Extreme hydrops fetalis and cardiovascularabnormalities in mice lacking a functionalAdrenomedullin geneKathleen M. Caron and Oliver Smithies*

Department of Pathology and Laboratory Medicine, University of North Carolina, Chapel Hill, NC 27599-7525

Contributed by Oliver Smithies, November 20, 2000

Adrenomedullin, a recently identified potent vasodilator, is ex-pressed widely and has been suggested to have functions rangingfrom reproduction to blood pressure regulation. To elucidate thesefunctions and define more precisely sites of Adm expression, wereplaced the coding region of the Adm gene in mice with asequence encoding enhanced green fluorescent protein whileleaving the Adm promoter intact. We find that Adm2/2 embryosdie at midgestation with extreme hydrops fetalis and cardiovas-cular abnormalities, including overdeveloped ventricular trabecu-lae and underdeveloped arterial walls. These data suggest thatgenetically determined absence of Adm may be one cause ofnonimmune hydrops fetalis in humans.

Adrenomedullin (Adm), a potent 52-aa peptide vasodilator,was isolated from a human pheochromocytoma on the

basis of its ability to elevate platelet cAMP levels (1). Recently,Adm has gained much attention in clinical settings because ofits elevated plasma levels in patients with various cardiovas-cular diseases (2). However, the variety of conditions associ-ated with elevated Adm levels, which include essential hyper-tension, septic shock, and normal pregnancy, suggest thatincreases in Adm are secondary to other primary events ratherthan causative. For example, Adm may be elevated in certainhypertensive disease states as a compensatory mechanism todecrease blood pressure, but genetic experiments to distin-guish between causation and compensation have not yet beenperformed.

On the basis of its nucleotide homologies with calcitoningene-related peptide (CGRP), calcitonin, and amylin, Adm isconsidered to be the fourth CGRP family member. The Admgene contains four exons spanning approximately 2 kb andcodes for two biologically active peptides: proadrenomedullinN-terminal 20 peptide (PAMP) and Adm. Recent studies toidentify the receptors for Adm and CGRP have led to thediscovery of a mechanism of signal transduction whereby aseries of single transmembrane spanning receptor activity-modifying proteins (RAMPs) interact with a common receptorto impart specificity of binding for either Adm or CGRP (3).The differential expression of the receptor activity-modifyingproteins may dictate the sites of function of the two ligands.

To explore the physiological functions of Adm in an in vivomodel, we used homologous recombination in embryonic stemcells to generate mice that are deficient for the coding regionof the Adm gene. At the same time, to provide a biologicalmarker for sites of Adm gene expression, we replaced the Admcoding region with a sequence coding for enhanced greenf luorescent protein (EGFP) while leaving the Adm promoterand Adm 59 untranslated region intact. Here we show that Admis essential for survival because Adm2/2 embryos die atmidgestation with extreme hydrops fetalis. In addition,Adm2/2 embryos have unusual cardiovascular defects that arebest characterized as overdeveloped ventricular trabeculaeand underdeveloped vascular smooth muscle in the largearteries. Because the combination of cardiovascular defectsand the severity of hydrops fetalis are, to our knowledge,

unlike any previously described pathology in the mouse, ourdata demonstrate a unique role for Adm in development andsuggest that the absence of Adm may be one cause of nonim-mune hydrops fetalis in humans.

Materials and MethodsGene Disruption by Homologous Recombination. Mice lacking Admwere generated with standard gene targeting methods (4). Thetargeting construct was made (Fig. 1a) with (i) a 3.7-kbBamHIySpeI genomic fragment that includes the Adm pro-moter; (ii) a 600-bp PCR-generated fragment that begins 59with the endogenous SpeI site, contains exon 1, intron 1, partof exon 2, and ends 39 at the endogenous initiator methionine;(iii) a 700-bp cDNA encoding EGFP (CLONTECH); (iv) a300-bp bovine growth hormone poly(A) addition region; and(v) a 1.3-kb XhoIySacI genomic fragment from downstream ofthe Adm gene. 129-derived SvEv embryonic stem cells (TC-1,a gift from Phil Leder, Harvard University, Boston, MA) wereelectroporated; G418yganciclovir-resistant colonies, initiallyidentified by the PCR-based assay depicted in Fig. 1 a and c,were expanded; and homologous recombination was con-firmed by Southern blot analysis with the use of an 800-bpSacIyAvr II probe external to the targeting construct sequence(Fig. 1 a and b). Male chimeras were mated to wild-type SvEvfemales to establish an isogenic Adm line, and all experimentswere conducted on the resulting isogenic line.

