Endothelial Nitric Oxide Synthase Membrane Targeting

6 downloads 0 Views 5MB Size Report
Acknowledgments-We are grateful to Drs. Lisa Robinson and Eva. Neer for helpful ... Hancock, J. F., Paterson, H., and Marshall, J. (1990) Cell 63, 133-139. 18.
THE JOURNAL OF BIOIOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 40, Issue of October 7,pp. 2501625020, 1994 Printed in U S A .

Endothelial Nitric Oxide Synthase Membrane Targeting EVIDENCE AGAINST INVOLVEMENTOF A SPECIFIC MYRISTATE RECEPTOR* (Received for publication, June 16, 1994, and in revised form, August 3, 1994)

Liliana BusconiS and Thomas MichelO From the Brigham andWomen$ Hosoital. , Cardiovascular Division, Harvard Medical School, Boston, Masskhusetts 02115

The endothelial isoform of nitric oxide synthase (ec- signaling,neurotransmissionand cell-mediated cytotoxicity NOS) is targeted to the particulate subcellular fraction (for reviews, see Refs. 1-3). Despite thediverse biological roles However, the as- of NO, the distinctNOS isoforms characterized to date appear by means of N-terminal myristoylation. sociation of ecNOSwith the particulate subcellular frac- to share many similarbiochemical features. However, the ention appears to be dynamicallyregulated, in that agonist dothelial cell NOS (ecNOS) is unique among these NOS isotreatment of endothelial cells induces translocation of forms in that it undergoes N-terminal myristoylation (4). Muthe enzyme from membraneto cytosol (Michel,T.,Li, G., tagenesis of the myristoylation consensus sequence of ecNOS and Busconi, L. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, cDNA yields a mutant (myr-ecNOS) cDNA which, when tran625242551. cDNA encoding wild-type and myristoyla- siently transfected into COS-7 cells, yields a protein which no tion-deficient mutant (myr-) ecNOS wastranscribed and longer undergoes myristoylation and is expressed in thecytosol translated in vitro, and we found that the recombinant (4,5). Thus, myristoylation is necessary for targeting ecNOS wild-type but not the myr- mutant protein undergoes expression to the particulate subcellular fraction. However, the myristoylation and is able to associate with biological role of myristoylation in subcellular targetingcomplex: is there membranes prepared from diverse cell sources. Treatare numerous examples of myristoylated proteins which unment of these cell membranes with heat or with trypsin did not affect their ability subsequently to serve as ac- dergo reversible association with intracellular membranes (6, ceptor membranes for the wild-type recombinant en- 7) and yet other myristoylated proteins which are found pricell cytosol (8). We have previously foundthat the zyme. The wild-typeecNOS, but not the myr- mutant, is marily in the subcellular distribution of ecNOS is dynamically regulated in able to form stable associations with phospholipid liposomes. We also exploredthe possibility that a polybasic endothelial cells, yet the other structural determinants of ecdomain within the ecNOS protein might serve as a sec- NOS membrane targeting remain poorly understood. Because ondary structural determinant for ecNOS membrane as- NO is a labile compound capable of reacting with diverse bisociation and constructed truncation mutants that flank omolecules, it seems likely that the site of intracellular NO a polybasic domainpresent in the ecNOS. These trunca- synthesis could importantly affect its signaling functions. The tion mutants, transcribed and translated in vitro or current studyexplores some of the structural determinantsfor transfected into COS-7 cells,undergomyristoylation ecNOS membrane targeting in order to gain further insight and are able to associate with biological membranes in a into the mechanisms whereby the ecNOS “myristoyl switch” (9) fashion indistinguishable from the wild type ecNOS. might be regulated. Taken together,these results indicate that ecNOS bindIn endothelial cells, the agonist bradykinininduces transloing to biological membranesis dependent upon interac- cation of the ecNOS from membrane tocytosol associated with tions of the N-terminal myristoyl moiety of ecNOS with enzymephosphorylation (9). The ecNOS appears similar to lipid components of the membrane, and this association othermyristoylatedproteins for which thecotranslational does not require a specific membraneprotein function- process of protein myristoylation has been found to be essening as a myristate receptor nor the presence of a poly- tially irreversible (10). Some myristoylated proteins appear to basic domain withinthe ecNOS. interact with specific saturable myristate receptors which determine membrane association (ll), yet in other proteins(1214), the myristatemoiety promotes association with biological The nitric oxide synthases (NOS)’ comprise a family of en- membranesprimarilythrough hydrophobic interactions bezymes involved in diverse cellular processes, including vascular tween the protein’s myristate and membrane phospholipids. The membrane association of some myristoylated proteins is * This work was supported in part by National Institutes of Health reversible and canbe regulated by protein phosphorylation (6) Grant HL46457 and by funds from the American Heart Association. The or by Ca2+ binding (14). In the present studies, we have excosts of publication of this article were defrayedin part by the payment plored the expression of ecNOS in heterologous systems to deof page charges. Thisarticle must therefore be hereby marked “uduertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate termine whether theassociation of this enzyme with the particulate subcellular fraction is receptor-dependent or, this fact. f Supportedby a fellowshipfrom the Consejo de Investigaciones alternatively, if hydrophobic interactions of the myristoylated Cientificas y Tecnicas (Argentina). protein with membrane phospholipids are sufficient to explain 9 Wyeth-Ayerst Established Investigatorof the American Heart Association. To whom correspondence should be addressed. Tel.: 617-732- its subcellular distribution. Other structural features of the ecNOS required for mem7376; Fax: 617-732-5132. The abbreviations used are: NOS, nitric oxide synthase; ecNOS, brane association have not been identified, although we note endothelial isoform of nitric oxide synthase;BAEC, bovine aortic endo- the presence in thededuced amino acid sequence of the ecNOS thelialcells; CHO, Chinese hamster ovary; bp, base pairb); PAGE, a polybasic domain similar to domainsimplicated in subcellupolyacrylamide gel electrophoresis; FPLC, fast protein liquid chromatography; MARCKS, myristoylated alanine-rich protein kinase C sub- lar targeting of other acylated proteins (15). Other myristoylated proteins contain polybasic domains that stabilize their strate.

