On the physiological importance of endoproteolysis of CAAX proteins: heart-specific RCE1 knockout mice develop a lethal cardiomyopathy.

Proteins terminating with a CAAX motif, such as the Ras proteins and the nuclear lamins, undergo post-translational modification of a C-terminal cysteine with an isoprenyl lipid via a process called protein prenylation. After prenylation, the last three residues of CAAX proteins are clipped off by Rce1, an integral membrane endoprotease of the endoplasmic reticulum. Prenylation is crucial to the function of many CAAX proteins, but the physiologic significance of endoproteolytic processing has remained obscure. To address this issue, we used Cre/loxP recombination techniques to create mice lacking Rce1 in the heart, an organ where Rce1 is expressed at particularly high levels. The hearts from heart-specific Rce1 knockout mice manifested reduced levels of both the Rce1 mRNA and CAAX endoprotease activity, and the hearts manifested an accumulation of CAAX protein substrates. The heart-specific Rce1 knockout mice initially appeared healthy but died starting at 3-5 months of age. By 10 months of age, approximately 70% of the mice had died. Pathological studies revealed that the heart-specific Rce1 knockout mice had a dilated cardiomyopathy. By contrast, liver-specific Rce1 knockout mice appeared healthy, had normal transaminase levels, and had normal liver histology. These studies indicate that the endoproteolytic processing of CAAX proteins is essential for cardiac function but is less important for the liver.

Prenylation of CAAX proteins is vitally important to eukaryotic cells (1)(2)(3)(4). Yeast lacking the shared ␣-chain of farnesyltransferase and geranylgeranyltransferase type I are not viable (12). In mammalian cells, prenylation of CAAX proteins is absolutely required for their proper targeting to membrane surfaces and for proper protein function (1)(2)(3)(4). The importance of protein farnesylation is clearly indicated by the fact that mice lacking farnesyltransferase die early during embryonic development (before embryonic day 6.5). 2 The geranylgeranylation of CAAX proteins is also critical for normal cellular function. Drugs that inhibit geranylgeranylation of CAAX proteins have pronounced effects on cell growth and can trigger apoptosis (13)(14)(15). Further underscoring the importance of protein geranylgeranylation, a cysteine protease from Yersinia pestis kills mammalian cells by clipping off the geranylgeranylcysteine from specific CAAX proteins (16).
While the importance of prenylation for protein function and cell viability is well documented (1)(2)(3)(4), the physiological importance of the second step of CAAX protein processing, the endoproteolytic trimming of the C terminus, has remained obscure. In yeast, RCE1 deficiency caused a partial mislocalization of Ras2p away from the plasma membrane, but there was no detectable effect on cell growth or vitality (5). In mammalian cells, Rce1 deficiency eliminated the endoproteolytic processing of the Ras proteins and led to a partial mislocalization of Ras proteins away from the plasma membrane (6,17). However, Rce1 deficiency had only a marginal effect on fibroblast cell growth (17).
Studies of Rce1 knockout mice have heightened the mystery surrounding the importance of Rce1-mediated endoproteolytic processing for mammalian cells (6). These knockout mice grow and develop normally until late in embryonic development. After embryonic day 15.5, however, most of them die and those that are born alive are small and survive for only a few weeks.
Why Rce1 knockout mice die remains enigmatic, since no pathology was observed in any of the major organ systems (6). The fact that most do not survive postnatally indicates that CAAX protein endoproteolysis is somehow important; however, the survival of some Rce1-deficient mice for a few weeks after birth indicates that the endoproteolytic processing is not nearly as important as prenylation for the vitality of mammalian cells and tissues.
Hematopoietic stem cells harvested from the livers of Rce1deficient embryos were capable of restoring normal levels of hematopoiesis when transplanted into lethally irradiated wildtype mice (6,18); thus, Rce1 deficiency did not appear to have significant deleterious effects on hematopoiesis. However, no one has yet examined the physiologic importance of Rce1 on tissues of adult mice, for example by examining tissue-specific knockout mice. Thus, the physiologic significance of Rce1 in adult animals is uncertain.
