a RESEARCH ARTICLE

The Chemokine Receptor CXCR7 Functions toRegulate Cardiac Valve RemodelingSangho Yu,1,2,3† Dianna Crawford,4†,‡ Takatoshi Tsuchihashi,1,2,3§ Timothy W. Behrens,4#

and Deepak Srivastava1,2,3*

CXCR7 (RDC1), a G-protein-coupled receptor with conserved motifs characteristic of chemokine recep-tors, is enriched in endocardial and cushion mesenchymal cells in developing hearts, but its function isunclear. Cxcr7 germline deletion resulted in perinatal lethality with complete penetrance. Mutantembryos exhibited aortic and pulmonary valve stenosis due to semilunar valve thickening, with occa-sional ventricular septal defects. Semilunar valve mesenchymal cell proliferation increased in mutantsfrom embryonic day 14 onward, but the cell death rate remained unchanged. Cxcr7 mutant valves hadincreased levels of phosphorylated Smad1/5/8, indicating increased BMP signaling, which may partlyexplain the thickened valve leaflets. The hyperproliferative phenotype appeared to involve Cxcr7 functionin endocardial cells and their mesenchymal derivatives, as Tie2-Cre Cxcr7flox/2 mice had semilunarvalve stenosis. Thus, CXCR7 is involved in semilunar valve development, possibly by regulating BMP sig-naling, and may contribute to aortic and pulmonary valve stenosis. Developmental Dynamics 240:384–393,2011. VC 2011 Wiley-Liss, Inc.

Key words: CXCR7; G-protein coupled receptor; chemokine receptor; endocardial cells; cardiac valve remodeling;semilunar valve development; BMP signaling

Accepted 17 December 2010

INTRODUCTION

CXCR7 (RDC1) is an evolutionarilyconserved, seven-transmembrane G-protein-coupled receptor (GPCR) withmotifs characteristic of chemokinereceptors, such as the DRY motif. It isexpressed in brain, heart, kidney,spleen, thymus, and various tumorsand associated vascular endothelialcells (Heesen et al., 1998; Maddenet al., 2004; Burns et al., 2006; Miaoet al., 2007). CXCR7 binds to CXCL12(SDF-1) with high affinity (Balaba-

nian et al., 2005; Burns et al., 2006),challenging the notion that CXCL12has a monogamous relationship withCXCR4, as suggested by the nearlyidentical phenotypes of CXCL12 andCXCR4 knockout mice (Nagasawaet al., 1996; Ma et al., 1998; Tachi-bana et al., 1998; Zou et al., 1998).CXCR7 also binds to another chemo-kine, CXCL11 (I-TAC) (Burns et al.,2006). However, the significance ofligand binding to CXCR7 is unclear,since CXCR7 does not appear totransduce any intracellular signaling,

such as Ca2þ mobilization or mitogen-activated protein kinase (MAPK) sig-naling, which are the hallmarks ofchemokine receptor activation (Burnset al., 2006; Proost et al., 2007; Bolda-jipour et al., 2008; Hartmann et al.,2008).Despite the lack of ligand-mediated

signaling, expression of and ligandbinding to CXCR7 has significant cel-lular and physiological consequences.During CXCR4-mediated migration ofprimordial germ cells (Boldajipouret al., 2008) and lateral line

Dev

elop

men

tal D

ynam

ics

1Gladstone Institute of Cardiovascular Disease, San Francisco, California2Department of Biochemistry & Biophysics, University of California, San Francisco, California3Department of Pediatrics, University of California, San Francisco, California4Center for Immunology, Department of Medicine, University of Minnesota Medical School, University of Minnesota, Minneapolis,Minnesota†Sangho Yu and Dianna Crawford contributed equally to this work.‡D. Crawford’s present address is Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320.§T. Tsuchihashi’s present address is Division of Cardiology, Department of Pediatrics, School of Medicine, Keio University, 35Shinanomachi Shinjuku-ku, Tokyo, Japan 160-8582.#T.W. Behrens’s present address is Genentech Inc., 1 DNA Way, South San Francisco, CA 94080.*Correspondence to: Deepak Srivastava, Gladstone Institute of Cardiovascular Disease, 1650 Owens Street, SanFrancisco, CA 94158. E-mail: dsrivastava@gladstone.ucsf.edu

DOI 10.1002/dvdy.22549Published online 19 January 2011 in Wiley Online Library (wileyonlinelibrary.com).

DEVELOPMENTAL DYNAMICS 240:384–393, 2011

VC 2011 Wiley-Liss, Inc.

primordium (Dambly-Chaudiereet al., 2007; Valentin et al., 2007) inzebrafish, CXCR7 creates an essentialCXCL12 gradient by sequesteringand internalizing CXCL12. CXCR7forms a heterodimer with CXCR4 andmodulates certain aspects of its func-tion (Sierro et al., 2007; Hartmannet al., 2008; Kalatskaya et al., 2009;Levoye et al., 2009). CXCR7 is alsoimportant in CXCR4-mediated trans-endothelial migration of tumor cells,but not in intra-tissue chemotaxis(Zabel et al., 2009). Overall, CXCR7 isan atypical chemokine receptor thatdoes not directly activate downstreampathways, but rather modifies orfine-tunes the activation of CXCR4 in

response to CXCL12. In certain envi-ronments, CXCR7 may also haveCXCR4-independent functions, as itpromotes cell survival and adhesioneven without ligand binding; themechanism is unknown (Raggo et al.,2005; Burns et al., 2006; Miao et al.,2007).

