Targeting the protein prenyltransferases efficiently reduces tumor development in mice with K-RAS-induced lung cancer

Abstract

RAS and RHO proteins, which contribute to tumorigenesis and metastasis, undergo posttranslational modification with an isoprenyl lipid by protein farnesyltransferase (FTase) or protein geranylgeranyltransferase-I (GGTase-I). Inhibitors of FTase and GGTase-I were developed to block RAS-induced malignancies, but their utility has been difficult to evaluate because of off-target effects, drug resistance, and toxicity. Moreover, the impact of FTase deficiency and combined FTase/GGTase-I deficiency has not been evaluated with genetic approaches. We found that inactivation of FTase eliminated farnesylation of HDJ2 and H-RAS, prevented H-RAS targeting to the plasma membrane, and blocked proliferation of primary and K-RAS^G12D-expressing fibroblasts. FTase inactivation in mice with K-RAS-induced lung cancer reduced tumor growth and improved survival, similar to results obtained previously with inactivation of GGTase-I. Simultaneous inactivation of FTase and GGTase-I markedly reduced lung tumors and improved survival without apparent pulmonary toxicity. These data shed light on the biochemical and therapeutic importance of FTase and suggest that simultaneous inhibition of FTase and GGTase-I could be useful in cancer therapeutics.

Many intracellular proteins, such as the RAS and RHO family proteins, are posttranslationally lipidated at a carboxyl-terminal CAAX motif. This process is called isoprenylation and is carried out by a pair of cytosolic enzymes, protein farnesyltransferase (FTase) and protein geranylgeranyltransferase type I (GGTase-I) (1). FTase and GGTase-I share a common α-subunit but have unique β-subunits that determine substrate specificity (2, 3). Some CAAX proteins, such as H-RAS, HDJ2, and prelamin A, are substrates for FTase, whereas others, such as RAP1A and RHOA, are substrates for GGTase-I (3). Protein isoprenylation facilitates membrane interactions, promotes protein–protein interactions, and can affect protein turnover (4).

The RAS proteins, by far the most thoroughly studied CAAX proteins, are involved in the pathogenesis of many forms of cancer (5). Because isoprenylation is essential for the plasma membrane targeting of the RAS proteins and their ability to transform cells (4), FTase inhibitors (FTIs) have been developed and tested as anticancer agents (6). FTIs showed efficacy in preclinical studies of malignancies, including those without RAS mutations (7–9). However, clinical trials of FTIs in humans have not been particularly successful (6). The mechanism by which FTIs inhibit cell growth is not entirely clear, but it likely involves interfering with the farnesylation of several CAAX proteins, in addition to RAS (6, 10, 11). Also, different FTIs have different properties (12–15), complicating efforts to define compound- versus mechanism-related effects.

A few years ago, Mijimolle et al. (16) attempted to address the functional relevance of FTase by generating mice with a conditional knockout allele for the gene encoding the β-subunit of FTase (Fntb). Cre-mediated recombination appeared to inhibit the farnesylation of HDJ2 and H-RAS, but only partially, and, most remarkably, H-RAS remained in the membrane fraction of cells. They also reported that Fntb-deficient fibroblasts grew in culture and that the development of K-RAS-induced tumors was unaffected by Fntb deficiency. These findings were surprising for several reasons. First, FTI treatment studies had suggested that the membrane association of H-RAS is utterly dependent on protein farnesylation (17). Second, a nonprenylated mutant of H-RAS (C186S) is found exclusively in the soluble, cytosolic fraction of cells (18). Third, FTI treatment of cells typically results in cell-cycle arrest (19, 20). Fourth, in mouse models, FTIs are efficacious against many tumors, including those without RAS mutations (9, 21).

A potential explanation for the differences between the genetic and pharmacologic studies is that FTIs might affect other proteins aside from FTase. Another is that the Fntb knockout allele generated by Mijimolle et al. (16) yielded a transcript with an in-frame deletion (22), and it is conceivable that this mutant transcript yielded a protein with some residual enzymatic activity.