The finding that Rce1 deficiency leads to a partial mislocalization of Ras proteins within mouse fibroblasts (6) has encouraged efforts to develop Rce1 inhibitor drugs, with the hope that such drugs might diminish the effects of mutationally activated Ras proteins in human cancers (19,20). A recent study indicated that Cre-mediated excision of Rce1 in mouse fibroblasts limits Ras-induced transformation of cells (17), but much more research is required to determine if Rce1 would be a useful therapeutic target. In any case, if Rce1 inhibitor drugs were shown to be effective in blocking tumor growth in vivo, it would be essential to document that they could be given safely. In that regard, the question of whether the absence of CAAX protein endoproteolysis would adversely affect the function of vital organ systems in vivo has never been addressed.
In the current study, we sought to assess the physiologic importance of the endoproteolytic processing of prenylated CAAX proteins. To address this issue, we created mice lacking FIG. 1. Deficiency of Rce1 in the livers of adult mice. A, Rce1 activity levels in liver extracts after intravenous injection of Cre adenovirus. B, retarded electrophoretic mobility of the Ras proteins in the liver of Rce1 flx/flx mice treated with Cre adenovirus. The higher band within the "Ras doublet band" contains Ras proteins that have not undergone endoproteolytic processing (6). C, Southern blot illustrating a near-complete excision of Rce1 in the liver of pI-pC-treated Rce1 flx/flx Mx1-Cre ϩ/ϩ mice. Liver DNA was cleaved with BamHI. The 5-kb BamHI fragment corresponds to the Rce1 flx allele (17). After induction of Cre expression with pI-pC, the coding sequences of Rce1 are excised, creating an Rce1 ⌬ allele (17). The BamHI fragment in the Rce1 ⌬ allele is 6.5 kb. When lower amounts of pI-pC are administered, the extent of recombination in the liver is incomplete, and small amounts of an 8.5-kb BamHI fragment are detected. In that allele, the coding sequences of Rce1 are completely excised, but the floxed neomycin-resistance marker is retained (17). D, reduced Rce1 activity levels in pI-pC-treated Rce1 flx/flx Mx1-Cre ϩ/ϩ mice (n ϭ 6), compared with nontreated Rce1 flx/flx Mx1-Cre ϩ/ϩ controls (n ϭ 6).
Rce1 in the heart, a tissue where Rce1 is expressed at very high levels (7). This study indicates that CAAX endoproteolysis is vitally important in the myocardium, as the absence of Rce1 in the heart caused a lethal-dilated cardiomyopathy.

EXPERIMENTAL PROCEDURES
Heart-specific Rce1 Knockout Mice-Mice with a "floxed" Rce1 allele (Rce1 flx ) were created in an earlier study to analyze the effect of an Rce1 gene excision on Ras-induced transformation of fibroblasts (17). In the Rce1 flx allele, the loxP sites flank all of the protein-coding sequences of the gene; thus, Cre-mediated recombination eliminates the possibility of translation of a functional protein. To create an Rce1 knockout allele (Rce1 ⌬ ), Rce1 flx/ϩ mice were bred with transgenic mice that expressed Cre in the germline (21). As expected from earlier studies (6), mice with a single copy of the knockout allele (Rce1 ⌬/ϩ ) were phenotypically normal and free of pathology over 18 months of observation. Mice hemizygous for an ␣-myosin heavy chain-Cre transgene (␣Myhc-Cre ϩ/o ) (22) were obtained from Dr. Michael Schneider (Baylor College of Medicine, Houston, TX).