Cardiac valve morphogenesis is acomplicated and delicate processinvolving myriad signaling events(Armstrong and Bischoff, 2004). Wefound that CXCR7 is expressed indeveloping and adult heart, especiallyin endocardial cells and their mesen-chymal derivatives. Given the com-plementary expression pattern ofCXCR4 and CXCL12 in the heart and

their identical loss-of-function cardiacphenotypes in mice, we speculatedthat CXCR7 is important for cardiacdevelopment. The consequences ofCXCR7 deletion have been reportedby two groups, but the phenotypeshave been divergent, leaving its func-tion in vivo unclear, particularlyregarding CXCR7’s effects on cardiacdevelopment.In this study, we generated Cxcr7

floxed mice and characterized theglobal and cell-specific consequences ofcomplete loss of Cxcr7 function. Com-plete deletion of Cxcr7 produced peri-natal lethality due to enlarged semilu-nar (aortic and pulmonary) valves andoccasional ventricular septal defects

Fig. 1. Expression pattern of Cxcr7 mRNA in developing mouse heart. A: Whole-mount in situ hybridization was carried out with digoxigenin(DIG)-labeled Cxcr7 riboprobe at E10.5. Cxcr7 mRNA is expressed in the prosencephalon, rhombencephalon, somites, neural tube, and heart. h,heart; nt, neural tube; pro, prosencephalon; rh, rhomboencephalon; so, somite. B: Section in situ hybridization image of sagittally sectioned P1mouse. Cxcr7 is expressed in various tissues, including brain, thymus, heart, stomach, and liver. Adr, adrenal gland; Br, brain; Cb, cerebellum; H,heart; Li, liver; Lu, lung; M, muscle; ONE, optic nerve ending; PU, penile urethra; R, rectum; Sk, skin; St, stomach; T, tooth; Th, thymus; Vb, verte-brae. C–F: Transverse sections made after whole-mount in situ hybridization. Cxcr7 is expressed in endocardial cells and endocardial cushionmesenchymal cells in both outflow tract (OFT) and atrioventricular canal (AVC) regions. Its expression in myocardial cells is much weaker than thatof endocardial cells. D and F are high-magnification images of OFT and AVC endocardial cushion regions, respectively. En, endocardium; ECC, en-docardial cushion; Myo, myocardium; A, atrium; V, ventricle; RV, right ventricle.

Dev

elop

men

tal D

ynam

ics

CXCR7 REGULATES SEMILUNAR VALVE PROLIFERATION 385

Fig. 2. Generation of targeted Cxcr7-null mice. A: Schematic representation of the wild-type Cxcr7 locus, the targeted allele, and the deletedlocus. Shaded regions represent exon 2 (exon 2, top); the cloning fragment that includes exon 2 and about 100 bp up- and downstream of exon 2(Ex2, middle); the 50 homologous arm, which was cloned from DNA immediately upstream of Ex2 (50 arm, middle and lower); and the 30 homolo-gous arm, which was cloned from DNA immediately downstream of Ex2 (30 arm, middle and lower). Black triangles indicate LoxP sites; arrowheadindicates primer sites used for genotyping. E, EcoRV; neo, neomycin-resistance gene cassette. B: Southern blot analysis of ES cell clones show-ing non-targeted (þ/þ) and targeted (þ/flox) clones. The 9.4-kb wild-type and 7.1-kb targeted allele fragments digested with EcoRV (E) were iden-tified with an upstream probe, as shown in A. C: PCR analysis of genomic DNA from E16.5 wild-type (þ/þ) embryos or embryos heterozygous (þ/�) or homozygous (�/�) for Cre-mediated deletion of the targeted locus. The 3.9-kb wild-type and 1.9-kb deleted products were from a single setof primers outside of exon 2, as shown in A. D: Genotypes of offspring from Cxcr7þ/� � Cxcr7þ/� crosses at weaning and at E16.5. E: Six of eightE18.5 mice from a single (Cxcr7þ/� � Cxcr7þ/�) litter, photographed less than 1 hr after cesarean birth. Identifier and genotypes are shown foreach mouse. F,G: Frontal views of E18.5 wild-type and Cxcr7�/� hearts. In some Cxcr7�/� hearts (22%), the aorta was misaligned and displacedto the right side (arrow). Ao, aorta; PA, pulmonary artery; ra, right atrium; la, left atrium; rv, right ventricle.

Dev

elop

men

tal D

ynam

ics

386 YU ET AL.

(VSDs). Endothelial-specific deletion ofCxcr7 mediated by Tie2-Cre caused asimilar, but less severe, cardiac pheno-type. The valve defects were due toincreased proliferation of valve mesen-chymal cells associated with increasedbone morphogenetic protein (BMP) sig-naling. Our results suggest thatCXCR7 is required for proper cardiacsemilunar valve morphogenesis andnormal BMP signaling.

RESULTS

Expression Pattern of CXCR7

in Mice

To examine the expression pattern ofCXCR7 in the embryonic heart, we per-formed whole-mount in situ hybridiza-tion. In E10.5 mouse embryos, Cxcr7was expressed at high levels in theprosencephalon, especially in the nasalprocess, and in a part of the rhomboen-cephalon and at lower levels in the neu-ral tube, somites, and heart (Fig. 1A).A similar pattern was observed inpostnatal day 1 (P1) mice (Fig. 1B). InE10.5 heart, Cxcr7 was mainly ex-pressed in endocardial cells and theirendocardial cushion mesenchymal cellderivatives, in both the outflow tract(OFT) and atrioventricular canal (AVC)regions (Fig. 1C–F) and to a lesserdegree in myocardial cells (Fig. 1C,E).A portion of OFT cushion mesenchymalcells are also derived from cardiac neu-ral crest cells and they also seemed toexpress Cxcr7 since all OFT cushioncells were Cxcr7þ, even though neuralcrest cells did not express Cxcr7.