In FTI-treated cells, K-RAS and N-RAS are alternately prenylated by GGTase-I (23–25). That finding prompted both pharmaceutical companies and academic laboratories to develop GGTase-I inhibitors (GGTIs) (26), which have shown promise in preclinical studies (27–31). The rationale for inhibiting GGTase-I is supported by genetic studies in mice: Inactivating the gene for the β-subunit of GGTase-I (Pggt1b) reduced tumor formation and prolonged survival in mice with K-RAS-induced lung cancer (32).

Because neither an FTI alone nor a GGTI alone inhibits the prenylation of K- and N-RAS, FTI/GGTI combinations and dual-prenylation inhibitors (DPIs) were developed (33, 34). DPIs and FTI/GGTI combinations block K-RAS prenylation in vivo, but only at high doses that are toxic in mice. However, some studies with FTI/GGTI combinations did not report significant toxicity (27, 35). Thus far, no one has used genetic approaches to study dual inhibition of FTase and GGTase-I.

In this study, we created a conditional knockout allele for Fntb and reevaluated the impact of Fntb deficiency on protein isoprenylation, cell proliferation, and the growth of K-RAS-induced tumors. We also bred mice homozygous for conditional knockout alleles in both Fntb and Pggt1b and assessed the effect of combined FTase/GGTase-I deficiency on the development of K-RAS-induced lung cancer.

Results

Generation and Validation of a Conditional FTase Knockout Allele.

To create a conditional knockout allele (Fntb^fl) for the β-subunit of FTase, we introduced loxP sites 1 kb upstream and 1 kb downstream of exon 1 (Fig. 1A). Mice homozygous for the conditional allele (Fntb^fl/fl) were healthy and fertile. Fntb^fl/fl mice were bred first with EIIa-Cre mice (36) to remove a floxed neo cassette and then with deleter-Cre mice (37) to produce mice harboring one conditional knockout allele and one knockout allele (Fntb^fl/Δ). Primary fibroblasts were cultured from E13.5 Fntb^fl/fl embryos. When these cells were incubated with Cre-adenovirus (adCre), they were converted to Fntb^Δ/Δ derivatives that had no detectable Fntb expression (Fig. 1B).

Fig. 1.

A conditional knockout allele for the β-subunit of FTase (Fntb^fl). (A) Schematic of the Fntb gene-targeting vector. LoxP sites were inserted 1 kb upstream and downstream of exon 1. Arrows show the locations of primers for genotyping. neo, neomycin-resistance cassette; tk, thymidine kinase cassette. (B) PCR and RT-PCR analyses demonstrating the inactivation of Fntb by treatment with adCre. Fntb^fl/+ and Fntb^fl/fl fibroblasts were incubated with adβgal or adCre, and genomic DNA and total RNA were isolated 4 days later. (C) Western blots of extracts from Fntb^fl/+ and Fntb^fl/Δ fibroblasts incubated with adβgal or adCre. An antibody against actin was used as a loading control. (D) Western blots showing the distribution of proteins in the membrane (mem) and cytosolic (cyt) fractions of adβgal- or adCre-treated fibroblasts. (E) (Upper) K-RAS western blot of extracts from Fntb^fl/flPggt1b^fl/fl fibroblasts that had been incubated with adβgal or adCre. (Lower) Western blot showing the distribution of K-RAS in the membrane and cytosolic fractions of Fntb^fl/flPggt1b^fl/fl fibroblasts incubated with adβgal or adCre. (F) N-RAS western blot of Fntb^fl/Δ fibroblasts treated with adCre and 10 μM GGTI for 4 days.

