|Home | About | Journals | Submit | Contact Us | Français|
Protein arginine N-methyltransferases have been implicated in a variety of processes, including cell proliferation, signal transduction, and protein trafficking. In this study, we have characterized essentially a null mutation induced by insertion of the U3βGeo gene trap retrovirus into the second intron of the mouse protein arginine N-methyltransferase 1 gene (Prmt1). cDNAs encoding two forms of Prmt1 were characterized, and the predicted protein sequences were found to be highly conserved among vertebrates. Expression of the Prmt1-βgeo fusion gene was greatest along the midline of the neural plate and in the forming head fold from embryonic day 7.5 (E7.5) to E8.5 and in the developing central nervous system from E8.5 to E13.5. Homozygous mutant embryos failed to develop beyond E6.5, a phenotype consistent with a fundamental role in cellular metabolism. However, Prmt1 was not required for cell viability, as the protein was not detected in embryonic stem (ES) cell lines established from mutant blastocysts. Low levels of Prmt1 transcripts (approximately 1% of the wild-type level) were detected as assessed by a quantitative reverse transcription-PCR assay. Total levels of arginine N-methyltransferase activity and asymmetric NG,NG-dimethylarginine were reduced by 85 and 54%, respectively, while levels of hypomethylated substrates were increased 15-fold. Prmt1 appears to be a major type I enzyme in ES cells, and in wild-type cells, most substrates of the enzyme appear to be maintained in a fully methylated state.
Methylation of arginine residues is one of many covalent modifications of eukaryotic proteins that occur concomitant with or shortly following translation. Two types of protein arginine methyltransferases have been classified according to their substrate specificity and reaction products (reviewed in reference 11). Type I enzymes catalyze the formation of NG-monomethylarginine and asymmetric NG,NG-dimethylarginine, while type II enzymes catalyze the formation of NG-monomethylarginine and symmetric NG,N′G-dimethylarginine. Most substrates for type I enzymes bind nucleic acid, usually RNA. These include heterogeneous nuclear RNA binding proteins (hnRNPs), which collectively contain 65% of the nuclear asymmetric dimethylarginine, as well as fibrillarin and nucleolin (19–21). The only known physiological substrate of symmetric (type II) arginine methyltransferase is myelin basic protein, a major protein component of the myelin sheath.
Genes encoding rat (PRMT1), human (HRMT1L2), and yeast (RMT1) type I enzymes have been characterized (12, 13, 17, 29). The mammalian genes appear to be ubiquitously expressed in all tissues (17, 29, 32). The yeast enzyme, which is not required for cell viability, accounts for over 85% of the protein dimethylarginine in the cell (12).
Type I enzymes have been implicated in a variety of processes, including cell growth control, signal transduction, and protein trafficking, but the biochemical and biological functions of arginine methylation have not been established. The enzymes preferentially methylate motifs rich in arginine and glycine (RGG boxes), a common feature of the RNA binding domains of hnRNPs (20). Arginine-methylated hnRNP A1 (24), but not Hrp1p (33), has lower affinity for RNA than the native protein, suggesting a potential mechanism for modulation of protein-RNA interactions. Levels of protein methylarginine may change in response to extracellular stimuli under circumstances in which biological responses are also suppressed by methyltransferase inhibitors. These include nerve growth factor-induced neurite outgrowth in PC12 cells (6) and mitogenic responses of lipopolysaccharide-treated B cells (16).
Interactions between the PRMT1 enzyme and potential signaling components have also emerged from yeast two-hybrid screens. The immediate-early gene product TIS21 (BTG2) and the leukemia-associated gene product BTG1 interact with PRMT1 and can modulate its enzymatic activity in vitro (17). TIS21 and BTG1 both belong to a family of mitogen-induced proteins implicated in negative regulation of the cell cycle. PRMT1 also binds to the cytoplasmic domain of the IFNAR1 chain of the alpha beta interferon receptor, while growth-inhibitory effects of interferon were suppressed by antisense oligonucleotides directed against the methyltransferase (1). Finally, a novel arginine methyltransferase (CARM1) associates with p160 coactivators and serves as a secondary coactivator of nuclear hormone receptors (4).
Other studies have identified a role for arginine methylation in protein trafficking. Shuttling of the yeast hnRNP-related proteins Np13p and Hrp1p between the nucleus and cytoplasm requires methylation by the Hmt1p methyltransferase (30). The human enzyme complements the shuttling defect, suggesting functional conservation between the two enzymes. Nuclear translocation of the large form of basic fibroblast growth factor may also depend on arginine methylation (23). In the presence of a methyltransferase inhibitor, basic fibroblast growth factor was not methylated and the protein did not localize to the nucleus.
