Portal de Periódicos da CAPES

ARP Is a Plasma Membrane-associated Ras-related GTPase with Remote Similarity to the Family of ADP-ribosylation Factors

1995; Elsevier BV; Volume: 270; Issue: 51 Linguagem: Inglês

10.1074/jbc.270.51.30657

1083-351X

Annette Schürmann, Silke Maßmann, Hans‐Georg Joost,

Ion channel regulation and function

The human and rat homologues of a novel Ras-related GTPase with unique structural features were cloned by polymerase chain reaction amplification and cDNA library screening. Their deduced amino acid sequences are highly homologous (97% identical amino acids; 88.3% identical nucleotides within the coding region) and comprise all six of the conserved motifs presumably involved in GTP binding. Because the sequences exhibit some similarity with members of the ADP-ribosylation factor (ARF) family (33% identity with ADP-ribosylation factor 1 (ARF1), 39% identity with ARF-like 3), the protein was designated ARP (ARF-related protein). In contrast to all other members of the ARF family, ARP lacks the myristoylation site at position 2 and comprises an insertion of 8 amino acids in the region between PM1 and PM2. mRNA was found in most rat tissues examined (skeletal muscle, fat, liver, kidney, spleen, testis, adrenals, ovary, thymus, intestine, and lung). Western blot analysis with antiserum against recombinant ARP showed a 25-kDa protein in membranes from rat liver, testis, and kidney. Thus, the protein appears to be posttranslationally modified for membrane anchoring. Unlike ARF, the ARP immunoreactivity was detected in plasma membranes but not in cytosol of fractionated 3T3-L1 cells. Recombinant ARP exhibited specific and saturable GTPγS (guanosine 5′-3-O-(thio)triphosphate) binding and, unlike ARF isotypes, GTPase activity in the absence of tissue extracts or phospholipids. Thus, the structural and functional characteristics of ARP indicate that it represents a novel subtype of GTPases, presumably exerting a unique function and possibly involved in plasma membrane-related signaling events. The human and rat homologues of a novel Ras-related GTPase with unique structural features were cloned by polymerase chain reaction amplification and cDNA library screening. Their deduced amino acid sequences are highly homologous (97% identical amino acids; 88.3% identical nucleotides within the coding region) and comprise all six of the conserved motifs presumably involved in GTP binding. Because the sequences exhibit some similarity with members of the ADP-ribosylation factor (ARF) family (33% identity with ADP-ribosylation factor 1 (ARF1), 39% identity with ARF-like 3), the protein was designated ARP (ARF-related protein). In contrast to all other members of the ARF family, ARP lacks the myristoylation site at position 2 and comprises an insertion of 8 amino acids in the region between PM1 and PM2. mRNA was found in most rat tissues examined (skeletal muscle, fat, liver, kidney, spleen, testis, adrenals, ovary, thymus, intestine, and lung). Western blot analysis with antiserum against recombinant ARP showed a 25-kDa protein in membranes from rat liver, testis, and kidney. Thus, the protein appears to be posttranslationally modified for membrane anchoring. Unlike ARF, the ARP immunoreactivity was detected in plasma membranes but not in cytosol of fractionated 3T3-L1 cells. Recombinant ARP exhibited specific and saturable GTPγS (guanosine 5′-3-O-(thio)triphosphate) binding and, unlike ARF isotypes, GTPase activity in the absence of tissue extracts or phospholipids. Thus, the structural and functional characteristics of ARP indicate that it represents a novel subtype of GTPases, presumably exerting a unique function and possibly involved in plasma membrane-related signaling events. INTRODUCTIONADP-ribosylation factors (ARF) 1The abbreviations used are: ARFADP-ribosylation factorGSTglutathione S-transferaseRACErapid amplification of cDNA endsbpbase pair(s)kbkilobase(s)HPLChigh pressure liquid chromatographyTLCthin layer chromatographyERendoplasmic reticulumDTTdithiothreitolPAGEpolyacrylamide gel electrophoresisPMplasma membraneGTPγSguanosine 5′-3-O-(thio)triphosphate. represent a subfamily of Ras-homologous GTPases (Kahn and Gilman, 1984; Bobak et al., 1989; Kahn et al., 1991) presumably involved in basic cellular functions, e.g. regulation of phospholipase D (Brown et al., 1993; Kahn et al., 1993; Cockcroft et al., 1994), exocrine secretion (Zeuzem et al., 1992b), vesicle traffic from ER to Golgi (Donaldson et al., 1991; Serafini et al., 1991; Kahn et al., 1992a), and endocytosis (D'Souza-Schorey et al., 1995). To date, six mammalian isotypes of ARF (ARF1-ARF6), three isotypes from yeast, and three from Drosophila have been identified (Lee et al., 1994a, 1994b; Kahn et al., 1995). All isoforms exhibit a high degree of overall structural similarity (65-96% identical amino acids) and share motifs determining common functional characteristics, e.g. the N-terminal myristoylation site. In addition, several