Expression of the Paired-Box Genes Pax-1 and Pax-9in Limb Skeleton DevelopmentELIZABETH E. LECLAIR, LAURA BONFIGLIO, AND ROCKY S. TUAN*Department of Orthopaedic Surgery, Thomas Jefferson University, Philadelphia, Pennsylvania

ABSTRACT Vertebrate Pax genes encode afamily of transcription factors that play impor-tant roles in embryonic patterning and morpho-genesis. Two closely related Pax genes, Pax-1 andPax-9, are associated with early axial and limbskeleton development. To investigate the role ofthese genes in cartilage formation we have exam-ined the expression profiles of Pax-1 and Pax-9 indeveloping chick limb mesenchyme in vivo and invitro. Both transcripts are detected by reversetranscription polymerase chain reaction andNorthern blotting throughout chick limb develop-ment, from the early bud stages (Hamburger-Hamilton 20–23) to fully patterned appendages(stage 30). Whole-mount in situ hybridizationreveals complex, nonoverlapping expression do-mains of these two genes. Pax-1 transcripts firstappear at the anterior proximal margin of thelimb buds, while Pax-9 is expressed more distallyat what will be the junction of the autopod andthe zeugopod. In situ hybridization to serial sec-tions of the girdles reveals that in the pectoralregion Pax-1 is expressed proximally in con-densed mesenchyme surrounding the junction ofthe developing scapula, humerus, and coracoid.In the pelvis, Pax-1 is expressed between thefemur and the developing acetabulum and alongthe ventral edge of the ischium; this transcriptwas also found in the distal hindlimb along theposterior edge of the fibula. Pax-9 transcriptswere not detected in the pectoral girdle at anystage, and only weakly in the pelvis along theventral ischial margin. In the distal parts of bothwings and legs, however, Pax-9 is strongly ex-pressed between the anterior embryonic carti-lages (e.g., distal radius or tibia) and the anteriorectodermal ridge. The expression of both geneswas strongest in undifferentiated cells of precar-tilage condensations or at the margins of differen-tiated cartilages, and was absent from cartilageitself. In micromass cultures of chondrifying limbbud mesenchyme expression of Pax-1 and Pax-9is maintained for up to 3 days in vitro, moststrongly at the end of the culture period duringchondrogenic differentiation. As seen in vivo,transcripts are found in loose mesenchyme cellsat the outer margins of developing cartilage nod-ules, and are absent from differentiated chondro-cytes at the nodule center. Taken together, these

investigations extend previous studies of Pax-1and Pax-9 expression in embryonic limb develop-ment while validating limb bud mesenchyme cul-ture as an accessible experimental system for thestudy of Pax gene function and regulation. Our invivo and in vitro observations are discussed withreference to 1) the relationship between somiticand limb expression of these two Pax genes, 2)what regulates this expression in different re-gions of the embryo, and 3) the putative cellularfunctions of Pax-1 and Pax-9 in embryonic skeleto-genesis. Dev Dyn 1999;214:101–115.r 1999 Wiley-Liss, Inc.

Key words: skeletal patterning; cartilage develop-ment; chondrogenesis; differentia-tion; sonic hedgehog; pelvic and pec-toral girdles

INTRODUCTION

Vertebrate skeletogenesis can be divided into axial,appendicular, and craniofacial development. Althoughcartilage and bone form in each area, these structuralproducts may have different cellular origins and differ-entiation pathways. The craniofacial bones, for ex-ample, rely largely on the migration and maturation ofthe embryonic neural crest, and most ossification pro-ceeds intramembranously without a cartilage interme-diate (LeDouarin et al., 1994; Fang and Hall, 1997;Hunt et al., 1998). In contrast, both axial and appen-dicular skeletal elements (including the vertebral col-umn, pectoral and pelvic girdles, and the limbs) arisefrom local condensations of mesodermal cells, andendochondral ossification of cartilaginous precursors isthe rule (Campbell and Kaplan, 1992; Hall and Miyake,1995). Given the mesodermal origin of both axial andappendicular skeletal structures and the apparent simi-larities of their condensation mechanisms, it is valu-able to ask if these two systems share common molecu-

Grant sponsor: National Institutes of Health; Grant numbers: HD15822, ES 07005, DE 11327, AR 07583, and HD 830180.

Laura Bonfiglio’s current address is Scientific Institute S. Raffaele-DIBIT, I-20132 Milan, Italy.

*Correspondence to: Dr. Rocky S. Tuan, Orthopaedic ResearchLaboratory, Department of Orthopaedic Surgery, Thomas JeffersonUniversity, 501 Curtis Building, 1015 Walnut Street, Philadelphia, PA19107. E-mail: Rocky.S.Tuan@mail.tju.edu

Received 7 July 1998; Accepted 7 October 1998

DEVELOPMENTAL DYNAMICS 214:101–115 (1999)

r 1999 WILEY-LISS, INC.

lar pathways involved in skeletal patterning and thecontrol of chondrogenesis.

Several vertebrate Pax genes influence early pattern-ing and differentiation of the axial skeleton. Membersof this gene family share a structural motif similar tothe DNA-binding region of the Drosophila pair-rulegene paired (Bopp et al., 1986). Evolutionary conserva-tion of this ‘‘paired-box’’ motif has led to the identifica-tion of nine murine Pax genes (Pax-1 through -9), avariety of homologues in other vertebrates (Dressler etal., 1988; Kessel and Gruss, 1990; Deutsch and Gruss,1991; Walther et al., 1991; Gruss and Walther, 1992;Hill and Hanson, 1992; Stuart et al., 1993; Wallin et al.,1993; Barnes et al., 1996a), and similar sequences inurochordates, cephalochordates, echinoderms, arthro-pods, and nematodes, among others (Burri et al., 1989;Holland et al., 1995; Glardon et al., 1997). Of the ninemouse Pax genes, all are expressed primarily duringembryonic development and most appear in ectodermalstructures or their derivatives. Only two, Pax-1 andPax-9, are expressed in mesodermal tissues that contrib-ute to the axial and appendicular skeletons in allvertebrate embryos (Gruss and Walther, 1992).