To generate Adm2/2 embryos, timed heterozygous matingswere set up, and the morning of vaginal plug detection wasconsidered to be embryonic day 0.5 (E0.5). To genotype theembryos, DNA was isolated from yolk sac tissue (E8.5–E12.5) orfrom a small piece of tail tissue (E12.5–E14.5) and analyzed bythe PCR-based strategy described above.

Expression Analysis by Using Real-Time Quantitative Reverse Tran-scription–PCR. Expression of the Adm gene was characterized byreal-time quantitative reverse transcription–PCR with the Se-quence detection system (Perkin–Elmer). Primers for Admamplification were 59-GAGCGAAGCCCACATTCGT-39 and59-GAAGCGGCATCCATTGCT-39. The probe for Adm detec-tion was 59-FAM-CTACCGCCAGAGCATGAACCAGGG-Tamra-39. All reactions included a b-actin internal standard. Theprimers used for b-actin amplification were 59-CTGCCTG-ACGGCCAAGTC-39 and 59-CAAGAAGGAAGGCTG-GAAAAGA-39. The probe for b-actin detection was 59-TET-

Abbreviations: Adm, Adrenomedullin; EGFP, enhanced green fluorescent protein; CGRP,calcitonin gene-related peptide; En, embryonic day n; H&E, hematoxylin/eosin.

*To whom reprint requests should be addressed at: Department of Pathology and Labo-ratory Medicine, 701 Brinkhous–Bullitt Building, CB #7525, University of North Carolina,Chapel Hill, NC 27599-7525. E-mail: jhlynch@med.unc.edu.

The publication costs of this article were defrayed in part by page charge payment. Thisarticle must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.§1734 solely to indicate this fact.

Article published online before print: Proc. Natl. Acad. Sci. USA, 10.1073ypnas.021548898.Article and publication date are at www.pnas.orgycgiydoiy10.1073ypnas.021548898

PNAS u January 16, 2001 u vol. 98 u no. 2 u 615–619

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

17,

202

0

CACTATTGGCAACGAGCGGTTCCG-Tamra-39. RNA wasisolated from E9.5 embryos with the use of the TRIzol Reagent(GIBCOyBRL). The reactions were performed with 5 mg totalRNA with minor differences from ABI 7700 manufacturer’sinstructions. Relative levels of Adm expression were determinedby the DDCt method and expressed as a percentage of wild type(Fig. 1d). All assays were repeated twice, each with duplicates.

Expression Analysis by Using GFP Fluorescence. Adult or embryonicheterozygote animals were euthanized, and organs or wholeembryos were fixed in 4% paraformaldehyde. Tissues wereembedded in Tissue-Tek OCT media (Sakura Finetek, Torrance,CA), and 8-mm frozen sections were analyzed under fluorescentmicroscopy. The slides were then stained with hematoxylin/eosin

(H&E) to histologically confirm sites of expression. To visualizeAdm expression during development, heterozygote embryoswere fixed in 4% paraformaldehyde and analyzed by fluorescentconfocal microscopy.

Histological Analysis. Embryos, dissected from the uterus at thedesired gestational times, were fixed in 4% paraformaldehyde,dehydrated, and embedded in paraffin. Five-micrometer-thicksections were mounted on slides and stained with H&E.

ResultsTargeted disruption of the Adm gene replaced the protein-coding region of the gene with a cDNA coding for EGFP but leftthe 59 regions of the gene intact, so that EGFP expression is Admspecific (Fig. 1a). The disrupted allele was detected by Southernblot analysis (Fig. 1b) and by PCR (Fig. 1c). To confirm that thetargeting inactivated the Adm gene, quantitative reverse tran-scription–PCR was performed on embryonic RNA, with the useof an Adm TaqMan probe and the ABI7700 Sequence Detectionsystem. Adm2/2 embryos express Adm mRNA at levels indis-tinguishable from zero, thus confirming complete loss of Admproduction; heterozygotes have half-normal expression (Fig. 1d).

Mice heterozygous for Adm showed no overt differences fromwild-type mice and were used to determine the pattern of Admexpression by observing fluorescence from its surrogate, EGFP.As expected from previous studies (5, 6), the adult adrenalmedulla showed strong fluorescence, but no fluorescence wasdetected in the zona fasciculatayreticularis of the adrenal cortex(Fig. 2a). Confocal microscopy of E9.5 Adm1/2 embryos showedmoderate expression in the heart and strong expression in thedeveloping vasculature (Fig. 2b). E13.5 Adm1/2 embryos expressAdm in the cardiac myocytes of the developing ventricle (Fig.2c). The vascular smooth muscle cells of the aorta and esophagusand the endothelial cells of the aorta and thoracic duct stronglyexpress Adm (Fig. 2d).

