Volume 307, number

3, 287-293

Q 1993 Federtltion of Europetin Biochemical

FIZBS I 1360 Sociclics 00145793K~SS.00

August 1992

Molecular cloning and characterization of human endothelial nitric oxide synthase Philip A. Marsden”, Keith T. Schappertb, Hai Sheine Chen ‘, Micltele Flowe&, Cynthia L. Sundell’, Josiah N. Wilcox’, Santiago Lamas’l and Thomas Micheld uRelfai Disisiott arzd Deparme~tt of Medicirw. Sr. Micltuci~~HospiraL Uttiwsity oJ’ Turonro. Toronto. Ont.. Canuda. “Departrtte/rr of Genetics, Hospiral for Sick Cltildretl, Uniwrsiry qf Torotm, Tornrrm Osi., Cm~du. ‘Depurtmnt of Medicine, Emory Unhersi\y, Arkma, GA, USA and dCurdiovuscufur* Division, Briphrttrt and Wotrtertk Hoq~irrrl, Hurwrd Medical School, Bosron, MA, WSA Received 8 June 1992 The constitutive calcium/calmodulin-dependent nitric oxide (NO) syn~hasc cxprcsscd in vascular endothclium shams common biochcmjcal and pharmacologic properties with ncuronal NO synthasc. However. rcccnl cloning und molecular charaacrizalion of NO synthase from bovine endothclial cells indicated the existence of a family of constitutivo NO synlhascs. Accordingly, we undertook molecular cloning and scqucncc analysis of human cndothelial NO synthasc. Complemcruary DNA clones prcdicl a protein of 1,203 amino acids sharing 94% identity with the bovine cndothelial protein, but only 60% identity with the rut brain NO synlhasc isolbrm. Northern blot analysis with an endothclial-dcrivcdcDNA identitied a 4.6-4.8 kb mRNA transcript in IWVEC und in situ hybridization localized transcripts to vascular endothelium but not neuronal tissue. Cyclic GMP; Endothelium-derived relaxing I’lctor: Endothclium: Vascular smooth muscle; Vasodilatation

I. INTRODUCTION Synthesis of nitric oxide (NO) by NO synthase derives from one of the two chemically equivalent guanidino nitrogens of t.-arginine, producing L-citrulline as a coproduct [1,2]. In cndothelial cells and neuronal tissue, NO synthase enzymatic activity is constitutively expressed but activation of the calcium/calmodulin sig nalling pathway is required for maximal activity [3,4]. Synthesis and release of NO is rapid and not dependent upon new protein synthesis. Dynamic control at the cellular level is through activation of specific cell surface receptors by calcium-mobilizing agonists. Constitutivc NO synthase contrasts with a pathway for NO synthesis evident in macrophages [S], Kupffer cells, hepatocytes, vascular smooth muscle [6], and mesangial cells [?I. Calcium/calmodulin-independent NO synthase activity is induced in these cell types by cytokines or bacterial wall products over a period of many hours. Induction of NO synthase activity is dependent upon new RNA and protein synthesis. Biological control for this NO synthase pathway is at the level of transcriptional activation [8] and appears to be controlled by the cytokine-response profile of the particular cell type. Sequence comdbbreviutiunx HUVEC, human umbilical vein endothclial cells; kb. kilobases; PCR, polymerase chain reaction; TNF-a, tumor necrosis factor-alpha. Corresporhce oclrlrcss:P.A. Marsdcn. Rm 7360, Medical Scicnccs Building, University of Toronto, I King’s College Circle, Toronto, Onlario MSS I AS, Canada, Fax: (I) (41G) 978-8765.