Assessing Cre-mediated Inactivation of Rce1 in the Heart-To assess the extent of Cre-mediated recombination within Rce1 in the heartspecific Rce1 knockout mice, BamHI-cleaved genomic DNA from the heart was analyzed with Southern blots (17). In addition, Rce1 expression in the heart was assessed by real-time quantitative PCR (QPCR). For these studies, total RNA was isolated from the heart with a TRI REAGENT protocol (Molecular Research Center, Cincinnati, OH). Reverse transcription of RNA (1 g) was performed with the SuperScriptII RNase H-Reverse Transcriptase Kit (Invitrogen, Carlsbad, CA); the resulting cDNA (50 ng) was used for real-time QPCR reactions, performed for 40 cycles (95°C for 15 s, 60°C for 60 s) on an ABI Prism 7700 Sequence Detection System (PE Applied Biosystems, Foster City, CA). The oligonucleotide primers were 5Ј-AAGGTTGTCCTGGCCCCTC-3Ј and 5Ј-TTTCGTAGCCAGGGCATGT-3Ј, and the probe was 5Ј-6FAM-GTTCTTGGGCCCGCTGCCTCACAG-TAMRA-3Ј (PE Applied Biosystems, Foster City, CA). Rce1 expression levels were normalized to those of glyceraldehyde-3-phosphate dehydrogenase (Rodent GAPDH Control Reagents, PE Applied Biosystems). Rce1 mRNA levels in the hearts of Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice were compared with those in the hearts of control Rce1 flx/⌬ mice.
Assessing Rce1 Activity in the Heart-CAAX endoprotease activity levels in heart tissue were measured with a coupled endoproteolysis/

FIG. 2. Deficiency of Rce1 in the hearts of adult mice. A, Southern blots showing
Cre-mediated excision of Rce1 in the heart, but not in the tail, of an Rce1 flx/flx ␣Myhc-Cre ϩ/o mouse. In the Rce1 flx/flx ␣Myhc-Cre ϩ/o mouse, note the appearance of the 6.5-kb Rce1 ⌬ band in the heart, but not in the tail. Also shown is a Southern blot of heart and tail DNA in an Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mouse. In the hearts of Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice, we typically observed a slight increase in the ratio of the 6.5-kb band to the 5.0-kb band; however, this increase was not always easy to discern because of the germline Rce1 ⌬ allele. B, QPCR studies demonstrating lower Rce1 mRNA levels in Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice (n ϭ 5) than in Rce1 flx/⌬ controls (n ϭ 5) (p Ͻ 0.01). C, Reduced CAAX endoprotease activity levels in Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice (n ϭ 5) compared with Rce1 flx/⌬ controls (n ϭ 5) (p Ͻ 0.01). D, retarded electrophoretic mobility of the Ras proteins in the heart of an Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mouse. The higher band within the Ras doublet band contains Ras proteins that have not undergone endoproteolytic processing (6). methylation assay (6,7). In this assay, the level of CAAX endoprotease activity in heart extracts was determined by measuring the ability of the extracts to cleave farnesyl-K-Ras and render it susceptible to methylation by Icmt, thereby producing K-Ras terminating with a farnesylcysteine [ 14 C]methyl ester. Hearts of mice were excised, rinsed with cold phosphate-buffered saline, cut into multiple pieces, and then homogenized and sonicated in a buffer (50 mM Tris-HCl, 1 mM EDTA, 100 mM NaCl, 5 mM MgCl 2, pH 7.4) containing protease inhibitors (Roche Applied Science, Nutley, NJ) (6,7). After centrifugation at 1000 ϫ g for 5 min and removal of tissue debris, heart extracts were collected. Extracts (300 g) were incubated for 30 min at 37°C in a total volume of 100 l in the presence of human recombinant farnesyl-K-Ras (4 M), the methyl donor S-adenosyl-L-[methyl-14 C]methionine (1 mCi, 80 Ci/mmol; Amersham Biosciences), and recombinant Icmt (8). The reaction was stopped by adding 100 l of 1 M NaOH containing 1% SDS. A portion of the mixture (135 l) was spotted onto a filter paper, which was wedged into the neck of a vial containing 5 ml of scintillation fluid. The vial was then capped and incubated overnight at room temperature. NaOH hydrolyzes the methyl esters, releasing [ 14 C]methanol, which rapidly enters the scintillation fluid. After removal of the filter paper, the radioactivity in the scintillation fluid is measured. As a control, we tested the incorporation of S-adenosyl-L-[methyl-14 C]methionine into nonfarnesylated K-Ras; the incorporation rates were invariably Ͻ5% of those into farnesyl-K-Ras.