genes encoding proteins with homology to the ARF isoforms have been cloned and designated ARF-like genes (Tamkun et al., 1991; Clark et al., 1993; Schürmann et al., 1994; Cavenagh et al., 1994). The products of these genes appear to lack ADP-ribosylation enhancing activity, and show different characteristics of GTP-binding (dependence on magnesium and phospholipids) than the ARF isoforms (Cavenagh et al., 1994). Although some of these proteins e.g. ARL4 (Schürmann et al., 1994) exhibit tissue and/or differentiation-specific expression, little is so far known about their cellular function. In the present paper, we describe the identification of a new GTPase which exhibits a remote similarity with other members of the extended ARF family. Since it differed in important characteristics from other ARF and ARF-like proteins, e.g. in the lack of a myristoylation motif, in a constitutive GTPase activity, and in its subcellular distribution, the protein was designated ARF-related protein (ARP).MATERIALS AND METHODSPreparation of RNA and Synthesis of cDNAMurine 3T3-L1 fibroblasts (Green and Kehinde, 1974) were obtained from American Type Culture Collection (Rockville, MD) and were differentiated as described (Weiland et al., 1990). Cells were frozen in liquid nitrogen and lysed with 4 M guanidine thiocyanate supplemented with 7% mercaptoethanol. Rat tissues (brain, heart, soleus muscle, adipose cells, liver, kidney, lung, spleen, intestine, testis, and thymus) were homogenized with a Polytron homogenizer in 4 M guanidine thiocyanate. The lysates were layered on a cesium chloride cushion (5.88 M) and centrifuged at 33,000 rpm (rotor SW40) for 22 h at 20°C. Pelleted RNA was dissolved with 300 μl of sodium acetate/Tris buffer and was neutralized by addition of 50 μl 2 M potassium acetate (pH 5.5). First-strand cDNA was synthesized with murine reverse transcriptase (First-strand cDNA synthesis kit, Pharmacia, Freiburg, Germany (FRG)) by oligo(dT) priming.Library Screening and DNA SequencingA cDNA fragment (150 bp) isolated by PCR-based cloning approach with primers matching PM1 and PM3 of ARF proteins (Schürmann et al., 1994) was used as a probe to screen λZap rat heart, rat brain, and human liver cDNA libraries (Stratagene, La Jolla, CA). Positive plaques were isolated, and inserts were subcloned into pBluescript (Stratagene). Deletions were generated by exonuclease digestion in both directions, and inserts were sequenced by the method of Sanger (T7 sequencing kit, Pharmacia).Northern Blot AnalysisSamples of total RNA (15 μg) were separated by electrophoresis on 1% agarose gels containing formaldehyde and transferred on to nylon membranes (Hybond, Amersham-Buchler, Braunschweig, FRG). Before transfer, gels were stained with ethidium bromide in order to ascertain that equal amounts of total RNA had been separated. Probes were generated with the Klenow fragment of DNA polymerase I and [32P]CTP from a cDNA clone (smg3/1; see below) by random oligonucleotide priming (Feinberg and Vogelstein, 1983). The nylon membranes were hybridized at 42°C, and blots were washed twice with 0.12 M NaCl, 0.012 M sodium citrate, 0.1% SDS and once with 0.015 M NaCl, 0.0015 M sodium citrate, 0.1% SDS.Expression of GST Fusion Proteins in Escherichia coliA construct comprising the cDNA of GST and rat ARP within a single open reading frame was prepared by subcloning of an NlaIV fragment of the complete ARP clone into the SmaI site of the vector pGEX. Similarly, an insert isolated from the incomplete clone by digestion with EcoRI and XhoI was subcloned into the SmaI site of pGEX after generating blunt ends with Klenow's fragment of DNA polymerase. Transformants of DH5α were produced, grown to an OD600 of 0.3-0.5 at 37°C and induced with isopropyl-1-thio-β-D-galactopyranoside for 18 h at 30°C. Cells were isolated by centrifugation, frozen, thawed, and lysed by sonication for 15 s. After treatment with Triton X-100 at a final concentration of 1%, the lysate was spun at 10,000 rpm for 10 min, and the supernatant was collected. GST-fusion proteins were adsorbed on glutathione-Sepharose by a batch procedure, and eluted with 10 mM glutathione in 10 mM Tris buffer (pH 7.4) or cleaved with thrombin (1 unit/100 μg of GST-ARP) by an incubation at 4°C for 2 h. Samples of the fusion proteins were separated by HPLC on a Sephasil C18 reversed phase column (4 × 250 mm, Pharmacia) in order to identify the bound nucleotides as described previously (Tucker et al., 1986).Assay of GTPγS BindingSamples of recombinant ARP (5 μg of protein) in a total volume of 100 μl Tris buffer containing EDTA (1 mM), DTT (1 mM), magnesium chloride (10 mM), NaCl (100 mM), and Triton X-100 (0.1%). Tracer GTPγS (approximately 400,000 cpm/sample) was added, and the samples were incubated for the indicated times at 30°C. Nonspecific binding was assayed with samples containing 0.1 mM unlabeled GTPγS. The incubation was terminated by addition 1 ml of ice-cold Tris buffer (pH 