Structural Analysis of the Ras Gtpase Activating Protein Catalytic Domain by Semirandom Mutagenesis: Implications for a Mechanism of Interaction with Ras-GTP1

Home , GTPase

[CANCERRESEARCH54. 5438-5444, October 15, 1994] Structural Analysis of the Ras GTPase Activating Protein Catalytic Domain by Semirandom Mutagenesis: Implications for a Mechanism of Interaction with Ras-GTP1

Lisa Hettich and Mark Marshall2

Department of Medicine, Division of Hematology/Oncology. Indiana UniversÃ¼ySchool of Medicine and Waither Oncology Center, Indianapolis. indiana 46202

ABSTRACT region of GAP in the regulation of cytoskeletal structure and cell adhe sion to the extracellular matrix (14). Expression of the GAP NH2- The bovine complementary DNA encoding the catalytic domain of Ras terminal SH2-SH3-SH2 region in Rat-2 fibroblasts resulted in a disrup GTPase activating protein was mutagenized semirandomly using a van tionofactinstressfibersandfocalcontacts.Thiseffectisprobablycaused atlon of the polymerase chain reaction. Sixty-four mutated codons were identified with seventeen of the mutations deleterious to Ras GTPase by the action of the GAP-associated protein, p190, and Rho (15-17). activating function. All of the inactivating single mutations affected the Changes in the cytoskeleton and loss of adhesion are important compo structure of the catalytic fragment as assessed by large decreases in nents of cell growth. soluble protein when expressed in Escherichia coli. Upon exnmlnation of In addition to its potential as a Ras effector molecule, GAP is the Ran binding properties of 10 of the mutants, only 1 was measurably currently the best model for examining GTP-dependent protein inter Impaired for I(as binding and 4 appeared to have increased affinity for actions with Ras. Ras-GTP has been shown to bind to GAP, while Ran. These results demonstrate that Ras binding and GTPase activation Ras-GDP does not (18). In addition, many mutations within the are two distinct properties of GTPase activating protein. Additionally, the so-called Ras effector region, which block cellular transformation, catalytic mechanism of GTPase activating protein Is much more sensitive also reduces the GAP sensitivity of the GTPase of the mutant or the to structural perturbation than is Ras binding. ability of the protein to bind GAP (19, 20). The primary Ras binding site of GAP has been localized to the carboxy-terminal 343 amino INTRODUCTION acids and is referred to as the GAP catalytic domain (21). This region Mutations in the ras proto-oncogenes have been implicated in the can be expressed as a stable truncated protein in Escherichia coli, initiation and proliferation of tumors in both animal carcinogenesis which can bind Ras-GTP and activate the Ras GTPase. The Ras models and human cancer (1). Oncogenic mutations result in a mod binding domain of GAP is conserved in other eukaryotic proteins ified form of the Ras protein, which is refractory to negative regula possessing Ras-GTPase activating function. It is not known if Ras tion, resulting in a constant proliferative signal (2, 3). Ras is a small binds a smallregionof GAPanalogousto the Ras effectorregionor Mr 21,000 membrane-associated protein which shares many biochem a larger area of the surface typical of most protein-protein interactions. ical characteristics with the a subunits of the prototypical heterotri The goal of this study has been to use a structurally unbiased meric 0-proteins (4). Ras binds both GDP and GTP with high affinity, approach in order to identify specific amino acid residues involved in possesses an intrinsic GTPase, and requires membrane association for the binding of GAP to Ras. Identification of discrete sites of protein normal function. Current evidence suggests that the primary mito interaction on the Ras protein suggests that analogous sites may exist genic effector of Ras is the Raf kinase. Ras-GTP binds to the Raf on GAP. The DNA encoding the catalytic fragment of GAP was serine/threonine kinase and is believed to assist in its activation (5â€”8). mutagenized using a semirandom method; mutant proteins were iden GAP3 is known to function as a negative regulator of Ras through the tified and biochemically characterized. Inactivating mutations were dramatic stimulation of the slow intrinsic GTPase of the Ras protein. found in 17 distinct codons, while mutations in 47 other codons had In addition to being a negative Ras regulator, GAP has also been little effect. All of the inactivating single codon mutations were found implicated as a component necessary for some Ras-dependent effects in to significantly reduce the quantity of soluble protein produced in mammalian cells and amphibian oocytes. In vitro evidence for a Ras E. coli, suggesting that gross conformational changes had occurred. dependent effector role for GAP has been demonstrated in at least three Only one of these mutants was found to have reduced binding affinity separate model systems. Ras and GAP cooperate in an SH2/5H3 domain for Ras. These results suggest that GAP interacts with Ras, either dependent manner to inhibit muscarinic gated potassium channel activa through a few specific contacts scattered throughout the primary tion in patch-clamped atrial membranes (9, 10). Activation of Xenopus amino acid sequence or through a large surface which is unaffected by 1@zevisp34 kinase (maturation promoting factor) by Ras has also been point substitutions. The activation of Ras GTPase by GAP is much suggested to be GAP-dependent (11), with Ras induction of germinal easier to disrupt than binding alone, indicating that GTPase stimula vesicle breakdown requiring the GAP SH3 domain (12). GAP has also tion requires a more conformationally defined structure which is not been implicated by expression-competition experiments as having posi required for binding. five transcriptional effects on the Ras-dependent polyoma enhancer (13). The most conclusive evidence for a signaling role for GAP comes from MATERIALS AND METHODS a series of experiments that showed involvement of the NH2-terminal Mutagenesis of the GAP gene Semirandom base misincorporations were Received 6/2/94; accepted 8/17/94. introduced into the bovine GAP complementary DNA by PCR using Taq The costs of publication of this article were defrayed in part by the payment of page polymerase (Perkin-Elmer; Ref. 22). The reaction conditions used were as charges. This article must therefore be hereby marked advertisement in accordance with recommended by the manufacturer except that the concentration of dATP was 18 U.S.C. Section 1734 solely to indicate this fact. I This work was supported in part by the Project Development Program, Research and limiting (40 @LMversus200 pM dGTP, dTFP, and dCTP). Reduction of dATP Sponsored Programs, Indiana University at Indianapolis. resulted in limited random misincorporation of deoxyribonucleotides when 2 To whom requests for reprints should be addressed, at Walther Oncology Center, ever an â€œAâ€•(deoxyadenosine)was required during primer extension of the lndiana University School of Medicine, 975 West Walnut Street, Room 501, Indianapolis, DNA. The addition of 0.1 mMMnCl2to the reaction further increased the rate 1N 46202-5121. of base misincorporation. Use of the oligonucleotide primers 5'-GCCCATA 3 The abbreviations used are: GAP, GTPase activating protein; PCR, polymerase chain reaction. AACFCCCAGTAAAG-3' and 5'-CGCO@GCAGAAUAGCfCACACAT 5438