Pax-1 is first expressed in the epithelial somites, theinitial morphological markers of axial segmentation(Deutsch et al., 1988; Deutsch and Gruss, 1991). Expres-sion is later restricted to the caudal half of the sclero-tome that condenses to form the intervertebral discsand the cartilaginous vertebral bodies. Pax-9 expres-sion, somewhat less abundant than that of Pax-1,appears in more lateral parts of the sclerotome thatform the cartilaginous pedicles and laminae of theneural arches (Neubuser et al., 1995). This repeatedpattern in segments of the vertebrate axis, echoing thesegmental expression of pair-rule genes in larval Dro-sophila, aroused initial speculation that metamerismin insect and mammalian embryos might share con-served molecular pathways (Dressler et al., 1988).Pax-1 and Pax-9 expression in regions of somite epithe-lialization, sclerotome differentiation, and cartilagematuration has likewise fueled interest in how thesegenes influence axial skeletal development and chondro-genesis (Wallin et al., 1994). Although their exact role isnot known, these genes appear to act downstream ofsonic hedgehog (shh), a gene strongly expressed in theembryonic notochord and floor plate (reviewed in Dahlet al., 1997; see also Discussion section). Disruptingthis pathway by mutation (Koseki et al., 1993; Neu-buser et al., 1995) or experimental manipulation (Brand-Saberi, 1993; Pourquie et al., 1993; Fan and Tessier-Lavigne, 1994; Johnson et al., 1994; Ebensperger et al.,1995; Fan et al., 1995) decreases the sclerotomal cellpopulation and prevents these cells from forming themesenchymal condensations required for chondrogenicdifferentiation.

This emerging pathway involving shh, Pax-1, andPax-9 in the somitic mesoderm is of additional interestbecause all three genes are also expressed in limb budmesoderm, which shows a similar sequence of cellular

events—condensation, differentiation, and cartilage ma-trix production—during appendicular skeletal develop-ment. Although limb expression of shh has been exten-sively described, the appendicular expression of Pax-1and Pax-9 has received relatively little attention com-pared to their axial expression (Deutsch et al., 1988;Dressler et al., 1988; Love and Tuan, 1993; Smith andTuan, 1994, 1995; Wallin et al., 1994; Ebensperger etal., 1995; Peters et al., 1995; reviewed in Brand-Saberiet al., 1996). Of four reports that mention Pax-1 andPax-9 in any vertebrate limb, only one of these hasfocused specifically on limb development. The mostdetailed study is by Timmons et al. (1994), who used insitu hybridization to map Pax-1 expression in histologi-cal sections of the embryonic mouse pelvic and pectoralgirdles. As part of a general survey of Pax-9 in mousedevelopment, Neubuser et al. (1995) also describedexpression in mouse limbs using both section andwhole-mount hybridizations. In the chick embryo, Eb-ensperger (1995) briefly described Pax-1 expression inday-4 and day-7 limb buds, citing only the ‘‘scapularregion’’ and ‘‘between the pelvic girdle and the femur.’’Peters et al. (1995) reported that Pax-1 was expressednear the ‘‘primordium of the pectoral girdle’’ at day 4.5and ‘‘in a sickle-shaped region dorsal to the cartilagi-nous scapula’’ at day 7. Pax-9 was not found in thepectoral region at either stage, and the development ofthe hindlimbs was not described. Both of these chickstudies utilized in situ hybridization of tissue sectionsfrom selected ontogenetic stages (e.g., day 4 vs day 7)and no whole-mount analyses were performed.

Given the limited published information on Pax-1and Pax-9 in vertebrate limb development, particularlyin the chick embryo, we set out to give a more completecharacterization of Pax-1 and Pax-9 gene expressionusing reverse transcriptase-polymerase chain reaction(RT-PCR), Northern blotting, and in situ hybridizationon a staged developmental series of limb buds (stages19–30) (Hamburger and Hamilton, 1951). Our analysescover both wing and leg structures, span the develop-mental period from first outward limb growth to fullypatterned appendages, and include the pectoral andpelvic girdles as well as the more distal cartilaginouselements. Using similar methods, we have analyzedPax-1 and Pax-9 expression in high-density micromasscultures of limb mesenchyme, a defined in vitro chondro-genic system, to explore the potential role of thesegenes in the cellular events leading to cartilage forma-tion.

RESULTSPax-1 and Pax-9 Expression in Embryonic LimbDevelopment

RT-PCR. We first screened total RNAs from succes-sive stages of chick limb development (stages 20–29)using RT-PCR. Total RNAs from 4- or 6-day-old chickembryos were used as positive controls (Peters et al.,

102 LeCLAIR ET AL.

1995; Barnes et al., 1996a). Negative controls included1) RNA from cultured chick embryo fibroblasts, 2) RNAfrom embryonic chick heart, and 3) omission of anycDNA template. All samples were processed simulta-neously, and separate RT-PCR reactions were run inparallel to detect Pax-1, Pax-9, or glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a constituitivelyexpressed transcript. Because preliminary amplifica-tion showed faint bands of greater molecular weightthan expected, all RNA samples were treated withDNase to eliminate genomic contamination (Fig. 1A).DNase-treated, positive-control RNAs processed withand without reverse transcriptase showed PCR prod-ucts only when reverse transcriptase was included inthe reaction mixture, confirming that the expectedproducts represent cDNAs produced from the RNA pool(Fig. 1B).

Pax-1 mRNAs were detected at all limb bud stagestested. Although we did not quantitatively assess thisgene expression, we noted band intensity increasingfrom a barely visible signal at stages 19–20 to succes-sively stronger signals in older limb buds (stages 29–30;Fig 1C). Pax-9 expression was first detected slightlylater in limb development, appearing as a faint band insamples from stages 21–22 (Fig. 1D). As with Pax-1, adevelopmental increase was observed, and this wasmaintained until the latest stage was examined. Noproducts were amplified in negative tissue controls orin the absence of cDNA. All the tissue RNA samples, butnot the blank, amplified equivalent strong bands forGAPDH (Fig. 1E).