Perinatal Lethality of

Cxcr7-Deficient Mice

To test the function of CXCR7 in mice,we generated mice with loxP sitesflanking the Cxcr7 gene (Cxcr7flox/flox)and used them to create ubiquitous orendothelial cell-specific Cxcr7 knockoutmice. A targeting vector was generatedto insert loxP sequences immediatelyupstream and downstream of Cxcr7exon 2 by homologous recombinationwith the Cxcr7 locus (Fig. 2A). Removalof exon 2, which contains the entirecoding region of Cxcr7, by Cre-medi-ated recombination would result in thecomplete loss of CXCR7 function. Afterconfirmation by Southern blot, cor-rectly targeted embryonic stem cell

clones (Fig. 2B) were injected intoblastocysts. Chimeric offspring werecrossed with wild-type C57BL/6 mice togenerate Cxcr7flox mice, which werebred with deleter transgenic mice(Schwenk et al., 1995) to allow Cre-mediated recombination in all cells.F1 mice were out-crossed with wild-type C57BL/6 mice to establish astably transmitting mouse line withheterozygous deletion of Cxcr7 exon 2(Cxcr7þ/�) (Fig. 2C).

When Cxcr7þ/� mice were inter-crossed, no Cxcr7�/�mice were found atweaning (Fig. 2D). A normal Mende-lian ratio was observed at E16.5 (Fig.2D) without abnormalities in size,color, or gross anatomy. About a thirdof E18.5 Cxcr7�/� mice delivered by ce-sarean birth at E18 were dead. Theremaining mice died shortly afterbirth; most were cyanotic and gaspedfor breath, but the phenotype was vari-able. For example, in a litter of eightmice, one of three Cxcr7�/� mice diedat or just before birth and appeared tohave a catastrophic circulatory failurein late development, as it was com-pletely white within a blood-filled em-bryonic sac (mouse 18-12, Fig. 2E). Thesecond appeared anatomically normalbut never breathed (mouse 18-10, Fig.2E). And the third gasped andremained cyanotic before dying within1 hr after birth (mouse 18-6, Fig. 2E).Some Cxcr7�/� mice appeared normaland robust at birth but then showedsigns of stress, such as gasping forbreath, and died suddenly.

CXCR7 Deficiency Causes

Thickening of Semilunar

Valves and Ventricular

Septal Defect

To analyze the phenotypes of Cxcr7�/�

mice in detail, we harvested E18.5embryos from intercrosses of Cxcr7þ/�

mice, and examined the internalorgans. In Cxcr7�/� embryos, the lungsappeared normal and were able toinflate. All other internal organs lookedgrossly normal except the hearts,which were slightly larger than age-matched wild-type hearts and occasion-ally had dilated atria or malalignmentof the aorta (Fig. 2F,G).

Transverse and coronal sections ofE18.5 Cxcr7�/� hearts revealed thick-ened semilunar valves (pulmonary and

aortic valves) in all embryos; atrioven-tricular (tricuspid and mitral) valveswere normal (Fig. 3A–J). One third ofthe embryos (n¼9) had VSDs in themembranous portion of the septum,and 22% had overriding aortas (Fig.3I). In contrast to a previous report(Sierro et al., 2007), we did not observeatrial septal defects or bicuspid semilu-nar valves. The perinatal lethalitymost likely resulted from insufficientblood flow to the body caused by semi-lunar valve stenosis, a condition thatcan cause severe cardiac dysfunction inhuman newborns as well.To determine when the semilunar

valve thickening occurred during devel-opment, heart sections were made atprogressive mouse embryonic stages.Until around E14.0, there was no appa-rent difference in the size of the OFTen-docardial cushions (Fig. 3K–M and O–Q). However, at E15.5, when wild-typevalve thinning into leaflets had alreadybegun, Cxcr7�/� pulmonary and aorticvalves were still thick, indicating defectsin semilunar valve remodeling (Fig.3N,R). Mitral and tricuspid valves werenormal in mutants. Since OFT endocar-dial cushion is composed of both endo-cardial- and neural crest–derived cells,while AVC endocardial cushion is almostexclusively derived from endocardialcells (Hutson and Kirby, 2007; Snarret al., 2008), this result may reflect func-tion of Cxcr7 in both cell types.

Increased Cell Proliferation

in Semilunar Valves of

Cxcr72/2 Embryos

Some neural crest cells originating atthe level of the posterior rhombencepha-lon migrate to the heart through pha-ryngeal arches 3, 4, and 6 and contrib-ute to various developmental events,such as outflow septation, great vesselalignment, and remodeling of pharyn-geal arch arteries (Hutson and Kirby,2003; Stoller and Epstein, 2005). Sincevalve thickening was observed only inOFT valves, and VSD and overridingaorta were present in some cases, wespeculated that CXCR7 is involved inthe migration of cardiac neural crestcells. However, whole-mount in situhybridization for Crabp1 mRNA, amarker of neural crest cells, at E10.5did not reveal any gross abnormalitiesbetween neural crest migration to

Dev

elop

men

tal D

ynam

ics

CXCR7 REGULATES SEMILUNAR VALVE PROLIFERATION 387

pharyngeal arches 3, 4, and 6 in wild-type and Cxcr7�/� embryos (data notshown).

To investigate whether the prolifera-tion or apoptosis of semilunar valvemesenchymal cells was affected inCxcr7�/� embryos at various develop-mental stages, we analyzed cell prolif-eration by phospho-histone H3 (pH3)immunofluorescence (Fig. 4A,B) andcell death by TUNEL assay. In wild-

Fig. 3.

Fig. 4.

Dev

elop

men

tal D

ynam

ics

388 YU ET AL.

type semilunar valves, the percentageof cells undergoing mitosis decreasedas embryos grew older. In contrast,Cxcr7�/� valves had �2-fold more pH3-positive semilunar valve cells thanwild-type valves at E14.0 and 3.5-foldmore at E15.5 (Fig. 4C). No differenceswere detected in the percentage ofpH3-positive cells in pulmonary valvescompared to aortic valves. The percen-tages of cells undergoing apoptosiswere almost identical in wild-type andCxcr7�/� valves (data not shown).