parison of recently isolated molecular clones for rat brain [4] and murine macrophage NO synthase [8,9] indicated that these proteins are the products of separate genes. The calcium/calmodulin-dependent NO synthases of endothelial cells and neuronal tissue, though similarly regulated and expressing identical cofactor requirements, localize to different subcellular compartments [I 0] and vary in apparent molecular weight [l I]. To determine whether these constitutively expressed NO synthases were encoded by distinct genes we have recently purified bovine cercbellar NO synthase, determined the amino acid composition of tryptic peptides, and used a PCR-based cloning strategy to isolate cDNA probes for bovine neuronal and bovine endothelial NO synthase [12]. Genomic southern analysis using neuronal and endothelial cDNAs identified unique restriction enzyme fragments indicating that these cDNA clones reflected the products of distinct genes. Furthermore, isolation and expression of a full-length cDNA for bovine endothelial NO synthase in eukaryotic transfectants resulted in the acquisition of calcium-stimulated NO synthase activity, as well as NADPH diaphorase activity. Sequence analysis of this full-length clone predicted a protein of 1205 amino acids that differed at numerous residues from the sequence determined for the purified bovine neuronal NO synthase. Moreover, bovine endothelial NO synthase showed only 50 and 60% homology with recently identified murine macrophage and rat brain NO synthases, respectively. These data were taken to indicate that bovine endothelial NO 287