8.0) containing 100 mM NaCl and 25 mM MgCl2, and the samples were filtered through nitrocellulose membranes (Sartorius, Göttingen, FRG; pore size 0.2 μm). The filters were washed four times with 1 ml of ice-cold washing buffer, dried, and counted in a water-compatible scintillation mixture (Ready Protein Plus, Beckman, Palo Alto, CA).Assay of GTPase ActivityGTPase activity was assayed by a previously described procedure (Tan et al., 1991). A sample of recombinant ARP (25 μg of protein) in 50 mM Tris, pH 8.0, 2 mM EDTA, 1 mM DTT, 10 mM MgCl2, 0.5 μM GTP, and 500 μg/ml bovine serum albumin was incubated with 25 nM [γ-32P]GTP (6000 Ci/mmol, DuPont NEN) in a total volume of 50 μl at 30°C for 60 min. Buffer A (88 μl; 50 mM Tris, pH 8.0, 5 mM MgCl2, 1 mM DTT), 10 μl of bovine serum albumin (10 mg/ml), 2 μl of 0.1 M ATP, 55 μl of H2O and 45 μl of [γ-32P]GTP-loaded ARP were mixed on ice and immediately transferred to 30°C. At 0, 5, 15, 30, and 60 min, triplicate aliquots (10 μl) were passed through nitrocellulose filters and washed with ice-cold buffer B (20 mM Tris, pH 8.0, 5 mM MgCl2, 10 mM NH4Cl, 0.1 M KCl, 1 mM 2-mercaptoethanol). The filters were dried, and the bound radioactivity was determined by scintillation counting. For direct identification of the bound nucleotides, the recombinant proteins were loaded with [α-32P]GTP (3000 Ci/mmol) and separated on nitrocellulose as described above. Filters were extracted with 2 M formic acid, and the eluates were separated by thin layer chromatography on polyethyleneimine cellulose (1 M lithium chloride, 1 M formic acid). The separated nucleotides were visualized by autoradiography.Antisera Production and ImmunoblottingA polyclonal antiserum against ARP was raised with the fusion protein GST-ARPΔ21. A polyclonal antiserum against the muscle- and fat-specific glucose transporter GLUT4 was raised with a peptide corresponding to its C-terminal sequence (STELEYLGPDEND). Polyclonal antiserum against Ha-Ras was a gift from Dr. K. Aktories, Universität des Saarlandes, Homburg/Saar, FRG. Rat tissues (brain, testis, liver, and kidney) were homogenized and centrifuged at 30,000 × g for 45 min. Membranes from rat heart and adipocytes were prepared as described previously (Hellwig and Joost, 1991). 3T3-L1 cells were homogenized and fractionated by differential centrifugation as described previously (Weiland et al., 1990; Ziehm et al., 1993). Samples of 15 μg of protein were separated by SDS-PAGE, transferred onto nitrocellulose, and incubated with the antisera in a dilution of 1:100 to 1:500. Immunoreactivity was detected with 125I-protein A as described previously (Schürmann et al., 1992). For blocking of the ARF-specific antiserum, 100 μl of serum was incubated with 100 μg of recombinant ARP at 4°C overnight.Expression of ARP in COS-7 CellsAn NlaIV fragment of the complete ARP clone was subcloned into the SmaI site of the mammalian expression vector pCMV. Transfection of COS-7 cells with this vector was performed by incubation with a calcium phosphate-DNA precipitate according to a previously described protocol (Wandel et al., 1994). Transfected cells were washed, homogenized, and a fraction of membranes was isolated by centrifugation at 30,000 × g for 45 min.RESULTSCloning of Rat and Human ARPPCR amplification of cDNA from 3T3-L1 cells or rat adipocytes with degenerate oligonucleotides matching the PM1 and PM3 motifs of ARF proteins yielded cDNA fragments of all known ARF isoforms, three ARF-like genes (Schürmann et al., 1994; Cavenagh et al., 1994), and an additional fragment of an unknown GTPase with remote similarity to the ARF family, later designated ARP (see below). The fragment was used for screening of a λZap cDNA library from rat heart. A single clone (approximately 0.7 kb) was isolated from the rat heart λZap library; it comprised a poly(A) tail and an open reading frame but lacked a translation start (the start of its sequence was nucleotide 194 in Fig. 1). This clone was later used to prepare a truncated fusion protein (GST-ARPΔ21). Additional 193 bp of the 5′-end of the cDNA as depicted in Fig. 1 were obtained by the RACE procedure. Later on, a second clone was isolated from a λZap rat brain library. This clone (3.5 kb, the start of its sequence is at nucleotide 67 in Fig. 1) contained the full reading frame including the translation start as well as a poly(A) tail at its 3′ end. Surprisingly, its sequence differed from that determined by the RACE procedure by an insertion of 6 bp (underlined, nucleotides 119-124). This difference appears to reflect an alternative splicing at the 3′ end of an exon, since there are two subsequent consensus motifs (CAG) for exon splicing. Based on this assumption, the boundary between exons 1 and 2 of the gene can be localized to nucleotide 125.A cDNA fragment of rat ARP was used for screening of a human liver cDNA library, and two clones with identical size (1.6 kb) were isolated. Fig. 2 depicts the nucleotide sequence, which harbors an open