25016

Subcellular Targeting of Endothelial NOS interactions with biological membranes (16-18). In addition to the fattyacyl modifications required for membrane association of these proteins, polybasic domains within thesemodified proteins have been identified which can markedly affect the stability of these associations. For the ecNOS, analysis of the deduced amino acid sequence of its cDNA reveals thepresence of a polybasic domain (RRKRR) in the middle of the protein (amino acid residues 629-633). The role, if any, of this domain in stabilizing or targeting the subcellular distribution of the ecNOS has not been explored; this polybasic domain is not evident in the other NOS isoforms (19). In the present studies, we analyze the pattern of subcellular distribution of ecNOS truncation mutants that flank this polybasic domain, using both transient transfection experiments and studies with in vitro transcribed and translated mutantcDNAs.

Myr- mutant G2-A

25017 TruncationTruncation mutant 2 mutant1

i

N terminus

C terminus

R

13 K 13 13633

-

100 aa FIG.1. Map of ecNOS mutants. This is a schematic of the ecNOS primary structure, showing the locations of modifications inecNOS that yield the mutants studied in this paper. Shown are the N-terminal glycine to alanine substitution (G2-A) to yield the myristoylation deficient (myr-)mutant, and the locations of the stopcodons of truncation mutant 1 and 2, which are located, respectively, at residues 652 and 628, to truncate C-terminal and N-terminal relative to the polybasic domain shown in expanded view a t residues 629-633. The reading frame for the wild-type ecNOS encodes a protein of 1205 amino acids.