The prevalence of NG,NG-dimethylarginine in RNA binding proteins and conservation among protein arginine N-methyltransferases underscore the potential biological importance of this posttranslational modification. However, a major issue arguing against a dynamic role for the type I enzymes in cell regulation concerns the possibility that arginine methylation is both constitutive and irreversible. While most substrates have not been characterized, some are known to exist only in a fully methylated state (18, 19). Moreover, no demethylase capable of removing dimethylarginine residues has been identified (11) and in the case of histones, turnover of dimethylarginine accompanies protein degradation (3).
Efforts to understand the biochemical function of mammalian arginine methyltransferases are complicated by several factors, including the existence of multiple enzymes and the fact that methyltransferase inhibitors nonspecifically target multiple processes in which S-adenosylmethionine serves as a methyl donor. In yeast, functional studies of arginine methylation have benefited greatly from genetic approaches that have led to the isolation of cells deficient in the enzyme. In principle, gene targeting strategies could be used for similar studies of the mammalian enzymes, assuming that the proteins are not required for cell viability.
The present study characterized a recessive early embryonically lethal mutation in a gene encoding the mouse ortholog of the human protein arginine N-methyltransferase 1 enzyme. The mutation was originally induced by gene entrapment in mouse embryonic stem (ES) cells and was selected during an in vitro screen for mutations in developmentally regulated genes (28). We now show that the enzyme is essential for early development, is a major source of arginine methyltransferase activity in ES cells, and yet is not required for cell viability. While proteins from mutant cells were significantly hypomethylated, most potential substrates in wild-type cells appeared to be blocked by prior methylation. The availability of Prmt1-deficient cells is expected to assist efforts to understand the function of this enzyme in normal cellular metabolism.
Decidual swellings containing embryonic day 6.5 (E6.5) and E7.5 embryos were dissected in ice-cold phosphate-buffered saline (PBS), fixed in PBS containing 4% paraformaldehyde and 0.2% glutaraldehyde overnight at 4°C, dehydrated in a graded ethanol series, cleared in xylene, embedded in Paraplast X-tra (Polysciences), and sectioned. Serial sections (7 μm) were collected on slides, stained with hematoxylin and eosin (Sigma), and mounted in Permount (Fisher Scientific).
E6.5 to E13.5 embryos were fixed in PBS containing fresh 2% paraformaldehyde and 2% glutaraldehyde for 30 min at 4°C, rinsed twice for 15 min each time and once for 1 h in ice-cold PBS, and stained overnight in PBS containing 0.02% NP-40, 0.01% sodium dodecyl sulfate (SDS), 2 mM MgCl2, 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, and 1 mg of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; Sigma), pH 7.2, per ml. Stained embryos were rinsed in PBS and visualized by dark-field microscopy.
Offspring bearing the 7.4.2 provirus were identified by Southern blot hybridization to a 1.3-kb StuI fragment that contained genomic sequences of the Prmt1 gene. E3.5 to E8.5 embryos were genotyped by PCR using a mixture of three primers, i.e., forward 7.4.2#1 (5′ATATCCTTTTGTGAGACCCC), reverse 7.4.2#2 (5′GGAAGGGCTCTGTCCTAA), and reverse lacZ#3 (5′CCTCTTCGCTATTACGCCAG). Each 25-μl PCR mixture contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 200 μM each deoxyribonucleoside triphosphate, 1.25 U of Amplitaq (Perkin-Elmer Cetus), and each primer at 0.2 μM. Reactions proceeded through 40 cycles of denaturation (95°C for 1 min), primer annealing (55°C for 1 min), and primer extension (72°C for 2 min), followed by a final 10-min extension at 72°C. The resulting PCR products were resolved by 1.2% agarose gel electrophoresis and visualized by UV light after staining with ethidium bromide. DNAs from mice and embryos were isolated as described previously (35).
Transcribed cellular sequences present in 7.4.2-βGeo fusion transcripts were cloned by rapid amplification of cDNA 5′ ends (5′RACE) using a 5′RACE kit (Gibco BRL) and following the manufacturer's instructions. First-strand cDNA was synthesized using 3 μg of total cellular RNA (5) and 0.2 μM reverse primer lacZ#3 (5′CCTCTTCGCTATTACGCCAG). After removal of the RNA template with RNase H (3.0 U), the cDNA was purified using BRL GlassMax spin cartridges and tailed with dCTP and terminal deoxynucleotidyltransferase. A 5-μl sample of tailed cDNA was amplified with 0.2 μM nested primer lacZ#2 (5′CTGCAAGGCGATTAAGTTGGG) and 0.2 μM Bethesda Research Laboratories anchor primer through 40 cycles of denaturation (95°C for 1 min) and primer annealing (55°C for 2 min) and extension (72°C for 3 min) and a 10-min extension at 72°C. The resulting PCR products were analyzed by Southern blot hybridization using a probe complementary to the first 20 nucleotides (nt) of the proviral long terminal repeat (LTR) (5′CCTACAGGTGGGGTCTTTCA). PCR products were gel purified (QIAquick gel extraction kit; QIAGEN), subcloned into the pCR plasmid vector (Invitrogen), and sequenced.
