Rem Is a New Member of the Rad- and Gem/Kir Ras-related GTP-binding Protein Family Repressed by Lipopolysaccharide Stimulation*

We report the cDNA cloning and characterization of a novel GTP-binding protein, termed Rem (for Rad and Gem-related), that was identified as a product of polymerase chain reaction amplification using oligonucleotide primers derived from conserved regions of the Rad, Gem, and Kir Ras subfamily. Alignment of the full-length open reading frame of mouse Rem revealed the encoded protein to be 47% identical to the Rad, Gem, and Kir proteins. The distinct structural features of the Rad, Gem, and Kir subfamily are maintained including a series of nonconservative amino acid substitutions at positions important for GTPase activity and a unique sequence motif thought to direct membrane association. Recombinant Rem binds GTP in a specific and saturable manner. Ribonuclease protection analysis found Rem to be expressed at comparatively high levels in cardiac muscle and at moderate levels in lung, skeletal muscle, and kidney. The administration of lipopolysaccharide to mice, a potent activator of the inflammatory and immune systems, results in the general repression of Rem mRNA levels in a dose- and time-dependent manner. Thus, Rem is the first Ras-related gene whose mRNA levels have been shown to be regulated by repression.

The Ras family of low molecular weight GTP-binding proteins has been implicated in a wide range of cellular processes, including cell growth and differentiation, intracellular vesicular trafficking, nucleocytoplasmic transport, and cytoskeletonal reorganization. To date, six subfamilies have been identified: Ras, Rho, Rab, Ran, ARF, and a newly described family composed of the Rad, Gem, and Kir proteins (1,2). These subfamilies are defined largely by primary sequence relationships but also by their regulation of common cellular functions. All GTPases of the Ras superfamily contain five well conserved amino acid motifs involved in guanine nucleotide binding and hydrolysis (1,3). These primary sequence motifs have been evolutionarily conserved and define a conserved structure whose importance has been confirmed through extensive mutational analysis (4). Therefore, the sequence of all GTPases share approximately 20 -30% amino acid identity, whereas the sequence identity is considerably higher within subfamilies (5). In addition, most family members share conserved COOH-terminal cysteine rich motifs needed for covalent modification by isoprenoid lipids (6). Prenylation is the initial step in the attachment of these proteins to the cytoplasmic leaflets of a variety of cellular organelles (7) and has been shown to be required for normal membrane localization and biological activity (8).
The Ras-related GTPases are thought to act as binary switching molecules, alternating between an active GTP-bound and an inactive GDP-bound structural state (4). They respond to external signals by exchanging GTP for constitutively bound GDP, thereby triggering intracellular signaling cascades. The signal is terminated when the protein hydrolyzes its bound GTP to GDP in a reaction that is stimulated by guanosine triphosphatase (GTPase)-activating proteins (GAPs) 1 (9). Progression through this GTPase cycle is regulated by additional regulatory proteins, including factors that stimulate guanine nucleotide exchange. These regulatory proteins may themselves possess intrinsic effector activity (9) and associate with specific GTPases largely through interactions with their G2 effector domains. A variety of G2 regions allow the Ras-related GTPases to interact with a wide range of cellular effector molecules (4). Thus, small GTPases are extremely versatile and found to regulate an array of cellular processes (2,9).