CXCR7 in Endocardial Cells

Is Important for Semilunar

Valve Morphogenesis

To identify the cell type in whichCXCR7 functions during valve devel-opment and to examine the role ofCXCR7 in the adult heart, we deletedCxcr7 specifically in endothelial andendocardial cells using the Tie2-Cretransgene. Tie2-CreCxcr7flox/� micesurvived normally after birth andwere fertile. However, at 6 months ofage, they had marked cardiac hyper-trophy (Fig. 5A). Cross-sections ofage-matched wild-type and Tie2-CreCxcr7flox/� hearts revealed muchthicker ventricular walls in mutanthearts, but no difference in the size ofthe ventricular cavity (Fig. 5B).

Echocardiographic analysis of cardiacfunction at 6 months of age showedno difference in fractional shorteningor ejection fraction between Tie2-CreCxcr7flox/� and wild-type mice (datanot shown), but the left ventricular pos-terior wall was about 1.4-fold thicker in

Tie2-CreCxcr7flox/� mice (Fig. 5C).These features are consistent with con-centric hypertrophy caused by pressureoverload (Jessup and Brozena, 2003).

To determine whether Tie2-CreCxcr7flox/� mice had semilunarvalve stenosis, we examined histologi-cal sections at the level of the semilu-nar valves. Indeed, the semilunarvalves of Tie2-CreCxcr7flox/� micewere thicker than those of wild-typecontrol mice, and the difference wasmore dramatic in the aortic valvescompared to the pulmonary valves(Fig. 5D–G). This result suggests thatCxcr7 expression in endocardial cellsand their valve mesenchymal deriva-tives is necessary to prevent thicken-ing of the semilunar valves.

Examination of the myocardialstructure of Tie2-CreCxcr7flox/� heartsat high magnification revealed greatlyenlarged cardiomyocytes, cellular dis-array, and increased intercellularspace (Fig. 5H,I). At 6 months of age,the mutant mice had substantial fi-brosis, marked by Masson’s trichromestaining, present in the myocardiumand around microvessels in the heart;little fibrosis was detected in wild-type mice (Fig. 5J,K).

BMP Signaling Is Enhanced

in Semilunar Valves of Cxcr7

Mutants

A hypomorphic allele of epidermalgrowth factor receptor (EGFR) alsoresults in thickening of semilunarvalves (Chen et al., 2000), and muta-tion of one of its ligands, heparin-

binding epidermal growth factor (HB-EGF), led to dysregulation of BMPsignaling and similar valve stenosisin mice (Jackson et al., 2003). Todetermine whether BMP signalingwas increased in Cxcr7�/� semilunarvalves, we performed immunostain-ing for phospho-Smad1/5/8 (pSmad1/5/8), which transduces BMP signal-ing. pSmad1/5/8 levels were about1.5-fold higher in Cxcr7�/� semilunarvalves than in wild-type valvesthroughout development (Fig. 6A–E).

DISCUSSION

This study shows that endothelialexpression of the chemokine receptorCXCR7 is important in limiting prolif-eration of cells during cardiac semi-lunar valve morphogenesis. Cxcr7germline deletion in mice caused peri-natal lethality, likely because of insuf-ficient blood supply to peripheral tis-sues as a result of severe aortic andpulmonary valve thickening and ste-nosis. Loss of CXCR7 also increasedBMP signaling in semilunar valvesand caused a sustained high prolifera-tive rate of valve mesenchymal cellsduring the valve remodeling stage.Tie2-CreCxcr7flox/� mice with endo-thelial cell-specific ablation of Cxcr7had similarly enlarged semilunarvalves later in life and developed mas-sive cardiac hypertrophy presumablydue to aortic valve stenosis.These results are consistent with

the first report of a Cxcr7 mutant phe-notype (Sierro et al., 2007), but notwith a subsequent independent Cxcr7knockout study (Gerrits et al., 2008).Perinatal lethality was observed inboth studies. However, Gerrits et al.(2008) did not observe semilunarvalve thickening or ventricular septaldefects and concluded that theenlarged hearts in Cxcr7-null micewere due to hyperplasia. In contrast,Sierro et al. (2007) observed aorticand pulmonary valve thickening, aswell as bicuspid aortic valves, a find-ing we did not observe. These discrep-ancies might reflect differences ingenetic background, knockout strat-egies, or effects on neighboring genes.Gerrits et al. (2008) knocked in the b-galactosidase (LacZ) transgene in theplace of Cxcr7 exon2, while we andSierro et al. (2007) replaced the exon2with a loxP-flanked conditional allele.

Fig. 3. Cxcr7�/� embryos have enlarged semilunar valves and occasional VSD or overridingaorta. A–J: H&E-stained sections of E18.5 embryos. Cxcr7�/� embryos have thick pulmonaryvalve (PV) and aortic valves (AoV) but normal tricuspid and mitral valves. VSD was observed in33% (n¼9) of cases (arrow in I). Transverse sections are shown in A, C, F, and H. Coronal sec-tions are shown in B, D, E, G, I, and J. Ao, aorta; PV, pulmonary valve; RV, right ventricle; LV,left ventricle; LA, left atrium; RA, right atrium; MV, mitral valve; TV, tricuspid valve. K–R: Coronalsections of hearts from wild-type and Cxcr7�/� embryos at various developmental stages. Thesizes and cell numbers of OFT endocardial cushions or semilunar valves were not significantlydifferent in wild-type and Cxcr7�/� embryos until E14. Difference in the size of semilunar valvesbecame apparent at around E15.5, when the valves undergo a remodeling process and formthin leaflets in wild-type hearts, but remained enlarged in Cxcr7�/� hearts.