Volu~ne 307, number 3

HUVEC NOS BAEC NOS

August 1992

FEES LE-l7-ERS ,

MGNLKS

MGNLKS

E E

50 50

PGPPCGLGLG LGLGLCGKQG P PGPPCGLGLG LGLGLCGKQG P

MYRIS HUVEC NOS BAEC NOS

PEGPKF PRYKNW PEGPKF PRVKNW

S ITYDT S ITYDT

HWEC NOS BAEC NOS

148 150

PKA HUVEC Nos BAEC NOS

HWEC NOS BAEC NOS

RES ELVFGAKQAW RNAPRCVGRI QWGKLQVFD RES ELVFGAKQAW RNAPRCVGRI QWGKLQVFD

GR GDFRIhiSQL AQEMF T'fICNHTKYATNRGNERSAI TVFPQ AQEMF TYTCNHIKYA TNRGNLRSAT 'I'VFPQ GR GDFRIWNSQL

248 250

MWEC NOS BAEC NOS

298

WWEC NOS BAEC NOS

348 350

300

HU'VECNOS BAEC NOS

TGTRNLC DPHRYNTLED YAVCMDLDTR TTSSLWKDKA IGTRNLC DPHRYNTLED VAVCMDLDTR TTSSLWKDKA

HWEC NOS BAEC NOS

VPPISGSLTP VFWQ VPPISGSLTP VFHQ

HIJVECNOS BAEC NOS

VANAVKTSAS LMG VANAVKTSAS LMG

NOS

BAEC NOS

398 400

848 450

HUVEC NOS BAEC NOS

MWEC

2SY

198 200

F LSPAFRYQPD PWKGSP I LSPAFRYQPD Pp!KGSP

ET G-QSYAQQL GRLFRKAFDP ET GRAQSYAQQL GRLFRKAFDP

54% 550

PPENGESFAA ALMEMSGPYN

598 600

VL WTSTFGNGD RVLCMDEYDV VSLEM

Volume 307, number 3 HWEC NOS BAEC NOS

HWEC NOS BAEC NOS

FEBS LETTERS (SSPRPEQHKS YKIRFN9@

August 1992

SDPLVSSWRR KRKESSNTDS AGALGTLRFd

C SDPLVSSWRR KRKESSNTDS AGALGTLRF

648 650

VFGLGSRAYP HFCAFARAVD TRLEELGGER LLQLGQGDEL CGQEEAFRG

698 700

FMN HU’VEC NOS BAEC NOS

IFSPKR SWKRQRYRLS fFSPKR SWKRQRYRLS

HUVEC NOS

VEN LQSSKSTRaT ILVRLD VEN LQSSKSTRAT ILVRLD

BAEC NOS

HWEC NOS BAEC NOS

798 800

A VEQLEKGSPG 6PP A VEQLEKGSPG GPP

UJVEC NOS BAEC NOS

848 850

898 900

FLAVIN HWEC NOS BAEC NOS

948 950

FAD-I HWEC NOS EAEC NOS

HUVEC NOS BAEC NOS

LTVA VLAYRTQDGL GPLHYGVCST WLSQL LTVA VLAYRTQDGL GPLHYGVCST WLSQL

DP VPCFTRGAPS DP VPCFIRGAPS

CPLVGPGTGI APFRGFWQER LHDIESKGLQ CTLVGPGTGI APFRGFWQER LHDTESKGLQ

998 1000

1048 1050

NADPH=R HWEC NOS BAEC NOS

RCSQLDHLYR DFX RCSQLDHLYR DEV

BWEC NOS BAEC NOS

EVHRVLCLER GHNF’VCGDVTMA

G VFGRVLTAFS REP G VFGRVLTAFS REP

KTYV QDILRTELAA

1098 1100

QTVQ RILATEGDME LDEAGDVTG QTVQ RILATEGDME LDEAGDVIG

1148 1150

KTW

QDILRTELAA

NABPH-A HWEC NOS BAEC NOS

LRDQQRYHED IFGLTLRTQE VTSRIRTQSF LRDQQRYHED TFGLTLRTQE VTSRIRTQSF

SLQE SLQE

RGA VPWAFDP

1203 1205

Fig. 1. Comparison or human and bovine constitutive endothelial nitric oxide synthases. The sequences arc numbered with rcspcct to the first potential initiator methioninc residue. Deduced amino acid sequence is denoted with the single letter code. Identical amino acids are framed. A consensus scqucnce CarN-terminal myristoylation is noted (MYRIS). A potential phosphorylation site for cyclic AMP-dependent protein kinase (PKA) and a putative binding site for calmodulin (CALCIUM-CALMODULIN) is indicated. Cofactor consensus sites inciudc; iiavin mononucleotide binding (FMN), R&n adeninc dinuclcotidc pysophosphate (FAD& il:: x-iiavin electron transport (FLAVIN), FAD-isoalioxazinc (FAD-I), nicotine adcninc dinucleotidc phosphate ribose (NADPH-R) and NADPM adcninc (NADPM-A). 289

FEBS LETTERS synthass represented a novel member of a family of constitutive calcium/dependent NO synthnscs [ 121. Endothelium-dependent vasorelaxation is suggested co be impaired in important diseases of the human vascular wall; atherosclerosis, diabetes, and hypertension, among others. The ability to dissect the pathophysiologic basis of these defects at the molecular level has been hampered by the lack of molecular clones for the human constitutive endothelial NO synthase. The isolation of cDNA clones encoding the human constitutive endothelial NO synthase is reported herein. 2. MATERIALS AND METHODS 2.1‘ Mure&rls Cell culture media were from Gibco, Grand Island, NY; cell culture plates from Costar. Cambridge, MA; human recombinant tumor nu. crosis factor-alpha (rHuTNFa, spcc. act. 9.8 x 10” U/mg) was a gift of Knoll Pharmaceuticals, Whippany, NJ; DNA restriclion and modifying enzymes from Pharmaciu LKB. Baie d’lJrfe, Quebec; Taq DNA polymerase from Pcrkin.Elmcr Cctus, Emeryvillc, CA; all other reagents were us described [I 31. 2-7, cDNA clo~rrs Recombinant bacteriophage clones were isolated by plaque hybrid. ization from a random-primed HUVEC Apt I1 cDNA library with cDNA probes labellcdwith [0P]dCTP (New England Nuclear, Wilmington, DE, spcc. act. 3000 Ci/mmol) by the random-primer method. Initial cDNA clones were isolated using a 400 bp bovine endothelial cD:\IA encoding amino acids 51G to 654 (bovine mid. region probe) of bovine endothclial NO spnthasc [12], A total of 5 crass-hybridizing phage clones were isolated from 5.9 x 10” plaques screened. To isolate flanking sequences a 360 bp HUVEC cDNA encoding amino acids 1105 to 1203 and 3’ non-coding sequence (3’ probe), and a 3GObp HUVEC cDNA encoding amino acids I3 to 134 (5’ probe) of buman endothelial NO synthase were utilized, In total, IO and 15 cross-hybridizing phagc clones were isolated from screens of 5.0 x IO’ plaques, respectively. Cross-hybridizing clones were plaque-puriiied, isolated by preparative agarosc gel clecuophoresis, sub. cloned into the &:coRI multiple cloning site of pBluescript I SK(-) (Srralagene, San Diego, CA), and subjected to dideony chum-termi. nalion sequence analysis with Scquenase 2.0 (US Biochemical Corp., Cleveland. OH). Oligonucleotide primers were synthesized on a. Pharmafia LKB Gene Assembler Plus. Deduced nucleotide sequences were confirmed on both strands of multiple overlapping phage clones. 2.3. Rapid uwtph~coriott 0J.S cDN.4 end (RACE prorocotj Modifications of the RACE protocol were utilized to isolate 50 nuclcotides of 5’ coding sequence [14]. Briefly, human umbilical vein cndothelinl cells (WUVEC) total cellr+tr RNA (5 ~6) was reverse transcribed with an anti-sense gene-specitic primer (S- GCG CiGG AAC TCC AGG CCC -3’) by murinc moloney leukemia virus reverse transcriptase (Gibco BRLlLifc Technologies). RNase H (BRL) was utilized to remove mRNA [ 151.First stand cDNA productwastailed with dATP by lerminal deouynucleotidc transferase (BRL). First round PCR amplification was performed in a total volume of SO,ul for 35 cycles at 45OC (Perkin-Elmer Cetus 480 thermocycler) (sense primers S’- GAC TCG AGT CGA CGA ATT CAA T,,,, -3’, 2 pmol; 5’- GAC TCG AGT CGA CGA ATT CAA -3’, 25 pmol; anti-sense gene+pecitic primer S- AAT TTC CAG CAG CAT GTT -3, 25 pmol). A second round of amplification was performed for 35 cycles at 55°C in a lotal volume of 5Oyl (sense primer 5’- GAC TCG AGT CGC CGA ATT CAA -3’, 35 pmol; antiscnse gene-specific primer 5’. TTC CCC CAC TGG ATC CGG CCC ACG CAG -3’, 35 pmol). Products ofthc desired size were isolated with prcparadve gel clectrophonsis, digested with EcoRI/&tn~Hl, and subcloned into pBluescript