Northern analysis. Our RT-PCR results were con-firmed by Northern blotting of total RNAs from anindependent collection of staged limb buds (Fig. 2).Hybridization with a Pax-1 cDNA probe detected asingle band at approximately 2 kb, the expected size ofthe chicken Pax-1 transcript (Barnes et al., 1996b).Pax-1 signal was no higher than background in theearliest limb buds (stages 19–20) but steadily increasedover later stages. A similar expression pattern wasobserved with the Pax-9 cDNA probe, which also hybrid-ized to a single band around 2 kb. Pax-9 signal firstappeared in limb bud samples at stages 22–23 andpersisted at slightly higher levels until stages 28–29.GAPDH levels in all lanes were roughly equal, as wasethidium bromide staining of the initial agarose gel.

Whole-mount in situ hybridization. The anatomi-cal profiles of Pax-1 and Pax-9 expression were nextassessed by whole-mount in situ hybridization (WISH).A complete series of embryos (stages 19–30) was pre-pared by first removing the heads and viscera. In somelarger specimens the wings and the legs were separatedby a transverse cut through the trunk. Both left andright limb buds remained attached to the trunk, andthese fragments were processed in toto, the trunkserving as an internal control for the known expressionof Pax-1 and Pax-9 in the epithelial somites and sclero-tome (e.g., Wallin et al., 1994; Peters et al., 1995).

Pax-1

Early expression pattern. After 3 to 4 days ofincubation the chick wing and leg buds are swollenridges of tissue projecting from the lateral margins ofthe trunk. At stages 19–20 Pax-1 expression is strong inthe entire series of trunk and tail somites, but not in thelimb buds (Fig. 3A,B). By stages 21–22, however, Pax-1-expressing domains were observed at the proximal,anterior margin of both wing and leg buds (Fig. 3C,D),whose subsequent development is separately describedbelow.

Fig. 1. RT-PCR detection of Pax-1 and Pax-9 mRNA in chick limbbuds. A: DNase treatment eliminates false-positive amplification. Lanes 1and 3: Untreated RNA (2) from whole embryo (day 6) amplified by PCRshows spurious bands from genomic contamination. Lanes 2 and 4:Parallel PCR of DNAse-treated samples (1) shows no amplification foreither set of gene-specific primers (Pax-1 or Pax-9). B: Amplified bandsare reverse-transcription (RT)-dependent. Lanes 1 and 3: RT-PCR ofDNase-treated RNA from whole embryo (day 6) shows amplification ofboth genes. Lanes 2 and 4: Parallel RT-PCR reactions lacking reversetranscriptase enzyme (2) show no amplification. C, D, E: Developmentalprofile of Pax-1, Pax-9, and GAPDH mRNAs in chick limb buds ofsuccessive stages (wings and legs pooled). Expression of both geneswas detected at early limb bud stages continuing through stage 29/30.Each gel represents a parallel RT-PCR reaction on aliquots of the samecDNA using different gene-specific primers. we4, whole embryo day 4(positive control tissue); 20–29, Hamburger-Hamilton stages of chick limbdevelopment; M, DNA marker VIII Boehringer Mannheim; CF, chickembryo fibroblasts (negative control tissue); B, water blank (no cDNAcontrol).

103Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

Wing buds (stages 21–29). At stage 21 the wingbuds are semicircular flaps extending from the lateralbody wall. Pax-1 is initially expressed proximally in acrescent-shaped band just dorsal to where the limb budmeets the flank (Fig. 3C, upper arrow). This bandextends caudally from the anterior margin of the wingand lies immediately ventral to the somites. Over thenext few stages (22–25) the wing extends distally,growing in length rather than breadth. Pax-1 expres-sion persists proximally at the anterior margin of thewing bud (Fig. 3E) while expanding somewhat mediallyand ventrally, so that both dorsal and ventral sides ofthe anterior proximal margin of the wing are darklystained.

At stages 26–27 the wing continues to elongate as thehandplate expands. Proximally, the embryonic shoul-der has developed as a rounded extension bulging fromthe body wall (Yander and Searls, 1980; Searls, 1986,1990). Pax-1 is strongly expressed in this region of thelimb in a cap-shaped domain covering the cranial partof the shoulder (Fig. 3F). Staining in this area extends

Fig. 2. Northern blot analysis of Pax-1 and Pax-9 expression in chicklimb buds. Total RNA samples (20 µg/lane) were hybridized with either achicken Pax-1, Pax-9, or GAPDH probe. The expression profile for bothPax-1 and Pax-9 is similar to that detected by RT-PCR. we, whole embryoday 4 (positive control tissue); 19–29, successive stages of chick limbdevelopment (wings and legs pooled); rRNA, ethidium bromide staining of28S ribosomal RNA.

Fig. 3. Localization of Pax-1 expression in developing chick limb budsby WISH. A: Stage 19 wing buds, lateral view. Pax-1 is expressed in thepharyngeal pouches and somites, but not in the wing bud at this stage.B: Stage 20/21. A faint staining of the anterior leg bud is just visible.C, D: Two views of the same specimen (stage 21/22) showing Pax-1expression at the anterior margin of both wing and leg buds. E: Stage 24.F: Stage 27/8 wing buds, lateral view. Pax-1 is expressed at the cranialmargin of the shoulder and in a craniocaudal stripe just dorsal to theshoulder. G: Stage 27 wing buds, ventral view. Expression is strong in thedeveloping intervertebral disks and the ventral, medial parts of the

shoulder girdle. The craniocaudal stripe is just visible laterally, above thewings. H: Stage 27 wings, cranial view. I: Stage 27 wings, dorsal view. Aconspicuous spot of Pax-1 expression is stained at the posterior margin ofeach wing. J: Stage 29 wings, cranial view. All three major areas of Pax-1expression (sclerotome, shoulder, and craniocaudal stripe) are stillpresent at this later stage (compare with Fig. H). K: Stage 27 leg buds,ventral view L: Stage 29 legs, dorsal view. Additional Pax-1 domainsappear on the posterior margin of the leg surrounding the developingfibula.