To assess the impact of Fntb deficiency on the isoprenylation of FTase substrates, we performed western blots of lysates from β-gal adenovirus- (adβgal) and adCre-treated Fntb^fl/Δ fibroblasts and control fibroblasts (Fntb^fl/+). HDJ2 and H-RAS in extracts of adCre-treated Fntb^fl/Δ cells exhibited reduced electrophoretic mobilities (Fig. 1C), characteristic of the nonfarnesylated forms of these proteins. The absolute levels of HDJ2 did not change, but the levels of H-RAS in adCre-treated Fntb^fl/Δ cells were 2- to 4-fold higher than in control cells, as judged by densitometry (Fig. 1C). The electrophoretic mobilities of K-RAS and N-RAS were unchanged, likely because these proteins are isoprenylated by GGTase-I (23–25). HDJ2 and H-RAS accumulated in the cytosolic fraction of Fntb^Δ/Δ cells, whereas K- and N-RAS remained associated with the membrane fraction (Fig. 1D). Inactivating a single Fntb allele in Fntb^fl/+ fibroblasts with adCre did not affect the electrophoretic mobilities or the membrane/cytosolic partitioning of HDJ2 and H-RAS (Fig. 1 C and D).

To determine whether K-RAS and N-RAS are geranylgeranylated in Fntb^Δ/Δ cells, we isolated Fntb^fl/flPggt1b^fl/fl fibroblasts and treated them with adCre to inactivate both Fntb and Pggt1b (Fntb^Δ/ΔPggt1b^Δ/Δ). In Fntb^Δ/ΔPggt1b^Δ/Δ cells, a substantial proportion of K-RAS accumulated in the cytosolic fraction and exhibited a reduced electrophoretic mobility (Fig. 1E). A similar shift in the mobility of N-RAS was observed in Fntb^Δ/Δ cells treated with a GGTI (Fig. 1F). Fntb^Δ/ΔPggt1b^Δ/Δ cells remained viable for a few days, but they underwent apoptosis and died within 4 days (Fig. S1 A and B). Inactivating Fntb and Pggt1b in cells expressing oncogenic H-RAS targeted to the plasma membrane by an amino-terminal myristoylation sequence also underwent apoptosis (Fig. S1 C and D).

Fntb Deficiency Blocks Proliferation of Primary and K-RAS^G12D-Expressing Fibroblasts.

Inactivating a single Fntb allele in Fntb^fl/+ fibroblasts with adCre did not affect cell proliferation (Fig. 2A). However, inactivating both alleles by treating primary Fntb^fl/Δ fibroblasts with adCre dramatically reduced proliferation (Fig. 2A). Similar results were found with primary and immortalized Fntb^fl/fl cells. Genotypically confirmed Fntb^Δ/Δ fibroblasts were large and flat and accumulated in the G₂M phase of the cell cycle (Fig. 2 B and C), but the number of apoptotic cells was low (Fig. S1A). Inactivating Fntb in primary fibroblasts increased p21^CIP1 levels and delayed serum-stimulated phosphorylation of AKT but did not affect levels of phosphorylated MEK and ERK1/2 (Fig. S2).

Fig. 2.

Inactivating Fntb stops fibroblast proliferation. (A) Proliferation of Fntb^fl/+ and Fntb^fl/Δ primary mouse fibroblasts incubated with adβgal or adCre. Data show a single cell line assayed in triplicate. Similar results were obtained with two different cell lines of each genotype. **P < 0.01; ***P < 0.001. (B) Photomicrographs of Fntb^fl/Δ cells incubated with adβgal or adCre. (C) Cell-cycle analysis of Fntb^fl/Δ fibroblasts incubated with adβgal or adCre. PI, propidium iodide. (D) PCR genotyping of genomic DNA from Fntb^fl/fl fibroblasts at various times after incubation with adβgal or adCre. (E) PCR genotyping of genomic DNA from individual clones of adβgal- or adCre-treated Fntb^fl/+ and Fntb^fl/fl fibroblasts. (F) Proliferation of primary Fntb^fl/+K^LSL and Fntb^fl/ΔK^LSL fibroblasts incubated with adβgal or adCre. Data represent the mean of a single cell line/genotype assayed in triplicate. Similar results were obtained in two independent experiments with two cell lines/genotypes. Inset shows PCR genotyping of genomic DNA demonstrating the activation of the K^G12D allele in cells incubated with adCre (ii and iv) but not in cells incubated with adβgal (i and iii). (G) Cell-cycle analysis of adβgal- and adCre-treated Fntb^fl/ΔK^LSL fibroblasts. The experiment was repeated three times with similar results.