The Rad, Gem, and Kir proteins are the first members of a new class of Ras-like GTPases (10 -12). Rad shares 61% identity at the amino acid level with Gem (11) and Kir (12), whereas the coding sequences of the Gem and Kir genes are 98% identical, differing significantly only in the 5Ј-untranslated sequences (12). The RGK proteins share structural features that are distinct from other Ras-related proteins. These include several nonconservative amino acid substitutions within regions known to be involved in guanine nucleotide binding and hydrolysis including unique G2 effector and G3 domains, extended NH 2 and COOH termini and a conserved COOH-terminal motif thought to mediate membrane association but lacking classical CAAX motifs needed to direct protein isoprenylation. In addition, the members of this Ras subfamily are subject to transcriptional regulation. Rad was found to be overexpressed in muscle of type II diabetes patients (10), and Gem expression is induced in mitogen-stimulated T-cells (11), whereas Kir expression is induced by oncogenic kinases (12). Although the cellular function of these proteins remains to be established, Rad has been shown to associate with skeletal muscle ␤-tropomyosin and the cytoskeleton of muscle cells (13) and to inhibit insulin-stimulated glucose uptake in a variety of cultured cell lines (14). This suggests a role for Rad in skeletal muscle * This work was supported in part by National Institutes of Health Grant EY 11231 and Grant IN-163 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  function and cytoskeletal organization. The deregulated expression of Gem prevents proliferation of normal and transformed 3T3 cells, suggesting that Gem is involved in regulating signaling pathways that influence cell growth (11). Finally, the cellular levels of Kir are dramatically increased in pre-B cells transformed by a select set of abl tyrosine kinase oncogenes (12). The correlation between Kir expression and the tumorigenic and metastatic potential of BCL/ABL and v-ABL transformed cells suggests that Kir may participate in the processes of invasion or metastasis. When expressed in yeast, Kir leads to the formation of pseudohyphae, a developmental transition normally induced by nitrogen starvation (15). Genetic analysis suggests that Kir acts upstream of the STE20 kinase and results in the activation of a mitogen-activated protein kinase cascade (15). These results are consistent with a model in which Kir, and perhaps other members of the RGK family, may regulate cellular signaling cascades by controlling the activity of mitogen-activated protein kinases that have yet to be determined.
In this report, we describe the initial characterization of a novel Ras-related GTP-binding protein, Rem, first identified using a degenerate PCR strategy. On the basis of structural criteria, Rem is a new member of the Rad, Gem, and Kir subfamily of Ras-related proteins. Rem mRNA is expressed predominantly in skeletal and cardiac muscle, lung, and kidney, and the bacterially expressed protein is shown to bind GTP in a specific and saturable manner. Because other members of the RGK family are transcriptionally regulated, we examined whether a similar method of regulation controlled the expression of Rem. Surprisingly, we find that in mice treated with lipopolysaccharide, a potent activator of cells of the immune and inflammatory systems, the levels of Rem mRNA are repressed in a dose-and time-dependent manner. Thus, Rem is the first Ras-related GTP-binding protein whose mRNA levels are regulated by repression.

EXPERIMENTAL PROCEDURES
General Methods-Standard molecular biology techniques were used (16). cDNA clones were subcloned to plasmid pBluescript II vectors (Stratagene) and sequenced by the dideoxy chain termination method (16) using the M13 universal primer or specific internal primers. Nicktranslated probes were synthesized using a labeling kit (Life Technologies, Inc.).
Reverse Transcriptase Polymerase Chain Reaction Identification and Cloning of Rem-The strategy used to generate the PCR-derived fragment of Rem from conserved amino acid sequences found within the Rad and Gem/Kir small GTP-binding proteins is outlined in Fig. 1. First strand cDNA from mouse testis and brain total RNA was a kind gift from Dr. Kevin Sarge of this department. The first strand cDNA was used for PCR with the degenerate PCR primers AT(A/C/T)AT(A/C/T)(C/ T)TNGTNGGNAA(C/T)AA and GT(T/C)TC(A/T/G) AT(A/G)AA(C/ T)TT(G/A)CA(A/G)TC(AG)AA. The resultant 105-base PCR product was blunt ended with T4 polymerase, and 5Ј-phosphate was added with polynucleotide kinase (17). The PCR product was then subcloned into the EcoRV site of pBluescript KS ϩ (Stratagene, La Jolla, CA). Sixty individual clones were sequenced using a Sequenase kit (U. S. Biochemical Corp.). Two PCR products, when sequenced, were found to contain a unique cDNA fragment that was used to design a Rem-specific oligonucleotide (Rem-3; see Fig. 1). The Rem-3 oligonucleotide was then used to screen an oligo(dT)-primed Uni-ZAP XR adult mouse testis cDNA library (the kind gift of Dr. Debra J. Wolgemuth, Center for Reproductive Sciences, Columbia University College of Physicians and Surgeons, New York, NY). Approximately 30,000 plaques were transferred in duplicate to filters that were then probed with 1 ϫ 10 6 cpm/ml of the end labeled oligonucleotide, TTCTACCGAGACTTCCCGGC. Filters were hybridized at 42°C in hybridization solution containing 6 ϫ SSC, 1 ϫ Denhardt's solution, 100 g/ml yeast tRNA, and 0.05% sodium pyrophosphate (18). The filters were washed for 2 h with 6 ϫ SSC and 0.05% sodium pyrophosphate at room temperature with three changes of wash buffer. Three positives were identified, and the largest (Rem 10-1, 1.1-kilobase insert) was plaque purified, rescued using phage mediated in vivo excision (according to manufacturer's protocol, Strat-agene), and sequenced. To obtain a larger cDNA clone, the mouse testis library was rescreened with 1 ϫ 10 6 cpm/ml of the nick-translated EcoRI/XhoI fragment of pRem 10-1. Duplicate filters were hybridized at 42°C in a 5 ϫ SSPE solution (18) containing 50% formamide. The filters were washed in 0.2 ϫ SSC and 0.1% SDS for 1 h at 60°C with three changes of wash buffer. Of five positives, the clone Rem 6-2 was plaque purified and rescued as a pBS plasmid because of its ability to hybridize to 5Ј-directed Rem-specific oligonucleotides.