The 189-nt 7.4.2 fusion transcript isolated via 5′RACE was used to screen an E8.5 mouse cDNA library (9). Two positive clones were identified out of a total of 106 plaques, and cDNA inserts of 1.1 kb and 600 nt were subcloned into the EcoRI site of pBluescript KS(−) (Stratagene) and sequenced. Subsequently, the 1.1-kb cDNA was used as a probe to rescreen the mouse E8.5 library and to screen an ES cell cDNA library (provided by B. Rosenberg, Massachusetts Institute of Technology). Altogether, 25 clones were isolated from the two libraries and sequenced.
Genomic DNA sequences flanking the provirus were cloned from a λFix II sv129 genomic library using the 189-nt 5′RACE product as a probe. Five clones contained a 1.3-kb StuI fragment that hybridized to 5′RACE sequences immediately adjacent to the provirus. The 1.3-kb fragment was subcloned into the EcoRV site of pBluescript KS(−) (Stratagene) and sequenced.
Total cellular RNA was isolated from various C57BL/6J mouse tissues (liver, heart, lung, kidney, ovary, brain, and spleen), whole wild-type embryos, and ES cells of the D3 line via the method of Chomczynski and Sacchi (5). A 15-μg sample of each RNA was denatured with formamide, electrophoretically separated on 1% formaldehyde agarose gels, transferred to a HyBond-N+ filter (Amersham), and hybridized to [α-32P]dCTP-labeled probes (10).
Reverse transcription (RT)-PCR was performed as previously described (15). Briefly, 20 μg of total cellular RNA was treated with RNase-free DNase (5 U; Gibco BRL), extracted with phenol-chloroform, precipitated in ethanol, and resuspended in 20 μl of H2O. A 5-μl portion of each RNA sample was reverse transcribed using a reverse oligonucleotide primer (5′CAGCAAACATGCAGAGGATGCCAGT to detect the 54-nt Prmt1 alternative exon or 5′GTGCACGCGCTGGTGGCTTACTTCAA to quantify Prmt1 transcripts in wild-type and mutant cells), 200 U of SuperScript II (Gibco BRL), and 500 μM each deoxyribonucleoside triphosphate. After treatment with 3.0 U of RNase H (Gibco BRL), sequences corresponding to the Prmt1 54-nt alternative exon were amplified using 2 μl of the first-strand cDNA, 0.2 μM forward primer (5′GCGAACTGCATCATGGAG), and 0.2 μM reverse primer (5′ACATCCAGCACCACCTTG) as described above. For quantitative analysis of Prmt1 transcripts, the PCRs used 5′GACAGCCATTGAGGACCGAC and 5′CCAATGCCTGCCTCATAA as forward and reverse primers, respectively.
An interspecific backcross [(C57BL/6J × Mus spretus)F1 × M. spretus] DNA panel from the Jackson Laboratory community resource (26) was used to map the Prmt1 locus. Genomic DNAs from C57BL/6J and M. spretus were digested with various restriction enzymes and blot hybridized to the 1.3-kb Prmt1 genomic probe to detect restriction fragment length polymorphism (RFLP) between the two strains. Filters containing MspI-digested genomic DNAs from the 94 backcrosses and two parental strains (BSS panel) were obtained from the Jackson Laboratory and hybridized to the 1.3-kb Prmt1 genomic probe. Each DNA was scored for the presence of the C57BL/6J allele, and the results were analyzed against the Jackson Laboratory database to determine the position on the existing map.
DNA sequencing reactions used the BigDye Terminator Cycle Sequencing Kit (ABI) and were analyzed on a Prism 377 DNA Sequencer. Templates were either subcloned into pBluescript (Stratagene) or sequenced directly from phage DNAs. Primers were either T3 and T7 or oligonucleotide primers corresponding to various regions of Prmt1. DNA sequences were compiled with Sequencher 3.0 software (Gene Codes Corporation), and sequence homology searches were performed using the BLAST algorithm (2).
ES cells lines were isolated from E3.5 blastocysts as described previously (35) and were maintained on 0.1% gelatinized tissue culture plates in ES cell medium (high-glucose Dulbecco modified Eagle medium [Gibco] supplemented with 15% preselected fetal bovine serum [Hyclone; heat inactivated at 55°C for 30 min], 0.1 mM 2-mercaptoethanol, and 100 mM nonessential amino acids [Gibco]) supplemented with 1,000 U of leukemia inhibitory factor (ESGRO; Gibco) per ml. Cultures were trypsinized every 2 days and replated at a 1:3 ratio with either a feeder layer of gamma-irradiated mouse embryo fibroblasts for maintenance of the cell line or for at least 3 generations without feeder layers for isolation of DNA, RNA, and cell extracts.