104 LeCLAIR ET AL.

ventrally and medially such that areas of Pax-1 expres-sion in both left and right wings nearly meet on theventral midline (Fig. 3G). Immediately dorsal to theshoulder, a second, separate domain of Pax-1 expres-sion is also visible as a craniocaudal stripe runningdorsal to the wing and ventral to the somites (Fig.3F–H). In the trunk, Pax-1 expression has becomeprogressively restricted to the rudiments of the interver-tebral disks, which stain intensely (Fig. 3G).

Slightly older wing buds (stages 27–28) show a novelPax-1 expressing domain on the posterior dorsal sur-face of each wing (Fig. 3I). Staining in this regionappears as a nearly circular spot just distal to thecrevice where the posterior edge of the wing meets thebody (the ‘‘posterior groove’’; Searls, 1990). Althoughsmall, these spots stain intensely and always appearsymmetrically on both wings. By stage 29, these do-mains elongate into dorsally situated stripes runningalong the posterior wing margin (not shown in Fig. 3).Anteriorly, the major Pax-1-expressing areas previ-ously described retain the same configuration: thesclerotomal cells in the axis, the cranial ‘‘shoulder pad’’on each wing, and a craniocaudal stripe along eachflank dorsal to the wings but ventral to the somites(Fig. 3J).

Leg buds (stages 21–29). At stage 21 each leg budshows a distinct round spot of Pax-1 expression proxi-mally, just dorsal to where the anterior leg bud meetsthe flank (Fig. 3D, lower arrow). Unlike the correspond-ing area in the wing bud, the initial leg patch is morecircular and extends approximately the width of asingle somite. Pax-1 in the leg at slightly later stages(22–25) persists in the same location with little change(Fig. 3E) save a slight ventral and medial expansion ofthe staining region. By stages 27–28 each leg has alarge irregular patch of Pax-1 expression at the proxi-mal junction where the leg meets the body (Fig. 3K).There is no indication of Pax-1 expression in the flankdorsal to this region, as seen in the wing. Although eachwing at this stage shows a posterior marginal spot, nocorresponding area of Pax-1 expression is found in thedeveloping leg. Distal areas of Pax-1 expression doappear at a more advanced stage of leg development. Bystage 29, two faint but distinct stripes appear in theposterior part of the leg zeugopod, running proximodis-tally and roughly parallel to each other on either side ofthe dorsoventral margin (Fig. 3L).

Pax-9

Young embryos (stages 20–22) show Pax-9 signal inboth the axial sclerotome and the pharyngeal pouchesbut not in the limb buds (not shown). By stage 23,staining appears at the anterior margin of the leg buds,about midway between the leg base and the apicalectodermal ridge (AER) (Fig. 4A). In slightly olderembryos these leg bud regions stain more intensely,

Fig. 4. Localization of Pax-9 expression in developing chick limb budsby WISH. All specimens were hybridized with the Pax-9 antisense probeexcept Fig. C, which shows a sense probe negative control. A: Stage23/24. Pax-9 signal appears in the developing pharyngeal pouches, thesomites, and the anterior margin of the leg buds. B: Stage 24/25.Expression is now visible on both dorsal and ventral sides of the anteriormargin in all four limb buds. C: Stage 27 legs, ventral view. Sense probecontrol showed no specific color development. D: Stage 27 legs, ventralview. In this view the Pax-9 signal in the limbs appears as paired, parallelstripes. This is due to the overlying apical ectodermal ridge (AER), whichis nonexpressing. Note continued Pax-9 staining in the trunk and tailsclerotome. E: Stage 28, lateral view. F: Stage 29, ventral view. As thewings and legs develop, Pax-9 expression persists at the anterior junctionof the autopod and the zeugopod in each limb.

105Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

while at the anterior margin of each wing bud lighterpatches are observed (Fig. 4B). Both dorsal and ventralsides of each limb are stained, so that the signalappears as two parallel stripes of Pax-9 expressionseparated by the anterior AER (Fig. 4D; sense probecontrol is shown in Fig. 4C). By stages 28–29, Pax-9-expressing cells in the leg are located exclusively at thejunction of the footplate and the zeugopod, immediatelyanterior to the developing ankle and digit I (Fig. 4E,F).In the wing there is a corresponding region anterior tothe wrist and digit II. Thus both wings and legsmaintain a similar expression pattern as each append-age develops its distinctive features.

Serial Sections of Whole-Mounts

Several embryos probed for Pax-1 using WISH werecryoembedded and serially sectioned in either trans-verse or frontal orientations. These sections confirmedgood penetration of the probes throughout the speci-mens, as shown by strong staining of the sclerotomalcells deep within the trunk around the notochord(Fig. 5A,B). Transverse sections of the limb bud regionsat stage 22 show Pax-1 strongly expressed in the dorsalpart of the lateral plate mesoderm just cranial to thewing buds (Fig. 5C) and the leg buds (not shown). Pax-1is expressed only in limb mesoderm, not the overlyingectoderm (Fig. 5D,E).

In situ hybridization of tissue sections. To furtherconfirm the cellular expression patterns of Pax-1 andPax-9, embryos of successive stages (21–30) were seri-ally sectioned and probed with 35S-labeled riboprobes.At least two embryos of each stage were completelysectioned, one in a longitudinal and one in a transverseorientation. Small groups of adjacent sections werearranged on sets of three slides. The first slide of eachset was probed for Pax-1, the second for Pax-9, and thethird was stained with Alcian blue and hematoxylin toassess the progress of chondrogenesis.

Pax-1: Pectoral region. At stages 21–22 Pax-1 isexpressed in the dorsal part of the lateral plate meso-derm just cranial to the wing buds (Fig. 6A). In slightlyolder embryos (stage 23) this expression becomes local-ized to the cranial and ventral parts of the embryonicshoulder (Fig. 6B). By stage 29, the chick shoulder jointconsists of a cartilaginous scapula, humerus, and cora-coid. Pax-1 expression was strongest cranial to andsurrounding the junction of these three elements(Fig. 6E), particularly in dense mesenchymal cellsoccupying the future joint interspace and surroundingthe embryonic perichondria. Expression was absentfrom the cartilages themselves. By following theseskeletal elements through sections farther along theembryonic axis, we observed expression continuingcaudally along the dorsal margin of the scapula andalso ventromedially surrounding the coracoid. Thislatter element is likely responsible for the deep stainingobserved externally on ventromedial parts of the shoul-der in whole-mounts (i.e., Fig. 3G).