To obtain a full-length cDNA clone, a bacteriophage cDNA library was constructed from mouse kidney. Poly(A) ϩ RNA was isolated using the Straight A's mRNA isolation system (Novagen, Madison, WI) and used to construct a random primed cDNA library (Directional RH Random Primer cDNA Library Construction System, Novagen). Approximately 30,000 plaques were transferred in duplicate to filters that were probed under high stringency with 1 ϫ 10 6 cpm/ml of the nicktranslated EcoRI/XhoI fragment of pRem 6-2. Of the four positive clones identified, pRem 5, which contained the largest and most 5Јextended insert (0.9 kilobase), was rescued and characterized by DNA sequencing. To construct a plasmid that contained the full Rem coding region, nucleotides 1-502 of pRem 5 were ligated to nucleotides 56 -1050 of pRem 6-2 using a unique NheI site. The resulting plasmid, pRemWT, was characterized by restriction mapping and sequencing.
RNase Protection Assays-Mouse total RNA was isolated using a STAT-60 kit (Tel-Test B, Friendswood, TX) according to the manufacturer's protocol. The PstI fragment of pRem 6-2 was subcloned into pBluescript KS ϩ (Stratagene, La Jolla, CA) to create the plasmid pRem PstI. Plasmids containing an 89-bp fragment of the ribosomal protein L32 (19) or a 303-bp fragment of mouse TNF-␣ cDNA subcloned in the vector pGEM-4 were a gift of Dr. Daniel Noonan of this department. Antisense radiolabeled riboprobes were prepared using linearized templates and a Maxiscript (Ambion, Austin, TX) kit according to the manufacturer's protocol. RNase protection assays were performed according to the method of Hobbs (19) with minor modifications. Briefly 20 g of total RNA was dissolved in 8 l of hybridization buffer (80% formamide, 0.3 M NaCl, 1 mM EDTA, 40 mM Pipes, pH 6.7). For LPS experiments, two protection assays were performed for each tissue. The first contained a mixture of Rem and L32 riboprobes, whereas the second contained the TNF-␣ and L32 riboprobe mix. 1 l (500,000 cpm) of each probe (in hybridization buffer) was added to each sample. Samples were overlaid with mineral oil, incubated for 1 min at 90°C, and subsequently incubated for 12-16 h at 56°C. Single-stranded RNA was digested as described by Hobbs (19). The reaction was stopped by the addition of an equivalent volume (110 l) of 4 M guanidine thiocyanate, 0.5% sodium N-lauryl-sarcosine, 25 mM sodium citrate, pH 7.0, 0.1 M ␤-mercaptethanol. Yeast tRNA (25 g) was added as carrier, and total RNA was precipitated by addition of isopropanol (225 l) followed by centrifugation (20). The pellet was dissolved in 10 l of loading buffer (80% formamide, 10 mM EDTA, 1 mg/ml bromphenol blue, 1 mg/ml xylene cyanol) and electrophoresed in a 5% acrylamide/8 M urea sequencing gel. The gel was dried and exposed to X-OMAT AR film (Eastman Kodak Co.) for the indicated time. The gel was quantitated using a Molecular Dynamics PhosphorImager SF (model 455A). Simultaneous measurement of the rpL32 transcripts, which encode the L32 ribosomal protein (19), served as an internal control for housekeeping gene levels.