CACI'G-3' amplified only the portion of the bovine GAP gene, which encodes (Qiagen) for 1 h at 4Â°C.Theagarose was then washed twice in batch with 50 the catalytic domain from codon 701 to 1044. The mixture of mutant GAP ml of buffer A, packed into a small column, and connected to a Pharmacia DNA fragmentswas subclonedfrom the PCR reactioninto the BamHI-PstI FPLC pump and detector. After further washing, bound proteins were eluted sites ofpU@8 in-frame with the lacZ gene to yield an E. coli expression library with a 0â€”0.250 M imidazole gradient. The A@-absothing peak was concen of GAP[701â€”1044]mutants.Approximately 80% of the codons in this GAP trated in a Centricon-lO concentrator (Amicon), washed twice with TED, gene fragment had an â€œAâ€•ora â€œr'inthe first or second position, making them assayed for GAP activity, and immunoblotted for GAP protein. Although susceptible to amino acid substitution. To enable affinity purification of wild-type GAP[701â€”1044Jwastypically purified to 90â€”95%homogeneity,the selected mutant proteins, mutated genes were subcloned as BamHI-PstI frag yield and purity of the mutant proteins were more variable. ments into the pQE-30 expression plasmid (Qiagen). The pQE-30 plasmid Binding of Ran to Mutant GAP PrOteins. The binding of Ras to catalyt provided for high-level expression of proteins fused to the 6x(histidine) affin ically impaired GAP mutants was measured using a kinetic competition assay ity tag. The F898S mutation was separated from a second mutation in the 4-26 (19) which has been previously adapted for comparing the relative Ras binding mutant by subcloning a restriction fragment containing the point change into aff@mityof normal and reduced activity NFl mutants (24â€”26). Basically, a the wild-type pQE-30-GAP[701-1044] plasmid. Standard recombinant DNA small quantity of [â€˜y-32P]GTP-boundRas protein (0.2 nM) was incubated in the manipulations were followed throughout this study (23). Plasmid DNAS were presence of sufficient GAP or GAP mutant to stimulate hydrolysis of completely sequenced using Sequenase (United States Biochemical). [y-32P]GTP bound to Ras (approximately 50â€”200 @g/mlfor most GAPs, Isolation and Preparation of GAP Mutants Single-colony isolates were including the wild-type control; 400â€”1,200 @g/mlformutants 4-33, 4-150, obtained from the mutated pUC8-GAP[701â€”1044]expressionlibraries by 4-119, and 4-16). Increasing concentrations of nonradiolabeled Ras[L61]-GTP transformation of RRllaciâ€•E.coli. Individual colonies were transferred into 5 (0.006â€”200 ELM)were added to the assay mixture described above. All re ml of LBA (Luria broth with 100 mg/ml ampidillin; Ref. 21) and grown agents, except for the purified GAP proteins, were premixed and incubated for overnight at 37Â°C.Each culture was split into 4.5 ml of stock culture for 2 mist at 30Â°C.Thereaction was initiated by the addition of GAP[701â€”1044] plasmid preparation and 0.5 ml for protein expression. GAP protein was made or mutant GAP. After a b-mm incubation (30 mm for the 4-16 and 4-119 by diluting the 0.5 ml sample into 4.5 ml of fresh LBA. After 1 hour of mutants), the reaction was terminated with 0.2 ml of 5% activated charcoal incubation at 37Â°C,protein expression was induced by the addition of 0.5 mr@i (Sigma Chemical Co.) in 50 mMNaH2PO4.Ras GTPase activity was measured isopropylthio-@-D-galactoside, followed by 4 more h of incubation. The cells by quantitating the amount of free [32Pjorthophosphate in the reaction super in each culture were pelleted and stored at â€”80Â°C.Bacterialpellets were natant following centrifugation to remove the charcoal. Competition was thawed and lysed on ice by the addition of 0.25 ml of TED buffer [50 mM expressed as the percentage of activity detected relative to the value Tris-HC1 (pH 73)-i mM EDTA-5 mM dithiothreitol] with 0.2 mg/mI lysozyme, obtained in the absence of competitor. Background counts measured in the 1 ,.@g/mleachof pepstatinA andleupeptin,and200 p.Mphenylmethylsulfonyl absence of both Ras competitor and GAP were subtracted from each sample fluoride. Cell lysis was completed by the addition of 0.1% Triton X-100. value. Chromosomal DNA was degraded by the addition of 10 mM MgCl2 and 10 @.&g/mlDNaseI(Boehringer Mannheim). Cell debris was removed by high speed centrifugation in a chilled microcentrifuge, and the clarified extracts RESULTS were used in GAP activity assays and immunoblots. Direct GAP Assays. The ability of each GAP mutant to stimulate Ras Semirandom Mutagenesis The DNA encoding GAP[701â€”1044] GTPase activity was measured by the nitrocellulose filter binding procedure was rnutagenized by a modified PCR. Using different reaction con (18, 19). Each 0.05 ml reaction contained approximately 0.1 ass of ditions, two different pUC8 expression libraries of GAP[701â€”1044J [â€˜y-32P]GTP-charged Ras protein, 20 mM 4-(2-hydroxyethyl)-1-piperazine mutants were constructed with different mutational frequencies. One ethanesulfonic acid (pH 7.5), 1 mM MgCI2, and 1 mg/ml bovine serum library designated â€œnumberfourâ€•averaged one point mutation every albumin. The reaction mixture was prewarmed for 2 mm at 30Â°Cprior to the 266 base pairs in the 1029-base pair GAP insert, while library â€œnum addition of 10 pi of clarified lysate. Samples were incubated an additional ber fiveâ€•hadapproximately one point mutation every 114 base pairs. 5 mm at 30Â°Cand then quenched by the addition of ice-cold 20 m@i Library four was used primarily for the identification of inactivating 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid and 250 @MMgC12. Samples were collected onto nitrocellulose filters (Schleicher and Schuell; mutations, while library five was used to identify mutations which did BA 85), washed extensively with quench buffer, and quantitatedby Ce not affect GAP activity. Mutant proteins were initially characterized renkov counting. Samples were done in duplicate. Lysates were prepared in crude extracts following induction. from cells transformed with pUCS, and pUC8-GAP[701â€”1044Jwere in Inactivating Mutations in the GAP Catalytic Domain. Approx cluded as controls in each assay series. Lysates, which were negative for imately 200 potentially mutant GAP[701â€”1044]proteins were examined GAP activity, were reconfirmed for authenticityby two additional sets of for the ability to stimulate Ras GTPase. Twenty-three clones were found assays performed on lysates prepared from cells which had been retrans to be inactive in the primary screen. Immunoblots of the induced E. coli formed with the purified mutant plasmid. lysates confirmed protein expression for each of the inactive clones. Immunoblot Analysis of Mutant GAP[701-10441 PrOtein Expression. Plasmid DNA was prepared from each inactive GAP clone positive for Mutant GAP clones which did not produce activity in cell extracts were expression and was completely sequenced to identify every mutation analyzed for the expression of GAP protein. Lysates used for the primary GAP assays were immunoblotted using denaturing polyacrylamide gel electrophore present in each gene. Twelve clones were found to contain only one sis as described previously. GAP[701â€”1044]wasdetected using the GAP codon change (not counting silent changes), and six were found to COOH-terminus-specific peptide antiserum 677 (21) and the ECL system contain two codon changes (Table 1). Of the six double mutants, five (Amersham). Because of variability associated with the low-titer 677 anti were shown to have one codon change which alone was nonphenotypic, serum, each immunoblot was repeated on fresh samples three times to confirm indicative that the remaining mutated codon was the inactivating lesion. expression. Of the eighteen inactivating mutants, nine had mutations effecting resi Purification of 6x(histidlne)-GAP[701-1044] Proteins. E. coli strain dues highly conserved among the RasGAP-related proteins (Fig. 1). @ was transformedwith0.5 ,.@gofpQE-30-GAP[701â€”1044]ormutant Mutants V893A, F898S, L9O1P, I902N, and I906T were all localized in GAPplasmidandplatedon a single LEA-plate.After 16 h of 37Â°Cincubation, the most highly conserved region of GAP. A variety of amino acid the confluent plate was washed into one liter of LBA and incubated for 1 h. substitutions were Obtained, ranging from conservative (N841S, V853A, Protein expression was induced by the addition of 0.5 mMisopropylthio-@3-D- galactoside, after which the culture was incubated an additional 4 h. Cell V893A, and H1O18R), loss or change in charge (E8260, E947K, and pellets were stored at â€”80Â°Cuntil lysis. The lysis of each sample was as D991G), and introduction of potential helix breakers (E826G, L845P, detailed above with the exception that 10 ml of 50 mMNaH2PO4(pH 7.0)-300 5877P, L9O1P, and D991G) to the most common class, loss of a hydro mM NaG (buffer A) was used. Following DNA degradation and centrifugation phobic sidechain (L732H, Y798H, L845P, F898S, I902N, 1906T, 195Th, of the lysate, the proteins were bound in batch to 1 ml of nickel-agarose beads and L995S). 5439