Even the less conspicuous regions of Pax-1 expres-sion were detected by our histological survey. Embryosstages 25–26 and older show intense Pax-1 at theshoulder but also laterally in two small, dorsoventrallydirected stripes immediately under the ectoderm of thebody wall (Fig. 6C,D, single arrowhead). This is thecross-sectional profile of the craniocaudal stripes ob-served in whole-mounts. In more caudal sections theposterior marginal spots appear as small patches ofPax-1-positive cells just under the dorsal ectoderm overthe developing humerus (Fig. 6D, double arrowheads).Probing complete serial sections of all specimens re-vealed no other Pax-1 signal in any wing structures.

Fig. 5. Cryosectional views of Pax-1 expression in WISH-stainedembryos. A: Stage 26 pelvic region, transverse section. Pax-1 positivecells (black) appear in the ventrally migrating sclerotome (s) on either sideof the neural tube (nt). B: A frontal section through the same region showsPax-1 staining surrounding the notochord (nc). C: Stage 22 wing buds,transverse section. Pax-1 is expressed in the sclerotome and strongly inthe lateral plate mesoderm slightly cranial to the wing bud. D: Stage 25 legbuds, transverse section. Pax-1 expression persists in the sclerotome andin the ventrolateral parts of the developing hip region (arrowhead).E: Detail of right lateral region in Fig. D. Positively stained cells are foundonly in the limb bud mesoderm, not the overlying ectoderm (arrow-head). d, dermatome; m, myotome; nc, notochord; nt, neural tube; s,sclerotome.

106 LeCLAIR ET AL.

Pax-1: Pelvic region. Pax-1 expression in the develop-ing leg bud initially parallels that of the wing, appear-ing in the lateral plate mesoderm of the developing hipregion (Fig. 7A–C). Just as Pax-1 in the pectoral regionis expressed at the cartilaginous junction of the wingand the girdle, in the pelvic region Pax-1 was detectedbetween the embryonic femur and the developing pel-vis, and in the pelvis itself along the ventrally directedischium (Fig. 7D). Very weak Pax-9 expression wasfound in adjacent sections surrounding the ischiumalone. This was the only cartilage where we saw Pax-1and Pax-9 coexpressed in the same region of the devel-oping limb skeleton. Even here, each gene transcript isconcentrated on a different side of the structure, Pax-1appearing predominantly on the ventral side and Pax-9appearing on the dorsal side.

The distal leg domains of Pax-1 expression, justvisible in whole-mounts of older embryos (stage 29,Fig. 3L), were clearly detected in sections of the develop-ing leg zeugopod. Signal is strongest along the posteriormargin of the leg dorsal to the midshaft of the embry-onic fibula (Fig. 7D) and weakly scattered betweenthe adjacent muscle bundles. More distally, this domainis restricted to a stripe of cells immediately underthe dorsal ectoderm near the tibiotarsal–metatarsaljoint. As in the wing, no transcripts were found in moredistal structures of the leg autopod.

Pax-9: Pectoral and pelvic girdles. Strong Pax-9expression was confined to the anterior midmargin ofthe wing and leg buds. Although weak expression wasfound in one area of the pelvis, as mentioned above, no

expression was found at any level of the pectoral girdle.Thus Pax-9 is largely restricted to distal structuresduring limb development. Expression is first detectedaround stage 23, most intensely in the developinghindlimb. Although Pax-9 expression in whole-mountsappears as parallel stripes on the dorsal and ventralsides of each limb bud, sections show a single, continu-ous population of Pax-9-expressing cells lying betweenthe anterior cartilages and the overlying AER(Fig. 7E,F). By stage 29, weak Pax-9 signal is localizedto loose mesenchymal cells anterior to the distal radialcartilage, the carpometacarpus and metacarpal II (notshown). In the leg, corresponding regions occur anteriorto the distal tibia, tibiale, and metatarsal I (Fig. 7F).Although the expression of this gene is strongest be-tween the AER and the anterior cartilages, additionalweak signal is scattered between the dorsal and ventrallimb muscle masses.

Pax-1 and Pax-9 Expression in LimbMesenchyme Cultures

That Pax-1 and Pax-9 are expressed throughoutembryonic chick limb development led us to analyze ifsuch gene expression is maintained during cartilagecell differentiation in vitro. Because both genes areexpressed in vivo by stage 23, we examined micromasscultures of mesenchyme cells prepared from embryos ofthat stage or slightly older (stage 25). Wing and legbuds were pooled for each stage and the dissociatedcells were plated at high density and cultured for 3days. Alcian blue staining showed multiple cartilagi-

Fig. 6. Localization of Pax-1 expression in the pectoral region by insitu hybridization to sections with 35S-labeled riboprobes. Brightfieldsections (left) are stained with Alcian blue and hematoxylin. Darkfieldsections (right) show the autoradiographic signal obtained by probingadjacent tissue sections. A: Wing transverse section, stage 21, showingearly expression in the lateral mesoderm cranial to the wing buds.B: Wing transverse section, stage 23. Expression is now located in thecranial part of the wing at the shoulder bulge. C: Wing transverse section,

stage 26/7. Expression continues in the shoulder. Also note craniocaudalstripes (arrowheads) on each side of the embryo trunk. D: Wing trans-verse section, stage 29. In this more caudal section the craniocaudalstripe (single arrowhead) and the posterior marginal spot (double arrow-head) are visible. E: Wing transverse section, stage 30/31. As the pectoralgirdle cartilages develop Pax-1 is most strongly expressed around thethree-way junction of the humerus, scapula, and coracoid. C, coracoid;h, humerus; s, scapula.

107Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

nous nodules after 2 days and a strong increase innodule formation after 3 days in culture (see Fig. 10),recapitulating the chondrogenic sequence followed bylimb mesenchyme in vivo.

RT-PCR. RT-PCR analyses showed that both Pax-1and Pax-9 were expressed in dissociated limb bud cellsand differentiating micromasses over the 3-day cultureperiod. In both stage-23 and stage-25 limb mesenchymewe observed similar Pax-1 expression profiles as afunction of time (Fig. 8). Expression was strongest infreshly dissociated cells immediately before plating.Cells harvested 24 hours after plating (day 1) showed adramatic decrease in signal down to barely detectable

levels. Over the next 2 days, however, band intensityincreased as Pax-1 expression in the cultured cellsappeared to rebound. Similar results were obtained forPax-9, although the signal was considerably weaker.Again, the strongest signals were observed in dissoci-ated cells, with slightly higher levels from the older(stage 25) limb mesenchyme. Pax-9 message was barelyamplified from cultures at days 1 and 2, but after 3 daysof growth a faint but detectable band was reestablishedin both sets of cells.

Northern analysis. Similar expression profiles ofPax-1 and Pax-9 mRNA were observed in a separate setof micromass cultures using Northern blotting. Freshly

Fig. 7. Localization of Pax-1 and Pax-9 expression in the pelvic regionand hindlimb by in situ hybridization to sections with 35S-labeled ribo-probes. Brightfield sections (left ) are stained with Alcian blue and/orhematoxylin. Darkfield sections (right ) show the autoradiographic signalobtained by probing adjacent tissue sections. A: Leg transverse section(stage 21) showing early expression in the lateral plate mesoderm justcranial to the leg buds. B: Leg transverse section, stage 23. Pax-1 signalappears in the trunk and tail sclerotome and also proximally where eachleg meets the body. C: Leg transverse section, stage 25/6. Later

expression is further localized to the medial and ventral parts of the hipregion. D: Adjacent transverse sections, stage 29 leg. Pax-1 is expressedaround the ventral part of the ischium, while Pax-9 is weakly expressedaround the dorsal side of the same cartilage element. E: Wing bud crosssection, stage 27. Pax-9 is expressed at the anterior margin of wing andleg buds just under the anterior AER. F: Cross-sectional views of older legbuds, stage 29. Pax-1 is expressed posteriorly around the embryonicfibula (f), while Pax-9 is expressed anteriorly surrounding the distal tibia (t)and the first metatarsal (1–4). fe, femur; il, ilium; is, ischium.

108 LeCLAIR ET AL.

dissociated limb bud cells (day 0) show clear expressionof Pax-1 and Pax-9 transcripts (Fig. 9). Cultures 24hours later (day 1) show weak signals barely abovebackground. Day-2 and day-3 cultures, however, haverenewed expression of both transcripts at levels roughlyequal to freshly dissociated cells. Note that this appar-ent increase in expression cannot be attributed tosimple cell multiplication in culture, as the RNA load ineach lane was normalized as estimated by opticaldensity and ethidium bromide staining.

In situ hybridization. Pax-1 and Pax-9 expressionin vitro was also confirmed by whole-mount in situhybridizations. Positive staining for both transcriptswas observed at all time points (days 1, 2, and 3) withthe most intense staining appearing in the oldestcultures (Fig. 10). No color development was observed

on equivalent cultures stained with Pax-1 sense probes.A faint but detectable purple stain was observed incultures stained with the Pax-9 sense probe, but thislevel was much lower than the signal obtained with theantisense message.

Fig. 8. RT-PCR detection of Pax-1 and Pax-9 expression in limb budmicromass cultures. After dissociation (day 0), expression of both genesinitially declines (day 1) but appears to rebound after several days inculture (2–3). Each gel represents parallel PCR reactions on the samecDNAs using different gene-specific primers. LBM, limb bud mesen-chyme; WE, whole embryo day 4 (positive control tissue); 0–3, days inculture; CF, chick embryo fibroblasts (negative control tissue); H, chickheart (negative control tissue); B, water blank (no cDNA control). Molecu-lar size markers are in the left most lane. * 5 500 bp.

Fig. 9. Northern blot analysis of Pax-1 and Pax-9 mRNAs in micro-mass cultures of stage 23 chick limb mesenchyme. Equivalent lanes ofRNA (20 µg) were probed with Pax-1/GAPDH or Pax-9/GAPDH probes,respectively. Expression of both Pax transcripts increases with time inculture (1–3 days) after an initial decline from the level found in freshlydissociated cells (day 0). CF, chick embryo fibroblasts (negative controltissue); rRNA, ethidium bromide staining of 28S ribosomal RNA.

Fig. 10. Detection of Pax-1 and Pax-9 expression in chick limb budmicromass cultures by WISH. Top two rows: Hybridization with Pax-1and Pax-9 antisense probes over successive days of micromass develop-ment shows increasing levels of Pax-1 and Pax-9 transcripts. Pax-1 andPax-9 sense probes show little or no signal. Bottom row: Alcian bluestaining for cartilage nodules. 1–3, days in culture.

Fig. 11. Localization of Pax-1 message in histological sections ofmicromass cultures hybridized with 32S-labeled probe. A: Alcian bluestaining of a sectioned cartilage nodule from a 3-day-old micromassculture. B: Brightfield image of a similar nodule stained with a Pax-1antisense probe, showing perinodular concentrations (arrows) of exposedsilver grains (black dots).

109Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

By day 3, when cartilage nodules were abundant, weobserved a concentration of positive cells in the undiffer-entiated, internodular areas, while nodular clusters ofchondrocytes showed only background signals. To in-crease the sensitivity of detection and to confirm thespatial localization suggested by the whole-mounts,histological cross sections of 3-day-old micromass cul-tures were hybridized with 35S-labeled probes for Pax-1and Pax-9. Despite the high background noise resultingfrom long exposure times, low hybridization signals forPax-1 and Pax-9 were detected in the central regions ofthe cultures, where cell density was greatest. Mostnoticeable was the absence of signal in large cartilagenodules. This was particularly true with our Pax-1probes where clusters of developed silver grains wereoften seen in the perinodular area surrounding thechondrocyte core (Fig. 11). A similar local concentrationof Pax-9 signal could not be clearly assessed, possiblydue to a combination of low intrinsic expression anddecreased detection sensitivity.