Fig. 1. Generation of Adm2/2 animals by homologous recombination. (a)Strategy to disrupt the Adm gene. (Top) Endogenous wild-type (WT) allele.(Middle) Linearized targeting vector. (Bottom) Targeted allele after homol-ogous recombination. Open boxes represent exons 1–4; the checkered regionin exon 4 codes for the mature Adm peptide; the initiator methionine of thepropeptide is indicated (Met). The solid black box depicts a bovine growthhormone poly(A) addition sequence downstream of the EGFP cDNA. Thelocations of primer sequences for PCR (p1, p2, and p3) are shown by smallarrows. The sequence used as a probe for the Southern-based detectionstrategy is indicated by a labeled line (PROBE). Restriction sites: A, AvrII; B,BamHI; Sc, SacI; Sp, SpeI; X, XhoI. Parentheses denote sites destroyed duringcloning. (b) Detection of the targeted allele by Southern blot analysis.Genomic DNA from E9.5 embryos was digested with AvrII and probed with theSacIyAvrII fragment (PROBE) depicted in the bottom line of a. (c) Detection ofthe targeted allele by PCR. Genomic DNA from embryos was amplified withthe primer sequences depicted in a. (d) Measurement of Adm RNA in total RNAextracts from E9.5 embryos by real-time quantitative reverse transcription–PCR. The relative quantity of Adm RNA in Adm1/2 and Adm2/2 embryos isexpressed as a percentage of total Adm RNA in Adm1/1 embryos. Error barsrepresent SEM.

Fig. 2. Sites of Adm expression determined by EGFP fluorescence. (a) Frozensection of adrenal gland from an adult Adm1/2. Note intense fluorescence inadrenal medulla (m) and the absence of fluorescence in adrenal cortex (c). (b)Confocal microscopy of an E9.5 Adm1/2 embryo. Moderate expression is seenin the heart (arrowhead h), and strong expression in the developing vascula-ture (arrowhead v) and forebrain (arrowhead b); a speck of nonspecificfluorescence has been covered with a black box. (c) Frozen section of an E13.5Adm1/2 embryo, showing expression in the developing left ventricle. Arrow-heads indicate the pericardial surface of the left ventricle. (d) Frozen sectionof an E16.5 Adm1/2 embryo, showing expression in the aorta (a), esophagus(oe), and thoracic duct (td).

616 u www.pnas.org Caron and Smithies

Dow

nloa

ded

by g

uest

on

May

17,

202

0

Mice homozygous for the targeted allele die in utero, asshown by our finding that among the progeny of Adm1/2 3Adm1/2 matings 32% were Adm1/1, 68% were Adm1/2, butnone were Adm2/2. Timed matings between Adm1/2 miceshowed that at E12.5 the Adm2/2 embryos are indistinguish-able from their Adm1/2 and Adm1/1 littermates. However, theAdm2/2 embryos die between E13.5 and E14.5, as evidencedby increased numbers of resorbing Adm2/2 embryos (Table 1).Early in this period, the yolk sac cavity becomes distended(Fig. 3a), and Adm2/2 embryos exhibit modest general edemawith a notable increase in the size of their thoracic cavitiesrelative to control littermates (Fig. 3b). By E14.5, Adm2/2

embryos suffer from extreme hydrops fetalis (Fig. 3c) thatencompasses the entire body and neural structures (Fig. 3d).Note that the yolk sacs have been removed from the embryosillustrated in Fig. 3 b–d. Histological examination of theAdm2/2 embryos (Fig. 3f ) confirms the enlargement of theirthoracic cavities and their extreme and generalized hydropscompared with control littermates (Fig. 3e).

Edema in mice embryos has previously been described inassociation with cardiovascular defects, although we find noreports of edema of this severity (7, 8). We therefore examinedthe condition of the heart in transverse sections of E13.5 andE14.5 Adm2/2 embryos. Fig. 4 illustrates that the E13.5 Adm2/2

heart (Fig. 4d) is typically only two-thirds the size of thewild-type heart (Fig. 4a) but shows signs of increased leftventricular trabecular development leading to smaller chambersize. By E14.5, this pathological overdevelopment becomes moremarked (Fig. 4 b and e): the compact zone appears thin andconvoluted, the number of trabeculae is increased, and themyocardium has a generally disorganized appearance (Fig. 4 cand f ). The mitral and tricuspid valves, intraventricular septum,and atria of Adm2/2 embryos show no obvious abnormalities atany time.