MATERIALSANDMETHODS pared by ultracentrifugation of cell homogenates for 1h a t 100,000 x g. Cell Culture a n d Dansfection-Cultures of bovine aortic endothelial The pelleted fraction was washed and resuspended by sonication in cells (BAEC), Chinese hamster ovary (CHO), and COS-7 cells were NTE buffer (100mM NaCI, 10 mM Tris-HC1, pH 7.4, 1 mM EDTA) and maintained a s describedpreviously(20).Transienttransfection of used a s acceptor membranes in incubations with in vitro transcribed COS-7 cells with ecNOS wild-type and mutant cDNAs was performed and translated ecNOS. As noted below, for some experiments, these exactly a s before (20). To create stable CHO cell lines expressing wild acceptor membranes were subjected atovariety of treatments: heating type and myr- ecNOS cDNA, the CHO cells were transfected using at 100 “C for 3 min, treatment with 500 pg/ml trypsin for 15 min a t previously describedcDNA constructs (4,20) in the vector pcDNAINeo 25 “C (followed by addition of aprotinin at 500 pg/ml), or incubation (Invitrogen) using the calcium phosphate method (21), and stable with trans1% Triton X-100 for 20 minutes, performed prior to assays of fectants were identified and cloned after 2-3 weeks of selection inG418. ecNOS membrane binding. For studiesof membrane binding of the in Expression of NOS was evaluated both by analyzing Western blots vitro synthesized ecNOS, a 2 0 4 aliquot of the freshly synthesized probed with a polyclonal antibody directed against the ecNOS and alsoreaction mixture was incubated with 30 pl of membrane suspension in by biosynthetic labeling with [35Slmethionine (TranssSS-label, ICN) fol- NTE buffer a t a final membrane protein concentration of 5 mg/ml a t lowed by immunoprecipitation with the ecNOS antiseruma s described 25 “C for 30 min; parallel controls were incubatedNTE with buffer only. previously in detail (4). G418-resistant CHO cells stably transfected Samples were then centrifugedfor 1h a t 100,000 x g to resolve soluble with the vector alone show no ecNOS protein eitherby Western blot or and particulate fractions, which were then denatured under identical immunoprecipitation analyses. For both the wild-type and myrecNOS conditions in Laemmli sample buffer and analyzedby SDS-PAGE and transfectants at least three independently cloned cell lines were char- analyzed by autofluorography on x-ray film or quantitated using a acterized, and each yielded results similar to those shown below. PhosphorImager (Molecular Dynamics). PCR-based Duncations-A polybasic domain was identified by visBinding of in Vitro Synthesized ecNOS to Liposomes-Liposomes ual inspection of the deduced amino acid sequenceof the ecNOS cDNA were prepared by sonication of phosphatidylcholine (12 mg/ml) and (20). Truncation mutants were generatedby PCR, using two different phosphatidylserine (4.2mg/ml) in NTEbuffer containing 0.2 M sucrose, antisense oligonucleotide primers which flank the polybasic domain; using conditions previously established to generate unilamellar lipothe “upstream” sense primer is in common for the two PCR reactions somescomprising80%phosphatidylcholine:20%phosphatidylserine and primes 5’ relative to a unique BspHI site in the ecNOS cDNA, the (22). Following in vitro transcription and translation of wild-type or two “downstream” primers, which flank either side of the polybasic mutant ecNOS cDNAs as described above, a 2 0 4 aliquot of the transdomain, each introducea stop codon and a new XbaI site. The sequence lation reaction was incubatedin a final volume of 50 pl made upby the of the upstream primer corresponds to