When adCre-treated Fntb^fl/fl cells were left on the culture plates for more than a week, cell growth gradually resumed. However, this growth was due to overgrowth by Fntb^fl/Δ cells (rather than Fntb^Δ/Δ cells) (Fig. 2D). Despite extensive efforts, we were unable to clone Fntb^Δ/Δ fibroblasts from adCre-treated Fntb^fl/fl cells. In more than 12 independent experiments, cell growth late after adCre treatment was invariably due to overgrowth by Fntb^fl/Δ cells (Fig. 2E). In contrast, we had no difficulty in obtaining stable Fntb^Δ/+ cell lines after treating Fntb^fl/+ cells with adCre (Fig. 2E).

To assess the impact of Fntb deficiency on the proliferation of cells expressing K-RAS^G12D, we isolated primary fibroblasts from Fntb^fl/Δ and Fntb^fl/+ embryos harboring an inducible oncogenic K-RAS allele (K^LSL) (38). The K^LSL allele is normally silent but K-RAS^G12D expression can be induced with Cre. Incubation of Fntb^fl/+K^LSL fibroblasts with adCre yielded Fntb^Δ/+K^G12D cells that proliferated more rapidly (Fig. 2F). In contrast, adCre treatment of Fntb^fl/ΔK^LSL cells (producing Fntb^Δ/ΔK^G12D cells) resulted in G₂M cell-cycle arrest (Fig. 2 F and G).

Inactivating Fntb Increases Survival of Mice with K-RAS-Induced Lung Cancer.

To determine the effect of an Fntb knockout on the development of a K-RAS-induced malignancy, we bred mice car-rying the K^LSL allele (K) and a lysozyme M-Cre allele (L) on Fntb^fl/Δ and Fntb^fl/+ backgrounds (mice harboring both alleles were designated KL). Littermate Fntb^fl/ΔL mice were monitored to assess the impact of Fntb deficiency in the absence of K-RAS-induced tumors; Fntb^fl/+L, Fntb^fl/+K, and Fntb^fl/ΔK mice were used as healthy controls (Ctr mice).

We previously showed that KL mice express K-RAS^G12D in most or all type II pneumocytes and develop lung cancer that is fatal by 25 days of age. Lung weight is increased ∼10-fold and alveoli are obliterated by diffuse hyperplasia and adenocarcinoma (32). In keeping with those findings, the maximum survival of Fntb^fl/+KL mice was 24 days (Fig. 3A). The survival of Fntb^fl/ΔKL mice was significantly longer, >100 days in some cases (P < 0.0001; Fig. 3A). The lung weights in 3-week-old Fntb^fl/ΔKL mice were lower than in Fntb^fl/+KL mice (P < 0.0001) but higher than in Fntb^fl/ΔL mice or Ctr mice (Fig. 3B). Moreover, histologic analyses revealed nearly complete obliteration of alveoli in Fntb^fl/+KL lungs, whereas Fntb^fl/ΔKL lungs retained areas of normal histology (Fig. 3C). Survival, lung weight, and lung histology in Fntb^fl/ΔL mice were indistinguishable from those of Ctr mice (Fig. 3 A–C).

Fig. 3.

Inactivation of Fntb reduces tumor development and prolongs survival of mice with K-RAS-induced lung cancer. (A) Kaplan–Meier curve showing increased survival in Fntb^fl/ΔKL mice (n = 16) compared with Fntb^fl/+KL mice (n = 10). All Fntb^fl/ΔL mice (n = 9) were alive at the end of the experiment (150 days). (B) Lung weight in 3-week-old Fntb^fl/+KL (n = 8), Fntb^fl/ΔKL (n = 7), Fntb^fl/ΔL (n = 6), and Ctr (n = 5) mice. (C) Hematoxylin/eosin-stained sections of lungs from 3-week-old mice. (Scale bars, 100 μm.) (D) Genotyping of wild-type and activated K^G12D alleles and the recombined Fntb^Δ allele by PCR amplification of genomic DNA from lung tissue. (E) Western blot showing the distribution of H-RAS and K-RAS in membrane and cytosolic fractions of lung lysates. (F) Western blots of protein lysates from the lungs of 3-week-old mice. Actin was used as a loading control.