DISCUSSION

To explore the dynamics of Pax-1 and Pax-9 expres-sion during appendicular skeletal development we havecharacterized the spatiotemporal profile of these tran-scripts over an extended period of chick limb morphogen-esis. To investigate the possible functional involvementof these genes during cartilage formation we have alsoexamined the maintenance and localization of Pax-1and Pax-9 expression in chondrifying mesenchymalcultures. These observations complement previous geneexpression studies in vivo while shedding new light onan experimentally accessible system for in vitro studyof Pax gene function and regulation.

During chick development both Pax-1 and Pax-9transcripts first appear in the limb buds at incubationday 4, Pax-1 at approximately stage 21 and Pax-9slightly later at stage 23. Both persist up to 3 days later(stages 29–30). Spatially, Pax-1 is expressed at theproximal anterior margin initially on the dorsal side ofthe limb bud, while Pax-9 appears more distally on bothdorsal and ventral sides of the anterior margin. Thisearly expression pattern is similar to that described forthe mouse embryo (Neubuser et al., 1995), where Pax-1is first expressed on embryonic day 10.5 in the anteriorproximal parts of both fore- and hindlimb buds. Pax-9appears slightly later (day 11.5) at the anterior proxi-mal margin of the hand and foot plates, on both dorsaland ventral sides of the limb. This initial similaritybetween the chick and mouse embryos implies thatearly Pax-1 and Pax-9 expression may be conserved invertebrate limb development.

Beyond this generality, however, the complex expres-sion patterns of these genes in both axial and limbskeletal development raise a number of interestingquestions.

What is the Relationship Between Somiticand Limb Expression of Pax-1 and Pax-9?

Several contrasts can be drawn between the axialand appendicular expression of these two related tran-scripts. In the axis, Pax-1 and Pax-9 are expressed ineach segment and are associated with nearly all of thedeveloping skeletal elements (e.g., vertebral bodies,intervertebral disks, and neural arches). In the limbs,Pax-1 and Pax-9 transcripts first appear in isolatedareas and are later confined to limited regions partiallysurrounding certain cartilages. Many cartilages showno detectable enrichment of Pax-1 or Pax-9 expression.Another contrast is that, although Pax-1 and Pax-9 areexpressed in largely overlapping areas of the sclero-tome, the same gene transcripts in limb mesenchymerarely appear in the same region and have a nearlydisjunct distribution (Fig. 12).

This developmental difference in localization withrespect to the skeleton may be related in part to thedifferent mesodermal precursors of the axis (paraxialmesoderm) and the limbs (somatopleural mesoderm),respectively. These systems are not totally separate,however, as chick somitic cells contribute to both thelimb musculature (Chevallier, 1975; Agnish and Koch-har, 1977; Gumpel-Pinot et al., 1984) and the proximalparts of the limb girdles (i.e., part of the cartilaginousscapula) (Chevallier et al., 1978). It is not known towhat extent paraxial and somatopleural mesodermshare molecular mechanisms for skeletal condensationand differentiation, or how much these systems mayhave diverged in the evolution and diversification of

Fig. 12. Generalized pattern of Pax-1 and Pax-9 expression in adevelopmental series of limb buds (wing or leg). Each limb bud is shownwith the anterior margin at the top. For a detailed description of specificwing and leg patterns, see the Results section. A: Stage 21. B: Stage 23.C: Stage 28/29.

110 LeCLAIR ET AL.

vertebrate limbs. Our analyses show that there areindeed widely separate, extraaxial domains of Pax-1and Pax-9 quite early in limb development, and thatexpression in older limbs grows stronger at a time whenin the axis expression of these genes is declining. Thissuggests that Pax-1 and Pax-9 may be independentlyregulated in the somitic mesoderm and the lateral platemesoderm by spatially confined, stage-specific signals.

What Regulates Pax-1 and Pax-9 Expressionin Different Regions of the Embryo?

A working model of Pax-1 regulation has been formu-lated for the axis (reviewed in Dahl et al., 1997). Thismodel features sonic hedgehog (shh), produced by noto-chord and floor plate cells, as a mediator of Pax-1 in thedeveloping sclerotome (Fig. 13A). Ectopically expressedshh or SHH protein can induce Pax-1 expression andexpand the sclerotome at the expense of dermamyo-tome (Brand-Saberi, 1993; Pourquie et al., 1993; Fanand Tessier-Lavigne, 1994; Johnson et al., 1994; Fan etal., 1995). In contrast, removing chick notochord andfloorplate decreases Pax-1 expression (Ebensperger etal., 1995) and notochord defective mouse mutants showlower Pax-1 levels and ventral vertebral defects (Kosekiet al., 1993). Pax-9 expression also declines in thecaudal axis of these mutant mice, implying that Pax-9

shares a similar dependence on midline structures forits sclerotomal expression (Neubuser et al., 1995). Thushedgehog signaling is likely an important regulator forboth these Pax genes in the embryonic axis.

Is appendicular expression of Pax-1 and Pax-9 regu-lated by the same molecular signals that control axialexpression? Unlike the axial situation where shh-expressing tissues are in close proximity to the scleroto-mal cells they influence, in the limb bud the source ofSHH and the sites of Pax expression are much fartherapart (Fig. 13B). In vivo, shh transcripts are restrictedto the posterior margin of the limb (Riddle et al., 1993)and SHH protein to the posterior half of the mesen-chyme (Lopez-Martınez et al., 1995). AppendicularPax-1 and Pax-9, however, first appear at the anteriormargin of the limb buds. Thus it remains to be seen ifthe early expression of limb bud Pax-1 and Pax-9 isdirectly regulated by long-distance shh signaling (e.g.,Fan and Tessier-Lavigne, 1994; Fan et al., 1995),indirectly regulated by shh via other molecular interme-diates, or regulated independently of the shh pathway.This last possibility has already been investigated inanother chick tissue, the pharyngeal endoderm, wherePax-1 and Pax-9 expression appear intrinsically con-trolled and do not depend on signals from surroundingstructures (Muller et al., 1996).