Quantifying Substrate Accumulation in Rce1-deficient Fibroblasts and Heart Tissue-To assess the level of accumulation of Rce1 substrates in Rce1 Ϫ/Ϫ fibroblasts, whole-cell lysates (50 g) from Rce1 ϩ/ϩ and Rce1 Ϫ/Ϫ fibroblasts were mixed on ice and incubated with S-adenosyl-L-[methyl-14 C]methionine at 37°C for 45 min. In this experiment, Rce1 substrates within the Rce1 Ϫ/Ϫ cells are cleaved by the Rce1 enzymatic activity in the Rce1 ϩ/ϩ lysates (23). The endoproteolytic cleavage reaction renders the proteins susceptible to Icmt-mediated carboxyl methylation; the amount of methylation can be quantified by measuring the amount of [ 14 C]methanol released after the addition of 100 l of 1 M NaOH. As controls, we measured the methylation in lysates (100 g) from either Rce1 ϩ/ϩ or Rce1 Ϫ/Ϫ fibroblasts. (In Rce1 ϩ/ϩ lysates, the amounts of methylation would be expected to be low, since CAAX proteins would be fully processed (i.e. cleaved and methylated) prior to the incubation; in Rce1 Ϫ/Ϫ lysates, the amount of methylation would be low because the absence of Rce1-mediated endoproteolysis precludes methylation of the isoprenylcysteine.) To assess the level of Rce1 substrates in hearts, we incubated total heart lysates (500 g) from Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice with S-adenosyl-L-[methyl-14 C]methionine at 37°C for 45 min. In this experiment, we relied on Rce1 in "non-cardiomyocytes" to cleave the Rce1 substrates in the Rce1-deficient cardiomyocytes, thereby rendering them susceptible to carboxyl methylation. The amount of methylation was quantified by measuring the base-releasable radioactivity. For controls, methylation was quantified in lysates from hearts of wild-type mice as well as mice deficient in Zmpste24, another integral membrane protease of the ER.
Statistical Analysis-Data are expressed as means Ϯ S.E., unless otherwise indicated. Differences between groups were analyzed with a two-tailed Student's t test (Primer of Biostatistics, Version 3.0, McGraw Hill, 1992).

RESULTS
In earlier studies, we showed that Cre-mediated excision of Rce1 eliminates the enzymatic activity responsible for the endoproteolytic processing of Ras proteins, causing retarded mobility on SDS-polyacrylamide gels (6). However, an Rce1 gene excision had little effect on the growth of cultured fibroblasts (17). In the current project, we sought to define the consequences of inactivating Rce1 in the tissues of adult mice. We injected Cre adenovirus into the internal jugular vein of Rce1 flx/flx mice, with the expectation that the adenovirus would be taken up almost exclusively by hepatocytes (after intravenous administration, adenovirus is taken up largely by the liver, Ref. 29). We further expected that Cre in the liver would excise the coding sequences of Rce1 and reduce CAAX endoprotease activity. Indeed, Cre adenovirus did cause a significant decrease in CAAX endoprotease expression in the liver (Fig. 1A). Western blot analysis of the Ras proteins in the liver revealed that approximately one-half of the Ras proteins had retarded electrophoretic mobility (Fig. 1B), indicating the complete loss of endoproteolytic processing in a significant fraction of the liver cells. We also inactivated Rce1 in the liver by breeding Rce1 flx/flx -Mx1-Cre ϩ/ϩ mice and then inducing Cre expression with injections of pI-pC (26 -28). As judged by Southern blots, the excision of Rce1 in the livers of the pI-pC-treated mice was complete (Fig. 1C), and Rce1 activity levels were dramatically reduced (Fig. 1D). Despite the absence of Rce1 in the liver DNA, the pI-pC-treated Rce1 flx/flx Mx1-Cre ϩ/ϩ mice gained weight and exhibited normal vitality over 2-3 months of observation. During this time, transaminase levels remained normal, and the histological appearance of the liver on hematoxylin/eosinstained sections was indistinguishable from that of wild-type mice (not shown).