Table1 DescriptionofrandomlyderivedinactivatingmutationsinGAP(701â€”1044J subcloned into the poly-histidine expression plasmid pQE-30. The No. of pQE-30 expression vehicle provided for the fusion of six histidine Mutant Mutation mutations Activityâ€• residues to the NH2-terminus of each GAP mutant, allowing purifi GAP[701â€”1044] None 1.00 cation of each protein using nickel-agarose affinity chromatography. 4-42 L732H 1 <0.02 Initially, each construct was induced and evaluated for protein cx 4-53 Y798H 1 0.02 4-137 E826G 1 <0.02 pression relative to wild-type GAP[701â€”1044]subcloned into pOE 4-113 N841S 1 0.02 30. In every case, mutant expression was 10% or less than the 4-35 L845P 1 <0.02 4-117 V853A 1 <0.02 wild-type GAP, consistent with the results obtained by immunoblot. 4-87 S877P 2â€• <0.02 The mutant proteins were purified and concentrated; protein concen 4-33 V893A 1 <0.02 tration was determined by Bradford assay and polyacrylamide gel 4-26 F898S 2b <0.02 4-98 L9O1P 2@' <0.02 electrophoresis, and GAP activity measured. Only the N841S, 4-141 1902N 2@' <0.02 V893A, 1906T, 195Th, and D991G GAP mutants were expressed in 4-150 1906T 1 0.02 sufficient quantity to purify to near homogeneity. The catalytic activ 4-177 E947K 1 <0.02 4-119 I957T 1 <0.02 ities of these five mutants were compared to the wild-type GAP[701â€” 4-16 D991G 1 <0.02 1044] control by measuring the stimulation of Ras GTPase by increasing 4-125 L995S 2b <0.02 4-44 H1O18R 1 <0.02 concentrations of each protein. As can be seen in Fig. 3, all five GAP 4-68 L732P 2 <0.02 mutants were significantly impaired in their catalytic activity when corn C873R pared to equal concentrations of wild-type GAP[701-1044]. Extended 4-111 E836K 3 ND 1862M incubation (30 ruin) of concentrated D991G and 195Th revealed weak M950T stimulation of Ras GTPase, although the activity of 195Th was extremely 4-124 T804A 3 ND difficult to detect. The partially purified Y798H, E826G, V853A, F858S, L1O2OP 01032R and E947K mutants were all weakly active, although insufficient protein 4-194 V934A 3 ND was Obtained to compare activity as a function of concentration. F943S L703P For mutants with measurable Ras-GTPase stimulating activity, it 4-23 Y720H 4 ND was possible to determine if the reduction in activity was due to a E722R decrease in Ras binding affinity. Using a kinetic competition assay, K724R K812R relative binding affinities for Ras were estimated for each of the 4-127 M715T 4 ND purified mutants (Table 2). Only one of the mutations, V853A, was M795V actually impaired for Ras binding. Surprisingly, the NM1S, F898S, S877P 191ST E947K, and D991G mutations actually increased the affinity of GAP (2BaseduponserialdilutionsofexpressingcellextractsandnormalizedtoGAP[701â€”for Ras. Binding affinity was increased approximately 6-fold with 10441. N841S and E947K and 16â€”80-fold with F898S and D991G. The b Second mutation probably not responsiblefor inactivatingphenotypebased upon analysis of nonphenotypic mutations. Second mutations: 4-26, I752T; 4-87, D987G; 4-98, mutant 195Th was measurably inhibited by a range of Ras[L61]-GTP I856M; 4-141, F943V; 4-125, K855E. The I752T mutation in 4-26 was later reverted to concentrations (1 to 100 p,M), but a reliable dissociation constant was wild-type without affecting the inactivated phenotype; ND, not done. not obtained due to the extreme reduction of GAP activity associated with this mutation. Complete inhibition of 195Th activity was repro ducibly observed with 1 to 10 @LMcompetitor,suggesting an increase In order to more accurately assess the degree of loss of catalytic in Ras binding affinity over wild-type GAP[701â€”1044J. activity for each mutant GAP protein, dilutions of each lysate were Noninactivating Mutations. Identifying nonessential amino acids made and assayed for GAP activity. The amount of activity relative to is an excellent method for locating nonbinding regions of a protein. the wild-type GAP[701â€”1044] control lysate was determined (Table An opposing data set of nonessential amino acids in the GAP catalytic 1). In most cases, there was no activity observed with each mutant domain was generated by identifying clones from library number 5, under the assay conditions. Only mutants Y798H, N841S, and 1906T which were fully active. Due to the high rate of mutation in this had weak amounts of detectable activity. Loss of GAP function could particular library, over 47 distinct codon changes were identified in 33 be accounted for by any of three reasons: (a) protein expression or clones, which had little or no effect upon GAP activity. A number of stability was significantly reduced, resulting in a corresponding drop false negatives originally identified in the number 4 library search also in detectable GAP activity; (b) the catalytic activity of the mutants contributed to the identification of nonessential amino acids. These was reduced; or (c) the ability to bind Ras was affected. To determine mutations are summarized in Table 3. The majority of nonessential the relative amounts of each mutant GAP protein in the lysates, a GAP residue changes were substitutions into positions which are not well carboxy-terminal antibody was used to immunologically quantitate conserved among the GAP family of proteins (32 nonconserved each mutant. The epitope recognized by the 677 antibody consists of codons, 15 semiconserved) and are shown in Fig. 1. As with the residues 968â€”981,which were unaffected by inactivating mutations. inactivating mutations, each mutant was tested multiple times to By immunoblot, each of the mutant GAP proteins was observed to be confirm its phenotype and to calculate the activity relative to wild present in low levels relative to the wild-type control (Fig. 2). Only type GAP[701â€”1044] (Table 3). Most of the mutants had near to N841S (4-113) was found to be expressed at a level approaching wild-type activity, with only four of the mutations resulting in proteins wild-type GAP, although over the course of four separate experi with less than 10% wild-type activity. Immunoblot analysis of se ments, reduced levels of N841S were routinely observed. Variations lected noninactivating mutant proteins showed expression levels corn in expression levels were observed with many of the mutants. Since parable to wild-type GAP[701â€”1044]as shown in Fig. 2. no activity and little protein was detected for many mutants, only a crude estimate of catalytic function could be made using cell lysates DISCUSSION necessitating purification of the mutant proteins. Mutant Purification and Binding Determinations. In order to Prior studies of GAP and neurofibromin have identified GAP purify the GAP proteins, ten of the single codon mutants were residues E774, A791, L899, R900, K932, 0935, K946, and G944 as 5440