290

August 1992

I SK(-). A library of 10J RACE-PCR products. translbrmed into DHlO/J clccirocompcieni cells (DRL), were screened with in situ calm ony lift hybridization using the above mentioned human 5’ cDNA probe. Multiple positive colonies were subjected to DNA sequence analysis to exclude PCRsL;ssociatednucleotidc incorporation errors. 2.4. RNA isolurion, rtorrltcrrt blorlitlg cm/ lt_dtri&urion HUVEC were isolated and maintained as described II6]. Northern analysis was performed as previously published [ 131.A 790 bp human endothelial cDNA corresponding to amino acids 380 to 643 (human mid-region probe) was lubclled with the random~primcr method (5 x 10bcpm/ml). flybridization and washes were performed under conditions of high stringency, Blots were rehybridizcd with a human @ tubulin cDNA under conditions of high stringency (0,75 kb EcoRl insert) (ATCC 77044) to control for the amount of RNA loaded per lane. A I.0 kb EroRIIP,srl restriction fragment, corresponding to 3’ coding and non-coding regions of human cytochrome P450 reductase. was also used in Northern analysis. This cDNA was isolated from an ~eoRt/X/~oI dT-primed AZAP II human vascular smooth muscle cDNA library. 4.0 x IO5plaques were screened with ihc human endothelial 3’ cDNA described above. VCS M13 helper phage allowed in-vitro phagemid rescue following manufacturer’s suggestions (Stratngene). Tissues wcrc collected at necropsy from an adult human male and adult baboon and fixed in 4% pnraformaldchydc with 0.1 M NaPO, (pbl 7.4) for 3-4 h at 4°C cyloprotccted in 15% sucrose-PBS overnight, embedded in OCT, frozen in liquid nitrogen, and stored at -70°C. Cryosections (7-10~111)were thaw-mounted ontoVectabond_ (Vector Laboratories. Burlingumc. CA) coated-slides, refrozen, and stored at -70°C with desiccant until use. Theabovc.mcntioned human mid-region probe was transcribed with T3 RNA polymerasc (Promepa) using [I-“‘SIUTP (Amrrsham. Arlington Heights, IL, apcc. acl, 1200 CiImmol) 10 produce kI~~-knglh anti-sense transcripts. Experiments were controlled by hybridizing sections whb lhe same cDNA probe transcribed in the sense orientation with T7 RNA polymerase (Promega). Studies were performed as previously dcscribed [17], Briefly, cryosections were pretreated with paraformaldchyde, proteinasc K (Sigma Chemical, St. Louis, MO), and prehybridized in 100 ~1 hybridization buffer (50% formamide, 0,3 M N&I, 20 mM Tris-HCI pH 8.0, 5 mM EDTA, 0.02% polyvinylpyrrolidine, 0.02% licoll, 0.02% bovine serum albumin, 10%dextran sulfale, and 10 mM dithiothreitol) at 42OC.Serial sections were hybridized with 6.0 x 10’ cpm ~‘5S]riboprobe/slidcat 55OC. After hybridization, the sections wcrc washed with 2 xSSC (I x SSC = 150 mM NaCI, 15mM Nacitrate. pH 7.0) with 10 mMP-mercaptoethanol and I mM EDTA, lreutcd will1 RNasc A (Sigma), again washed in the same buffer, followed by a high slringency wash in 0.1 x SSC with IO mM pa mercaptoeihanol and 1 mM EDTA, at 55°C. Slides were then washed in 0.5 x SSC and dehydrated in graded alcohols containing 0.3 M NH,Ac. Sections were dried, coaled with NTB2 nuclear track ernul~ sion (Eastman Kodak, Rochcstcr. NY), and exposed in the dark at 4°C for 4 to IO w’ecks.Alier development. the sections wcrc counter. stained with humatoxylin and eosin to aid in cell identilication,