Samples (200 μg of protein each) were mixed with an equal volume of 2× Laemmli loading buffer (27), boiled for 5 min, and fractionated by SDS–10% polyacrylamide gel electrophoresis. Proteins were transferred onto PolyScreen polyvinylidene difluoride membrane (NEN Life Sciences) in Tris-glycine buffer (27). Polyclonal serum from rabbits immunized against recombinant rat PRMT1 (32) was diluted 1:3,000 in TBST (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.05% Tween 20) containing 5% milk and bound to proteins for 1 h. Membranes were washed three times for 10 min each time in TBST, incubated with a 1:10,000 dilution of peroxidase-conjugated anti-rabbit secondary antibody (Santa Cruz Biotechnology, Inc.) in TBST for 30 min, and detected by enhanced chemiluminescence assay (Amersham).
For protein extraction, cells were harvested at 75% confluency, washed twice, scraped into ice-cold PBS, pelleted, and resuspended in 1 ml (per 10-cm-diameter plate) of ice-cold reaction buffer (50 mM Tris-HCl [pH 7.6], 0.1 mM EDTA, 0.1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride). The cell suspensions were sonicated on ice with two 15-s bursts using a microtip sonicator (XL2015; Heat Systems) at a setting of 4.0 and clarified by centrifugation at 10,000 × g for 30 min, and the supernatant was flash frozen in an ethanol–dry-ice bath. Protein concentrations were determined by a modified Lowry assay (DC Protein Assay; Bio-Rad).
Reactions to measure methyltransferase activity were performed in reaction buffer supplemented with 50 μM S-adenosyl-l-[methyl-3H]methionine (Amersham; adjusted to 5,000 dpm/pmol with unlabeled AdoMet), 100 μM R3 peptide (21), and either wild-type or mutant cell extract. Reactions to measure the methylation status of proteins in cell extracts were performed in reaction buffer containing 50 μM S-adenosyl-l-[methyl-3H]methionine and either equal amounts of wild-type and mutant cell extracts or wild-type cell extract alone. Methylation reaction mixtures contained 140 μg of total protein in 140 μl and were incubated at 36°C. Aliquots (20 μl) were withdrawn at different intervals, added to 20 μl of 2× SDS sample buffer, and heated to 100°C for 5 min to stop the reaction. The samples were diluted into 1 ml of PBS, and the incorporated methyl-3H was determined via a filter binding assay (20).
Analysis of methylated arginine derivatives in total cell lysates was performed by the Vanderbilt-Ingram Cancer Center Peptide Sequencing and Amino Acid Analysis Shared Resource. Briefly, samples were hydrolyzed with 6N HCl (110°C for 18 h) in vacuo in a Waters PicoTag apparatus, derivatized by the Waters AccQ Tag amino acid analysis method, and separated on a Waters AccQ Tag C18 column (3.8 by 150 mm) developed with a 1:40 dilution of eluent A (Waters). Methylated arginine standards (NG-monomethylarginine, asymmetric NG,NG-dimethylarginine, and symmetric NG,N′G-dimethylarginine; Sigma) were analyzed under the same conditions, and their elution profiles were compared to the elution profiles of cell protein hydrolysates.
The 7.4.2 mutation was induced by the U3βGeoSupF retrovirus in mouse ES cells and identified during an in vitro screen for inserts into developmentally regulated genes. The provirus expresses a β-galactosidase–neomycin phosphotransferase fusion protein encoded by transcripts that initiate in the flanking cellular DNA. The 7.4.2 mutation was chosen for germ line transmission because expression of the β-galactosidase fusion gene was induced during differentiation in vitro (28). Undifferentiated 7.4.2 ES cells stained mostly white but showed an increase in lacZ expression following differentiation into embryoid bodies. lacZ expression was also developmentally regulated in vivo (Fig. (Fig.1).1). β-Galactosidase activity was not detected in blastocyst stage embryos (E3.5) but was induced in postimplantation embryos as early as E6.5. At E6.5, diffuse staining was evident throughout the whole embryonic portion of the egg cylinder, and by E7.5, staining appeared strongest along the midline of the neural plate and in the forming head fold (Fig. (Fig.1A).1A). Diffuse staining was also visible throughout the rest of the neural plate and in the primitive streak. E8.5 embryos stained mostly in the areas corresponding to the developing central nervous system, including both the brain and spinal cord regions. Intense staining was visible in the anterior-most aspect of the developing brain (prosencephalon) and continued caudally through the mesencephalon, rhombencephalon, and fusing neural folds (Fig. (Fig.1B1B to D). Neural-tube staining was restricted to the lateral sides, and the floor plate did not stain at all. Staining was also apparent in the primitive streak, brachial arches, notochord, and heart anlage. Again, diffuse staining was visible throughout the embryos. Between E9.5 and E13.5, this diffuse staining pattern became even more apparent; however, darker staining was still visible in the telencephalon, neural tube, heart, and first and second brachial arches (Fig. (Fig.1E).1E). In addition, staining could be seen in the developing forelimb bud. At all of the developmental stages examined, X-Gal staining was widespread in the embryo but absent from extraembryonic tissues, including the yolk sac, extraembryonic ectoderm, and ectoplacental cone.
