Isoprenylcysteine carboxyl methyltransferase deficiency in mice.

After isoprenylation, Ras and other CAAX proteins undergo endoproteolytic processing by Rce1 and methylation of the isoprenylcysteine by Icmt (isoprenylcysteine carboxyl methyltransferase). We reported previously that Rce1-deficient mice died during late gestation or soon after birth. We hypothesized that Icmt deficiency might cause a milder phenotype, in part because of reports suggesting the existence of more than one activity for methylating isoprenylated proteins. To address this hypothesis and also to address the issue of other methyltransferase activities, we generated Icmt-deficient mice. Contrary to our expectation, Icmt deficiency caused a more severe phenotype than Rce1 deficiency, with virtually all of the knockout embryos (Icmt-/-) dying by mid-gestation. An analysis of chimeric mice produced from Icmt-/- embryonic stem cells showed that the Icmt-/- cells retained the capacity to contribute to some tissues (e.g. skeletal muscle) but not to others (e.g. brain). Lysates from Icmt-/- embryos lacked the ability to methylate either recombinant K-Ras or small molecule substrates (e.g. N-acetyl-S-geranylgeranyl-l-cysteine). In addition, Icmt-/- cells lacked the ability to methylate Rab proteins. Thus, Icmt appears to be the only enzyme participating in the carboxyl methylation of isoprenylated proteins.

. These post-isoprenylation processing steps may help target CAAX proteins to membrane surfaces within cells (1).
The endoprotease and methyltransferase steps have attracted interest because they offer a potential means for modulating the activity of CAAX proteins, many of which participate in cell signaling (1). Several groups have hypothesized that inhibiting the endoprotease or the methyltransferase might retard the growth of tumors caused by mutation-activated Ras proteins (1,2,6,7). At this point, however, testing such hypotheses appears to be a few years away. No specific high affinity inhibitors suitable for animal testing have been developed, either for the endoprotease or for the methyltransferase. Just as importantly, neither the spectrum of substrates nor the physiologic importance of the two processing steps has been explored fully. This is particularly the case for the methyltransferase.
To define the physiologic relevance of the post-isoprenylation processing steps, our laboratory generated and characterized mice lacking the endoprotease Rce1 (8). Membranes from Rce1deficient embryos and cells were completely unable to carry out the endoproteolytic processing of Ras and a host of other CAAX proteins. Surprisingly, the consequences of knocking out Rce1 in the mouse were relatively mild. Although most of the Rce1 knockout mice died before birth, the embryos remained viable until late in gestation, and as late as embryonic day 18.5 many were normal in size, appeared healthy, and had no obvious histologic abnormalities. A few of the Rce1 knockout mice were born and lived for a few weeks.
A methyltransferase for mammalian CAAX proteins, isoprenylcysteine carboxyl methyltransferase (Icmt), has been identified recently (5) and shown to be located in the endoplasmic reticulum (5,9). We used gene-targeting techniques to produce a mouse embryonic stem (ES) cell line lacking both Icmt alleles and documented that membranes from those cells lacked the ability to methylate recombinant K-Ras (10). It is important to note, however, that existing reports have raised the possibility that certain isoprenylated proteins might be methylated by other enzymatic activities, at least in some cell types. For example, Giner and Rando (11) concluded that there were distinct methyltransferase activities for the two classes of carboxyl-methylated isoprenylated proteins, the CAAX proteins and the CXC Rab proteins (11). That conclusion was based on a variety of reciprocal inhibition studies with different methyltransferase substrates and inhibitors.
The goal of the current study was to generate Icmt knockout mice, both to define the physiologic consequences of Icmt deficiency and to further define biochemical roles of Icmt. Based in part on the report of an additional methyltransferase activity (11), our a priori prediction was that Icmt-deficient mice might be affected less severely than the Rce1-deficient mice and might even be viable. We also predicted that Icmt-deficient cells might retain the capacity to methylate the CXC-containing Rab proteins. As outlined in this report, both of those expectations were dashed by our experimental results.