mutations P T GE HR A/SP R phenotype 0 0 00 L@ 0 00 0 bOAP .GSLR----V RARYSM---EK IMPEEEYS- EFXELIL.QKE LHWYALSH- -VC--GQ- -- DRTLLASILL KIFLH-EKLE &mGAP1 .GSLR----l nlnYta--dh vfPlatYd- dlmnLlLesv dqrpitvSa- -VsilGelvs gkTevAqpLv rlFtHtEria NFl .dtL a etvla---dr ferlvElv- tnmg-d-QgE LpiamALanv vpC--sQ-- - -wdeLArvLv tlFds-rhLl IRA1 .afLRvfidi vtnYpvnpEK hemdkmlaid dPlkyliknp ilaffg-S1- -aC- -sp--a DvdLyAggfL naFdt-rnas IRA2 . Aktdl--gK leaadkf 1-- ry--.tlehpq L-ssfg-aa- -vC-pas--- DidayAagLi rtaFet-rnat sari EhL1

mutations G TRHI A G PYG PG phenotype 0 ooAo o 0 0OL@ 00 bGAP SLLIJCTLNDR EISMEDEAT7 LFRATTLAST LMEQYMK-ATATQFVHHALKDSILRI-MES KQ----SCE LS PSKLEXNE dGAP1 piik-aLaDh EIShltdpTr iFRgnTLvSk ntlldeaNr-lsglhylHqtLr pvlsql-vae Kk-----*p(@ idPSKikdrs NFl yqLLwnlnfsk EvelaDsmqT LFRgnsLASk iMtfcfK-vy gatylqklL DplLRIvitS sdwqhv*SfE vdPt rLEpsE IRA1 hiLvteLlkq Elkraarsdd ilRrnscAtr alslYtrâ€”srgnkyliktLr pvlqgl-.vdn Ke--- -*sfE idkmK-pgsE IRA2 hivvaqLikn Elekssrp'1\5ilRrnscAtr slsmlar-sk gneylirtLq plikkI-iqn rd----ffE ieklKpE-ds sari lsL,fqmvltt Efeatsdvls L1RAnTpvSr mlttYtrrgp gqaylrsiLy qclndv-aih pdlqld*ryl vntgqLspsE homology block Block 1