Experimental Animals-The tissues of adult female C57BL6/C3H mice were kindly provided by Dr. Mark Kindy of this department. Lipopolysaccharide stimulation was achieved by intraperitoneal injection (10 g/animal) to paired animals for each time point. Lipopolysaccharide was from Escherichia coli 0111:B4 (DIFCO). Control animals were not injected. At the indicated times the animals were treated with metafane and euthanized by cervical dislocation, and tissues were harvested immediately and quick frozen on dry ice. RNA was isolated and used for ribonuclease protection analysis as described above. These studies were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Recombinant Protein Production-Recombinant Rem was expressed as a glutathione S-transferase (GST)-fusion protein. A Rem PCR product containing a 5Ј BamHI restriction site was generated using oligonucleotides 5Ј-CGCGGATCCATGACTCTTAACACGCA and 3Ј-TTC-TACCGAGACTTCCCGGC, sequenced to verify the cDNA, and subcloned to BamHI/NheI-digested pRemWT to create pRem express. The BamHI/XhoI fragment of pRem express was subsequently cloned in-frame to pGEX-KG (21) to create pRem GEX. GST-Rem was produced in BL21DE3 cells upon isopropyl-␤-D-thiogalactopyranoside addition, purified on glutathione agarose beads (Sigma), and removed from GST by thrombin cleavage as described (22). Thrombin cleavage was deemed to be Ͼ95% efficient as judged by Coomassie Blue staining after SDS-polyacrylamide gel electrophoresis analysis and Western blotting with anti-GST antibodies. A bacteria protein co-purified with Rem and was resistant to extensive high salt washes of the glutathioneagarose beads (see Fig. 4C). Thrombin was removed by incubation with benzamidene-Sepharose 6B (Pharmacia Biotech Inc.) for 1 h at 4°C with mixing. The protein was then dialyzed against 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, and 5% glycerol and stored in multiple aliquots at Ϫ70°C.
GTP Binding Assays-GTP binding to Rem was determined with the rapid filtration assay (23). Rem (1 g) or bovine serum albumin (1 g) was incubated in binding buffer (20 mM Tris, pH 7.5, 50 mM NaCl, 0.1% Triton X-100, 1 mM dithiothreitol, 40 g/ml bovine serum albumin, and 1 M GTP␥S (0.45 Ci/sample)) with the indicated concentration of Mg 2ϩ or EDTA at 22°C. At the indicated times, aliquots of 100 l were withdrawn in duplicate, and the reaction was stopped by addition of 400 l of ice-cold wash buffer (20 mM Tris, pH 7.5, 50 mM NaCl, and the indicated Mg 2ϩ concentration) and immediately filtered through BA 85 nitrocellulose filters (Schleicher & Schuell) followed by washing with 12 ml of ice-cold wash buffer. The radioactivity remaining on the filters was determined by scintillation counting. Nonspecific background was determined by performing the binding assay in the absence of added protein. To assess the specificity of GTP binding, cold ribonucleotides were added to a final concentration of 20 M, and the binding reaction was allowed to proceed for 1 h prior to rapid filtration analysis.

PCR Cloning and Nucleotide Sequence of Mouse Rem-
We sought a comprehensive method to identify additional members of the RGK family of low molecular weight GTP-binding proteins. A PCR-based strategy using degenerate oligonucleotide primers designed to exploit amino acid differences between the guanine nucleotide-binding domains of the RGK family and other Ras-related GTP-binding proteins was used to isolate novel RGK-related cDNA fragments from mouse testis cDNA ( Fig. 1 and see "Experimental Procedures"). The 5Ј oligonucleotide was designed to partially overlap the G4 guanine nucleotide-binding site motif IILVGNK, whereas the 3Ј oligonucleotide recognized a portion of the G5 guanine nucleotide-binding region and a flanking region of amino acids highly conserved within the RGK family FDCKFIET. PCR generated a pool of cDNAs of the approximate predicted size (105 bp), which were subcloned and sequenced. Sequencing of 60 individual clones revealed that approximately 90% of the products were identical to the previously characterized Rad, Gem, or Kir genes (10 -12). Two of the remaining clones encoded a previously unknown sequence that was homologous to but clearly different from RGK genes (Fig. 1). To emphasize this relationship we designated this clone Rem, for Rad and Gem-related.