PCR genotyping of genomic DNA from Fntb^fl/ΔKL lungs revealed activation of the K-RAS^G12D allele and a recombined Fntb allele (Fig. 3D). We harvested protein lysates from lung tissue from 3-week-old Fntb^fl/+KL and Fntb^fl/ΔKL mice and performed western blots with antibodies against the CAAX proteins, H-RAS, HDJ2, and prelamin A, which serve as markers of cellular FTase activity (39). In lung extracts of Fntb^fl/ΔKL mice, a substantial proportion of H-RAS accumulated in the cytosolic fraction, whereas K-RAS remained in the membrane fraction (Fig. 3E). Approximately 50% of the HDJ2 in Fntb^fl/ΔKL lungs exhibited a reduced electrophoretic mobility (Fig. 3F). Also, reduced FTase activity resulted in an accumulation of nonfarnesylated prelamin A (Fig. 3F). There was no difference in levels of phosphorylated ERK1/2 in lung extracts from Fntb^fl/+KL and Fntb^fl/ΔKL lungs (Fig. 3F).

Simultaneous Inactivation of Fntb and Pggt1b Reduces Tumor Load.

To assess the effect of combined Fntb and Pggt1b deficiency on K-RAS-induced tumors, we bred Fntb^fl/ΔPggt1b^fl/ΔKL mice and control KL mice in which neither prenyltransferase was inactivated. Littermate Fntb^fl/ΔPggt1b^fl/ΔL mice were analyzed to determine the impact of Fntb/Pggt1b deficiency in the absence of K-RAS-induced tumors. Fntb^fl/+Pggt1b^fl/+L, Fntb^fl/ΔPggt1b^fl/+L, and Fntb^fl/+Pggt1b^fl/ΔL mice (collectively designated Ctr) were used as healthy controls.

Lung weight and histology were indistinguishable in 3-week-old Fntb^fl/ΔPggt1b^fl/ΔKL and Ctr mice (Fig. 4 A and B). But in 3-week-old KL mice, lung weight was 10-fold higher than in Ctr mice (as a result of tumor burden), and the maximum survival was 24 days (Fig. 4 A–C). The levels of phosphorylated ERK1/2 in lung lysates were lower in Fntb^fl/ΔPggt1b^fl/ΔKL mice than in KL mice (Fig. 4D), as were K_i-67 levels (as judged by immunofluorescence confocal microscopy) (Fig. 4E). Fntb^fl/ΔPggt1b^fl/ΔKL mice survived far longer than KL mice (median, 170 vs. 22 days; P < 0.0001) (Fig. 4C), but all eventually developed tumors and were euthanized. Lung histology and survival were indistinguishable in Fntb^fl/ΔPggt1b^fl/ΔL and Ctr mice (Fig. 4 B and C).

Fig. 4.

Simultaneous inactivation of Fntb and Pggt1b in mice with K-RAS^G12D-induced lung cancer. (A) Lung weight of 3-week-old KL (n = 4), Fntb^fl/ΔPggt1b^fl/ΔKL (n = 6), Fntb^fl/ΔPggt1b^fl/ΔL (n = 5), and Ctr (n = 9) mice. BW, body weight. (B) Representative hematoxylin/eosin-stained lung sections of mice from the experiment shown in A. (Scale bars, 100 μm.) (C) Kaplan–Meier curve showing survival of KL (n = 12), Fntb^fl/ΔPggt1b^fl/ΔKL (n = 10), and Fntb^fl/ΔPggt1b^fl/ΔL (n = 9) mice. Black tick marks indicate healthy Fntb^fl/ΔPggt1b^fl/ΔL mice that were euthanized for tissue analyses. (D) Western blots of protein extracts from the lungs of 3-week-old KL and Fntb^fl/ΔPggt1b^fl/ΔKL mice showing levels of phosphorylated ERK1/2. Total ERK1/2 was used as a loading control. (E) Confocal immunofluorescence micrographs showing expression of K_i-67 (pink) in cells from KL, Fntb^fl/ΔPggt1b^fl/ΔKL, and Ctr mice. Nuclei were visualized with DAPI (blue).