mutations G STLP ART VV H P AA S PN TP C phenotype 0 Â£@0OA Aoo oo o A o@ AM @o o bGAP DVN*ThLAHLL NILSELVEKI FMASEIIJPPT LRYIYGCLQK -SVQHK--WPT -N'flIIRTRVVSGFVFLRLIC PAILNPRMFN dOAP 1 aVd*TNLhnLq dyvervfEal tksadrcPkv LcqlfhdL-r ecageh-fPs -NrevRysW SGF1FLRffa PAILgPk1Fd NFl sle*eNqrnLL qmtekffhal issSsefPPq LRsvchCLyq -vVsqr-fP- -qnsi-gaVg Sa-mFLRfIn PAlvsPyeag IRA1 nsekntbdife kymtrL@idaI tssiddfPie Lvdlcktiyri -aasvn-fPe -yayi-a--V gsFVFLRfIg PAlvsPdseN IRA2 Dae*rqielfv kymnELlEsI snsvsyfPPp LfYlcqniyK -vaceK-fPd -haii-a--a gsFVFLRffC PAlvsPdseN sari @J*ersAqLL lltkrfldav insideiPyg iRwvck-Ldr -nltnrlfPs isdsticsli gGFfFLRfvn PAlisPqtsm homology block Block 2 Block 3A mutations TPVAM LK TR R TS GPG P N phenotype 0 0 0 0 0 0 A tb o 0 0 000 0 0 bOAP IISDSPSPIA ARTLTLVAKS VQNLANLIVEFGAKEPYMEGVNPFIKSN-KH RMIMFLDELIG NVPEIJPD'VFEHSR-TDLC-R dGAP1 ltterldaqt sRTLTLisKt iQsLgNLVss NFl ildkkPkPri eRgLkL,rnsKi 1Qs1ANhV1F -tKEehMrpf NdFvKSNfda arrfFLDias dcPt-sDavn HS1-sfis-d IRA1 Ilivthah-d rkpfitlAlv iQsLANgrEn ifKkdilvsk eeFlKtc-sd ki fnFL.sELc kiPtâ€”nnfTv nvR-eD. ... IRA2 Il-Dishise kRTf1s1AKV iQNiANgsEn fsrwPalcsq kdFlKec-sd Ri frFLaELc r-td-rtidi qvRâ€”TD.... sari lldscPSdnv rkThatiAKi iQsvANâ€”gtsstKthldvsf qPmlKey-ee kvhnlLrkLG NvgdffealE ldayiaLskk homology block Block 3B

mutations G SY R G G phenotype A Ao A 0 0 bOAP -DLA---ALH EICVAHSDEL -RTLSNERGA QQHVLKKIJLAITELLQQKQN QYTKTNDVR dGAP1 NFl gnvl---ALH rllwnnqeki gayLSsnRdh kavgrrpfdk matLL IRA1 IRA2 sari -sLAlemtvn EIyltHeiiL -enLdNlydp dsHVhliLqe lgE Fig. 1. Mutational summary of the conserved Ras GAP catalytic domain. Bovine GAP (residues 701â€”1044)was aligned with the five known Ras GAP homologues using the method described by Wilbur and Lipman (33). Amino acid substitutions are listed above the location in the sequence where they were introduced. Below each substitution is a symbol denoting the in vitro activity of the mutation: o, active; i@,catalytically impaired. Below the sequences are shown blocks of significant homology among the RasGAP proteins (29). Protein designations: bGAP, bovine RasGAP residues 701â€”1044;dmGAP1,Drosophila Gapi residues 455 to 701; NFl, human neurofibromin residues 1194â€”1521;IRA1,Saccharomyces cerevisiae Iraip residues 1503â€”1791;IRA2,S. cerevisiae Ira2p residues 1666â€”1936;sari,Schizosaccharomycespombe GAP residues 169â€”472(34â€”37). being essential for catalytic activity (26â€”29).Although Ras binding (20). Since previous mutagenesis of conserved residues in GAP and data has not been reportedfor most of these mutations, residues L899 and neurofibromin has resulted only in catalytic impairment and not reduction K932 have been shown to decrease catalytic activity without reducing the in Ras binding, we chose to use a semirandom approach to identify amino affinity for Ras-GTP. However, a recent report suggested that mutations acid residues in the GAP catalytic domain essential for binding Ras. at the position analogous to K932 in NFl (K1423) might prevent binding In our analysis, 17 of 64 codon changes were deleterious to the to Ras (25). The reason for the discrepancy is unclear but could be related normal function of GAP. Without exception, every inactivated mutant to the instability of NFl protein mutated at this position. Similarly, found in this study failed to match all the criteria predicted for a catalytically impaired neurofibromin mutants with substitutions corre protein binding site mutant. Each inactivating mutation resulted to sponding to GAP residues K946 and G944 reverted Ras-transformed some degree in destabilization or insolubility of the GAP catalytic fibroblasts, suggesting that the mutant NFl proteins were competing with fragment when expressed in E. coli. Examination of every mutation, the Ras effector for binding to Ran (28). including silent changes, within each defective GAP mutant failed to Our primary goal was to identify, if possible, a localized binding identify any suboptimal codon usage which might have resulted in surface for Ras by mutagenically eliminating the amino acid side reduced expression in E. coli. We conclude that these mutations chains involved in the interaction. In searching for such mutations in disrupted intramolecular interactions required for maintaining the GAP, we predicted that these residues would be exposed to the solvent stability of the folded protein. The loss in catalytic activity observed and not be important for maintaining the tertiary structure of the with many mutants may have been more a result of global distortion catalytic domain. Such mutations would be analogous to those found of the native conformation than targeting of a specific active site. The in the Ras effector loop, which are located on the surface of the naturally occurring neurofibromin K1423E mutant (K932 in GAP) is protein and are available for side chain interactions. Mutation of Ras very similar to our inactivating GAP mutants in that it is catalytically in this region blocks biological activity and effects interaction with impaired, binds Ras, and is structurally destabilized (26). Random GAP without reducing protein stability in either E. coli or animal cells mutagenesis of the NFl/GAP-related domain has also resulted exclu 5441