A Rem-specific oligonucleotide probe was designed (Rem-3) and used to screen a mouse testis cDNA library, from which several clones were isolated ( Fig. 1; see "Experimental Procedures"). The largest of these cDNA clones was Rem 6-2, which extended from a polyadenylation tract to nucleotide 441 (nucleotide positions refer to the final sequence of the cDNA that can be retrieved from GenBank U91601). Preliminary ribonuclease protection assays using a portion of the Rem 6-2 cDNA indicated that Rem was abundantly expressed in kidney (data not shown). Based on the high levels of expression in kidney, a random-primed mouse kidney cDNA library was constructed and screened with a probe generated from the Rem 6-2 clone to obtain the 5Ј end of the Rem cDNA (see "Experimental Procedures"). Four additional cDNA clones were isolated, one of which (Rem 5) extended from nucleotides 1 to 881. These overlapping cDNA clones (Rem 6-2 and Rem 5) were spliced together to form a full-length cDNA. The 1.5kilobase cDNA sequence included a portion of the 5Ј-untranslated region, an 891-nucleotide (219 -1109) open reading frame with a putative initiator methionine in a region that matched the Kozak sequence motif (24), and a large 3Ј-untranslated region followed by a polyadenylate tail. No other methionines were observed 5Ј to the putative start codon in any other reading frame. In addition, 5Ј to the Kozak sequence and 3Ј to the in-frame stop codon, stop codons were observed in all three reading frames. Analysis of the open reading frame revealed significant identity at the nucleotide level with members of the Ras superfamily.
Mouse Rem Protein-The deduced amino acid sequence of Rem is shown in Fig. 2. This cDNA predicts a protein of 297 amino acids with a calculated molecular size of 32,893 Da. A data base search was performed to determine the degree of homology of Rem with the RGK family and to identify additional G-proteins with structural similarities. The Rem protein contains a core sequence (amino acids 84 -246) that is highly related to members of the Ras superfamily of small GTP-binding proteins. The highest degree of similarity was with mouse and human Gem, Kir, and Rad (46.7-47.2% sequence identity),

FIG. 1. Selective amplification of a novel Rad-and Gem/Kir-related cDNA fragment by PCR.
Oligonucleotide primers-1 (2304-fold degenerate, sense strand) and -2 (192-fold degenerate, antisense strand) were based on the sequences, respectively, of the conserved G4 and G5 guanine nucleotide-binding regions found within all Ras-related GTP-binding proteins. Primer-2 encodes a sequence that is highly conserved within the Gem/Kir and Rad proteins but not among other Ras-related GTPases. These primers were used in PCR with 5 l of cDNA synthesized from mouse testis poly(A) ϩ RNA. The resulting 105-bp products of this reaction were analyzed, and two were found to encode a novel cDNA fragment as described under "Experimental Procedures." A 20-bp Rem-specific screening oligonucleotide (primer-3) was synthesized on the basis of the nucleotide sequence of these PCR-derived products and used to screen a mouse testis cDNA library as described under "Experimental Procedures." The presence of multiple nucleotides at a given position in the primers is indicated; the symbol N in the nucleotide sequence indicates that all four nucleotides were used.
but there was also a high degree of homology with additional Ras-like GTPases, including 22.8, 25.9, and 22.3% identity to dictyostelium RasA (25), human Rap-2A (26), and Rap-2B (27). To assure ourselves that Rem was a new member of the RGK Ras subfamily, we performed a general comparative protein analysis among the different Ras-like subfamilies (data not shown). On the basis of this comparison, it is evident that Rem is a novel member of the RGK Ras-related subfamily. Fig. 2 depicts the alignment of Rem with a select subset of these proteins. The greatest similarity exists in regions that correspond to the guanine nucleotide-binding domains conserved in all Ras family members. The Rem protein exhibits all five of the domains (G1-G5) that have been shown to take part in both guanine nucleotide binding and the catalytic functions of the Ras protein superfamily (4). Although both the NH 2 -and COOH-terminal extensions past the Ras core region are divergent, the COOH-terminal 10 amino acids of Rem are highly conserved in Rad, Gem, and Kir proteins (Fig. 2). This region does not contain a typical CAAX, XXCC, or CXC prenylation site (where A is an aliphatic amino acid and X is any amino acid) present in almost all Ras family members, although it does contain a conserved cysteine residue at position 7 from the COOH terminus. This conserved COOH-terminal motif may therefore represent a novel lipid modification site or direct the association of these proteins with membranes by interaction with an anchoring protein.