To determine whether farnesylation and geranylgeranylation were inhibited in lungs of Fntb^fl/ΔPggt1b^fl/ΔKL mice, we performed western blots on lung lysates from 3-week-old mice. Approximately 25% of HDJ2 exhibited a reduced electrophoretic mobility, characteristic of the nonfarnesylated protein. Protein geranylgeranylation was also inhibited, as the nonprenylated (np) form of RAP1A accumulated in lung lysates (Fig. 5A). Moreover, a large proportion of K-RAS accumulated in the cytosolic fraction of lung lysates from Fntb^fl/ΔPggt1b^fl/ΔKL mice, suggesting that many cells lacked both FTase and GGTase-I activity (Fig. 5B). Immunofluorescence and confocal microscopy revealed prelamin A staining in lung sections from Fntb^fl/ΔKL and Fntb^fl/ΔPggt1b^fl/ΔKL mice, whereas np-RAP1A staining was detected only in the lungs of Fntb^fl/ΔPggt1b^fl/ΔKL mice (Fig. 5C). Some cells in lungs from Fntb^fl/ΔPggt1b^fl/ΔKL mice exhibited strong staining for both prelamin A and np-RAP1A (Fig. 5C and Fig. S3A). These double-positive cells were also identified in lungs of Fntb^fl/ΔPggt1b^fl/ΔL mice, and some of those cells were type II pneumocytes because they were also positive for SP-C (Fig. S3B). Moreover, K-RAS was identified in the cytosolic fraction of lung lysates from Fntb^fl/ΔPggt1b^fl/ΔL mice (Fig. S3C). Despite the presence of cells that apparently lacked both FTase and GGTase-I activity, we found no evidence of apoptosis in lung sections of Fntb^fl/ΔPggt1b^fl/ΔL mice (Fig. S4).

Fig. 5.

Reduced FTase and GGTase-I activities in lungs of Fntb^fl/ΔPggt1b^fl/ΔKL mice. (A) Western blots of protein extracts from lungs of 3-week-old KL and Fntb^fl/ΔPggt1b^fl/ΔKL mice. Actin was used as a loading control. (B) Western blot showing the distribution of K-RAS in the membrane and cytosolic fractions of KL and Fntb^fl/ΔPggt1b^fl/ΔKL mice. (C) Confocal immunofluorescence microscopy showing prelamin A (red) and np-RAP1A (green) expression in lung sections from 3-week-old KL, Fntb^fl/ΔKL, Fntb^fl/ΔPggt1b^fl/ΔKL, and Ctr mice. Nuclei were stained with DAPI (blue). The specificity of prelamin A staining was established with lung sections of Zmpste24^−/− mice (where prelamin A accumulates due to a defect in the conversion of farnesyl-prelamin A to mature lamin A). Arrows indicate cells positive for both prelamin A and np-RAP1A. Asterisk, alveolus. (Scale bars, 25 μm.) (D) Recombination efficiency of the “floxed” K^LSL, Fntb^fl, and Pggt1b^fl alleles determined by quantitative PCR of genomic DNA from lung biopsies of 3-week-old Fntb^fl/ΔPggt1b^fl/ΔKL mice (n = 5) and from lung tumors of 28-week-old Fntb^fl/ΔPggt1b^fl/ΔKL mice (n = 2).

In the immunohistochemistry analyses, we detected cells in lungs of Fntb^fl/ΔPggt1b^fl/ΔKL mice that were positive for np-RAP1A but not prelamin A (Fig. S3A) and cells that were negative for both. We suspect that incomplete recombination in the Fntb and Pggt1b alleles underlies the development of tumors in older Fntb^fl/ΔPggt1b^fl/ΔKL mice. Indeed, the percentage of genomic DNA with an inactivated Fntb allele fell from 18% in 3-week-old mice to 5% in the tumors of 198-day-old mice. Simultaneously, the percent inactivation of the Pggt1b gene fell from 20% in 3-week-old mice to 10% in tumors of 198 day-old mice (Fig. 5D).