3. RESULTS AND DlSCUSSlON The nucleotide sequence(s) reported in this paper has been submitted to the GenBank/EMBL Data Bank with accession number(s) M95296. Nucleotide coding sequences are 90% identical comparing human and bovine endothelial constitutive NO synthases. The methionine-encoding sequences near the 5’ end of the open reading frame have neighbouring nucleotide sequences consistent with eukaryotic translational start

Volum

FEBS LETTERS

307. number 3

I

2

ENDOTHELIAL NC)SYNTHASE

I,,, #gQ&k,

P450 CYTOCHROME REDUCTASE

: .-‘: 4!!mlww ,: .~.,>I ,,.’.-:

3

: -

28s

- 28s

Fig. 2. Northern blot annlysis of constitutivc human cndothclial nitric cfkct of TNF-a. In the upper panel total cellular RNA (IO pgllane) isolated from HUVEC; control (lunc I), TNF-a (100 ndml, 4 h) (lane 2). TNF-1: (100 ngml. 24 h) (lane 3). Nylon membranes were probed with a ‘zP-labellcd constitutivc human endothelial NO synthase cDNA. In the lower panel membranes wcrc reprobed with a >‘P-lubclled human cytochromc P4SOreductasc cDNA.

oxide synthase;

sites AGTAAC(ATG)G [18]. The deduced amino acid sequence of human endoehelial NO synthase, comprising 1203 amino acids, predicts a protein with a mw of 133 kDa. This is in close agreement with lhe predicted molecular mass of bovine endothelial NO synthase (1205 amino acids) [ 121 as well as the molecular weight of 135 kDa suggested from purification of the enzyme from bovine endothelial cells [l I]. Indicated in Fig. I is an amino acid sequence comparison of the human and bovine endothelial NO synthases. The sequence is numbered with respect to the first potential initiator methionine codon. Cellular fractionation has indicated that endothelial NO synthase may be localized to the particulate fraction ofendothelial cell preparations [IO]. This contrasts with the cytosolic localization of constitutive neuronal and inducible macrophage NO synthases. A consensus motif for N-terminal myristoylation is noted in both human and bovine endothelial sequences [19], Such a motif is not present in neuronal and macrophage N termini. Binding sites for cofactors and nucleotides are highly conserved between human and bovine endothelial NO synthases, indeed overall amino acid sequence identity is 94%. This contrasts with a predicted amino acid identity of 50 and GO%with murine macrophage and rat neuronal NO synthase, respectively. Constitu-