no single residue is significantly responsible for the binding interac tion with Ras or a few, as yet uncharacterized, residues mediate Ras binding. Many of the inactivating mutations identified affected residues L@ predicted to be important to structure/function based upon their con servation among the GAP homologues. L732H, E826G, L845P, F898S, L9O1P, 1902N, 1906T, and 195Th all had impaired catalytic activity. These residues appear to be conserved because of their role in defining the tertiary structure of the protein. The cluster of inacti vating changes in the highly conserved and catalytically important FLR...PA...P region (amino acids 898â€”909)demonstrates the efficacy of the randomized mutagenic method. We also found that mutations in the neighboring residues V893, F898, and 1906 reduced catalytic Fig. 2. InStabilityofthe mutated GAP catalytic domain in E. coli. Bacteria expressing function but not binding. A synthetic peptide corresponding to this GAP[701â€”10441,inactivemutants of GAP[701â€”1044),anda pUC8 vector control were region of bovine GAP was previously shown to inhibit the ability of lysed and an equal amount of soluble protein (â€”15 @g)separatedon a sodium dodecyl sulfate polyacrylamide geL Proteins were transferred to nitroceilulose and immunoblotted GAP to stimulate Ras GTPase (30). This peptide also promoted with an antibody specific for a region of GAP carboxy-terminus not affected in the mutant proteins analysed. Table3 DescriptionofrandomlyderivedmutationswithinGAP(701â€”1044Jwhichhave a negligible effect on activity. No. of Mutant Mutation mutations Activityâ€• 5-74 L703P 5 1.00 4-156 M711T 0.25 2 .@â€˜ 5-27 E712G 2 ND 5-12 K713E 3 0.15 4-63 H740R 1.00 4-48 T747S 1 1.00 5-12 1747A 3 0.15 5-18 T747A 1 2.00 4-143 L749P 0.33 5-26 K755R 2 0.67 I 4-189 R773G 1 0.50 4-168 M795T 2 0.07 4-69 0797K 2 1.00 0.0 0.1 0.2 0.3 0.4 0.5 5-40 K800I 0.33 5-56 T8O2A 2.00 GAP concen@ation(mg/mi) 4-191 S821G 1.00 Fig. 3. Activity profiles for homogeneous purified GAP mutants. Purified mutant 5-30 S821G 1 1.00 2 GAP[701â€”1044] proteins were serially diluted and tested for the ability to stimulate the 5-7 58210 0.15 intrinsic GTPase of H-Ras-GTP. A 100% value indicates complete hydrolysis of Ras 5-6 S824P 3 0.67 GTP. Each protein used in this experiment was purified to near (80% or better) homo 4-49 @25Y 1.00 2 geneity. â€¢,GAP[701â€”1044];, N841S; 0, V893A; A, 1906T;0, I957T; +, D991G. 4-69 S830P 1.00 5-48 E833G 3 2.00 4-85 D837G 1.00 5-74 A843T 5 1.00 5-48 H844L 3 2.00 Table 2 Binding affinity ofRas for different GAP(701â€”1044Jmutantproteins 5-29 K855R 2 2.00 Relative binding affinities for each @roteinareestimated by adding increasing con 5-48 K855R 3 2.00 centrations of Ras[L61]-GTP to a Ras- 2P-GTP hydrolysis assay containing sufficient 5-72 K855E 2 2.00 mutant GAP protein to give measurable hydrolysis. The IC50 obtained for GAP[701â€” 4-84 1856T 1.00 1044] is in the micromolar range due to its expression in E. coli. To confirm the potency 4-66 A859V 2 1.00 oftheRas[L611-GTPproteinused,acompetitionprofilewasmadewiththeneurofibromin 5-6 1862V 3 0.67 GAP-relateddomain,whichhadannat.Clone IC50of40 5-62 Y869H 0.67 5-74 T890A 5 1.00 IC50â€•GAP[701â€”1044]# Mutation 5-72 N908P 2 2.00 None 80Â±11@ 5-26 F912C 2 1.00 4-53 Y798H 175 4-173 1914T 2 0.50 4-137 E826G 140Â±72 p.M 5-6 S918P 3 0.67 4-113 N8415 16Â±5p.at 5-74 S918P 5 1.00 4-117 V853A >200 pM 4-142 A923V 1.00 4-33 V893A 80 Â±2pat 5-41 V930A 2 0.33 4-26 F898S 5Â±2pat 4-173 V934M 2 0.50 4-150 I906T 200Â±20 4-30 F943L 2 1.00 4-177 E947K 12Â±2p.ai 4-168 S959R 2 0.07 4-119 1957T <80 pMâ€• 5-55 K961R 1.00 4-16 D991G 1 Â±0.6pat 5-12 1965T 3 0.15 5-3 L968S 0.09 a IC5@,s represent the average of at least three separate experiments. 5-29 2 2.00 b @etoextremelyweak GAP activity, an exactIC@ wasnotobtainedfor thismutant. E976G 5-41 L977P 2 0.33 4-66 D979G 2 1.00 5-7 S984P 2 0.15 sively in the identification of inactivating mutants which destabilized 5-74 D987N 5 1.00 the protein.4 The failure of this and other studies to identify mutations 4-43 H996Y ND 4-30 E1028G 2 1.00 in GAP which significantly affect Ras association suggests that either 5-27 R10440 2 ND a Based upon serial dilutions of expressing cell extracts and normalized to 4 F. McCormick, personal communication. GAP[701â€”1044]. 5442