Fig. 4. Increased proliferation of cardiac semilunar valve mesenchymal cells in Cxcr7 mutants.A,B: Cells undergoing mitosis were stained with pH3 antibody (red) at various embryonic stages.Nuclei were counterstained with DAPI (blue). There were more proliferating cells in semilunarvalves in Cxcr7�/� embryos from E14 onward. Shown are representative sections at E15.5. C:Quantitative analysis of pH3 immunostaining (pH3-positive cells/total valve cells � 100) showedthat the percentage of cells undergoing mitosis in Cxcr7�/� valves was increased �2-fold at E14and �3.5-fold at E15.5. *P < 0.01.

Dev

elop

men

tal D

ynam

ics

CXCR7 REGULATES SEMILUNAR VALVE PROLIFERATION 389

No predicted microRNAs are in thislocus, making it less likely that thereare differences in the deletion ofpotentially embedded genes.

Semilunar valve thickening in Tie2-CreCxcr7flox/� mice confirms the endo-thelial origin of this phenotype. How-ever, the less severe phenotype ofTie2-CreCxcr7flox/� mice compared togermline Cxcr7�/� mice suggests thatCxcr7 function in other cell types mayalso be important. Given the Cxcr7expression in neural crest–derivedcushion mesenchymal cells and lackof overriding aorta and VSD in Tie2-CreCxcr7flox/� mice, Cxcr7 may alsohave a critical function in neuralcrest–derived cushion cells during

Fig. 5.

Fig. 6.

Dev

elop

men

tal D

ynam

ics

390 YU ET AL.

OFT maturation and septum forma-tion. Alternatively, the imperfectrecombination induced by Tie2-Cremight have prevented elicitation ofthe low penetrance VSD and overrid-ing aorta (33 and 22%, respectively)phenotypes in Tie2-CreCxcr7flox/�

mice. Further study is required topinpoint the cellular origin of eachphenotype and to elucidate the Cxcr7function in neural crest–derived cellsusing neural crest–specific Cre mice.

Disruption of EGFR signaling path-ways in mice also causes cardiac valvehypertrophy due to increased prolifer-ation at the late stage of valve devel-opment (Chen et al., 2000; Jacksonet al., 2003). There is evidence thatEGFR signaling antagonizes BMPsignaling (Kretzschmar et al., 1997;Nonaka et al., 1999; Lo et al., 2001),and pSmad1/5/8 levels may beincreased in Hbegf knockout valves(Jackson et al., 2003), consistent witha link between EGFR signaling andCxcr7. EGFR activation antagonizesBMP signaling by phosphorylatingSmad proteins at the linker regiondistinct from phosphorylation sites byBMP receptor (Kretzschmar et al.,1997) or by stabilizing the Smad tran-scriptional co-repressor TGIF (Loet al., 2001). Both effects are medi-ated by the MAPK Erk. In one study,Hbegf mRNA was downregulated inCxcr7�/� semilunar valves, corrobo-rating the relationship betweenEGFR signaling and CXCR7 (Sierroet al., 2007). It is unclear how CXCR7regulates Hbegf expression. In ourstudy, treating CXCL12 to primary

sheep aortic valve mesenchymal cells,which express Cxcr7, or Cxcr7-trans-fected COS1 cells did not activateEGFR signaling (data not shown).

CXCR7 shares the same ligand,CXCL12, with CXCR4 but ligandbinding to CXCR7 does not activateclassical chemokine pathways such asCa2þ mobilization or MAPK signaling(Burns et al., 2006; Proost et al.,2007; Boldajipour et al., 2008; Hart-mann et al., 2008). Although CXCR7can modulate the function of CXCR4through ligand sequestration or het-erodimerization (Dambly-Chaudiereet al., 2007; Sierro et al., 2007; Valen-tin et al., 2007; Boldajipour et al.,2008; Levoye et al., 2009; Zabel et al.,2009), semilunar valve thickening is aunique phenotype of Cxcr7 knockoutmice that is not observed in eitherCxcr4 or Cxcl12 knockout mice, whereVSD is the major cardiac phenotype.In contrast, semilunar valve thicken-ing is caused by both Cxcr7 deficiencyand disruption of EGFR signaling,although VSD and overriding aortaare unique to Cxcr7�/� mice. Thus,CXCR7 may influence semilunarvalve morphogenesis and ventricularseptum formation/OFT maturationthrough two separate mechanisms. Itwill be interesting to test whetherCXCR7 activates different pathwaysin endocardial- and neural crest–derived cushion mesenchymal cells.

Valve defects are among the mostcommon forms of human congenitalheart disease and typically involveabnormal thickening of the aortic orpulmonary valves. The consequences

can range from neonatal cardiac fail-ure to cardiac hypertrophy later inlife, depending on the severity of thestenosis. Numerous signaling path-ways have been linked to valve forma-tion (Armstrong and Bischoff, 2004).Disruption of any of these pathwaysduring pregnancy could lead to fatalconditions, underscoring the impor-tance of elucidating the exact mecha-nisms of valve formation. For exam-ple, the absence of Ptpn11, which isinvolved in a signaling pathway medi-ated by the EGFR, results in dysplasticoutflow valves (Chen et al., 2000). Inhumans, mutations of PTPN11, whichencodes the protein tyrosine phospha-tase Shp-2, cause Noonan syndrome,characterized by pulmonic valve steno-sis (Tartaglia et al., 2001). Similarly,human mutations in NOTCH1 (Garget al., 2005) cause thickened aorticvalve leaflets and can result in aorticvalve stenosis early or later in life. Itwill be interesting to determinewhether CXCR7 or pathways that itregulates also contribute to human aor-tic or pulmonary valve disease.