reading frame of 603 codons. Its deduced amino acid sequence is highly homologous to that of rat ARP (Fig. 3A, 195 (97%) identical and 6 differing amino acids, 5 of them conservative substitutions), indicating that the cDNA indeed represented the human homologue. Within the coding region, the nucleotide sequences of rat and human ARP were 88% identical. In the untranslated 3′-region, the overall homology was 71%, reaching even 91% over a stretch of 45 bp (underlined sequence of nucleotides 738-783 in the human cDNA).Fig. 2Nucleotide and deduced amino acid sequence of human ARP as determined with a full-length cDNA clone isolated from a human liver cDNA library. Amino acids are given in one-letter code above the respective codons. A stretch of 45 nucleotides in the untranslated 3′-region that is highly homologous (91% identical nucleotides) to the rat sequence is underlined.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Fig. 3Sequence alignments of ARP and other Ras-related GTPases. Panel A, alignment of the amino acid sequences of human and rat ARP. The alignment of the deduced amino acid sequences was performed with the PALIGN program (open gap cost 7, unit gap cost 1). Identical amino acid residues were highlighted by vertical lines (|). Panel B, alignment of rat ARP with rat ARL3: The alignment was performed as in panel A. Amino acids identical with residues that are conserved in all other members of the ARF family are outlined by asterisks. The conserved motifs of GTP-binding (PM1-PM3, G1-G3) are depicted above the sequences. Panel C, dendrogram of an alignment of rat ARP with other members of the Ras superfamily. The alignment was performed with the CLUSTAL program (gap penalty 5, open gap cost 10, unit gap cost 10). Accession numbers of the compared isoforms are as follows: rat Ha-Ras, P20171; canine Rho1, P24406; rat Rab1, P05711; rat Rab2, P05712; rat Rab4A, X06890, human Ran, P17080; human ARF1, P10947; human ARF6, P26438; rat ARL1, X76920; rat ARL3, X76921; rat ARL4, X77235; mouse SarA, P36536; yeast CIN4, L36669.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Structure of ARPThe deduced amino acid sequence of 201 residues contains all conserved motifs presumably involved in GTP binding (Fig. 1, boxed). A data base search revealed that the closest relatives of the protein are members of the ARF family (ARF-like 3 with 39%, and ARF1 with 33% identity). An alignment of ARF with ARL3 (Fig. 3B) indicated that the sequence homology is mainly restricted to the GTP-binding motifs. However, in particular the PM1, PM3, and G2 motifs are characteristic for the ARF family (Fig. 3B, residues conserved within the ARF family are highlighted by asterisks). Although the G3 motif is usually less well conserved, the motif in ARP (CSA) matches those previously found in the ARF family (TCA, SCA, TSA, CSA). In addition, the sequence of ARP contains a tryptophan residue (Trp-86) 6 amino acids after PM3. The presence of this residue appears specific for the ARF family, since none of the other Ras homologues contains a tryptophan in or near that position. On the basis of this structural comparison, we designated the novel G-protein ARP, for ARF-related protein, because it appeared to represent a distant relative of the ARF family.It should be noted that the sequence of ARP exhibited several differences to other members of the ARF family. Most strikingly, it lacks the conserved myristoylation site, a glycine in position 2. Furthermore, the loop region between PM1 and PM2 is longer than that in other ARFs because of an insertion of 8 additional amino acids. In addition, the C terminus is longer than that of other ARFs. A multiple alignment of the amino acid sequence of ARP with prototypes of the other Ras-homologous GTPases confirmed that ARP is a distant relative of the ARF family. As is illustrated by the dendogram of the alignment (Fig. 3C), ARP is located on a branch of the ARF family, its similarity to the other ARFs being higher than that of the GTP-binding proteins SarA and CIN4, but considerably lower than all previously identified ARF-like proteins.Expression of ARP in E. coli and COS-7 CellsConstructs comprising the cDNA of glutathione S-transferase and either one of the two rat ARP cDNA clones in the expression vector pGEX were prepared. Recombinant fusion proteins (GST-ARP and GST-ARPΔ21) were isolated from E. coli DH5α and were partially purified with glutathione-Sepharose. On SDS-PAGE (Fig. 4), the purified fusion protein GST-ARP appeared as a band migrating at an apparent molecular mass of approximately 50 kDa, which corresponds with the sum of the masses of GST (26 kDa, Fig. 4, lane 2) and ARP (calculated molecular mass 23 kDa plus 1.5 for an N-terminal extension introduced by the cloning strategy). The preparations of the fusion proteins always contained GST (Fig. 4, lanes 3 and 5), which was probably generated by proteolysis at the fusion site during isolation of the proteins. An apparently