bp 973-991 of the coding se- addition either NTE buffer alone or with 12.5 plof liposomes plus 17.5 quence and is located 332 bp 5’ to the BspHI site. The downstream pl of NTE a t 20 “C for 30 min. This mixture was diluted 4-fold and then (antisense) primer that wasdesigned to yield a truncation located 3’ resolved on a Superose-12FPLCcolumn(Pharmacia Biotech Inc.) relative to the polybasic domain (“truncation 1”;652 amino acids) ex- which had been pre-equilibrated in Tris-buffered saline containing 1% tends from bp 1975 and has the sequence5‘-GCTCTAGATCAGAACA- bovine serum albumin and 0.025% Tween 20; 500-pl fractions were CACAGAACCTGAG-3’. The other antisense primer that was designed collected, immunoprecipitated with ecNOS antiserum (4), resolved on to yield a truncation located 5’ relative to the polybasic domain (“trun- SDS-PAGE, and analyzed on a PhosphorImager (MolecularDynamics). cation 2”; 628 amino acids) extends from bp 1903 and has the sequence 5’-GCTCTAGATCACCAGGAGGACACCAGCGG-3’. PCR was camed RESULTS out for 25 cycles at 94 “C x 1min, 55 “C x 1min, and 72 “C x 1.5min in the presence of 10% dimethyl sulfoxide using pcDNAINeo-ecNOS as Stable Expression of Dansfected Wild-type andmyr- ecNOS template. The PCR fragment was digested with BspHI and XbaI and in CHO Cells-CHO cells were transfected with cDNA encodcloned into pcDNAINeo-ecNOS (20) which had been previously digested ing wild-type and myristoylation mutant (myr-) ecNOS in the with these same restriction enzymes. These truncation mutants each vector pcDNAINeo, and, after selection with G418, numerous thus preserve the amino-terminal portion of the ecNOS from the beginning of the coding sequence, but terminate at new stop codons intro- resistant clones were obtained. Several of these G418-resistant duced either sideof the polybasic domain. Aschematic representation of clones stably expressed ecNOS mRNA and functional protein, these ecNOS mutants is shown in 1. Fig. The nucleotide sequence of the and we documented (by Western blot analysis) that the level of PCR-generated portionof the mutantcDNAs was determined using the ecNOS expression in these clones was similar to thatfound in dideoxy method with Sequenase (U.S. Biochemical Corp.) according to BAEC; sham-transfected cells had no detectable ecNOS (data the manufacturer’s protocols. not shown). The subcellular distributionof wild-type and myrIn Vitro Dunscription and Danslation of Wild-type and Mutant ecNOS-Different cDNA clones encoding wild-typeor mutant ecNOS in mutant NOS after biosynthetic labelingwith [35Slmethionineis shown in Fig. 2. More than 90% of wild-type ecNOS is found in pcDNAINeo (0.5pg)weretranscribedandtranslatedinvitroin a the membranefraction; by contrast, in CHO cell lines expresscoupled reticulocyte lysate system (TNT system, Promega, Madison, WI) following the manufacturer’s protocols, using T7 RNA polymerase ing the myr- mutant ecNOS, the enzyme is localized to the in a total volume of 50 pl; proteins were labeled using either [’‘Slmecytosol. These data are consistent with our studies of COS-7 thionine(ICN Radiochemical; 40 pciheaction) or I:’Hlmyristic acid cells transiently transfected with these constructs and docu(Amersham Corp.; 30 pCi/reaction). ment that thesubcellular targeting of recombinant ecNOS dePreparation a n d Assay of Biological Acceptor Membranes-BAEC, pends upon enzyme myristoylationbut not on the specific hetCOS-7, or CHO cells were grown to confluence, then harvested and sonicated as described previously (4). Membrane fractions were preerologous cell expression system studied.