Simultaneous Inactivation of Fntb and Pggt1b Inhibits Tumorigenesis in a Second K-RAS-Induced Lung Tumor Model.

To further investigate how inactivating Fntb and Pggt1b affects tumor development, we administered adCre to Fntb^fl/ΔPggt1b^fl/ΔK mice, K mice heterozygous for one or both of the conditional alleles, and healthy Ctr mice (38). AdCre results in K^G12D expression in a limited number of cells, leading to a limited number of tumors that can be characterized in number, grade, and lesion area. Eight weeks after administration of 5 × 10⁷ pfu adCre, K mice had large tumors that were easily visible on the surface of the lungs (Fig. 6 A–C). However, the lung surface of adCre-treated Fntb^fl/ΔPggt1b^fl/ΔK mice was nearly indistinguishable from that of Ctr mice, and Fntb^fl/ΔPggt1b^fl/ΔK mice had 76% fewer tumors and 79% smaller lesion area than K mice (Fig. 6 A–C). The lung lesions in K mice ranged from atypical adenomatous hyperplasia and epithelial hyperplasia to adenocarcinoma, the most common lesion (identified in 7 of 8 mice; Fig. 6D). The lung histology of Fntb^fl/ΔPggt1b^fl/ΔK mice ranged from entirely normal to the presence of epithelial hyperplasia (Fig. 6D) and small adenomas; adenocarcinoma was observed in only 1 of 11 mice.

Fig. 6.

Inactivation of Fntb and Pggt1b reduces tumor burden in a second K-RAS^G12D-induced lung cancer model. (A) Photographs of lungs 8 weeks after inhalation of adCre. Note the extensive lesions (white areas) on the surface of lungs from K mice and the relatively normal appearance of the lungs from Fntb^fl/ΔPggt1b^fl/ΔK mice. (B and C) Tumor number (B) and surface area (C) in lungs from K (n = 8) and Fntb^fl/ΔPggt1b^fl/ΔK (n = 11) mice 8 weeks after inhalation of adCre. (D) Hematoxylin/eosin-stained sections of typical lesions in lungs from K and Fntb^fl/ΔPggt1b^fl/ΔK mice. (Scale bars, 100 μm.)

Discussion

In this study, we produced mice with a conditional knockout allele for Fntb and showed that inactivation of Fntb eliminated farnesylation of HDJ2 and H-RAS, prevented membrane targeting of H-RAS, and blocked the proliferation of primary, immortalized, and K-RAS^G12D-expressing fibroblasts in vitro. Moreover, inactivating Fntb in mice with K-RAS-induced lung cancer reduced tumor growth and improved survival. In addition, simultaneous inactivation of Fntb and Pggt1b had a strong antitumor effect.

Our findings differ significantly from those of Mijimolle et al. (16). In the latter study, the Fntb knockout did not affect H-RAS membrane association or stop fibroblast proliferation, nor did it affect the development of K-RAS-induced tumors in mice. When their report was published, it was provocative because it challenged both the widely accepted notion that H-RAS association with membranes depends on protein farnesylation and the view that HDJ2 prenylation depends on FTase. The explanation for the differences between our studies and theirs is unknown; however, we suspect that the differences might relate to the fact that Mijimolle et al.’s allele yielded an unexpected splicing event. Rather than generating a transcript with a nonsense mutation, the mutation yielded a short in-frame deletion in the Fntb transcript (22). In contrast, the recombination event in our Fntb allele deleted the promoter and exon 1 and created a bona fide null allele.