homogeneous preparation of ARP (lane 4) was obtained by cleavage of the adsorbed fusion protein with thrombin at the fusion site. Cleaved recombinant ARP was devoid of GST-activity (data not shown) and migrated as a single band (Fig. 4A, lane 4) at approximately 27 kDa. For unknown reasons, the fusion protein generated with the truncated clone (GST-ARPΔ21) could not be cleaved with thrombin. This fusion protein was used for generation of a polyclonal antiserum against ARP.Fig. 4Expression of rat ARP in E. coli (A) or COS-7-cells (B). A, full-length or truncated ARP-cDNA were subcloned into the pGEX expression vector and transformed into E. coli DH5α, and fusion proteins were generated as described. The proteins were separated on SDS-PAGE and stained with Coomassie Blue. GST, preparation of GST obtained with bland vector; GST-ARP, fusion protein eluted from the affinity column with glutathione; ARP, recombinant ARP, which was isolated from the affinity column by thrombin cleavage; GST-ARPΔ21, fusion protein generated with the incomplete cDNA clone and eluted with GST. B, the full-length ARP-cDNA was subcloned into the mammalian expression vector pCMV and COS-7 cells were transfected as described. Cells were homogenized, and a particulate fraction of total membranes was isolated. Proteins were separated by SDS-PAGE and transferred on to nitrocellulose membranes. Immunochemical detection was performed with antiserum raised against GST-ARPΔ21.View Large Image Figure ViewerDownload Hi-res image Download (PPT)The cDNA of ARP was subcloned into the mammalian expression vector pCMV, and COS-7 cells were transiently transfected with the construct. Cells were homogenized, and a pellet of total membranes was assayed for ARP immunoreactivity. Indeed, as is illustrated in Fig. 4(panel B), a 25-kDa protein was detected with serum against GST-ARPΔ21 in cells transfected with the construct; it was not present in cells transfected with blank vector. The protein was not detected with serum that was blocked with recombinant ARP generated by thrombin cleavage of GST-ARP (not shown).GTP Binding and GTPase Activity of Recombinant ARPPreliminary experiments indicated that the recombinant ARP binds [35S]GTPγS in a saturable and specific manner. As is illustrated in Fig. 5(panel A), this binding is dependent on the ambient magnesium concentration; maximum binding was observed at 1 mM magnesium chloride. The time course of binding (illustrated in panel B of Fig. 5) reached its maximum after 60 min and appeared slower than that of other GTP-binding proteins assayed in our lab (Rab4). 2A. Schürmann, S. Maßmann, and H.-G. Joost, unpublished data. The fusion protein GST-ARP (data not shown) bound similar amounts of GTP, whereas GST (data not shown) failed to bind any tracer in a specific, GTP-inhibitable manner. Furthermore, addition of liposomes failed to alter GTPγS binding of ARP (data not shown).Fig. 5Specific binding of GTPγS, nucleotide exchange, and GTPase activity of recombinant rat ARP. Panel A, magnesium dependence of GTPγS binding. Samples of 5 μg of recombinant ARP were incubated with tracer GTPγS and the indicated magnesium concentrations, and bound tracer was separated by filtration on nitrocellulose membranes after 60 min. Panel B, time course of GTPγS binding. Samples of 5 μg of recombinant ARP (open circles) or 10 μg of GST-ARPΔ21 (filled circles) were incubated with tracer GTPγS and 10 mM magnesium, and bound tracer was separated by filtration on nitrocellulose membranes after the indicated times. Nonspecific binding as determined with samples containing 100 μM unlabeled GTPγS was less than 300 cpm/sample. Binding to GST was determined in separate series and was not distinguishable from nonspecific binding. Panel C, nucleotide exchange and GTPase activity of ARP. Samples of 1.25 μg of recombinant ARP were loaded with [γ-32P]GTP (circles) or [35S]GTPγS (triangles) for 60 min, and the decrease in bound tracer in the presence (filled symbols) or absence (open symbols) of 1 mM GTP or GTPγS was assayed at the indicated time points. All assays were performed in the presence of 10 mM magnesium chloride without added phospholipids. Panel D, analysis of the nucleotides bound to recombinant ARP. ARP was loaded with [α-32P]GTP, and bound nucleotides were isolated after the indicated times and separated by TLC.