Subcellular Targeting of Endothelial NOS

25018

HCM

Mr

-

HCM

-

205 -@

205-e

11697

11697

6 6 4

myrwild-type

wild-type trunc. 1

Mr

trunc. 2

0

6 6 +

4 4 -

FIG.2. Subcellular distribution of wild-type and myr- ecNOS in CHO cells. Shown in this figure are the resultsof SDS-PAGE and autofluorography of biosynthetically labeled CHO cells stably transfected with wild-typeor myristoylation-deficient (myr-j mutant ecNOS cDNA and immunoprecipitated. Thecells were biosynthetically labeled with 13sSlmethionine for 3 h in a six-well tissue culture dish (50 pCi/ well), harvested, and sonicated to yield the cell homogenate (H); twothirds of the homogenate was then subjected to ultracentrifugation to yield cytosol (C) and membrane (Mj fractions, and each fraction was immunoprecipitated withecNOS antiserum andresolved on a 7% polyacrylamide gel a s described previously (4), subjected to autofluorography, and exposed overnight on Kodak XAR film using an intensifying screen. The M, of molecular weight markers is shown in kilodaltons. Results shown in this figure are representative of three independently isolated clones studied for each of the wild-type and mutant stable transfectants.

Subcellular Targeting of ecNOS Duncation Mutants--Two ecNOS truncation mutants were constructed using PCR-based mutagenesis to yield two mutants (trunc. 1 and trunc. 2) that preserve theN terminus of the protein and endat residues 652 and 628, respectively. A schematic representation of these ecNOS mutants is shown in Fig. 1.These two ecNOS truncation mutants were constructed to explore the role of a polybasic domain in ecNOS targeting and flank either side of this domain. When transiently transfected into COS-7 cells, these mutants showed no evidence of NO synthetic activity (data not shown), either by NADPH diaphorase staining(23) or by direct enzyme assays of conversion of [3H]arginine to [3H]citrulline (24), despite robust expression of mutant protein (Fig. 3). Importantly, as shown in Fig. 3, the truncation mutants were similarly targeted to the particulate subcellularfraction. This finding argues against an essential role of this polybasic domainin ecNOS subcellulartargeting. We have previously shown that the myr- mutant ecNOS remains in thecytosol of transiently transfectedCOS-7 cells (41, documenting a n essential role for myristoylation in membrane targeting. In Vitro Danscription and Dunslation of Wild-type and Mutant ecNOS-In order to validate that the in vitro synthesis of ecNOS yieldsprotein which has undergone cotranslational modifications similar to ecNOS expressed in cellular systems, we studied thein vitro myristoylation of wild-type and mutant ecNOS. Fig. 4 shows the results of in vitro transcription and translation of wild-type and mutantecNOS in the presence of [3H]myristic acid. There islabeled myristate incorporation into wild-type ecNOS, as well as into the two ecNOS truncation mutants. These data document that the wild-type ecNOS undergoes the same cotranslational modification when synthesized in vitroas we have previously documented in intactcells; a similar pattern is observed for two truncation mutants (described above). An important negative control shows that the myr- mutant undergoes no [3Hlmyristateincorporation following mutagenesis of its myristoylation consensus sequence. Association of in Vitro Synthesized ecNOS with Biological Membranes-Fig. 5 shows an experiment in which in vitro transcribed and translated ecNOS was incubated with different biological membranes and then sedimented by ultracentrifugation. In the absence of added membranes, the in vitro synthesized ecNOS remains soluble after ultracentrifugation. By contrast, the incubation of the ecNOS with membranes

H C M H C M H C M FIG.3. Subcellular distribution of wild type and truncated ecNOS in transfected COS-7 cells. COS-7 cells were transfected with cDNA encoding wild-typeor truncated ecNOS cDNA and analyzed72 h later following biosynthetic labeling with [:%]methionine, subcellular fractionation, and immunoprecipitation as described for Fig. 1. Results of immunoprecipitations from cellular homogenate( H ) ,cytosol (C), and membrane ( M ) fractions are shown for wild-type ecNOS, truncation mutant 1 (encoding amino acids 1-652), and truncation mutant 2 (residues 1-628). The doublet band present upon immunoprecipitation of wild-type ecNOS in transientlytransfected COS-7 cells has been previously noted (4); although this "extra" immunoprecipitated band is of unknown origin, it is seenonly upon ecNOS cDNA transfection and its immunoprecipitation can be specifically blocked with the immunogen peptide, and its presence on these gelsdoes not affectthe interpretation of the resultsshown. Shown here is an overnight exposureof an autofluorogram of SDS-PAGE on a 7% polyacrylamide gel. This experiment was repeated twice with equivalent results.

-

116+ 97

6 6 -

4 4 +

FIG.4. In vitro transcription and translation of wild type and mutant ecNOSin the presence of [SH]myristicacid. cDNA corresponding to the wild-type, truncation mutants 1 and 2, and myristoylation-deficient (myr-) ecNOS was transcribed and translated in vitro in the presence of [3Hlmyristicacid as describedinthetext.The samples were directly analyzed by SDS-PAGE and autofluorography; the film shown here was exposed for 3 days using an intensifying screen. This experiment was repeatedonce, with identical results.

prepared from BAEX or from (untransfected) COS-7 or CHO cells led in each case to the formation of a sedimentable complex of the labeled enzyme. These data suggest that the association of the ecNOS with biological membranes does not appear to be determined by the cell type of origin. Furthermore, as shown in Fig. 6, the association of the in vitro synthesized ecNOS with biological membranes is not affected by prior treatment of these membranes with heat denaturation or trypsin, although detergent(Triton X-100) treatment of the membranes impaired theirability to serve as acceptors for the ecNOS. A key control for these experimentsis to study theassociation of the myr- mutant ecNOS with biological membranes under identical conditions. In the lower panel of Fig. 6 are shown results obtained when the in vitro synthesized myr- mutant ecNOS is incubated in thepresence or absence of biological membranes. The myr- mutant ecNOS remains almost exclusively in the supernatant, and there practically is no association of the myrecNOS with acceptor membranes in vitro under anyconditions, in marked contrast to the wild-type enzyme. By contrast, as

25019

Subcellular Targeting of Endothelial NOS

minus BAEC CHO COS-7 memb.