EXPERIMENTAL

PROCEDURES

Generation of Conditionally

Targeted Cxcr7 Mice

All experimental procedures wereapproved by the Institutional AnimalCare and Use Committee at the Uni-versity of California San Francisco. Aconditional knockout targeting vectorwas designed to insert loxP sequencesabout 100–200 base pairs from eitherside of exon 2 of the Cxcr7 locus. Thelinearized targeting vector was elec-troporated into 129/SvEv embryonicstem (ES) cells (Dr. B. Koller, Univer-sity of North Carolina, Chapel Hill,NC). ES clones were screened by PCRand verified by Southern blot. Cor-rectly targeted ES cells were injectedinto C57BL/6 mouse blastocysts. Chi-meric offspring were crossed withC57BL/6 mice, and agouti offspringwere screened for the targeted allele,Cxcr7flox. Deleter Cre transgenic micewere used to allow Cre-mediatedrecombination in all cells (Schwenket al., 1995). Tie2-CreCxcr7flox/� micewere generated by crosses ofCxcr7flox/flox � Tie2-CreCxcr7þ/� orCxcr7flox/� � Tie2-CreCxcr7flox/þ. Tie2-

Fig. 5. Endothelial-specific deletion of Cxcr7 causes cardiac semilunar valve stenosis andsecondary hypertrophy. A: Image of whole hearts taken from 6-month-old wild-type or Tie2-CreCxcr7flox/� mice. Endothelial-specific deletion of Cxcr7 caused marked hypertrophy. B: Trans-verse sections of wild-type and Tie2-CreCxcr7flox/� mice. The mutant heart had a thicker ventricu-lar wall than the wild-type heart but a similarly sized ventricular cavity. C: Left ventricular posteriorwall (LVPW) thickness, measured by echocardiography, was �1.4-fold thicker in Tie2-CreCxcr7flox/�

hearts than in wild-type hearts (n¼3 per group, *P<0.05). D–G: Heart sections of 6-month-old wild-type or Tie2-CreCxcr7flox/� mice were stained with hematoxylin-eosin. Tie2-CreCxcr7flox/� mice hadaortic valve (AoV) stenosis similar to that in Cxcr7�/� mice but a milder pulmonary valve (PV) pheno-type. H,I: Magnified images (20�) of wild-type and Tie2-CreCxcr7flox/� heart sections stained withhematoxylin-eosin. Myocardial structure was disarrayed and the myocytes were not well packed inmutant hearts. J,K: Masson’s trichrome staining showed more collagen deposition (blue) in Tie2-CreCxcr7flox/� hearts than in wild-type hearts.

Fig. 6. CXCR7 deficiency enhances BMP signaling in semilunar valves. A–D: Immunohisto-chemistry with anti-phospho-Smad1/5/8 (pSmad1/5/8) antibody (brown) shows increased BMPsignaling in Cxcr7�/� semilunar valves. Nuclei were counterstained with hematoxylin. E: Quanti-tation of pSmad1/5/8 immunostaining shows that the BMP signaling was increased throughoutcardiac valve development in Cxcr7�/� mice. *P < 0.05.

Dev

elop

men

tal D

ynam

ics

CXCR7 REGULATES SEMILUNAR VALVE PROLIFERATION 391

CreCxcr7flox/þ mice were used as wild-type controls. Tie2-Cre mice (Koniet al., 2001) were purchased from theJackson Laboratory (West Grove, PA).

In Situ Hybridization

Section in situ hybridization for P1mouse was performed by Phylogeny(Columbus, OH). Mouse tissue wasfrozen, cut into 6–12mm sections,mounted on gelatin-coated slides, andfixed in 4% formaldehyde. The cRNAprobe for Cxcr7 was synthesized invitro according to the manufacturer’sconditions (Ambion, Austin, TX) andlabeled with 35S-UTP (Amersham,Pittsburgh, PA). Sections were hybri-dized overnight at 55�C with 35S-la-beled cRNA probe (50–80,000 cpm/ml).Whole-mount in situ hybridization forCxcr7 and Crabp mRNAs was per-formed with E10.5–E11.0 embryos(n¼4 and 3, respectively) as described(Kwon et al., 2009). Embryos werefixed in 4% formaldehyde overnightand kept in 100% methanol at �20�Cuntil use. Hybridization was per-formed at 65�C overnight. Riboprobeswere labeled with DIG, and embryoswere stained according to the manu-facturer’s instruction (Roche, India-napolis, IN).

Histology

Embryonic hearts at various stageswere embedded in paraffin, cut into8–10-mm sections, deparaffinized, andstained with hematoxylin and eosinfor histological analysis (n¼4 forE11.5, n¼3 for E12.5, n¼3 for E14.0,n¼3 for E15.5, and n¼4 for E18.5).For Masson’s trichrome staining, 6-month-old hearts from wild-type andTie2-CreCxcr7flox/� mice were embed-ded in paraffin and transverselysectioned (n¼3 for each group).Deparaffinized sections were dippedsequentially into Weigert’s iron hema-toxylin working solution for 10 min,Biebrich scarlet-acid fuchsin solutionfor 15 min, phosphomolybdic-phos-photungstic acid solution for 15 min,aniline blue solution 5–10 min, and1% acetic acid solution for 2–5 min.