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Unlike the recombinant ARP, the truncated fusion protein (GST-ARPΔ21) bound only minute amounts of tracer GTP during a 180 min incubation period. However, full GTP binding was observed after incubation with tracer for 24 h. Furthermore, HPLC analysis of the native recombinant proteins indicated that both ARP and GST-ARPΔ21 were loaded with GTP/GDP in a ratio of 1:3 after isolation from the E. coli.Panel C of Fig. 5 illustrates experiments designed to assess the nucleotide exchange and the GTPase activity of recombinant ARP. ARP was loaded for 60 min with [35S]GTPγS (triangles) or [γ-32P]GTP (circles), and the nucleotide exchange or hydrolysis, respectively, was assayed as the decreases in bound radioactivity. Fig. 5(panel C) illustrates the time course of these experiments. In the absence of added unlabeled nucleotide, the GTPγS binding was stable for 60 min, whereas a slow exchange was visible when 1 mM unlabeled GTPγS (filled triangles) was added. A marked release of radioactivity from the recombinant ARP was observed (kinetic constant 0.093 min−1), when the protein was loaded with [γ-32P]GTP (open circles), probably reflecting the release of phosphate from the GTP. Since it was previously reported that preparations of recombinant proteins may contain bacterial GTPases (Welsh et al., 1994), we ran parallel samples containing excess GTP (filled circles) in order to inhibit these potential contaminants. However, the addition failed to inhibit the decrease in tracer binding to the recombinant ARP. Moreover, the decrease in bound GTP and the increase in bound GDP could be demonstrated directly by TLC analysis of the bound nucleotides (Fig. 5, panel D). Thus, it is concluded that the release of radioactivity from the bound [γ-32P]GTP indeed reflects the intrinsic GTPase activity of ARP.Tissue Distribution of ARP in RatNorthern blot analysis (Fig. 6) of total RNA from rat tissues showed two transcripts of an approximate size of 1.3 and 3.1 kb. The inaccuracy of the size determination considered, these transcripts match reasonably well with the size of the two clones isolated from the rat cDNA libraries (0.94 and 3.5 kb). Highest mRNA levels were found in testis, thymus and adrenals, low levels were detectable in most other tissues investigated including skeletal muscle, fat, liver, spleen, ovary, intestine, and lung. Two transcripts (1.5 and 4 kb) were detected in total mRNA from mouse 3T3-L1 cells, with no apparent change during the course of their differentiation (data not shown).Fig. 6Northern blot analysis of ARP mRNA in different rat tissues. Total RNA from the indicated tissues was hybridized with a probe generated from a cDNA fragment comprising the coding region of ARP. B, total brain; H, heart; M, skeletal muscle; A, fat cells; Li, liver; K, kidney; S, spleen; T, testes; Ad, adrenal gland; O, ovary; Th, thymus; I, intestine; Lu, lung. As judged from ethidium bromide staining (data not shown) the amounts of RNA in each lane were essentially identical with the exception of kidney (partial degradation) and brain (lower amounts).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Antisera were raised in rabbits against the fusion protein GST-ARPΔ21 and were tested in preliminary experiments with COS-7 cells transfected with the ARP cDNA. As is illustrated in Fig. 4, Fig. 5, Fig. 6, Fig. 7, the immune serum reacted with a 25-kDa protein in membranes from transfected COS-7 cells, testis, liver, and kidney. This signal corresponded with that of recombinant ARP (right lanes in Fig. 7), and was blocked by preincubation of the serum with recombinant ARP (not shown here; see also Fig. 8). Other nonspecific bands, i.e. the additional band at 10 kDa, were not blocked by recombinant ARP. By longer exposure (not shown), a weak signal was detected in membranes from adipocytes; no immunoreactivity was present in heart. By a rough comparison with the immunoreactivity of recombinant ARP, the amounts of ARP present in the membranes of kidney and testis can be estimated to approximately 5 ng/15 μg of membrane protein. The apparent molecular mass of the specific immunoreactivity correlates reasonably well with the calculated value (23 kDa), indicating that ARP is considerably bigger than all previously identified members of the ARF family. As anticipated, the ARP immunoreactivity in membranes from rat tissues migrated at a smaller apparent molecular weight than recombinant ARP, which, due to the cloning strategy, has an extended N terminus. A second antiserum (not shown) detected more nonspecific bands but reacted with the same rARP-inhibitable immunoreactivity at 25 kDa.Fig. 7Immunochemical detection of ARP in membranes from rat tissues. Samples of membrane proteins (15 μg/lane) from adipocytes, heart, testis, kidney, liver, and brain were separated by SDS-PAGE (12% gels), transferred on to nitrocellulose membranes, and probed with specific antiserum. Note that the apparent molecular mass of recombinant ARP is approximately 2 kDa higher because of an extension of the N terminus due to the cloning strategy.