Tissue Distribution of Rem mRNA-Ribonuclease protection analysis revealed that Rem was expressed at detectable levels in every mouse organ examined (Fig. 3). The highest basal levels were detected in cardiac and skeletal muscle with slightly lower levels of Rem mRNA in lung and kidney. Low levels of mRNA were identified in spleen and brain with barely detectable levels in several additional tissues. The significance of the abundance and distribution of Rem message in these tissues is unclear. The tissue distribution of Rem contrasts with that of both Rad and Gem; Gem mRNA is most abundant in kidney, lung, and spleen, whereas Rad is expressed in cardiac and skeletal muscle and lung (10,11).
Rem Is a GTP-binding Protein-The close homology between Rem and Ras in the regions associated with the binding of GTP suggested that Rem was a GTP-binding protein. To determine whether the Rem cDNA coded for a protein possessing GTPbinding activity, recombinant Rem was expressed and affinity purified (see "Experimental Procedures"). The recombinant protein migrated with an apparent molecular mass of 39 kDa, which is in close agreement with the calculated molecular mass of the Rem cDNA of 33 kDa. Binding of GTP was then assayed with [ 35 S]GTP␥S, and bound GTP was separated by rapid filtration on nitrocellulose. As seen in Fig. 4A, Rem binds GTP␥S rapidly and in a Mg 2ϩ -dependent manner upon incubation under standard nucleotide exchange conditions. Nucleotide binding was saturable and completely blocked by excess (20-fold) unlabeled GTP and GTP␥S, whereas pyrimidine nucleotides do not compete to any extent (Fig. 4B). ATP also showed slight inhibition of GTP binding to Rem at high concentrations. These GTP-binding properties closely resemble those of recombinant Rad and indicate that Rem specifically associates with guanine nucleotide (28).

FIG. 2. Comparison of the amino acid sequences of the mouse Rem (mRem), mGem, human Rad, and mN-Ras proteins.
The alignment was performed with the CLUSTAL W1.6 program (33). Hyphens represent gaps introduced for optimal alignment. Numbers indicate residue numbers. Amino acid residues that are conserved in at least three of the four proteins in the alignment are placed in shaded boxes. Consensus sequences for GTP-binding regions are labeled G1-G5 and indicated in italics (4). The mouse Rem cDNA and predicted protein sequence may be retrieved from GenBank using accession number U91601; mGem, mouse Gem (accession number U10550); hRad, human Rad (accession number L24564); mN-Ras, mouse N-ras (accession number P08556).

Repression of Rem mRNA by Lipopolysaccharide-
The mRNA levels of both Gem and Kir are transcriptionally induced, respectively, in mitogen-stimulated T-cells and abl tyrosine kinase transformed B-cells (11,12). To determine whether Rem was regulated in a similar fashion, we characterized the level of Rem mRNA expressed in a variety of mouse tissues after exposure to the potent immune and inflammatory system stimulant, LPS (29,30). Surprisingly, intraperitoneal injection of LPS caused a potent and transient decrease in the levels of Rem mRNA in cardiac muscle, skeletal muscle, kidney, spleen, and thymus (data not shown). To further characterize the lipopolysaccharide-stimulated repression of Rem mRNA levels in mouse tissues, its time course was determined in cardiac muscle (see "Experimental Procedures"). As is illustrated in Fig. 5, Rem mRNA was repressed 8.5-fold below basal levels upon LPS treatment (10 g/animal), with repression first detected within 2 h and maximal repression achieved within 4 h of LPS treatment (Fig. 5B). Rem repression was sustained to 8 h and returned to basal levels by 20 h. The degree of repression and its duration were both dose-and mouse straindependent (data not shown). During this same time course, the inflammation marker TNF-␣, a cytokine that is known to be markedly stimulated by LPS treatment (30), was analyzed (Fig. 5A). The time course of TNF-␣ expression somewhat preceded that of Rem with maximal expression at 2 h. In addition, TNF-␣ stimulation was transient with return to basal levels of mRNA by 8 h after LPS stimulation. Because LPS is known to stimulate cells of the monocytic lineage as well as lymphocytes, we examined the level of Rem mRNA in primary T cells, B cells, and macrophages as well as in the macrophage cell lines RAW 264.7 and P388D1 (data not shown). However, mRNA was not detected in these cells, suggesting that Rem is not expressed in cells of the immune system. DISCUSSION We report here the identification of a 33-kDa GTP-binding protein, Rem, a novel member of the Ras family of GTP-binding proteins whose mRNA levels are rapidly and markedly decreased by LPS in a variety of mouse tissues. The predicted amino acid sequence of Rem is most highly homologous to the Rad, Gem, and Kir gene products, and by virtue of this similarity we include Rem as the newest member of the RGK subfamily of Ras-related small G-proteins. Rem is an interesting gene both by virtue of its rapid repression by LPS and because the amino acid sequence indicates several novel features that distinguish it from the majority of Ras-related proteins.