Initial characterization of 48 offspring from intercrosses between 7.4.2 heterozygotes revealed that none of the viable offspring were homozygous for the provirus. Out of 48 offspring genotyped, 34 heterozygotes and 14 wild-type pups were recovered, a ratio of approximately 2.5:1. This situation strongly suggested that the 7.4.2 provirus disrupted a gene essential for murine development and that homozygous embryos die in utero. As 7.4.2 mice have been maintained for over 10 generations, the embryonically lethal phenotype appears to be tightly linked to the provirus.
In order to determine the time of embryonic death, 187 embryos obtained by crossing 7.4.2 heterozygotes were dissected from the uterine decidua between E6.5 and E10.5 and genotyped by PCR (Fig. (Fig.2B).2B). Since the sequences flanking the provirus were cloned (see below), wild-type and mutant alleles could be easily distinguished based on the size of the PCR product (Fig. (Fig.2A).2A). No homozygous embryos were recovered at E10.5 and E8.5, although a large number of empty resorption sites were observed (between 16 and 30% of the litter). Heterozygous embryos were morphologically indistinguishable from their wild-type littermates. The ratio of wild-type to heterozygous embryos indicated that 7.4.2 homozygotes die before E8.5, accounting for the large proportion of resorption sites. Only one homozygous embryo was recovered at E7.5. This embryo was much smaller than its wild-type and heterozygous littermates and resembled a severely disorganized and degenerated egg cylinder. Two homozygous embryos recovered at E6.5 also displayed an obvious developmental delay in embryonic portions of the embryo, while extraembryonic tissues seemed to develop normally. Other small and partially resorbed embryos dissected at the E6.5 and E7.5 stages were genotyped as heterozygotes. However, this probably reflects contamination of homozygous mutant embryos with maternal tissue, accounting for the lower-than-expected yield of 7.4.2 homozygotes at this stage. Since we were able to recover morphologically normal homozygous mutant blastocysts (E3.5), which were present at the expected ratio, it is likely that embryonic death occurs shortly after implantation (E4.5) but before the onset of gastrulation (E6.5).
To better document mutant phenotypes, histologic sections were prepared from embryos present in uterine decidual swellings at E6.5 to E7.5 (Fig. (Fig.3;3; Table Table1).1). Similar studies of embryos from wild-type litters served to establish the nature and frequency of nonspecific abnormalities unrelated to the 7.4.2 mutation. At E6.5, 28% of the sectioned embryos derived from heterozygous crosses displayed severe defects, a number much higher than that observed in litters from wild-type crosses (about 5%). Morphological abnormalities observed among the majority of aberrant embryos were consistent and different from the ones observed in the control litters. The presumptive 7.4.2 homozygotes were generally retarded in growth, especially in embryonic portions of the conceptus, and displayed signs indicating that they were undergoing resorption. Egg cylinders of these mutants were severely attenuated and often fragmented and lacked an organized two-layered cellular structure. No proamniotic cavities, ectoplacental cavities, or amniotic folds were observed in aberrant embryos. However, the mutant embryos appeared to develop normal ectoplacental cones and extraembryonic membranes and appeared to make good contact with the uterus. At E7.5, 23% of the decidui contained aberrant embryos (versus 5% in control litters). While the majority of the resorption sites were empty, some contained remnants of retarded and disintegrated egg cylinders. In summary, it appears that 7.4.2 homozygotes implant into the uterus but fail to develop beyond the early egg cylinder stage. Embryonic lethality of these mutants occurs shortly after implantation, and resorption is initiated prior to gastrulation.
Transcribed cellular sequences appended to proviral βGeo sequences were cloned by 5′RACE. This exploits the fact that cell-virus fusion transcripts initiate in the flanking cellular DNA and terminate at the poly(A) site in the 5′ LTR. The resulting 189-nt 5′RACE product was sequenced (Fig. (Fig.4A)4A) and used to probe Northern blots of cellular RNA from 7.4.2 cells. A single 1.3-kb transcript was detected (data not shown). The 5′ RACE fragment was subsequently used to isolate cDNAs of the mutated gene and genomic DNA sequences surrounding the 7.4.2 provirus.
The 5′ end of the consensus cDNA sequence (see below) matched regions of the 5′RACE product and also included a small region located in the genomic DNA sequence near the site of virus integration (Fig. (Fig.4).4). The first 110 nt of the 5′RACE product (shown as open and shaded boxes in Fig. Fig.4A)4A) and a 54-nt region of the flanking genomic sequence (shown as a shaded box in Fig. Fig.4B)4B) were both contained in the cDNA sequence. The 54-nt genomic sequence is flanked by consensus splice site sequences, further suggesting that it is an exon. The 5′RACE product contains this exon (shaded box), as well as a 56-nt sequence (open box, Fig. Fig.4A)4A) that is presumably derived from one or more upstream exons. The last 79 nt of the 5′RACE product (extending to the provirus integration site) match the genomic DNA sequence immediately upstream of the provirus and represent transcribed intron sequences that extend into the provirus.