Inactivating Fntb in fibroblasts induced a G₂M arrest associated with large flattened cells, up-regulated p21^CIP1, and reduced serum-stimulated phosphorylation of AKT. Previously, we showed that inactivation of Pggt1b induces a G₁ arrest associated with cell rounding and up-regulation of p21^CIP1. The cell rounding and cell-cycle arrest could be overcome, at least temporarily, by expressing farnesylated mutants of RHOA and CDC42, suggesting that a limited number of geranylgeranylated proteins are important for those phenotypes. The creation of a bona fide knockout allele for the Fntb allele opens the door to performing similar experiments to determine whether geranylgeranylated versions of FTase substrates reverse the phenotypes of Fntb-deficient cells.

Fntb deficiency reduced tumor growth and improved survival in mice with K-RAS-induced lung cancer, similar to the effects of inactivating Pggt1b (32). Because K-RAS remains prenylated and associated with membranes in both Fntb- and Pggt1b-deficient cells, these studies support the notion (40) that the therapeutic effects of FTIs and GGTIs are independent of the RAS proteins.

We hypothesized that simultaneous inactivation of Fntb and Pggt1b would limit tumor growth more effectively than inactivation of either gene alone—in part because this approach would be predicted to inhibit K-RAS prenylation and membrane association. The simultaneous inactivation strategy was effective, at least to an extent, because a substantial proportion of K-RAS in lysates from Fntb/Pggt1b-deficient fibroblasts, lung tissue, and lung tumors accumulated in the cytosolic fraction. As we predicted, the simultaneous inactivation of Fntb and Pggt1b had a far greater inhibitory effect on K-RAS-induced tumors than inactivation of either gene alone. However, the main effect of inhibiting FTase and GGTase-I is likely independent of the RAS proteins because both control and myristoylated H-RAS-transfected fibroblasts underwent apoptosis after inactivation of both Fntb and Pggt1b.

The main concern surrounding the combined inhibition of FTase and GGTase-I has been toxicity (33, 34). Whereas the simultaneous inactivation of Fntb and Pggt1b clearly induced cell death in fibroblasts, the inactivation of both genes in type II pneumocytes in Fntb^fl/ΔPggt1b^fl/ΔL mice did not produce pulmonary disease phenotypes and did not induce apoptosis. Moreover, Fntb^fl/ΔPggt1b^fl/ΔKL mice appeared healthy for several months despite widespread expression of K-RAS^G12D in the lung. Some cells in Fntb^fl/ΔPggt1b^fl/ΔKL lungs clearly lacked both enzymes, as a large proportion of K-RAS in lung lysates was found in the cytosolic fraction. Also, immunochemical studies revealed that some cells were positive for both prelamin A and np-RAP1A. It is intriguing that some cells in the lung were apparently viable despite the absence of both FTase and GGTase-I. However, our genetic approach does not allow us to address the impact of FTase and GGTase-I deficiency in every lung cell (such as lung stem cells), and we cannot rule out the possibility that a more widespread inactivation of these enzymes would be toxic.

In summary, our findings support the idea that farnesylation is essential for H-RAS membrane association, HDJ2 prenylation, and fibroblast proliferation in vitro. In mice harboring a mutationally activated form of K-RAS in the lung, blocking protein farnesylation retarded tumor growth and improved survival. Our results also support the idea that simultaneous inhibition of FTase and GGTase-I could be therapeutically useful. Finally, the experimental approach described here should be useful for dissecting the in vivo importance of protein farnesylation and geranylgeranylation in other cell types in a variety of diseases.

Materials and Methods

A Conditional Knockout Allele for Fntb.

A 2.2-kb fragment spanning promoter sequences, exon 1, and parts of intron 1 was amplified by PCR from the genomic DNA of strain 129/OlaHsd embryonic stem (ES) cells. The fragment was cloned into pNB1, which contains a polylinker flanked by loxP sites. The floxed fragment was excised and cloned into pKSloxPNTmod (41) upstream of a floxed neo cassette. Finally, 5′- and 3′-flanking arms were amplified by PCR and cloned upstream and downstream, respectively, of the floxed exon 1 fragment and neo cassette. The gene-targeting vector was electroporated into 129/OlaHsd ES cells, and targeted clones (identified by Southern blotting with flanking probes) were used to produce germline-transmitting chimeric mice. A detailed description of all other methods appears in SI Materials and Methods.