associated with mutational activation of the various Ras isoforms reflect the unique ability of each of the Ras proteins to affect the differentiation program of progenitor or stem cells that differentiate along distinct lineages.
The proposed model for activated Kras function in stem cells or partially committed progenitors, as an initiating step in human oncogenesis, would seemingly imply that “mature” tumors might continue to express stem cell markers. However, while Oct3/Oct4 expression in human tumors has previously been reported (60, 83-85), the expression is typically seen only in a very small fraction of tumor cells (53), possibly cancer stem cells. This can be explained by either of two mechanisms. First, it is possible that a small fraction of cells within a tumor maintain stem cell characteristics and these cells are needed to continuously “replenish” the bulk of the tumor with progeny cells that exhibit a more differentiated phenotype. However, it is also possible that an activated Kras allele is needed to expand the stem cell population early in tumorigenesis, and subsequently, mutant Kras is selected for its ability to contribute to other aspects of tumor progression, while its role in stem cell maintenance is diminished. Indeed, activated Kras has been shown to promote proliferation, survival, and invasive properties in a variety of non-stem cell contexts (29, 31, 38, 70, 71, 74, 78).
It is worth noting that the observed effects of the various mutationally activated Ras proteins on stem cell differentiation do not necessarily reflect a normal requirement for Ras proteins in stem cell maintenance or differentiation. However, the ability of KrasV12 to expand stem cells could be related to the reported studies describing a unique requirement for endogenous Kras, but not Hras or Nras, in mouse embryonic development (46, 49, 68). Despite the ubiquitous expression of Ras isoforms in various tissues, the lack of correlation between expression and malignancy, and the fact that the various isoforms interact with the same constellation of effectors (2, 13, 17, 21, 33, 51, 73, 78), distinct cellular consequences of activating the various Ras family members have been reported (67, 89, 96). This may reflect differential activation of effectors and/or differential signal intensity/duration (24, 25, 62, 63). Indeed, Ras effectors themselves can exhibit seemingly opposing effects, depending on the cellular context (22, 62, 63, 75, 94, 101). Such opposing activities are consistent with our findings that some Ras effectors can mediate distinct (and seemingly opposing) phenotypic consequences in the F9 model.
Such differences in isoform-dependent Ras signaling appear to largely involve the differential subcellular localization/processing of Ras and its effectors (15, 18, 30, 35, 41-43, 58, 61, 69, 80). For example, a recent report indicates that oncogenic Hras-induced senescence is mediated by the endoplasmic-reticulum-associated, unfolded protein response (26). Our studies with chimeric Ras isoforms and the Raf effector containing carboxy-terminal motifs that cause distinct subcellular localization of signaling complexes support a critical role for compartmentalized signaling in the differential biological activity of the Hras and Kras isoforms (Fig. (Fig.10).10). However, it is also possible that the carboxyl termini of the various Ras proteins also contribute to their distinct biological effects through unique interactions with cellular proteins that have yet to be identified. Such a possibility is supported by our findings that the Kras chimera containing the extreme C terminus of Hras can maintain stem cell characteristics in RA-treated F9 cells and that the activated Kras-4A splice isoform induces apoptosis (Fig. (Fig.1010).
The Raf, PI-3 kinase, and RalGDS proteins are important Ras effectors that contribute to Ras-mediated proliferation and survival or differentiation (50). We observed that inhibiting PI-3 kinase can prevent either differentiation or self-renewal, depending upon the Ras isoform activated. Interestingly, it has also been reported that Ras isoform-dependent transformation potential correlates with PI-3 kinase activation (52). Similarly, we observed that PI-3 kinase activity mediates expression of the POU transcription factor Oct3/Oct4, a homeobox gene, which has direct consequences for self-renewal or differentiation of F9 cells. Moreover, our analysis of Kras effector domain mutants further supports a role for the Raf and PI-3 kinase pathways in endodermal progenitor expansion and reveals an additional important role for RalGDS activation, which has recently emerged as an important Ras effector in the context of human tumorigenesis (37, 40, 77). Interestingly, specific activation of RalGDS has been shown to prevent Hras-mediated differentiation of PC12 and myeloid cells (36, 64).
Our findings with the Raf-KTail chimera and the pharmacologic MEK inhibitor revealed both MEK-independent and MEK-dependent functions of activated Kras in stem/progenitor cells. Interestingly, these findings are consistent with recent studies indicating that KrasV12-induced expansion of mouse colonic epithelial stem cells in vivo appears to be MEK dependent, while maintenance of the undifferentiated state of these cells is MEK independent (K. Haigis and T. Jacks, MIT, personal communication). Taken together, our signaling studies reveal complex and context-dependent roles for three key Ras effectors in the various aspects of stem/progenitor cell maintenance, proliferation, and differentiation (Fig. (Fig.10).10). However, many additional Ras effectors have been identified, and it certainly remains possible that some of those additionally contribute to the distinct functions of activated Hras and Kras in this setting. In summary, the identification of a unique role for Kras in promoting stem/progenitor cell expansion and an inhibition of differentiation to primitive endoderm provides a potential explanation for the high frequency of Kras mutations in tumors of endodermal origin. Finally, the observed unique ability of Kras to expand a stem/progenitor cell population indicates a potentially important role for Kras activation in the initiation of tumorigenesis.
We are grateful to S. Strickland for the F9 cells, to D. Solter for providing antibodies to SSEA1 and SSEA3, to M. Rothenberg and A. Singh for the Kras effector constructs, to M. Betson and S. Sharma for critical reading of the manuscript, and to members of the Bernards and Settleman laboratories for helpful discussions.
This work was supported by NIH R01 CA109447-03 to J.S.
Published ahead of print on 2 February 2008.