Analysis of the Rem cDNA sequence indicates that it encodes a protein of 297 amino acids with a deduced molecular mass of 32,893 Da. The regions of highest homology between Rem and Ras family members correspond to the five regions of Ras that constitute the core consensus involved in GTP binding (G1-G5; Fig. 2) (4), whereas the flanking NH 2 -and COOH-terminal sequences are unrelated to Ras. These extended regions account for the larger molecular weight of Rem when compared with the majority of members of the Ras family. The guanine nucleotide consensus sequences GX 4 GK(T/S) (G1 motif), NKXD (G4 motif), and EXSA (G5 motif) are present and well conserved in Rem (Fig. 2). However, Rem, along with other proteins in the RGK subfamily, contains a number of divergent amino acids at highly conserved positions that are known to be critical for normal GTP hydrolysis. Positions 89, 135, and 137 in Rem, which correspond to amino acids 12, 59, and 61 in Ras, are substituted relative to most of the previously characterized small GTPases. These three positions when changed to a variety of amino acids confer oncogenicity by rendering Ras defective for GTP hydrolysis or insensitive to GAP stimulation, thereby causing Ras proteins to be maintained in a GTP-bound active state (3,31,32). However, whereas mutagenesis studies have suggested that amino acid substitutions within this conserved catalytic region impair GTPase activity, it appears that some deviation from the consensus structure can be tolerated. Indeed, recent studies have found that Rad exhibits detectable GTP hydrolysis that is stimulated by a cellular GAP activity (28). Studies to characterize the biochemical properties of recombinant Rem are ongoing.
Rem contains an unusual G3 (DXXG) region, which participates in the binding of both Mg 2ϩ and GTP in Ras GTPases. The sequence, DTWE (residues 133-136), contains the invariant aspartic acid residue but lacks the highly conserved glycine thought to play a critical role in the conformational change Ras-related GTPases undergo in response to differences in the bound guanosine nucleotide (GTP or GDP) (3,4). These structural changes suggest that Rem may use a different mechanism to bind and hydrolyze the ␥-phosphate of GTP. Nevertheless, we are able to demonstrate that Rem binds GTP in a specific and saturable manner. Structural studies will be needed to determine whether these amino acid substitutions result in alterations to the active site.
The sequence of the G2 domain in Rem is unique and diverges significantly from the same domain in Rad, Gem, Kir, and other Ras-related proteins. This region is termed the effector domain and mutational analysis has shown it to be responsible for GAP binding by Ras-related proteins (9). This difference is significant because the G2 domains of GTPases play a central role in defining effector protein interactions (4) and are often perfectly conserved within a specific subfamily but not among different subfamilies. Thus, the Rem, Rad, Gem, and Kir proteins constitute the first Ras subfamily to lack a FIG. 3. Tissue distribution of mRNA for mouse Rem. Upper panel, 20 g of total RNA extracted from a variety of mouse tissues was subjected to ribonuclease protection assay as described under "Experimental Procedures." Radiolabeled, antisense RNA probes corresponding to Rem (231 nucleotides) and ribosomal protein L32 (89 nucleotides, internal control) were hybridized and subjected to RNase treatment, and the RNA-protected probes were polyacrylamide gel electrophoresisresolved and visualized by autoradiography (16 h exposure). Lower panel, quantitation of the protected RNA fragment was determined using a PhosphorImager, and the RNA concentrations per lane were normalized against the expression of the housekeeping gene rpL32 as described under "Experimental Procedures." The data are representative of three separate ribonuclease protection assays.
conserved effector domain and therefore may not be regulated by known GAP-like enzymes.