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Fig. 8Immunochemical detection of ARP in membrane fractions from 3T3-L1 cells. Differentiated 3T3-L1 cells were fractionated as described. Membrane fractions were separated by SDS-PAGE and transferred on to nitrocellulose membranes. Parallel blots of the same membrane preparation were probed with ARP antiserum (upper left), blocked antiserum (upper right), and antisera against Ha-Ras and the glucose transporter GLUT4 (lower panel) as markers of plasma membranes or intracellular microsomes, respectively. The reproduction of the blot probed with Ha-Ras antiserum was cut for rearrangement of the lanes in the same order as the other panels.View Large Image Figure ViewerDownload Hi-res image Download (PPT)In order to assess the subcellular distribution of ARP, 3T3-L1 cells were homogenized and fractionated by differential centrifugation. With this previously established procedure (Weiland et al., 1990; Ziehm et al., 1993), fractions of plasma membranes (PM), ER-enriched high density microsomes, Golgi-enriched low density microsomes, and cytosol were isolated. ARP immunoreactivity (Fig. 8) was detected mainly in the plasma membrane fraction of 3T3-L1 cells; only traces were present in high-density microsomes and no immunoreactivity was detected in the cytosol. The signals were blocked by preincubation of the antiserum with recombinant ARP (Fig. 8, upper right). These data indicate that ARP is exclusively associated with membranes and does not occur in a soluble, cytoplasmic form. In low density microsomes, a 25-kDa band was visible, but this immunoreactivity appeared not inhibited by blocking of the serum with recombinant ARP (upper right). However, since the antiserum reacted with several nonspecific bands in the low density microsomes, we cannot fully exclude that ARP is also present in these microsomes. As is illustrated in the lower panel of Fig. 8, the purity of the membrane fractions was assessed with Western blots of Ha-Ras (marker for plasma membranes) and glucose transporter GLUT4 (marker for low density microsomes). The pattern of the subcellular distribution of these markers and of ARP indicates that the ARP immunoreactivity in the plasma membrane fraction reflects a specific association rather than a cross-contamination of membrane fractions.DISCUSSIONOn the basis of structural and functional criteria (Kahn et al., 1992b), the novel GTP-binding protein ARP is a Ras-related GTPase with remote similarity to the family of ADP-ribosylation factors. It contains all six motifs (PM1-PM3, G1-G3) involved in nucleotide binding (Valencia et al., 1991), and a recombinant protein isolated from E. coli indeed binds and hydrolyses GTP. Its closest relatives are members of the ARF family (ARL3, 39% identity; ARF1, 33%), whereas the homology to other Ras-related G-proteins is considerably lower (Ras, 17.5% identical amino acids; Rab, 21.9%; Ran, 20.9%; Rho, 24.4%). Furthermore, ARP harbors motifs and residues that are typical for the ARF family: The PM1 motif GLDNAGKTT, the PM3 motif WDXGGQ, the conserved tryptophan in position 86 between the PM3 and G2 motifs, and the G2 motif ANKQD (instead of GNKQD as in all other Ras-related proteins). However, the sequence of ARP differs from that of other members of the ARF family by a lack of the myristoylation site (glycine 2), a striking insertion of 8 amino acids between PM1 and PM2, and an extended C terminus that is highly charged but lacks the lysine residues present in all other ARF isoforms. Also, as is discussed below, the new GTPase exhibited striking functional differences (GTPase activity, subcellular distribution) to the other members of the ARF family. In order to emphasize these differences, we decided to designate the protein ARF-related protein (ARP).A striking feature of ARP is the lack of a known lipid modification signal in its sequence. In analogy to other Ras-related GTPases, and on the basis of the Western blot analysis, which demonstrated exclusive association of ARP to membranes, it is reasonable to assume that the protein is co-translationally or post-translationally modified for membrane anchoring. Other GTPases have C-terminal motifs for farnesylation (e.g. Ras) or fatty acid acylation (e.g. Rab), or are myristoylated at glycine 2 (all so far known relatives of ARF). None of these motifs are apparent in ARP. Thus, the nature of membrane attachment remains to be determined. At present, we consider the possibility that cysteine 186 is a site of palmitoylation, in analogy to G-protein α-subunits or Ras (Casey, 1995). In spite of the lack of further data, it appears reasonable to assume that targeting and membrane anchoring of ARP is fundamentally different from that of ARP isotypes.A second striking difference between ARP and other members of the ARF family is its high GTPase activity. ARF and ARF-like proteins do not hydrolyze GTP in the absence of tissue extracts containing GTPase activating factors and phospholipids (Randazzo and Kahn, 1994; Makler et al., 1995). In contrast, a marked decrease in bound tracer from ARP preloaded with [γ-32P]GTP was detected in the absence of any added tissue extracts or phospholipids. This activity does not appear to be due to bacterial contamination of the recombinant protein, since it was not inhibitable by excess unlabeled GTP. Furthermore, other recombinant GTPases, e.g. ARL4, which we isolated by the same procedure were essentially devoid of GTPase activity. 