MATERIALS AND METHODS
Generation of Icmt-deficient Mice-A sequence replacement genetargeting vector designed to replace exon 1 of the mouse Icmt gene with a neomycin-resistance gene (10) was electroporated into 129/SvJae ES cells, and targeted cells (Icmtϩ/Ϫ) were identified on Southern blots with a 5Ј-flanking probe (10). Two clones, each with a single neo integration, were used to produce Icmtϩ/Ϫ mice. Timed matings of Icmtϩ/Ϫ mice were performed to assess the viability of homozygotes (IcmtϪ/Ϫ) at different stages of development. The genotype of each embryo was determined by Southern blot analysis (10). IcmtϪ/Ϫ fibroblasts were produced from mouse embryos as previously described (8).
Production of Chimeric Mice from IcmtϪ/Ϫ ES Cells-To assess the contribution of Icmt-deficient ES cells to different tissues, two independent lines of IcmtϪ/Ϫ ES cells (10) were injected into C57BL/6 blastocysts. Ten male chimeric mice were obtained; all were 35-75% chimeras as judged by coat color. At 8 weeks of age, the mice were sacrificed, and genomic DNA was purified from multiple tissues and analyzed by Southern blot with a 32 P-labeled Icmt probe. The ratio of IcmtϪ to Icmtϩ bands in each tissue was determined by phosphorimager.
Measurement of Icmt Activity in Embryo Lysates-Embryos were harvested and immediately placed in ice-cold buffer A (50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2, 1 mM EDTA, 100 mM NaCl), supplemented with a protease inhibitor mixture (Complete Mini, Roche Molecular Biochemicals). The embryos were homogenized with a Polytron and then centrifuged at 500 ϫ g for 5 min to remove debris. The protein concentration of the homogenate was determined with a Bio-Rad DC Protein Assay (Bio-Rad, Hercules, CA). To measure Icmt activity, lysates (40 -100 g) were incubated with 10 M S-adenosyl-L-[methyl-14 C]methionine (55 Ci/mol, Amersham Pharmacia Biotech) and 50 M of either N-acetyl-Sgeranylgeranyl-L-cysteine (AGGC, Biomol) or N-acetyl-S-farnesyl-Lcysteine (AFC, Biomol). Recombinant farnesyl-K-Ras (7) and recombinant geranylgeranyl-Rab proteins (described below) were also tested as methyl-accepting substrates. The total volume for the methylation reactions was 50 l. After a 30-min incubation at 37°C, the methylation reaction was stopped by adding 50 l of 1.0 M NaOH containing 0.1% SDS. Most of the reaction mixture (90 l) was spotted onto a pleated 2 ϫ 8-cm filter paper wedged in the neck of a 20-ml scintillation vial containing 5 ml of scintillation fluid (ScintiSafe Econo 1, Fisher). The vials were capped and incubated at room temperature for 5 h to allow the [ 14 C]methanol (formed by base hydrolysis of methyl esters) to diffuse into the scintillation fluid (4). The filter papers were then removed, and the vials were counted for radioactivity. Methyltransferase activity (pmol/mg total cell protein/min) was calculated after subtracting the background level of methylation in control reactions (lysates and S-adenosyl-L-[methyl-14 C]methionine but no isoprenylated substrates). For the Rab methylation assays, control reactions also contained Rab1A, a CC Rab protein that does not undergo carboxyl methylation (12).
Expression and Purification of Recombinant Rab Proteins-Recombinant Rab proteins, Rab escort protein (REP), and Rab geranylgeranyltransferase (RabGGTase) were purified as previously described (13)(14)(15). Briefly, bovine Rab3B, murine Rab3D, and human Rab6 were expressed as N-terminal histidine-tagged fusion proteins in Escherichia coli with pET14b (Novagen, Madison, WI) as the expression vector for Rab3B and Rab6 and pRSET (Invitrogen, Carlsbad, CA) for Rab3D. The proteins were purified by affinity chromatography using a Ni 2ϩ -Sepharose resin. RabGGTase was expressed in Sf9 insect cells by coinfection with baculoviruses coding for both the ␣and ␤-subunits and then purified by cation exchange chromatography followed by gel filtration chromatography. REP was also expressed in Sf9 cells as a C-terminal histidine-tagged fusion protein and was purified by Ni 2ϩ -Sepharose affinity chromatography.