The aorta and carotid artery of the Adm2/2 embryos are alsoabnormal (Fig. 5 c and d); they have markedly thinner wallsand are irregular in shape when compared with wild type (Fig.5 a and b). However, we observed no significant differences inthe external appearance of the smaller vessels of the paws andtail (Fig. 2 b and d). The thinner major arteries and the cardiacpathology of the Adm2/2 embryos demonstrate that cardio-vascular development is abnormal in the absence of Adm.

DiscussionBecause our Adm2/2 animals express EGFP in place of Adm,we comment on a recent report that extremely high levels ofa GFP transgene expressed in the heart of adult mice causesdilated cardiomyopathy (9). For several reasons, it is highlyunlikely that the abnormal trabecular development in theventricles of our Adm2/2 embryos is caused by EGFP toxicity.First, the level of EGFP reported to be toxic appears muchgreater than in our mice, as judged by f luorescence. Second,our phenotype manifests itself over 24–48 h during embryonicdevelopment, as opposed to a several-month progression in theadult transgenic mice. Third, other investigators have previ-ously shown that GFP is nontoxic in mice (10–12), and thatmoderately expressed cardiac-specific GFP transgenes donot cause adverse effects during embryogenesis or adulthood(13, 14).

Nonimmune hydrops fetalis has an incidence of 1:3,000 birthsand a high fetal mortality rate. One-quarter of these cases areassociated with cardiovascular defects such as arrhythmias andmalformations. Fifteen percent are of unknown etiology. Here

Table 1. Genotype of embryos from Adm1y2 3 Adm1y2 matings

Age

Genotypes of embryos

Total1y1 1y2 2y2

E9.5 37 53 23 113E10.5 11 15 6 32E12.5 6 9 5 20E13.5 11 14 7 32E14.5 12 30 18* 60E17.5 4 9 1 14P10 47 101 0 148

*Dying or resorbed embryos.

Fig. 3. Adm2/2 embryos have massive generalized edema. (a) Appearance of E14.5 Adm2/2 (Left) and Adm1/1 (Right) embryos in their yolk sacs shortly afterdissection from the uterus. Note that the yolk sac (arrowhead ys) of the Adm2/2 embryo is distended with fluid, and its blood vessels are thinner than those ofits wild-type littermate. (b) The thoracic cavity of E13.5 Adm2/2 embryos is considerably enlarged (Left) compared with a wild-type littermate (Right). The blacklines indicate the width of the thoracic cavity just above the diaphragm. (c) Severe hydrops fetalis is apparent in the E14.5 Adm2/2 embryos (Left), readily visibleafter their yolk sacs are removed. The arrowheads (s) indicate the outer skin layer of the embryos. (d) The extreme hydrops of the Adm2/2 embryos, dissectedaway from their yolk sacs, is completely general, as indicated by the uniformly swollen back and head. The arrowheads (s) again indicate the outer skin layer.(e and f ) H&E stain of transverse sections through E14.5 Adm1/1 (e) and Adm2/2 ( f) embryos. The Adm2/2 embryo has a fluid-filled thoracic cavity, and the tissuesexternal to the rib cage are markedly swollen (e and f, 31).

Caron and Smithies PNAS u January 16, 2001 u vol. 98 u no. 2 u 617

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

17,

202

0

we describe a type of nonimmune hydrops fetalis that is asso-ciated with cardiovascular defects and is caused by a single genedefect. Cardiovascular defects have been reported in mice havinga wide variety of disrupted genes [RXR a (8, 15), a 4 integrin(16), PDGF a (17), BARK 1 (18), Hand-1 (19, 20), PPAR g (21),calreticulin (22), jumonji (23), etc.] likely reflecting the manyways in which cardiac anomalies can be induced. However, theconsequences of these gene deficiencies differ in two fundamen-tal ways from the Adm2/2 phenotype. First, any observed edematends to be mild, localized, and subcutaneous. Second, the heart

defects are generally characterized as involving hypoplasia in-dicative of impaired myocardial growth and subsequent ventric-ular dilation. In contrast, Adm2/2 hearts show evidence oftrabecular hyperplasia that can progress sufficiently to causeventricular occlusion. Adm, secreted by neonatal rat cardiomy-ocytes, inhibits protein synthesis and collagen production whenit is added to cultured cardiomyocytes (24) or cardiac fibroblasts(25). In line with these findings, our observations suggest thatabsence of Adm in the developing heart leads to abnormallyenhanced growth. However, in apparent contrast, our studiesalso show that the absence of Adm results in less than normalvascular smooth muscle cell development in large arteries, whichis unlikely to be secondary to low blood pressure caused by apoorly functioning heart because it does not occur in other micewith cardiac failure.