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1994 American Association for Cancer Research. RANDOM MUTAGENESIS OF RASGAP guanine nucleotide exchange, suggesting that the peptide interacted 3. Gibbs, J. B., Sigal, I. S., Poe, M., and Scolnick, E. M. Intrinsic GTPase activity directly with the Ras protein. Since mutations within this region of distinguishes normal and oncogenic ras p21 molecules. Proc. Natl. Acad. Sci. USA, 81: 5704â€”5708,1984. GAP do not disrupt Ras binding, it is possible that the conserved site 4. Downward, J. The ras superfamily of small GTP-binding proteins. Trends Biochem. is not significantly involved in the primary association of Ras and 5th., 15: 469â€”472, 1990. GAP but in a secondary interaction which leads to Ras GTPase 5. Zhang, X. F., Settleman, J., Kyriakis, J. M., Takeuchi-Suzuki, E., Elledge, S. J., Marshall, M. S., Bruder, J. T., Rapp U. R., and Avruch, J. Normal and oncogenic activation. p2lras proteins bind to the amino-terminal regulatory domain of c-Raf-1. Nature This study has identified a new subregion of GAP which is important (Lond.), 364: 308â€”313, 1993. for function. Four inactivating mutations (E8260, N841S, L845P, and 6. Warne, P. H., Viciana, P. R., and Downward, J. Direct interactionof Ras and the amino-terminal region of Raf-1 in vitro. Nature (Lond.), 364: 352â€”355,1993. V853A) were found flanking a section of GAP strongly predicted to be 7. Vojtek, A. B., Hoilenberg, S. M., and Cooper, J. A. Mammalian Ras interacts directly an external loop (data not shown). Each of these positions is moderately with the serine/threonine kinase Raf. Cell, 74: 20@-214,1993. to highly conserved, although this area has not been designated as 8. Van Aelst, L, Barr, M., Marcus, S., Polverino, A., and Wigler, M. Complex formation between RAS and PM and other protein kinases. Proc. NaIl. Acad. Sd. significantly conserved among the members of the GAP family (29). USA, 90:6213â€”6217,1993. These mutations effect catalysis much like the other three subregions 9. Yatani, A. K., Okabe, K. Polakis, P., Halenbeck, K., McCOrmick,F., and Brown, A. M. rasp2l and GAP inhibitcouplingofmuscarinic receptoisto atrialK+ channels Cell,61: shown here to be important.The V853A mutation in this region was the 769â€”776,1990. only one identified in this study which reduced affinity for Ras-GTP. 10. Martin, 0. A., Yatani, A., Clark, R., Conroy, L, Polakis, P., Brown, A. M., and However, it cannot be firmly concluded that these residues are McCormick, F. GAP domains responsible for ras p21-dependent inhibition of mus carmnicatrialK+ channelcurrents.Science(WashingtonDC), 255: 192â€”194, directly involved in GTPase activation or Ras binding since they 1992. also destabilize the structure of the protein. The resulting changes 11. Dominguez, I., Marshall, M. S., Gibbs, J. B., Garcia de Herreros, A., Comet, M. E., in protein conformation could be significant enough to influence Graziani, 0., Diaz-Meco, M. T., Johansen, T., McCormick, F., and Moscat, J. Role of GTPase activating protein in mitogenic signalling through phosphatidylcholine interaction between Ras and GAP mediated by distant sidechain hydrolysing phospholipase C. EMBO J., 10: 3215â€”3220,1991. interactions. 12. Duchesne, M., Schweighoffer, F., Parker, F., Clerc, F., FrObert,Y., Thang, M. N., and Tocque, B. Identification of the SH3 domain of GAP as an essential sequence for While most of the GAP mutants examined had dissociation con Ras-GAP-mediated signaling. Science (Washington DC), 259:525-528,1993. stants for Ras similar to the wild-type GAP[701â€”1044] control, four 13. Schweighoffer, F., Barlat, I., Chevallier-Multon, M. C., and Tocque, B. Implication actually bound Ras with higher affinity than wild-type GAP. How of GAP in Ras-dependent transactivation of a polyoma enhancer sequence. Science (Washington DC), 256: 825â€”827,1992. ever, due to the structural effects of these mutations, interpretation of 14. McGlade, J., Brunkhorst, B., Anderson, D., Mbamalu, 0., Settleman, J., Dedhar, S., the binding data is difficult. It is possible that the protein preparations Rozakis-Adcock, M., Chen, L B., and Pawson, T. The N-terminal region of GAP consist of a mixture of active and inactive polypeptides. A high regulates cytoskeletal structure and cell adhesion. EMBO J., 12: 3073-3081, 1993. proportion of inactive molecules would require less Ras-GTP to 15. Ridley, A. J., and Hail, A. The small GTP-binding protein rho regulates the assembly inhibit the assay, giving an artifactually low 50% inhibitory concen of focal adhesions and actin stress fibers in response to growth factoN Cell, 70: 389â€”399,1992. tration. 16. Settleman,J.,Narasimhan,V.,Foster,L C.,andWeinberg,R.A. Molecularcloning While many of the conserved GAP residues remain to be analyzed of cDNAs encoding the GAP-associated protein pl9O: implications for a signaling for their role in GAP function, our data allow specific conclusions to pathway from ras to the nucleus. Cell, 69: 539â€”549,1992. 17. Settleman, J., Albright, C. F., Foster, L C., and Weinberg, R. A. Association between be made. The activation of Ras GTPase is a function distinct from the GTPase activators for Rho and Ras families. Nature (Lond.), 359: 153â€”154, initial binding to Ras-GTP. This is not surprising in light of the 1992. observation that the Ras D33N and D38N/D38E effector mutations 18. Vogel, U. S., Dixon, R. A.. Schaber, M. D., Diehl, R. E., Marshall, M. S., Scolnick, E. M., Sigal, I. S., and Gibbs, 3. B. Cloning of bovine GAP and its interaction with also block GTPase activation by GAP without significantly de oncogenic ras p21. Nature (Lond.), 335: 90â€”93,1988. creasing GAP binding (31). The activation of Ras GTPase by GAP 19. Schaber, M. D., Garsky, V. M., Boylan, D., Hill, W. S., Scolnick, E. M., Marshall, M. S., Sigal, I. S., Gibbs, B. ms interaction with the GTPase-activating protein was much easier to disrupt than binding alone, indicating that (GAP). Proteins, 6: 306â€”315,1989. GTPase activation requires a smaller, more conformationally de 20. Marshall, M. S. The effector interactions of p2lras. Trends Biochem. Sd., 18: fined structure which is independent of binding. At least one 250â€”254,1993. 21. Marshall, M. S., Hill, W. S., Ng, A. S., Vogel, U. S., Schaber, M. D., Scolnick, kinetic study has suggested that GAP accelerates the isomerization E. M., Dixon, R. A. F., Sigal, I. S., and Gibbs, I. B. A C-terminal domain of GAP of Ras-GTPto a conformation where GTE can be rapidly hydro is sufficient to stimulate ras p21 GTPase activity. EMBO J., 8: 110@â€”1110, lyzed (32). 1989. 22. Leung, D. W., Chen, E., and Goeddel, D. A method for random mutagenesis of a The major finding of this study is that binding of Ras to GAP is not defined DNA segment using a modified polymerase chain reaction. Technique mediated by a conserved, contiguous stretch of Ras-like effector (Phila.), 1: 11â€”15,1989. 23. Maniatis@ T., Fritsch, E. F., and Sambrook, J. Molecular Cloning: A Laboratory residues. While our results might also suggest the existence of an Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory, 1982. extremely small and compact binding site which was missed by the 24. Li, Y., Bollag, G., Clark, R., et a!. Somatic mutations in the neurofibromatosis 1 gene randomized procedure, this and other mutagenic studies have ruled in human tumors. Cell, 69: 275-281, 1992. 25. Poullet, P., Lin, B., Esson, K, and Tamanoi, F. Functional significance oflysine 1423 out most conserved residues as being involved in Ras binding. Protein of neurofibromin and characterization of a second site suppressor which rescues association occurring through many hydrophobic and hydrogen bond mutations at this residue and suppresses RAS2Val-19-activated phenotypes. Mol. interactions across a large surface region would explain why all of the Cell.Biol.,14: 815â€”821,1994. 26. Wiesmuller, L, and Wittinghofer, A. Expression of the GTPase activating domain of known Ras effector proteins, such as yeast adenylyl cyclase, Raf, and theneurofibromatosistype1(NFl) geneinEscherichiacoliandroleoftheconserved GAP, show little or no obvious sequence similarity within their Ras lysine residue. J. Biol. Chem., 267: 10207â€”10210,1992. 27. Skinner, R. H., Bradley, S., Brown, A. L, Johnson, N. J. E., Rhodes, S., Stammers, binding domains. D. K., and Lowe,P. N. Use of the Glu-Glu-PheC-terminalepitopefor rapid purification of the catalytic domain of normal and mutant ras GTPase-activating proteins. J. Biol. Chem., 266: 14163â€”14166,1991. ACKNOWLEDGMENTS 28. Nakafuku, M., Nagamine, M., Ohtoshi, A., Tanaka, K., Toh-E, A., and Kaziro, Y. Suppression of oncogenic ras by mutant neurofibromatosis type 1 genes with single We thankPaul Polakis for helpfulcomments. amino acid substitutions. Proc. NatL Aced. Sd. USA, 90: 6706-6710, 1993. 29. Gutmann, D. H., Boguski, M., Marchuk, D., Wigler, M., Collins, F. S., and Ballester, R. Analysisof the neurofibromatosistype1 (NFl) GAP-relateddomainby site REFERENCES directed mutagenesis. Oncogene, 8: 761â€”769,1993. 30. Rubinfeld, B., Wong, G., Bekesi, E., Wood, k, Heimer, E., McCormick, F., and 1. Bos, J. L. ras oncogenes in human cancer: a review. Cancer Rca., 49: 4682â€”4689, Polakis, P. A synthetic peptide corresponding to a sequence in the GTPase activating 1989. protein inhibits p2lras stimulation and promotes guanine nucleotide exchange. tnt. J. 2. Barbacid, M. ras genes. Annu. Rev. Biochem., 56: 779â€”827,1987. Pept. Protein Res., 38: 47â€”53,1991. 5443