August 1992

tive human endothelial NO synthase reveals no significant sequence similarity over the initial 226 residues of neuronal NO synthase [4]. Northern blot analysis (Fig. 2, upper panel) indicated the most prominent mRNA transcript to be 4.6 to 4.8 kb in size. We and others have previously demonstrated that proinflammatory cytokines, such as TNF-a, induce a pathway for NO synthesis in endothelial cells [20,21]. Addition of human recombinant TNF-a (100 @ml) for 4 and 24 h failed to increase levels of endothelial NO synthase mRNA transcripts. l[nfact, such treatment dccreased levels of constitutive cndothelial NO synthase mRNA when assessed at 24 h (n = 3). NO synthases show a high degree of homology to cytochrome P45O reductase [4]. This is especially evident at the CORM termini of the molecules, regions of putative nucleotide and cofactor interactions. A human cytochrome P450 reductase cDNA was utilized to control for the amount of RNA loaded per lane (Fig. 2, lower panel). Similar results were obtained when blots were reprobed with a human P-tubulin cDNA (results not shown). Therefore, the increase in endothelial NO synthase activity induced by prolonged treatment with TN&LXis likely to involve post-transcriptional modification of NO synthase. induction of co-regulatory processes, or the induction of a distinct endothelial NO synthase isoform. Anti-sense cRNA probes were utilized under conditions of high stringency for in situ hybridization studies. These studies were performed to assess the cellular expression of endothelial constitutive NO synthase. Shown in Fig. 3A is localization of NO synthase mRNA to vascular endothelium within a human intercostal artery. In Fig. 3B constitutive endothelial NO synthase mRNA was also localized in representative sections of baboon cerebellum solely to endothelial cells of blood vessels and not to neurons. In contrast to findings obtained with anti-sense riboprobes no hybridization signal was detected using a sense riboprobe (data not shown). Endothelial cell identity was confirmed by UIcs europucus lectin immunohistochemistry [ 171and in-situ hybridization using a von Willcbrand factor cRNA probe on adjacent serial sections [22]. Sucl9 studies demonstrated specific hybridization to endothelial NO synthase mRNA and not neuronal NO synthase mRNA. There are striking areas of regional amino acid homology among the NO synthase family. Functional assessment of putative regulatory and functional domains are under active investigation. For instance, whether N-terminal myristoylation of endothelial NO synthase targets the protein to membranes remains to be determined. Functional expression of a full-length cDNA for human endothelial NO synthase will be an important component of such studies. Bovine endolhelial NO synthase cDNAs clearly confer calcium-activated NO synthase activity in a hetcrologous system [ 121.We believe that the marked sequence identity as well as equivalent size and tissue distribution of mRNA transcripts, all 291

Volume 307. number

3

FEBSLETTERS

August 1992

Fig. 3. Rcprescnlative sections of in situ localization ofconstitutive endothclial NO synlhase mRNA within n human intercostal artery (A) and baboon cerebellum (B). (A) NO synthasc mRNA was detected in the cndothcljal cells of a human intercostal artery by in situ hybridization using an endothclinl NO synthase [a-J5S]UTP-labelled cRNA probe. (B) NO synthasc expression was localized within baboon cerebellum solely to cndothelial cells of blood vessels and not neuronal tissue. Similar results wcrc obtained on serial sections hybridized to a human von Willcbrand factor [a-“S]UTP-labelled cRNA probe. Serial sections hybridized with ;Icorresponding sense human cndothulial NO synthase [a-35SIUTP-labclled cRNA probe did not show hybridization. Autoradiographs were photngraphcd using a combination of bright field illumination and polarized epi-illumination. A. 1250x; B. 416%

suggest that these cDNA clones encode the human and bovine constitutive endothelial isoforms. Indeed, the current study suggests that isoforms may be highly conserved across species. The isolation of human sequences may now allow direct assessment of endothclial NO synthase mRNA expression in important vascular dis292

eases of man and permit studies of genomic organization and transcriptional regulation. The c%pCrttCChniCal USSiStEnCe Or Wei Ch.?tIg Lu is gratcrully iicknowlcdged. We thank S.H. Orkin for the HUVEC /2gtl I CDNA library. M.F. is ;I Medical Research Council or Canada/ ICI ..I“k/lrJ,~l~,l~o,/l~,/i,S!