Preparation of Gernanylgeranylated REP⅐Rab Complexes-Geranylgeranylated Rab⅐REP complexes (REP⅐RabGG) were formed in vitro by incubating Rab proteins with RabGGTase, and geranylgeranylpyrophosphate (GGPP) in the presence of limiting amounts of REP (16). Rab protein ( Northern Blot Analysis-A 262-base pair 32 P-labeled Icmt cDNA probe (spanning exons 1-3 of the Icmt gene) was hybridized to a mouse multiple-tissue poly(A) ϩ RNA blot (CLONTECH, Palo Alto, CA); hybridization and washing were performed as described previously (7). The blot was exposed to x-ray film for 72 h at Ϫ80°C.

RESULTS
Biochemical Analysis of Icmt-deficient Embryos-Icmtϩ/Ϫ mice were produced from two independent lines of ES cells. Genotyping of 21-day-old offspring from Icmtϩ/Ϫ intercrosses revealed that about two-thirds (58 of 82) were heterozygotes, and the remainder were wild-type. Genotyping of embryos revealed that IcmtϪ/Ϫ embryos constituted 25% of the litter until embryonic day 10.5 (E10.5, Fig. 1B). By E11.5, there were only a few viable IcmtϪ/Ϫ embryos; those embryos had beating hearts and red blood cells but were far smaller than heterozygous and wild-type mice (Fig. 1C). Virtually all of the IcmtϪ/Ϫ mice died by E12.5, several days before the first of the homozygous Rce1 knockout embryos started to die (8). Histologic studies of IcmtϪ/Ϫ embryos did not reveal a specific cause of death.
To gain insights into the importance of Icmt in the formation of different organs, we generated male chimeric mice (n ϭ 10) with two lines of IcmtϪ/Ϫ ES cells (10) and performed Southern blots to assess the relative capacities of Icmtϩ/ϩ and IcmtϪ/Ϫ cells to populate different tissues ( Fig. 2A). The IcmtϪ/Ϫ cells contributed significantly to the development of skeletal muscle, as shown by the 1:1 ratio of IcmtϪ and Icmtϩ band intensities in the genomic DNA of that tissue but made a negligible contribution to the formation of the brain. The IcmtϪ band was virtually undetectable in brain, and the Icmtϩ/IcmtϪ band ratio was about 8:1. High Icmtϩ/IcmtϪ ratios were also present in liver and testis. Interestingly, the extent to which IcmtϪ/Ϫ cells contributed to the formation of each tissue appeared to be inversely correlated with the normal levels of Icmt expression in that tissue. Thus, as documented by a Northern blot (Fig. 2B), Icmt expression in wild-type mice was quite low in skeletal muscle, but high in the brain and liver. Measurements of enzyme activity in wild-type mice were in general agreement with the Northern blot results, with high activity levels in the brain, testis, and liver and low levels in skeletal muscle (Fig. 2C).
To determine whether IcmtϪ/Ϫ embryos retained the capacity to methylate isoprenylated proteins, perhaps through a redundant enzymatic activity, we tested the capacity of lysates of IcmtϪ/Ϫ embryos to methylate farnesyl-K-Ras (Fig. 3A). No enzymatic activity above background levels was identified. We considered the possibility that a redundant enzymatic activity might not be able to methylate K-Ras and therefore tested the ability of IcmtϪ/Ϫ lysates to methylate two small molecule substrates, N-acetyl-S-geranylgeranyl-L-cysteine and N-acetyl-S-farnesyl-L-cysteine (Fig. 3B). The results of those studies were identical to the results of the K-Ras experiments, loss of activity in lysates of IcmtϪ/Ϫ embryos. We also measured methyltransferase activity against small molecule substrates in lysates from Icmtϩ/Ϫ embryos and tissues from adult Icmtϩ/Ϫ mice (liver, brain, and heart). Activities were invariably reduced by 50% (data not shown), not less than 50%, as would be the case if there were redundant methyltransferase activities.