Mr

-

Mr 205-

'

205-b

I16+

116-

6 6 -

97

91

s

P S P S P

s

P

nnnn

~~

~~

-

"

4 4 -

6 6 -

-

MEMBRANES:

4 4 +

-k

- +

"

TRUNC. 1 TRUNC. 2

unnu

s

s

FIG.7. Membrane binding of in vitro synthesized ecNOS trunmutants. ecNOS truncation mutant cDNAs were transcribed FIG.5. Binding of in vitro synthesized ecNOS to biological cation and translated in vitro in the presence of [%]methionine and then membranes. Membrane fractions were prepared from different cul- incubated in the absence (""7 or presence ("+") of COS-7 cell memtured cells and used a s acceptor membranes for the ["SSlmethionine- branes, then pelleted by ultracentrifugation to yield supernatant ( S ) labeled in vitro transcribed and translated wild-type ecNOS as deand pellet (P) fractions which were analyzedby SDS-PAGE and autofluscribed inthetext. Following incubation of the labeled in vitro orography as in Fig. 5. Shown are results for the two truncation musynthesizedecNOSwithmembranes from bovine aorticendothelial tants ( n u n c . 1 and n u n c . 2 ) which flank the polybasic domain, as cells (BAEC)or from untransfected CHO or COS-7 cells (as shown) or discussed in the text. This experiment was repeated twice with identical in theabsence of added membranes ("minus memb.")the sampleswere results.

P S

P

P S

P

subjected to ultracentrifuguation. The supernatant (S)and pellet (P) fractions were denatured in Laemmli sample buffer and analyzed by SDS-PAGE and exposed by autofluorography; a n overnight exposureis shown.

s

Mr

P S P S

P

s

P S

---n-

I

I

P "0-ecNOS-wt

205-

WILD-

116-

TYPE

91 66 +

MINUS liposomes

"m-

my-

4 4 -

MYRMUTANT 4 4 -

Membranes: YES Treatment: none

NO

b

--0-

PLUS liposomes

ecNOS-wt myr-

none heat trypsin Triton

FIG.6. Binding of wild-type and myr- ecNOS to denatured membranes. Wild-type ormyr- mutant ecNOS cDNAs were transcribed and translated in vitro in the presence of [35Slmethionineand then incubated in the absence ("NO")or presence ("YES")of COS-7 cell membranes which had undergone various treatments as described in the text. The samples were then pelleted by ultracentrifugation to yield supernatant (S) and pellet (P)fractions which were analyzedby SDSPAGE and autofluorography as in Fig. 4. The conditionsfor heat, trypsin, and Triton X-100 treatments of the COS-7 membranes priorto their incubation with the i n vitro synthesized ecNOS are given in the text. This experiment was replicated four times with equivalent results.

shown in Fig. 7, both of the truncation mutantswe synthesized i n vitro were able to interact with acceptor membranes and remained soluble in their absence. I n Vitro Bindingof ecNOS toLiposomes-To further explore the nature of the interaction of ecNOS with membranes, we studied theassociation of i n vitro synthesized ecNOS with synthetic phosphatidylcholine/phosphatidylserineunilamellar liposomes. The extremely slow sedimentation propertiesof such liposomes led to our decision to analyze the formation of proteoliposome complexes using gel filtration FPLC (25),as shown in Fig. 8. Using a n experimental approach similar to thatused for the study of interactions of ecNOS with biological membranes, the in vitro synthesized ecNOS was incubated with

5

10

IS

20

fraction number FIG.8. Superose-I2 chromatographyof wild-type and myr-ecNOS in presence and absenceof phospholipid liposomes.Wildtype or myr- mutant ecNOS cDNAs were transcribed and translatedin vitro in the presence of [:"S]methionine and then incubated without (upper panel)or with (lower panel)phosphatidylcholine-phosphatidylserine liposomes a s described in the text. The reaction mixtures were then resolved by Superose-12 gel filtration FPLC, and individualfrac(toreduce the sample tions werecollected and then immunoprecipitated volume) and analyzed by SDS-PAGE; a PhosphorImager was used to quantitate the principal radioactive band, seen a t 135 kDa and corresponding to ecNOS. The ordinate represents arbitrary units from the PhosphorImager analysis. The position of the void volume of the column is noted a s V,,. These data were replicated twice in separate experiments.