The 1.3-kb genomic DNA fragment hybridized to a single 1.3-kb StuI fragment in DNA from wild-type mice (Fig. (Fig.4C).4C). An additional 10.5-kb fragment was detected in DNA of mice heterozygous for the 7.4.2 provirus. A fragment of the same size also hybridized to a virus-specific probe, consistent with insertion of an intact βGeo provirus (Fig. (Fig.4C).4C). In addition, no rearrangements were observed in either the virus or flanking cellular DNA.
The composite cDNA sequence of the gene disrupted by the 7.4.2 provirus is depicted in Fig. Fig.5.5. Open reading frames of 353 or 371 amino acids are present, depending on the absence or presence of the 54-nt exon. Although the first ATG in the composite cDNA sequence is in a favorable context for translation initiation, there is no in-frame upstream stop codon. Comparison of the composite cDNA sequence to 112 murine and 209 human expressed sequence tags obtained from the National Center for Biotechnology Information dbEST database revealed none that extended beyond the 5′ end of the composite cDNA (data not shown). The 54-nt exon was present in 5 out of 13 cDNAs from the E8.5 library but only in 1 of 10 cDNAs from the ES cell library.
A BLAST search of the nucleic acid databases with the composite 7.4.2 cDNA sequence returned matches with rat (PRMT1), human (HRMT1L2), and yeast (RMT1/HMT1) protein arginine N-methyltransferases. The rat cDNA lacked the 54-nt sequence corresponding to the presumed alternative exon (17). However, this sequence was present in two out of three variants of the human gene (29). The mouse and rat sequences are identical at the protein level, except for the 18-amino-acid segment missing from the rat protein that corresponds to the 54-nt alternative exon. The mouse and human proteins are 95% identical, differing at only two internal positions (E118 to V and H179 to Y) and 10 amino-terminal residues present in the mouse (and rat) protein but absent from human HRMT1L2 (Fig. (Fig.5).5). The mouse protein also shares 45% overall identity with the Saccharomyces cerevisiae RMT1 enzyme (12) and 11% identity with the Escherichia coli L11 methyltransferase (34). This high degree of phylogenetic conservation was especially evident in regions containing consensus methyltransferase motifs (Fig. (Fig.5).5). Based on these sequence comparisons, we concluded that the gene disrupted by the 7.4.2 provirus is the murine ortholog of the rat and human protein arginine N-methyltransferases (PRMT1 and HRMT1L2, respectively). The 7.4.2 gene was provisionally named Prmt1 for mouse arginine N-methyltransferase 1.
The pattern of β-galactosidase expression suggested that Prmt1 is regulated in a tissue- and/or stage-specific manner. Analysis of total cellular RNAs from several mouse tissues (liver, heart, lung, kidney, ovary, brain, and spleen); E8.5, E9.5, and E11.5 mouse embryos; and the D3 ES cell line by Northern blot hybridization revealed that Prmt1 encodes a 1.3-kb transcript present in all of the RNAs examined (Fig. (Fig.6A).6A). In general, the transcript seems to be more abundant in ES cells and embryos than in adult tissues. Prmt1 transcript levels also varied among all of the adult tissues tested, with the highest expression in the ovaries and uterus and the lowest in the liver. Of the three brain regions examined, only the cerebellum displayed a slightly higher transcript level.
In order to determine the expression pattern of the transcripts containing the 54-nt alternatively spliced exon, RT-PCR analysis was performed using the RNAs described above and primers flanking this exon. RT-PCR revealed that transcripts containing the message encoded by the 54-nt exon were uniformly present in ES cells, embryos, and all of the adult tissues tested (Fig. (Fig.66B).
The chromosomal location of the mouse Prmt1 locus was determined by interspecific backcross analysis using the Jackson Laboratory interspecific backcross panel (C57BL/6JEi × SPRET/Ei)F1 × SPRET/Ei. This mapping panel has been initially typed for over 450 chromosomal and X-linked loci spread among the autosomes, as well as the X chromosome, and the resulting genetic map was anchored to published maps by mapping known loci, simple sequence length polymorphisms, and loci defined by endogenous retroviruses. In order to identify a suitable RFLP for use in mapping studies, C57BL/6JEi and M. spretus genomic DNAs were digested with a panel of restriction enzymes and analyzed by Southern blot hybridization using the 1.3-kb Prmt1 probe. Subsequently, an MspI RFLP was used to map the Prmt1 locus on mouse chromosome 7 between Fig1 and D7Bwg0826e (Fig. (Fig.7).7). No known mutation or disease phenotype is linked to the Prmt1 locus.