Volume 307, number 3

FEBS LETTERS

Pharma Fellowship rccipicnt. J.W. was supported by a grant from lhc Development Fund ofthc Emory University Dcpartmcnt orfvlcdicinc. T.M. is a recipient ofa Clinician Scientist Award from the American Heart Association and is supprted by a Grant-in-Aid Born the Amcrican Heart Association National Centre and NIH Grant HL 46457. P.M. isa recipient ofa Kidney Foundation ofCanada Scholarship and is supported by the Kidney Foundation of Canada and the J.P. Bickcll Foundation.

REFERENCES 111Tgnarro, L.J. (1990) Annu. Rev. Phurmucol. Toxicol. 30, 535SGO.

PI Moncada, S., Palmer, R.M.J. and Higgs, E.A. (1991) Pharm.

Rev. 43, 109-142. 131 Bredt, D.S. and Snyder, S.H. (1990) Proc. Natl. Acad. Sci. USA 87, 682-685. [41 Bredt, D.S., Hwang. P,M.. Glatt, C.E.. Lowcnstein. C.. Reed, R.R. and Snyder, S.H. (1991) Nature 351, 714-715, 151Stuchr, D.J. and Nathan, CF. (1989) J. Exp. Med. 169. 15431555. [61 Beasley, D., Schwartz. 1.H. and Brenner, B.M. (!991) J. Clin. Invest. 87, 602-608. I71Marsdcn, P.A. and BaIlermann. B.J. (1990) 1. Exp, Med. 172, 1843-1852. [B] Xic, Q., Cho, H.J., Caluycay. J.. Mumrord. R,A., Swidcrck. KM., Lee, T.D., Ding, A., Troso, T. and Nathan, C. (1992) Science 256. 225-228.

Augusl1992

191 Lyons. CR., Orloff, G.J. and Cunningllam, J.M. (1992) J. Biol. Chem. 267.6 370-374. [IO] i3ojc. K.M. and Fung. H. (1990) J. Pharmacol. Exp. Ther. 253. 20-26. [l I] Forstcrmtinn. U., Pollock. J.S.. Schmidt, H.H.H.W., Hellcr. M. and Murnd, F. (1992) Proc. Natl. Acad. &i. USA 88,1788-1792. [I?] Lamas, S., Marsden, P.A.. Li, G.K., Tcmpst, P. and Michcl, T. (1992) Proc. Natl. Acad, Sri. USA, 89.63486352. [I31 Marscn. P.A. and Brcnner. B.M. (1992) Am. J. Physiol. (Cell) 262, CBS4-CB61. [I41 Frohman, M-A.. Dush, M.K. and Martin, G.R. (1988) Proc. Natl. And. Sci. USA 85, 8998-9002. [I51 Jain. R.. Gomcr. R,M. and Murtagb, J.J. (1992) BioTcchniques 12, 55-59. 1161Kourembanas, S., Marsdcn, P.A., McQuillan, L.P. and Failer, D.V. II9911 .I. Clin. Invest. 88. 1054-1057. [I 71 Wildx, J.N., Smith. K.M., Williams, L.T.. Schwartz, SM. and Gordon. D. (19831J. Clin. Invest. 82. 1134-l 143. 1181Kourk. M. ti99l)‘l. Cell. Biol. 115, a87-903. [I91 Magee. T. and Hanley. M. (1988) Nature 335. 114-115. [20] Lamas, S., Michel. T., Brcnner, B.M. and Marsden, P.A. (1991) Am. J. Physiol. (Ccl1 Phyaiol.) 261, C634-0341. [21] Radomski, M.W., Palmer, R.M.J. and Mnnwdd, S. (1990) Proc. Natl. had. Sci. USA 87, 10043-10047. [22] Shelton-lnloes, B.B.. Chcbah, F.F., Mannucci, P.M.. Fedcrici, A.B. and Sadler, J.E. (1987) J. Clin. Invest. 79, 1459-1465.

293