The COOH terminus of Rem lacks one of the typical CAAX, XXCC, or CXC prenylation motifs found on most Ras-like proteins and required for the attachment of Ras to the plasma membrane (6,7). Although Rem, Rad, Gem, and Kir do not contain classic prenylation motifs, they all contain a highly conserved COOH-terminal motif. This region contains a conserved cysteine residue seven amino acids from the COOH terminus that may serve as a site for lipid modification. The COOH-terminal domain also contains an extended and highly conserved polybasic region. Because polybasic domains have been implicated in providing the binding energy necessary for membrane association of Ras-related proteins (34), it seems likely that this region may also be involved in membrane association. Indeed, the conserved COOH-terminal sequence of Gem has been reported to be required for plasma membrane localization (11). In addition, when the Kir protein is expressed in yeast, an intact COOH terminus is essential to induce pseudohyphal formation (15). However, the mechanism by which the COOH terminus could direct membrane localization is unclear. Whether the process relies on direct membrane of a ribonuclease protection assay using total RNA (20 g/lane) isolated from mouse cardiac muscle and hybridized with either Rem or TNF-␣ and rpL32 antisense riboprobes as described under "Experimental Procedures." Mice were injected with LPS (10 g/animal), and treatment continued for the indicated times. Rem and TNF-␣ message levels were quantitated using a PhosphorImager, normalized against endogenous rpL32 mRNA variations, and plotted as relative RNA levels. The data shown are representative of a typical experiment repeated five times. binding or interaction with an anchoring protein remains to be determined.
The most surprising finding of the current study was the rapid and transient suppression of Rem mRNA levels in a variety of mouse tissues upon LPS treatment. In cardiac muscle, decreases in Rem mRNA are detected within 2 h of LPS injection, reach their lowest levels after 4 h, and return to basal levels by 20 h. Although several previously characterized Raslike genes (35)(36)(37) including the Rad, Gem, and Kir genes are known to be regulated by transcriptional induction, to our knowledge, Rem is the first Ras-related GTP-binding protein to be shown to undergo suppression of mRNA levels in response to stimulation. It is not known whether this regulation is at the transcriptional level or due to differences in mRNA stability. We are presently searching for a cell line in which to examine these issues.
LPS treatment is known to act primarily by stimulating monocytes, neutrophils, and endothelial cells (38). This results in the rapid production of a variety of cytokines, primarily the production of TNF-␣ and interleukin-1 by activated macrophages (29,30). However, additional cytokines as well as a variety of diverse and potent cellular regulators including arachidonic acid derivatives and nitric oxide are also triggered by LPS treatment (38). The modulation of Rem mRNA by LPS and the previous finding that Gem expression was regulated in immune cell populations by a variety of mitogenic stimuli led to our analysis of Rem expression in immune cells. However, ribonuclease protection studies failed to detect Rem mRNA in immune cell populations. The wide tissue distribution and relative expression levels of Rem does however correlate with tissues that contain a high proportion of vascular cells. Because LPS is known to regulate vascular endothelial cells, we are currently examining whether Rem is expressed in this cell type.
The biochemical function of Rem proteins remains to be defined and will be the focus of future work. The unique structure of Rem, its enrichment in tissues with a large number of vascular endothelial cells, its ability to specifically bind GTP, and its regulation by LPS suggest that it may control cellular pathways in endothelial cells. Although a large number of GTP-binding proteins have been identified, they have diverse cellular functions, and the presence of this biochemical activity in itself does not define a cellular role for Rem. However, the extensive information gained from the biochemical characterization of Ras and other GTP-binding proteins may facilitate studies of Rem. Indeed, recent studies with Kir expression in yeast suggest that members of the RGK Ras subfamily act as molecular switches controlling specific aspects of cell physiology through a mitogen-activated protein kinase cascade. Additional studies using dominant-negative and constitutively active forms of Rem may help to answer whether it regulates such a putative pathway.