In principle, the Prmt1 mutation could be a cell-lethal defect in which mutant embryos persist beyond implantation due to maternal stores of mRNA, enzyme, or methylated substrates. Although most maternal RNA is rapidly degraded following the first cell division, the Prmt1 enzyme or methylated substrates could persist for several days. To determine whether Prmt1 is required for cell viability, we sought to derive a homozygous mutant ES cell line from preimplantation blastocysts. In our experience, ES cell lines are easier to derive from 129sv mice than from the C57BL/6 mice that harbor the 7.4.2 mutation. Therefore, the Prmt1 mutation was crossed for 3 generations into 129sv mice. F3 mice heterozygous for the provirus were interbred, and blastocysts were collected at E3.5 and cultured as described in Materials and Methods. Individual cell lines were established from 9 out of 40 blastocysts, of which 5 were homozygous for the Prmt1 mutation. All mutant cell lines appeared morphologically similar to other ES cell lines and grew with doubling times similar to that of wild-type ES cells (approximately 12 h; data not shown).
U3 gene trap vectors were designed to disrupt cellular gene expression by terminating transcription at one of two poly(A) sites (one in each LTR) carried by the provirus (14). Previous studies found high levels of a 4.7-kb βGeo transcript in 7.4.2 cells, approximately the size expected for Prmt1-βGeo fusion transcripts that terminate in the 5′ LTR. To test whether the 7.4.2 provirus disrupted expression of the Prmt1 gene, homozygous mutant cell lines (Fig. (Fig.8A)8A) were analyzed by Northern and Western blotting. Transcripts of 1.3 kb were detected in the wild-type and heterozygous cell lines, but not in the two mutant cell lines, using Prmt1 cDNA sequences downstream of the site of provirus integration as a probe (Fig. (Fig.8B).8B). Similarly, the Prmt1 protein was detected in extracts from wild-type and heterozygous cells but not in those from homozygous mutant cells (Fig. (Fig.8C).8C). Levels of Prmt1 transcripts in mutant cells were estimated by RT-PCR after mixing of different amounts of wild-type RNA with a constant amount of mutant RNA (Fig. (Fig.8D).8D). The PCR signal obtained after mixing of mutant RNA with a 100-fold dilution of wild-type RNA was clearly greater than that obtained with mutant RNA alone, whereas the 250-fold dilution did not enhance the amplification of Prmt1 sequences. This suggests that mutant cells express approximately 1% of the wild-type level of Prmt1 transcripts. We conclude that the 7.4.2 provirus severely disrupts Prmt1 expression and that the protein is not required for cell viability.
Although Prmt1 is orthologous to the predominant arginine methyltransferase expressed in mammalian cells, other enzymes are known to have type 1 activity (32). In addition, loss of Prmt1 could lead to a compensatory increase in the expression or activity of other methyltransferases. We therefore examined total methyltransferase activities in wild-type and Prmt1 mutant cells. As shown in Fig. Fig.9A,9A, Prmt1−/− cells expressed approximately eight times less methyltransferase activity than Prmt1+/+cells, as assayed with an optimal synthetic-peptide substrate (21). This reduction is comparable to the contribution by Prmt1 orthologs to the total methyltransferase activity in other cell types, as estimated biochemically (12, 31). Therefore, we conclude that Prmt1 is the major type I enzyme in ES cells and that loss of Prmt1 activity does not result in a major compensatory increase in other methyltransferases.
Mutations in RMT, a type I enzyme of yeast, result in increased levels of hypomethylated substrates (12). Therefore, two types of experiments were performed to examine the methylation status of proteins in Prmt1−/− cells. First, total proteins extracted from Prmt1−/− and Prmt1+/+ cells were tested for methyl acceptor activity when mixed with wild-type cell extracts. As shown in Fig. Fig.9B,9B, the methyl acceptor activity of total protein isolated from mutant cells was approximately 15 times greater than that observed for proteins from wild-type cells. The fact that proteins from wild-type cells proved to be such poor substrates suggests that most potential substrates are blocked by prior methylation, as has been observed with nucleolin and fibrillarin, which exist in a fully methylated state developmentally in vivo (18, 19). Second, the methylarginine content of cellular proteins was measured following acid hydrolysis and high-pressure liquid chromatography. As shown in Fig. Fig.9C,9C, total levels of asymmetric dimethylarginine were approximately twofold lower in Prmt1-deficient cells than in wild-type cells (54% ± 1.5% reduction based on three independent pairs of samples) whereas levels of monomethylarginine and symmetric dimethylarginine were slightly increased.