Immunostaining and TUNEL

Assay

Paraffin-embedded embryos of wild-type and Cxcr7�/� mice were cut atvarious stages either transversely orcoronally. For immunostaining, anti-bodies for phospho-histone H3(Upstate, East Syracuse, NY) andphospho-Smad1/5/8 (Cell Signaling,Danvers, MA) were used at 1:100dilution. For pH3 staining, a FITC-conjugated secondary antibody wasused, and the slides were mountedwith Vectashield containing DAPI(Vector Laboratories, Burlingame,CA). For pSmad1/5/8 staining, ahorseradish peroxidase–conjugatedsecondary antibody was used, and thesections were stained with the ABCstaining system (Vector Laboratories).TUNEL assays were performed withan in situ cell death detection kit(Roche). Number of animals used:pH3 staining (n¼3 for E11.5, E12.5,E13.5, and E14.0; n¼5 for E15.5);pSmad1/5/8 staining (n¼3 for E11.5and E15.5, n¼6 for E12.5 and E14.0);TUNEL assay (n¼2 for E11.5, n¼3 forE12.5, n¼4 for E14.0 and E15.5). Forwild-type and mutant mice, the samenumbers of animals were used.

Echocardiography

Systolic function and ventricular wallthickness were assessed with M-modeand power Doppler mode in 6-month-old wild-type and Tie2-CreCxcr7flox/�

mice (n¼3 for each group); mice wereanesthetized with 1.75% isoflurane.Core temperature was maintained at37–38�C, and scans were performedin a random-blind fashion. Eachmouse underwent three separatescans on three different days. Valueswere averaged from three scans.

Statistical Analysis

Statistical analysis was performedwith the Student’s t-test and P < 0.05was considered statisticallysignificant.

REFERENCES

Armstrong EJ, Bischoff J. 2004. Heartvalve development: endothelial cell sig-naling and differentiation. Circ Res 95:459–470.

Balabanian K, Lagane B, Infantino S,Chow KY, Harriague J, Moepps B, Are-

nzana-Seisdedos F, Thelen M, Bachel-erie F. 2005. The chemokine SDF-1/CXCL12 binds to and signals throughthe orphan receptor RDC1 in T lym-phocytes. J Biol Chem 280:35760–35766.

Boldajipour B, Mahabaleshwar H, Kar-dash E, Reichman-Fried M, Blaser H,Minina S, Wilson D, Xu Q, Raz E. 2008.Control of chemokine-guided cell migra-tion by ligand sequestration. Cell 132:463–473.

Burns JM, Summers BC, Wang Y, Meli-kian A, Berahovich R, Miao Z, PenfoldME, Sunshine MJ, Littman DR, KuoCJ, Wei K, McMaster BE, Wright K,Howard MC, Schall TJ. 2006. A novelchemokine receptor for SDF-1 and I-TAC involved in cell survival, cell adhe-sion, and tumor development. J ExpMed 203:2201–2213.

Chen B, Bronson RT, Klaman LD, Hamp-ton TG, Wang JF, Green PJ, MagnusonT, Douglas PS, Morgan JP, Neel BG.2000. Mice mutant for Egfr and Shp2have defective cardiac semilunar valvu-logenesis. Nat Genet 24:296–299.

Dambly-Chaudiere C, Cubedo N, GhysenA. 2007. Control of cell migration in thedevelopment of the posterior lateralline: antagonistic interactions betweenthe chemokine receptors CXCR4 andCXCR7/RDC1. BMC Dev Biol 7:23.

Gerrits H, van Ingen Schenau DS, Bak-ker NE, van Disseldorp AJ, Strik A,Hermens LS, Koenen TB, Krajnc-Franken MA, Gossen JA. 2008. Earlypostnatal lethality and cardiovasculardefects in CXCR7-deficient mice. Gene-sis 46:235–245.

Hartmann TN, Grabovsky V, Pasvolsky R,Shulman Z, Buss EC, Spiegel A, NaglerA, Lapidot T, Thelen M, Alon R. 2008.A crosstalk between intracellularCXCR7 and CXCR4 involved in rapidCXCL12-triggered integrin activationbut not in chemokine-triggered motilityof human T lymphocytes and CD34þcells. J Leukoc Biol 84:1130–1140.

Heesen M, Berman MA, Charest A, Hous-man D, Gerard C, Dorf ME. 1998. Clon-ing and chromosomal mapping of anorphan chemokine receptor: mouseRDC1. Immunogenetics 47:364–370.

Hutson MR, Kirby ML. 2003. Neural crestand cardiovascular development: a 20-year perspective. Birth Defects Res CEmbryo Today 69:2–13.

Hutson MR, Kirby ML. 2007. Model sys-tems for the study of heart developmentand disease. Cardiac neural crest andconotruncal malformations. Semin CellDev Biol 18:101–110.

Jackson LF, Qiu TH, Sunnarborg SW,Chang A, Zhang C, Patterson C, LeeDC. 2003. Defective valvulogenesis inHB-EGF and TACE-null mice is associ-ated with aberrant BMP signaling.Embo J 22:2704–2716.

Jessup M, Brozena S. 2003. Heart failure.N Engl J Med 348:2007–2018.

Kalatskaya I, Berchiche YA, Gravel S,Limberg BJ, Rosenbaum JS, HevekerN. 2009. AMD3100 is a CXCR7 ligand

Dev

elop

men

tal D

ynam

ics

392 YU ET AL.

with allosteric agonist properties. MolPharmacol 75:1240–1247.

Koni PA, Joshi SK, Temann UA, Olson D,Burkly L, Flavell RA. 2001. Conditionalvascular cell adhesion molecule 1 dele-tion in mice: impaired lymphocytemigration to bone marrow. J Exp Med193:741–754.

Kretzschmar M, Doody J, Massague J.1997. Opposing BMP and EGF signal-ling pathways converge on the TGF-beta family mediator Smad1. Nature389:618–622.

Kwon C, Qian L, Cheng P, Nigam V,Arnold J, Srivastava D. 2009. A regula-tory pathway involving Notch1/beta-cat-enin/Isl1 determines cardiac progenitorcell fate. Nat Cell Biol 11:951–957.