liposomes, and the samples were then chromatographed on a Superose-12 FPLC column. In the absence of liposomes, the wild-type and myr- mutant ecNOS comigrated in the included volume of the Superose-12 column (see Fig. 8). However, following incubation with liposomes, the wild-type ecNOS now migrated in thevoid volume of the column consistent with its association with theadded liposomes. The elution profile of the

25020

Subcellular Targeting of Endothelial NOS

myr- ecNOS was not significantly affected by the coincubation with liposomes,

( E ) , or may belocated in mid-molecule, as foundfor the M C K S protein (131,or the polybasic domain maybe at the C terminus, as described for K-ras (17). Aside from its myristoyDISCUSSION lation consensus sequence (4), the ecNOS has no other obvious The subcellular targetingof many myristoylated proteinsis acylation consensus sequences (26). However, a polybasic dodynamically regulated and isoften complexly determined. The main in the ecNOS molecule is similar to domains that stabilize myristoylated ecNOS translocates from the endothelial cell the association of other acylated proteins with plasma memmembrane to cytosol following agonist treatment; this subcel- branes (15). The present study shows that two ecNOS truncalular translocation of ecNOS is associated with itsphosphoryl- tion mutants flanking this polybasic domain are both myrisation. NO is highly labile inbiological fluids, and thesubcellu- toylated and bind equally well to biological membranes invitro; lartranslocation of ecNOS mightimportantly affect the these truncation mutants are appropriately targeted t o the cellular signaling roles of its product. It is therefore particu- particulatesubcellular fraction intransienttransfection of larly important to characterize further the basis for membrane COS-7 cells. It remains possible that this polybasic domain association of the ecNOS. plays some secondary role in ecNOS targeting or stability in A major question in the regulation of ecNOS subcellular tarsitu in endothelialcells. However, our experiments in two sepageting is whether the protein associates with the membrane rate heterologous expression systems indicate that this polybasolely through hydrophobic interactions with membrane phossic domain is not necessary for membrane association. pholipids, as has been found for the myristoylated proteins Our data both in intact cells and in cell-free systems are C subMARCKS (myristoylated alanine-rich protein kinase consistent with the hypothesis that myristoylation is the major recoverin strate; Ref. 12) and for the signal-transducing protein determinant for ecNOS membrane association, and this asso(14).Alternatively, the ecNOS might interact with a specific ciation appears tooccur through hydrophobic interactions with myristate receptor protein,as has been proposed for the myristoylated oncogene product pp60"" (11).As we have alsofound membrane lipids. However, although myristoylation is necesfor ecNOS, the membrane association of these myristoylated sary for ecNOS membrane targeting, it is not sufficient to exof ecNOS within the proteins can be reversibly regulated, for example, by phospho- plain the agonist-dependent translocation rylation of MARCKS (6) or pp60""" (71, or by Ca2+ binding to cell (9). I t is plausible that the dynamic regulation of ecNOS membrane association in endothelial cells is importantly afrecoverin (14). involving other regions of For the ecNOS, we have previously shown that myristoyla- fected through allosteric mechanisms tion is absolutely requiredfor targeting to the particulate sub-the ecNOS protein or perhapsby reversible covalent modificacellular fraction, and our present studies show that this myr- tions of the ecNOS molecule such as phosphorylation or palmiistoylation-dependent targeting can similarly take place both toylation. in other heterologous cell expression systems and in uitro. A Acknowledgments-We are grateful to Drs. Lisa Robinson and Eva variety of biological acceptor membranes appear