We previously described methods for using lacZ-based vectors to screen for mutations in regulated genes (25, 28). The present study characterized one of these mutant cell lines, termed 7.4.2, generated by an inserted U3βGeo provirus. The 7.4.2 cell line showed an increase in lacZ expression upon differentiation of ES cells into embryoid bodies (28). We show here that the gene disrupted by the provirus encodes the murine arginine methyltransferase 1 enzyme (EC 18.104.22.168). Embryos homozygous for the mutation arrest in development prior to E6.5, indicating that protein methylation is required for early postimplantation development. However, cell lines derived from homozygous mutant embryos are viable despite the fact that Prmt1 enzyme levels and the extent of protein methylation are significantly reduced.
cDNAs encoding Prmt1 were characterized as part of our analysis of the 7.4.2 mutation. As previously reported for the human gene, Prmt1 transcripts are alternatively spliced to generate coding sequences for proteins of 353 or 371 amino acids. Overall, murine Prmt1 shares 100 and 95% sequence identity with the rat and human proteins, respectively. Phylogenetic conservation of the two proteins suggests that each has a distinct function, possibly related to substrate specificity or regulation of enzyme activity. Finally, Prmt1 was mapped to mouse chromosome 7 in a region syntenic with the region of human chromosome 19q, where the human gene had been previously mapped (29). Neither region is associated with disease loci in mice or humans.
Our analysis of Prmt1 expression confirms earlier reports describing widespread expression of the orthologous rat and human genes in all of the tissues examined (17, 29, 32). In addition, all of the tissues examined expressed both spliced forms of the murine gene. Expression of the inserted β-galactosidase marker was highest in developing neural structures. This is potentially significant in light of previous studies showing high levels of methyltransferase activity in the developing brain that decline after birth (22). However, homozygous mutant embryos die close to the time when the Prmt1 gene is first induced, as assessed by β-galactosidase staining, and well before the onset of neural development. As early embryonic death appears to be a common consequence of mutations disrupting basic cellular processes (7), the phenotype is consistent with the important role Prmt1 is thought to play in RNA metabolism. For example, the timing of embryonic death is similar to that observed with a mutation in Fug1, which encodes the RAN GTPase-activating protein (8), and is earlier than that observed with a mutation in hnRNP C, a highly abundant, ubiquitous constituent of nuclear riboprotein complexes (35). Both RAN GAP and hnRNP C appear to function in RNA biogenesis and transport.
By comparing the levels of enzyme activity in extracts from wild-type and Prmt1-deficient cells, we showed that Prmt1 accounts for over 85% of the total arginine methyltransferase activity in ES cells, as assayed using an optimal (21) synthetic RGG substrate. However, Prmt1 is responsible for just over 50% of the normal steady-state levels of NG,NG-dimethylarginine present in cellular proteins. Therefore, other arginine methyltransferases, presumably with different substrate specificities, appear to make significant contributions to the total dimethylarginine content of the cell. Such enzymes could include PRMT3 and HRMT1L1, type I-related enzymes that modify RGG substrates poorly, if at all (29, 32).
The methylation status of cellular proteins was also assessed by testing their capacity to be methylated in vitro. Prmt1, present in wild-type cell extracts, was incubated with proteins isolated from either wild-type or Prmt1-deficient cells in the presence of S-adenosyl-l-methyl-[3H]methionine. While proteins from wild-type cells had negligible acceptor activity, proteins from Prmt1-deficient cells were significantly hypomethylated. This has two implications. First, the function of Prmt1 appears to be nonredundant, since other cellular enzymes compensate for no more than a small part of the lost Prmt1 activity. Second, as previously observed with nucleolin and fibrillarin (18, 19), most potential substrates in normal cells appear to be blocked by prior methylation. Our analysis does not exclude the possibility that individual substrates are present in a hypomethylated state in ES cells or that the methylation status may change in response to different physiological conditions. Hypomethylated proteins may also exist only transiently, for example, during a specific step in a biochemical process. However, this latter possibility would require an arginine demethylase, an activity that has not yet been identified in mammalian cells. Experiments to study potential dynamic changes in protein methylation are in progress.
In summary, this report describes the first mutation in a mammalian arginine methyltransferase. These enzymes have been implicated in a wide variety of cellular processes, including cell proliferation, signal transduction, and protein trafficking. The availability of Prmt1-deficient cells will assist efforts to identify physiological substrates and to understand the function of the enzyme in normal cellular metabolism.
We thank Harvey Herschman for the gift of anti-Prmt1 antibody, Abudi Nashabi for technical assistance, Eric Howard for measurements of protein methylarginine content, and Lucy Rowe, Mary Barter, and Lois Maltais of The Jackson Laboratory for assistance with gene mapping and nomenclature.
This work was supported by Public Health Service grants (R01HG00684, R01GM51201, and R01RR13166 to H.E.R.) and by a grant from the Kleberg Foundation. Additional support was provided by an NCI Cancer Center Support Grant (P30CA42014) to the Vanderbilt-Ingram Cancer Center. M.J.R. was supported by a Medical Scientist Training Grant (5T32-GM07347).