Consistent with the apparent absence of a redundant methyltransferase activity, there was a substantial accumulation of methyltransferase substrates in lysates from IcmtϪ/Ϫ embryos (i.e. an accumulation of cellular proteins that could be methylated by the yeast ortholog of Icmt, Ste14p, Fig. 3C).
An earlier study (11) concluded that distinct S-adenosylmethionine-dependent methyltransferase activities were responsible for the methylation of the CAAX and CXC groups of isoprenylated proteins. That result would predict that lysates from IcmtϪ/Ϫ embryos would retain the ability to methylate CXC Rab proteins. This was not the case. Lysates from IcmtϪ/Ϫ cells were incapable of methylating three different CXC Rab proteins, although those proteins were readily methylated by lysates from Icmtϩ/ϩ cells (Fig. 4). The importance of Rab methylation by Icmt remains obscure, but one obvious possibility is that it is important for membrane targeting. In preliminary experiments, we have used cell fractionation experiments and the expression of GFP⅐Rab fusions to assess the localization of Rab6 (a Golgi CXC Rab protein) in IcmtϪ/Ϫ and Icmtϩ/ϩ cells but did not observe noticeable differences (data not shown). DISCUSSION Icmt catalyzes the formation of a carboxyl methyl ester on the isoprenylcysteine of CAAX proteins. The methylation reaction is the last of three sequential CAAX-box modifications and the most subtle, at least from the perspective of the primary structure of the protein. Methylation changes the molecular mass of the protein by a mere 14 daltons versus several hundred for both the isoprenylation and endoprotease steps. We had predicted that Icmt-deficiency might produce a relatively mild phenotype. First, deletion of the methyltransferase gene in yeast (i.e. STE14) has little impact apart from its effect on the mating pheromone a-factor (17), and a-factor apparently does not exist in mammals. Second, Rce1 deficiency produced a relatively mild phenotype, with some knockout mice surviving for a few weeks after birth. Thus, we expected that murine Icmt deficiency would produce a similarly mild phenotype, or perhaps even milder given the studies suggesting the existence of additional Icmt-like activities in mammalian cells (11,18). Our a priori prediction was not upheld. Icmt deficiency yielded a more severe phenotype, with most IcmtϪ/Ϫ embryos dying between E10.5 and E11.5. Importantly, our biochemical studies with embryo lysates did not uncover a residual or redundant Icmt-like activity, and the lysates manifested a striking increase in Ste14p substrates.
The IcmtϪ/Ϫ embryos probably died because Icmt-deficient cells failed to grow and contribute to the formation of various organs. Southern blots of tissues from chimeric mice generated with IcmtϪ/Ϫ cells revealed that Icmt-deficient cells are se- verely defective in their capacity to contribute to the formation of certain organs (e.g. liver and brain) although they retained the ability to contribute to the formation of others (e.g. skeletal muscle). 2 We doubt that this finding was spurious, for several reasons. First, similar results were obtained with two lines of IcmtϪ/Ϫ ES cells. Second, there was a reasonably strong inverse correlation between normal levels of Icmt expression and the ability of IcmtϪ/Ϫ ES cells to contribute to the formation of a tissue. Third, the Icmt chimeric mouse experiments were performed in parallel with studies with Zmpste24Ϫ/Ϫ ES cells, 3 which robustly populated all of the tissues of chimeric mice. 4 Why were the developmental abnormalities more severe in IcmtϪ/Ϫ embryos than in Rce1Ϫ/Ϫ embryos? One possibility is simply that Icmt has more substrates than Rce1. Indeed, our experiments revealed for the first time that the CXC Rab proteins (which are not processed by Rce1) are methylated by Icmt. Rab proteins terminating in CXC and CC are geranylgeranylated at both cysteines (12). The CXC Rab proteins, but not the CC Rab proteins, are then carboxyl-methylated (12, 19 -23). For example, Rab3a and Rab4, which terminate in Cys-Ala-Cys and Cys-Gly-Cys, respectively, are carboxyl methylated (19,20), whereas Rab1A and Rab2, which terminate in Cys-Cys, are not (12,23). Interestingly, replacement of the CXC terminus of Rab3a with CC abolishes methylation, whereas the opposite result is obtained when the CC terminus of Rab1A is replaced with a CXC sequence (12).