The extreme hydrops of Adm2/2 embryos requires comment. Forthe above reasons, it is unlikely to be solely because of cardiac orvascular defects. Likewise, it is unlikely that abnormal vascularpermeability could account for the hydrops, because gene disrup-tion models with vascular permeability defects do not show thesame degree of edema (26–28). Abnormalities in lymphatic vesseldevelopment could theoretically cause the severe edema, andlymphatic vessels sprout from veins around E12.5 in mice, which isthe time of the hydrops onset in our mutants. Furthermore, we findthat Adm is expressed in the thoracic duct of Adm1/2 embryos (Fig.2d). However, current data are not yet sufficient to allow a chain ofevents to be specified that leads to the hydrops of Adm2/2 embryos.Nevertheless, our overall data suggest that a genetically determinedabsence of Adm could be one cause of nonimmune hydrops fetalisin humans.

We thank J. Cross, N. Freedman, N. Maeda, H. Rockman, W. Samson,K. Sulik, and members of the Smithies and Maeda laboratories forhelpful advice and discussions, and R. Bagnell, H. S. Kim, N. Shaw, andX. Chen for assistance with our experiments. This work was supportedby grants from the National Institutes of Health to O.S. (GM20069 andHL49277) and to K.M.C. (HL10344).

Fig. 4. Adm2/2 embryos have developmental heart defects. Transverse sections through the hearts of Adm1/1 (Upper) and Adm2/2 (Lower) embryos werestained with H&E. (a) E13.5 Adm1/1 littermate. (d) E13.5 Adm2/2 embryo. The heart of the Adm2/2 embryo is smaller than the heart of its wild-type littermate.(b) E14.5 Adm1/1 littermate. (e) E14.5 Adm2/2 embryo. The heart of the Adm2/2 embryo is still smaller, and the left ventricle is significantly occluded withincreased trabeculae compared with its Adm1/1 littermate. (c) E14.5 Adm1/1 littermate. ( f) E14.5 Adm2/2 embryo. Higher-power magnification of the leftventricle at E14.5 shows an increase in the number of trabeculae, a thinner, convoluted compact zone (cz), and a generally delaminated appearance of theAdm2/2 heart compared with wild type. c, endocardial cushion; s, septum; t, tricuspid valve; m, mitral valve; rv, right ventricle; lv, left ventricle; ch, chamber; cz,compact zone (a, b, d, and e, 32; c and f, 310).

Fig. 5. Adm2/2 embryos have thin arterial walls. Transverse sectionsthrough vessels of Adm1/1 (Upper) and Adm2/2 (Lower) embryos at E13.5were stained with H&E. (a and c) aorta; (b and d) carotid artery. Note thethin vascular walls (approximately three cells thick) of the Adm2/2 vesselscompared with wild-type vessels (approximately five cells thick). a, aorta;c, carotid artery (310).

618 u www.pnas.org Caron and Smithies

Dow

nloa

ded

by g

uest

on

May

17,

202

0

1. Sakata, J., Shimokubo, T., Kitamura, K., Nakamura, S., Kangawa, K., Matsuo,H. & Eto, T. (1993) Biochem. Biophys. Res. Commun. 195, 921–927.

2. Hinson, J. P., Kapas, S. & Smith, D. M. (2000) Endocr. Rev. 21, 138–167.3. McLatchie, L. M., Fraser, N. J., Main, M. J., Wise, A., Brown, J., Thompson,

N., Solari, R., Lee, M. G. & Foord, S. M. (1998) Nature (London) 393, 333–339.4. Koller, B. H., Hagemann, L. J., Doetschman, T., Hagaman, J. R., Huang, S.,

Williams, P. J., First, N. L., Maeda, N. & Smithies, O. (1989) Proc. Natl. Acad.Sci. USA 86, 8927–8931.