To further explore the physiologic importance of CAAX protein endoproteolysis, we created mice lacking Rce1 in the heart, where Rce1 is expressed at very high levels (7). To generate the heart-specific Rce1 knockout mice, we bred Rce1 flx/flx mice harboring a Cre transgene under the control of the ␣Myhc promoter (22,30). As expected, Southern blots of genomic DNA from the hearts of Rce1 flx/flx ␣Myhc-Cre ϩ/o mice revealed incomplete recombination ( Fig. 2A), which was not surprising, since the majority of cells within the heart are not myocytes (31,32).
We suspected that our best chance to achieve a near-complete inactivation of Rce1 in cardiac myocytes would be to breed ␣Myhc-Cre ϩ/o mice harboring one Rce1 knockout allele (Rce1 ⌬ ) and one "floxed" allele (Rce1 flx/⌬ ␣Myhc-Cre ϩ/o ). In those myocytes, Cre-mediated inactivation of a single Rce1 flx allele would eliminate all Rce1 expression. The presence of the Rce1 ⌬ allele in the genomic DNA obviously made it more difficult to discern Cre-mediated inactivation of the Rce1 flx allele with Southern blots (Fig. 2A). However, the inactivation of Rce1 in the heart could be documented easily with QPCR studies and by measuring CAAX endoprotease activity levels. QPCR revealed significantly fewer Rce1 transcripts in Rce1 flx/⌬ ␣Myhc-Cre ϩ/o hearts than in Rce1 flx/⌬ control hearts (Fig. 2B). Consistent with these data, we also found a significant reduction in CAAX endoprotease activity levels in heart extracts from Rce1 flx/⌬ -␣Myhc-Cre ϩ/o mice (Fig. 2C). The reduced Rce1 mRNA levels and CAAX endoprotease activity levels demonstrate that the ␣Myhc-Cre transgene was effective in inactivating Rce1 in the heart. The less-than-complete elimination of Rce1 transcripts and endoprotease activity was not surprising in view of the fact that the ␣Myhc-Cre transgene is expressed only in the myocytes of the heart and not in other cell types (e.g. endothelial cells and fibroblasts). As expected, we observed a doublet Ras band in the heart tissue from Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice (Fig. 2D), suggesting a complete loss of endoproteolytic processing in a significant fraction of cardiac myocytes.
Several weeks before their deaths, most of the heart-specific Rce1 knockout mice were listless and had ruffled fur. Although we did not observe edematous extremities in the mice, we always noted a sizable accumulation of pleural and peritoneal fluid at autopsy, and the hearts were invariably enlarged (Fig.  4A). Echocardiography showed dilated left ventricles (not shown). Histological sections revealed dilatation of all four chambers of the heart, and organized thrombi were occasionally noted within the left atrium (Fig. 4B). The left ventricular musculature was thin and dystrophic, and there were increased amounts of collagen between ventricular myocytes (Fig. 5). Heart tissue from Rce1 flx/flx , Rce1 ϩ/ϩ ␣Myhc-Cre ϩ/o , and Rce1 flx/⌬ control mice was histologically normal.
The striking cardiomyopathy in the Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice (along with the normal heart histology and normal survival in the control groups) suggested that the absence of Rce1mediated endoproteolytic processing was mechanistically related to the development of cardiomyopathy. However, such a mechanism would clearly be open to criticism without firm biochemical evidence that Rce1 substrates actually accumulate in the hearts of Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice. We therefore developed a coupled endoproteolysis/methylation assay to document the accumulation of Rce1 substrates in Rce1-deficient cells and tissues. This assay measures the ability of Rce1 to cleave accumulated Rce1 substrates, rendering them susceptible to Icmt-mediated methylation.