Our current studies show that the methylation of the CXC Rab proteins is carried out by Icmt, the same methyltransferase that is responsible for CAAX protein methylation. Thus, Icmt almost certainly has more substrates than Rce1, which has no role in Rab protein processing. If these additional substrates (i.e. the Rab proteins) lie at the root of the more severe developmental defects in Icmt deficiency, one might predict that the methylation of Rab proteins has a significant impact on their function. In the case of Rab6, we did not observe a detectable effect of methylation on the intracellular localization of the protein, but we caution against overinterpreting those results. Those experiments did not assess other potential effects of Rab methylation such as effects on protein function or stability.
A second potential explanation for the more severe developmental problems in the IcmtϪ/Ϫ embryos is that the position-ing of the carboxylate anion on CAAX proteins has a profound influence on the binding of isoprenylated proteins to membranes, protein partners, or both. Both Rce1 and Icmt deficiency eliminate the carboxyl methylation of CAAX proteins and thereby leave the protein with a C-terminal carboxylate anion rather than an ␣-methyl ester. However, the position of the carboxylate anion differs, being within the isoprenylcysteine residue in the setting of Icmt deficiency and located three amino acids downstream in the setting of Rce1 deficiency. Methylation itself is clearly important for membrane binding: methylating N-acetyl-S-farnesyl-L-cysteine increases the partitioning of that molecule into the organic phase of an n-octanol/ water mixture (24), and methylating farnesylated CAAX peptides increases their binding to synthetic liposomes (25). The impact of carboxylate anion positioning (i.e. on the prenylcysteine or on the end of the C-terminal tripeptide) on lipidbinding properties has not been studied in a similar fashion. However, it is easy to imagine that a more vicinal carboxylate anion (i.e. with Icmt deficiency) might be more potent in inhibiting the association of the isoprenyl lipid with membranes.
The positioning of the carboxylate anion might also affect prenylation-dependent protein-protein interactions. Chen et al. (26) have shown that the binding of K-Ras to microtubules is dependent on the structure of the C terminus. Farnesylated K-Ras binds to microtubules, but this binding is eliminated by the Rce1-mediated release of the last three amino acids. Binding to microtubules is restored by carboxyl methylation. Thus, in the case of K-Ras, the precise structure of the C terminus affects protein-protein interactions. These data, along with the aforementioned considerations regarding membrane binding, have made the "carboxylate anion positioning" hypothesis quite intriguing. Of course, the effect of carboxylate anion positioning could differ for different CAAX proteins.
A third potential explanation for the more severe phenotype in Icmt-deficient mice compared with Rce1-deficient mice is that some CAAX proteins might undergo endoproteolytic processing in the absence of Rce1. According to this hypothesis, the milder phenotype of Rce1 deficiency could reflect the fact that Rce1 processes fewer CAAX protein substrates than Icmt. This possibility is reasonable, given that a second yeast protein, Afc1p (Ste24p), 3 assists Rce1p in the removal of the -AAX from the farnesylated mating pheromone a-factor. Thus far, however, there has been no direct demonstration that the murine ortholog of AFC1, Zmpste24, 3 has any role in CAAX protein processing (1).
The development and characterization of Icmt-deficient mice has clarified the role of Icmt in mouse development and in the processing of isoprenylated proteins. Just as importantly, these studies have suggested new hypotheses. For example, it will now be of interest to determine the importance of methylation in Rab protein function and to test the carboxylate anion positioning hypothesis. In addition, the production of Icmt-deficient fibroblasts opens the door to addressing the importance of carboxyl methylation in the transformation of fibroblasts by mutation-activated forms of the Ras proteins.