Levoye A, Balabanian K, Baleux F, Bach-elerie F, Lagane B. 2009. CXCR7 heter-odimerizes with CXCR4 and regulatesCXCL12-mediated G protein signaling.Blood 113:6085–6093.

Lo RS, Wotton D, Massague J. 2001. Epi-dermal growth factor signaling via Rascontrols the Smad transcriptional co-repressor TGIF. Embo J 20:128–136.

Ma Q, Jones D, Borghesani PR, Segal RA,Nagasawa T, Kishimoto T, Bronson RT,Springer TA. 1998. Impaired B-lympho-poiesis, myelopoiesis, and derailed cere-bellar neuron migration in CXCR4- andSDF-1-deficient mice. Proc Natl AcadSci USA 95:9448–9453.

Madden SL, Cook BP, Nacht M, WeberWD, Callahan MR, Jiang Y, DufaultMR, Zhang X, Zhang W, Walter-Yohrl-ing J, Rouleau C, Akmaev VR, WangCJ, Cao X, St Martin TB, Roberts BL,Teicher BA, Klinger KW, Stan RV,Lucey B, Carson-Walter EB, Laterra J,Walter KA. 2004. Vascular gene expres-sion in nonneoplastic and malignantbrain. Am J Pathol 165:601–608.

Miao Z, Luker KE, Summers BC, Beraho-vich R, Bhojani MS, Rehemtulla A,

Kleer CG, Essner JJ, Nasevicius A,Luker GD, Howard MC, Schall TJ.2007. CXCR7 (RDC1) promotes breastand lung tumor growth in vivo and isexpressed on tumor-associated vascula-ture. Proc Natl Acad Sci USA 104:15735–15740.

Nagasawa T, Hirota S, Tachibana K,Takakura N, Nishikawa S, Kitamura Y,Yoshida N, Kikutani H, Kishimoto T.1996. Defects of B-cell lymphopoiesisand bone-marrow myelopoiesis in micelacking the CXC chemokine PBSF/SDF-1. Nature 382:635–638.

Nonaka K, Shum L, Takahashi I, Takaha-shi K, Ikura T, Dashner R, NuckollsGH, Slavkin HC. 1999. Convergence ofthe BMP and EGF signaling pathwayson Smad1 in the regulation of chondro-genesis. Int J Dev Biol 43:795–807.

Proost P, Mortier A, Loos T, Vandercap-pellen J, Gouwy M, Ronsse I, SchutyserE, Put W, Parmentier M, Struyf S, VanDamme J. 2007. Proteolytic processingof CXCL11 by CD13/aminopeptidase Nimpairs CXCR3 and CXCR7 bindingand signaling and reduces lymphocyteand endothelial cell migration. Blood110:37–44.

Raggo C, Ruhl R, McAllister S, Koon H,Dezube BJ, Fruh K, Moses AV. 2005.Novel cellular genes essential for trans-formation of endothelial cells by Kapo-si’s sarcoma-associated herpesvirus.Cancer Res 65:5084–5095.

Schwenk F, Baron U, Rajewsky K. 1995.A cre-transgenic mouse strain forthe ubiquitous deletion of loxP-flankedgene segments including deletion ingerm cells. Nucleic Acids Res 23:5080–5081.

Sierro F, Biben C, Martinez-Munoz L,Mellado M, Ransohoff RM, Li M, WoehlB, Leung H, Groom J, Batten M, Har-vey RP, Martinez AC, Mackay CR,Mackay F. 2007. Disrupted cardiac de-

velopment but normal hematopoiesis inmice deficient in the second CXCL12/SDF-1 receptor, CXCR7. Proc Natl AcadSci USA 104:14759–14764.

Snarr BS, Kern CB, Wessels A. 2008. Ori-gin and fate of cardiac mesenchyme.Dev Dyn 237:2804–2819.

Stoller JZ, Epstein JA. 2005. Cardiac neu-ral crest. Semin Cell Dev Biol 16:704–715.

Tachibana K, Hirota S, Iizasa H, YoshidaH, Kawabata K, Kataoka Y, KitamuraY, Matsushima K, Yoshida N, Nishi-kawa S, Kishimoto T, Nagasawa T.1998. The chemokine receptor CXCR4is essential for vascularization of thegastrointestinal tract. Nature 393:591–594.

Tartaglia M, Mehler EL, Goldberg R,Zampino G, Brunner HG, Kremer H,van der Burgt I, Crosby AH, Ion A,Jeffery S, Kalidas K, Patton MA,Kucherlapati RS, Gelb BD. 2001. Muta-tions in PTPN11, encoding the proteintyrosine phosphatase SHP-2, causeNoonan syndrome. Nat. Genet. 29:465–468.

Valentin G, Haas P, Gilmour D. 2007. Thechemokine SDF1a coordinates tissuemigration through the spatially re-stricted activation of Cxcr7 and Cxcr4b.Curr Biol 17:1026–1031.

Zabel BA, Wang Y, Lewen S, BerahovichRD, Penfold ME, Zhang P, Powers J,Summers BC, Miao Z, Zhao B, Jalili A,Janowska-Wieczorek A, Jaen JC, SchallTJ. 2009. Elucidation of CXCR7-medi-ated signaling events and inhibition ofCXCR4-mediated tumor cell transendo-thelial migration by CXCR7 ligands. JImmunol 183:3204–3211.

Zou YR, Kottmann AH, Kuroda M, Taniu-chi I, Littman DR. 1998. Function ofthe chemokine receptor CXCR4 in hae-matopoiesis and in cerebellar develop-ment. Nature 393:595–599.

Dev

elop

men

tal D

ynam

ics

CXCR7 REGULATES SEMILUNAR VALVE PROLIFERATION 393