31. Marshall,M.S., andHenich,L A. CharacterizationofRaseffectormutantinterac 35. Tanaka, K., Nakafuku, M., Satoh, T., Marshall, M. S., Gibbs, J. B., Matsumoto, K. tions with the NFl-GAP related domain. Oncogene, 8: 425â€”431,1993. Kaziro, Y., and Toh-E, A. S. cerevisiae genes IRA) and IP.A2 encode proteins that 32. Moore, K. J., Webb, M. R., and Eccieston, J. F. Mechanism of GTP hydrolysis by may be functionally equivalent to mammalian ras GTPase activating protein. Cell, p2lN-ras catalyzed by GAP: studies with a fluorescent GTP analogue. Biochemistry, 60: 803â€”807,1990. 32: 7451â€”7459,1993. 36. Gaul, U., Mardon, 0., and Rubin, 0. M. A putative Ras GTPase activating protein 33. Wilbur, W. J., and Lipman, D. J. Rapid similarity searches ofnucleic acid and protein acts as a negative regulator of signaling by the Sevenless receptor tyrosine kinase. databanks. Proc. Natl. Acad. Sci. USA, 80: 726â€”730,1983. Cell, 68: 1007â€”1019,1992. 34. Wang, Y., Boguski, M., Riggs, M., Rodgers, L, and Wigler, M. Sari, a gene from 37. Ballester, R., Marchuk, D., Boguski, M., Saulino, k, Letcher, R., Wigler, M., and Schizosaccharomyces pombe encoding a protein that regulates rasl. Cell. Regul., 2: Collins, F. The NFl locus encodes a protein functionally related to mammalian GAP 453â€”465,1991. andyeastproteins.Cell,63: 851â€”859,1990.

5444

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1994 American Association for Cancer Research. Structural Analysis of the Ras GTPase Activating Protein Catalytic Domain by Semirandom Mutagenesis: Implications for a Mechanism of Interaction with Ras-GTP

Lisa Hettich and Mark Marshall

Cancer Res 1994;54:5438-5444.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/54/20/5438

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/54/20/5438. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.