tobe equally Neer for helpful discussions. efficacious in binding the in uitro synthesized myristoylated ecNOS, arguingagainstthe tissue-specificexpression of a REFERENCES membrane receptor for the myristoylated ecNOS. The binding 1. Moncada, S., Palmer, R.M. J., and Higgs, E. (1991) Pharmacol. Reo. 43, 109-142 of ecNOS to biological membranes is not destroyed by boiling 2. Marletta, M. (1993) J. Biol. Chem. 268, 12231-12234 nor by trypsin treatmentof the membranes. Using gel filtration 3. Nathan, C. (1992) FASEB J. 6,3051-3064 FPLC, we also found that only the wild type but not themyr4. Busconi, L., and Michel, T. (1993)J . Biol. Chem. 268, 8410-8413 5. Sessa, W. C., Barber, C. M., and Lynch, K. R. (1993) Circ. Res. 72,921-924 mutant ecNOS is able to bind to phospholipid liposomes, docu6. Thelem, M., Rosen, A., Naim, A. C., and Aderem, A. (1991) Nature 351,320menting themyristoylation-dependent association of wild-type 322 7. Walker, F., deBlaquiere, J., and Burgess, A. W. (1993) J. Biol. Chem. 268, ecNOS withsynthetic phospholipids. Takentogether,these 19552-19558 data support the hypothesis that the myristoylated ecNOS is 8. Towler, D. A,, Gordon, J. I., Adams, S. P., and Glaser, L. (1988) Annu. Reu. targeted to membranes not by virtue of interactions with any Biochem. 57,69-99 9. Michel, T.,Li, G., and Busconi, L. (1993) Proc. Natl. Acad. Sci. U. S. A. Bo, specific myristate-protein receptor, but rather on the basis of 6252-6255 hydrophobic interactions between the myristoylated protein 10. Gordon, J. I., Duronio, R. J., Rudnick, D. A,, Adams, S . P., and Gokel, G . W. (1991) J. Biol. Chem. 266,8647-8650 with membrane phospholipids. 11. Resh, M. (1989) Cell 58,281-286 This finding does not address how these hydrophobic inter- 12. George, D. J., and Blackshear, P. J. (1992) J . B i d . Chem. 267, 24879-24885 actions mightbe modulated in the cell to promote translocation 13. Taniguchi, H., and Manenti, S. (1993) J. Bid. Chem. 268,9960-9963 of the ecNOS, but our datado suggest that the factors regulat- 14. Zozulya, S., and Stryer, L. (1992) Proc. Natl. Acad. Sci. U.S. A. 89, 1156911573 ing ecNOS membrane association are intrinsic to the ecNOS 15. Resh, M. (1994) Cell 76,411-413 itself. It is very unlikely thatecNOS translocation involves the 16. Silverman, L., and Resh, M. (1992) J. Cell Biol. 119,415-423 17. Hancock, J. F., Paterson, H., and Marshall, J. (1990) Cell 63, 133-139 demyristoylation of ecNOS, as this protein modification (which 18. Baudier, J., Deloulme. J. C., VanDorsselaer, A,, Black, D., and Matthes, H. W. occurs cotranslationally) is usually irreversible (10, 151, and D (19 1) J. Biol. Chem. 266, 229-237 our own data' do not support thedemyristoylation of the cyto- 19. Nathan, C., and Xie, Q. (1994) J. Biol. Chem. 269, 13725-13728 20. Lamas, S., Marsden, P. A., Li, G. K., Tempst, P., and Michel, T.(1992) Proc. solic ecNOS. Thus, myristoylation is necessary but not suffiNatl. Acad. Sci. U.S. A. 89, 63484352 21. Ausubel, F., et al. (eds) (1993) Current Protocols in Molecular Biology, Greene cient to explain the subcellular traffic of the ecNOS. Press, New York Manymyristoylated,palmitoylated,orisoprenylated pro- 22. Arbogast, L. Y.,Rothblat, G. H., Leslie, M. H., and Cooper, R. A. (1976) Proc. teins have secondary domains which stabilize membrane assoNatl. Acad. Sci. U. S. A. 73, 36803684 ciation, including polybasic domains, as well as additional sites 23. Scherer-Singler,U., Vincent, S. R., Kimura, H., and McGeer, E. G. (1983) J . Neurosci.-Methods 9,229-234 for protein acylation(reviewed in Ref. 15). These polybasic 24. Bredt, D. S., and Snyder, S. H. (1990) Proc. Natl. Acad. Sei. U. S. A. 87, 682-685 domain implicated in subcellular targeting maybe located at 25. Kotkow, K. J., Furie, B., and Furie, B. C. (1993) J . Biol. Chem. 268, 15633the protein's N terminus, as for src and src-related proteins 15639

L. Busconi and T. Michel, unpublished observations.

26. Mumby, S. M., Kleuss, C., and Gilman, A. G. (1994) Proc. Natl. Acad. Sei. U.S.A. 91,2800-2804