To show that this strategy was feasible, we first asked if Rce1 substrates accumulate in Rce1-deficient fibroblasts. Indeed, they did (Fig. 6A). When whole-cell lysates from Rce1 ϩ/ϩ and Rce1 Ϫ/Ϫ fibroblasts were incubated in the presence of S-adenosyl-L-[methyl-14 C]methionine, the accumulated Rce1 substrates in the Rce1 Ϫ/Ϫ cells were cleaved, allowing them to be methylated (Fig. 6A). The methylation could be readily quantified with a base-hydrolysis assay. In contrast, incubating Rce1 ϩ/ϩ lysates with S-adenosyl-L-[methyl-14 C]methionine yielded negligible levels of methylation, reflecting the fact that most CAAX proteins in those cells had already been cleaved and methylated (Fig. 6A). Similarly, incubating Rce1 Ϫ/Ϫ lysates in the presence of S-adenosyl-L-[methyl-14 C]methionine yielded very low levels of methylation because the absence of Rce1 prevents methylation (Fig. 6A).
Next, we tested whether Rce1 substrates accumulate in the hearts of Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice. For these studies, we relied on the fact that many cells in Rce1 flx/⌬ ␣Myhc-Cre ϩ/o hearts (e.g. endothelial cells and fibroblasts) express Rce1 at wild-type levels. Thus, when fresh Rce1 flx/⌬ ␣Myhc-Cre ϩ/o lysates were incubated with S-adenosyl-L-[methyl-14 C]methionine, we predicted that the Rce1 from the endothelial cells and fibroblasts would cleave the accumulated substrates in the Rce1-deficient cardiomyocytes, allowing methylation to occur. Indeed, this prediction was upheld. We observed far more methylation in the lysates from Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice than in lysates from wild-type controls and from mice lacking another ER endoprotease, Zmpste24 (Fig. 6B). These data clearly show that Rce1 substrates accumulate in hearts of Rce1 flx/⌬ ␣Myhc-Cre ϩ/o mice.
We hypothesized that absent endoproteolysis of the Ras proteins might result in impaired growth factor-mediated activation of the Ras effector Erk1/2, contributing to the development of cardiomyopathy. However, growth factor-mediated activation of Erk1/2 was no different in Rce1 Ϫ/Ϫ and Rce1 ϩ/ϩ cells (Fig. 7A). We were concerned that intrinsic genetic differences between independent lines of Rce1 Ϫ/Ϫ and Rce1 ϩ/ϩ fibroblasts might conceivably have prevented us from observing subtle differences in Erk1/2 activation. Accordingly, we repeated the Erk1/2 activation experiments in parental Rce1 flx/flx cells and Rce1 ⌬/⌬ cells that were derived from them (i.e. two cell lines that were genetically identical except for the difference in Rce1 expression). Once again, however, we observed no differences in Rce1-expressing and Rce1-deficient cells in Erk1/2 activation in response to serum or epidermal growth factor (Fig. 7B). DISCUSSION While the physiologic importance of protein prenylation for cell homeostasis has become axiomatic (1, 2), the physiologic relevance of the subsequent step in CAAX protein processing, endoproteolysis, has remained mysterious (4). In the current study, we demonstrate that the endoproteolytic processing of CAAX proteins is critically important in the heart. Mice lacking Rce1 expression in cardiac myocytes develop a dilated cardiomyopathy, beginning as early as 3 months of age. By 10 months of age, most of the heart-specific Rce1 knockout mice had died. Pathologic studies uncovered findings that are common to many forms of dilated cardiomyopathy, including chamber enlargement, ventricular thinning, dystrophic myocytes, mural thrombi, and an increased amount of fibrous tissue between myocytes (33). The clear-cut importance of Rce1 in the heart contrasts with the findings in hematopoietic cells and in hepatocytes. Hematopoietic stem cells from Rce1 Ϫ/Ϫ embryos rescued lethally irradiated recipient mice and manifested normal long-term "repopulating potential," even in competitive repopulation experiments, suggesting that Rce1-mediated protein processing is not particularly important for the vitality of hematopoietic cells (6,18). Also, in the current study, we found that Rce1 deficiency in the liver had no effect on liver histology or LFTs over several months of follow-up.
There is little doubt that the cardiomyopathy in the heartspecific Rce1 knockout mice is mechanistically related to the absence of Rce1-mediated endoproteolysis. The heart-specific Rce1 knockout mice exhibited cardiomyopathy and reduced survival, whereas the littermate experimental controls were normal. Also, we demonstrated a clear-cut accumulation of Rce1 protein substrates in heart lysates from the heart-specific Rce1 knockout mice. Thus, our results allow us to conclude, for the first time, that Rce1-mediated processing of CAAX proteins is physiologically important in vivo. The absence of Rce1 leads to an accumulation of Rce1 substrates in the heart and cardiomyopathy.