3A. Schürmann and H.-G. Joost, unpublished data. It should also be noted that the binding of GTPγS to ARP proceeded relatively slowly, equilibrium being reached after 30-60 min. Thus, ARP appears to require the presence of an exchange factor in addition to a GTPase-activating protein in order to function as a fast GTP-dependent switch. However, it has to be taken into account that the GTP-binding assays were conducted in an artificial in vitro system, and with protein that certainly lacked post-translational modification. Recently, it has been shown that myristoylation of ARF1 accelerates GDP exchange and markedly decreases dissociation of GTP (Franco et al., 1995).The truncated ARP (GST-ARPΔ21), which we had prepared with the incomplete cDNA clone bound GTP in a specific manner but with a very low association rate. Since the native GST-ARPΔ21 is loaded with unlabeled GTP/GDP when isolated from E. coli, the finding indicates that the truncation had markedly reduced the nucleotide exchange rate and suggests that the N terminus of ARP is essential for nucleotide exchange. Previous experiments with ARF constructs comprising N-terminal modifications have suggested that the N terminus of ARF1 is essential for its GTPase activity, and might therefore interact with ARF GTPase-activating protein (Randazzo et al., 1994). ARPΔ21, however, seems to have significant GTPase activity, since the native recombinant protein was isolated in its GDP-bound form. It should be noted that truncated ARP lacks a basic residue (Lys-15), which is conserved in all members of the ARF family and corresponds with Lys-15 in ARF1. Furthermore, almost all other previously described Ras-homologous GTPases harbor a basic residue corresponding with lysine 5 in Ras in their N terminus. Thus, it is tempting to speculate that it is the lack of a basic residue preceding the PM1 motif that causes the marked reduction in the nucleotide exchange rate in GST-ARPΔ21.The subcellular distribution of ARP in 3T3-L1 cells is remarkable in that it was predominantly associated with the plasma membrane fraction of 3T3-L1 cells. This subcellular distribution resembles that of Ha-Ras, which we used as a marker for the plasma membrane fraction. It thereby differs strikingly from that of ARF, which has been found in the cytosol and in Golgi-derived vesicles of other cells (Randazzo et al., 1994; Serafini et al., 1991). Thus, since some structural elements, the targeting, and the membrane anchoring of ARP are fundamentally different from that of ARF isoforms, ARP appears to exert a unique, yet unknown, function. From its subcellular distribution we derive the speculative working hypothesis that it is involved in plasma membrane-related signaling events. INTRODUCTIONADP-ribosylation factors (ARF) 1The abbreviations used are: ARFADP-ribosylation factorGSTglutathione S-transferaseRACErapid amplification of cDNA endsbpbase pair(s)kbkilobase(s)HPLChigh pressure liquid chromatographyTLCthin layer chromatographyERendoplasmic reticulumDTTdithiothreitolPAGEpolyacrylamide gel electrophoresisPMplasma membraneGTPγSguanosine 5′-3-O-(thio)triphosphate. represent a subfamily of Ras-homologous GTPases (Kahn and Gilman, 1984; Bobak et al., 1989; Kahn et al., 1991) presumably involved in basic cellular functions, e.g. regulation of phospholipase D (Brown et al., 1993; Kahn et al., 1993; Cockcroft et al., 1994), exocrine secretion (Zeuzem et al., 1992b), vesicle traffic from ER to Golgi (Donaldson et al., 1991; Serafini et al., 1991; Kahn et al., 1992a), and endocytosis (D'Souza-Schorey et al., 1995). To date, six mammalian isotypes of ARF (ARF1-ARF6), three isotypes from yeast, and three from Drosophila have been identified (Lee et al., 1994a, 1994b; Kahn et al., 1995). All isoforms exhibit a high degree of overall structural similarity (65-96% identical amino acids) and share motifs determining common functional characteristics, e.g. the N-terminal myristoylation site. In addition, several genes encoding proteins with homology to the ARF isoforms have been cloned and designated ARF-like genes (Tamkun et al., 1991; Clark et al., 1993; Schürmann et al., 1994; Cavenagh et al., 1994). The products of these genes appear to lack ADP-ribosylation enhancing activity, and show different characteristics of GTP-binding (dependence on magnesium and phospholipids) than the ARF isoforms (Cavenagh et al., 1994). Although some of these proteins e.g. ARL4 (Schürmann et al., 1994) exhibit tissue and/or differentiation-specific expression, little is so far known about their cellular function. In the present paper, we describe the identification of a new GTPase which exhibits a remote similarity with other members of the extended ARF family. Since it differed in important characteristics from other ARF and ARF-like proteins, e.g. in the lack of a myristoylation motif, in a constitutive GTPase activity, and in its subcellular distribution, the protein was designated ARF-related protein (ARP).