5. Kapas, S., Martinez, A., Cuttitta, F. & Hinson, J. P. (1998) J. Endocrinol. 156,477–484.

6. Satoh, F., Takahashi, K., Murakami, O., Totsune, K., Sone, M., Ohneda, M.,Sasano, H. & Mouri, T. (1996) Neurosci. Lett. 203, 207–210.

7. Paszty, C., Mohandas, N., Stevens, M. E., Loring, J. F., Liebhaber, S. A., Brion,C. M. & Rubin, E. M. (1995) Nat. Genet. 11, 33–39.

8. Sucov, H. M., Dyson, E., Gumeringer, C. L., Price, J., Chien, K. R. & Evans,R. M. (1994) Genes Dev. 8, 1007–1018.

9. Huang, W. Y., Aramburu, J., Douglas, P. S. & Izumo, S. (2000) Nat. Med. 6,482–483.

10. Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T. & Nishimune, Y. (1997)FEBS Lett. 407, 313–319.

11. Godwin, A. R., Stadler, H. S., Nakamura, K. & Capecchi, M. R. (1998) Proc.Natl. Acad. Sci. USA 95, 13042–13047.

12. Barrow, J. R., Stadler, H. S. & Capecchi, M. R. (2000) Development (Cam-bridge, UK) 127, 933–944.

13. Fleischmann, M., Bloch, W., Kolossov, E., Andressen, C., Muller, M., Brem, G.,Hescheler, J., Addicks, K. & Fleischmann, B. K. (1998) FEBS Lett. 440,370–376.

14. Xian, M., Honbo, N., Zhang, J., Liew, C. C., Karliner, J. S. & Lau, Y. F. (1999)J. Mol. Cell. Cardiol. 31, 2155–2165.

15. Sapin, V., Dolle, P., Hindelang, C., Kastner, P. & Chambon, P. (1997) Dev. Biol.191, 29–41.

16. Yang, J. T., Rayburn, H. & Hynes, R. O. (1995) Development (Cambridge, UK)121, 549–560.

17. Schatteman, G. C., Motley, S. T., Effmann, E. L. & Bowen-Pope, D. F. (1995)Teratology 51, 351–366.

18. Jaber, M., Koch, W. J., Rockman, H., Smith, B., Bond, R. A., Sulik, K. K., Ross,J., Jr., Lefkowitz, R. J., Caron, M. G. & Giros, B. (1996) Proc. Natl. Acad. Sci.USA 93, 12974–12979.

19. Riley, P., Anson-Cartwright, L. & Cross, J. C. (1998) Nat. Genet. 18, 271–275.20. Firulli, A. B., McFadden, D. G., Lin, Q., Srivastava, D. & Olson, E. N. (1998)

Nat. Genet. 18, 266–270.21. Barak, Y., Nelson, M. C., Ong, E. S., Jones, Y. Z., Ruiz-Lozano, P., Chien,

K. R., Koder, A. & Evans, R. M. (1999) Mol. Cell. 4, 585–595.22. Mesaeli, N., Nakamura, K., Zvaritch, E., Dickie, P., Dziak, E., Krause, K. H.,

Opas, M., MacLennan, D. H. & Michalak, M. (1999) J. Cell Biol. 144, 857–868.23. Lee, Y., Song, A. J., Baker, R., Micales, B., Conway, S. J. & Lyons, G. E. (2000)

Circ. Res. 86, 932–938.24. Tsuruda, T., Kato, J., Kitamura, K., Kuwasako, K., Imamura, T., Koiwaya, Y.,

Tsuji, T., Kangawa, K. & Eto, T. (1998) Hypertension 31, 505–510.25. Horio, T., Nishikimi, T., Yoshihara, F., Matsuo, H., Takishita, S. & Kangawa,

K. (1999) Eur. J. Pharmacol. 382, 1–9.26. Puri, M. C., Rossant, J., Alitalo, K., Bernstein, A. & Partanen, J. (1995) EMBO

J. 14, 5884–5891.27. Suri, C., Jones, P. F., Patan, S., Bartunkova, S., Maisonpierre, P. C., Davis, S.,

Sato, T. N. & Yancopoulos, G. D. (1996) Cell 87, 1171–1180.28. Kim, K., Drummond, I., Ibraghimov-Beskrovnaya, O., Klinger, K. & Arnaout,

M. A. (2000) Proc. Natl. Acad. Sci. USA 97, 1731–1736. (First PublishedFebruary 4, 2000; 10.1073ypnas.040550097)

Caron and Smithies PNAS u January 16, 2001 u vol. 98 u no. 2 u 619

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

May

17,

202

0