Because the Ras proteins are by far the most extensively studied CAAX proteins, and because they have such a well documented role in cell signaling (34 -38), it is tempting to speculate that the effects of Rce1 deficiency in the heart are somehow related to the absence of Ras endoproteolytic processing and reduced Ras signaling. Indeed, this scenario seems quite plausible, particularly since the expression of an activated H-Ras transgene in mice causes myocardial hypertrophy (39) and the expression of a dominant-negative H-Ras transgene in mice causes a lethal dilated cardiomyopathy (see also cardiogenomics.med. harvard.edu/groups/proj1/pages/ras_home.html). 3 However, despite the potential attractiveness of the "Ras signaling hypothesis," we would caution against the blanket assumption that diminished Ras signaling is responsible for the "Rce1 cardiomyopathy." First, we did not observe any effect of Rce1 deficiency on the activation of downstream Ras effectors by growth factors. These results are consistent with the finding of normal MAP kinase activation by growth factors in hematopoietic cells (18). Second, and probably more importantly, Rce1 has dozens of protein substrates, and it is quite conceivable that the elimination of endoproteolysis of some of the non-Ras substrates underlies the pathology in the heart-specific Rce1 knockout mice. For example, the endoproteolytic processing of lamin B1 (a key structural component of the nuclear lamina) does not occur in the absence of Rce1 (40). Very subtle genetic defects in another nuclear envelope protein, lamin A/C, clearly cause dilated cardiomyopathy (41), so one could easily imagine that absent endoproteolytic processing of lamin B1 could affect heart function. One could make similar arguments about the potential involvement of many other non-Ras CAAX proteins in the pathology of Rce1deficient hearts.
The proposition that Rce1 deficiency might cause cardiomyopathy by interfering with the processing of many different CAAX protein substrates would not be surprising, given what has been learned about the antitumor effects of the farnesyltransferase inhibitor drugs. Although farnesyltransferase inhibitor drugs were initially designed to interfere with the processing of the Ras proteins, it is now very clear that they inhibit tumor growth largely by interfering with the processing of non-Ras protein substrates (42,43). These drugs inhibit tumor growth in vitro regardless of whether the tumors contained an activated Ras protein (44); they even suppress the growth of K-Ras-induced tumors without perceptibly reducing K-Ras prenylation (45).
There are many dozens of CAAX protein substrates with a myriad of important roles in cell biology. In the case of the farnesyltransferase inhibitors, it would obviously be desirable to identify the precise CAAX protein substrate or substrates that underlie the antitumor effects of these drugs. However, despite many years of effort by many prominent laboratories in industry and academia, the identity of the CAAX protein substrates that mediate the therapeutic effects of the farnesyltransferase inhibitors has never been clearly defined. Similarly, it would be very difficult to identify the precise CAAX protein substrates that mediate the development of cardiomyopathy in heart-specific Rce1 knockout mice.
In summary, our studies show, for the first time, that the endoproteolytic processing of prenylated CAAX proteins is physiologically important in mammals. Heart-specific Rce1 knockout mice develop a dilated cardiomyopathy, with most mice dying by 10 months of age. This finding would probably need to be borne in mind when considering the therapeutic application of Rce1 inhibitor drugs. FIG. 7. Normal serum-stimulated activation of Erk1/2 in Rce1 ؊/؊ (A) and Rce1 ⌬/⌬ (B) fibroblasts. Fibroblasts were serumstarved overnight. The next morning, serum-containing medium was added to the cells (time 0). At various time points, cells were harvested and analyzed by immunoblotting with an antibody against phosphorylated Erk1/2. The blots were stripped and incubated with an antibody recognizing total Erk1/2 as a loading control. The experiment was repeated three times with similar results. Also, similar results were obtained when Erk1/2 was stimulated with epidermal growth factor rather than serum.