What Do Pax-1 and Pax-9 Regulate?

Very little is known about Pax gene function at thecellular level (Mansouri et al., 1996; reviewed in Dahl,1997). Several recent studies, however, show that Pax-1is involved in somite patterning, sclerotome differentia-tion, and vertebral morphogenesis. For example, misex-pression of Pax-1 after antisense oligonucleotide treat-ment leads to abnormal axial segmentation and ventralvertebral defects in chick embryos (Barnes et al., 1996a,1996b; Smith and Tuan, 1995) that are phenotypicallysimilar to mouse Pax-1 mutants (Balling et al., 1988;Chalepakis et al., 1991; Gruss and Walther, 1992;Wallin et al., 1994). Thus Pax-1 and (by extension)Pax-9 are thought to have an effect on the proliferation,commitment, or differentiation of prechondrogenic cells(Wallin et al., 1994). One possibility is that Pax genesevoke cell division before organogenesis, their expres-sion declining once a critical cellular mass is reached(Dahl et al., 1997). Supporting evidence comes fromcultured cells (which can be transformed into tumors byoverexpression of Pax-1) (Maulbecker, 1993) and theobservation that Pax-1 and Pax-9 are expressed in theembryonic pharyngeal endoderm and intervertebraldiscs, both highly mitotic sites (Fan et al., 1995).

Although our investigations show that Pax-1 andPax-9 are strongly expressed in chick limb mesenchymeover a long developmental period, in close associationwith major elements of the limbs and limb girdles, therole of these genes in limb development is still equivo-cal. In support of a proliferative function for Pax genesin the limb we observe that Pax-1, in particular, ispresent around large joints (i.e., the massive scapulo-humerus–coracoid joint) where greater cell numbers

Fig. 13. Schematic diagram of Pax-1 regulation in the axial skeletonand the limb. A: In the axis, tissues expressing sonic hedgehog protein(SHH) such as notochord and floorplate (dark gray) are in close proximityto Pax-1-expressing cells (light gray) in the ventral half of the epithelialsomite. B: In the limb bud, SHH is expressed in the posterior part of themesenchyme, directly opposite from the early anterior expression ofPax-1. AER, apical ectodermal ridge.

111Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

may be required. In contrast, we note that neitherPax-1 or Pax-9 are expressed in the most mitotic regionof the limb bud, the distal progress zone immediatelyunder the AER. In areas that do express these tran-scripts, it is not known if they support above-averagelevels of mitosis. Because many chick cartilages arenegative for both Pax-1 and Pax-9 (particularly thedistal ones), these genes are clearly not involved orrequired in all limb chondrogenesis. For example, Pax-1knockout mice with severe vertebral defects developapparently normal pelvic girdles and show only a slightpectoral anomaly in the embryonic acromion process(Timmons et al., 1994). This result implies that the axisis more sensitive than the limbs to the dose of Pax-1message, or that Pax-1 function in the limbs is some-how shared with unknown regulatory factors. Indeed,the lack of a limb phenotype in knockouts suggests thatcompensatory factors might be upregulated when Paxgene expression is absent from conception. Alternately,Pax-1 expression may be entirely vestigial if otherpathways have supplanted Pax function in the evolu-tion of vertebrate limbs.

The chick limb bud now presents another system fordeciphering these and other questions. To previous invivo investigations we add the observation that Pax-1and Pax-9 mRNAs persist in limb mesenchyme in vitro,and that these transcripts are temporally and spatiallyassociated with the progress of chondrogenic differentia-tion. The restriction of expression to the internodularareas is consistent with the situation in the embryo,where Pax-1 and Pax-9 are expressed in loose mesen-chyme cells surrounding precartilage condensations,are downregulated at the onset of chondrocyte differen-tiation, and are absent from mature cartilages. Thatthe micromass system can in part mimic the onset,duration, and localization of these messages duringchondrogenesis may prove a useful tool for dissectingregulatory aspects of these Pax genes, for example, bytesting potential upstream inhibitors or stimulators(e.g., hedgehogs). Likewise, downstream effects may beidentified by manipulating the levels of Pax-1 andPax-9 mRNAs (e.g., by retroviral expression vectors orantisense oligonucleotide treatment) (Stott and Chuong,1997; Stott et al., 1998) and observing the effects on cellproliferation, cell–cell interactions, or synthesis of extra-cellular matrix. Because many cellular and molecularaspects of chondrogenesis in vitro can be quantified,even subtle regulatory influences on the chondrogenicprocess might be detected by these methods. In vivo,local alterations of Pax transcripts by antisense treat-ment or surgical manipulation of hedgehog-rich tissues(i.e., notochord and the posterior limb bud margin)(Riddle et al., 1993) may further unravel the causalcascade upstream and downstream of Pax-1 and Pax-9expression.

Conclusion

The embryonic chick limb shows striking patterns ofPax-1 and Pax-9 expression over an extended develop-

mental period. In this report we have characterized theexpression of these genes in vivo, providing stage-by-stage descriptions of their dynamic patterns in relationto the cartilaginous skeleton. We have also explored themaintenance of this gene expression in chondrifyingcultures of limb bud mesenchyme, demonstrating itspersistence for up to 3 days in vitro. Given the experi-mental legacy of limb bud studies and renewed interestin molecular patterning mechanisms, the developingchick limb bud and its parallel cell-culture systemsshould prove useful tools for deciphering the role ofthese Pax genes in vertebrate skeletal development.

EXPERIMENTAL PROCEDURESRNA Extraction

Chick embryos (Hamburger-Hamilton stages 19–30)were dissected into cold Ringer’s solution and the wingand leg buds removed close to the flank. Isolated limbsbuds were kept in sterile calcium- and magnesium-freesaline with glucose (CMFSG) on ice during dissection.Limb buds from each stage were homogenized in Tri-Reagent (Sigma, St. Louis, MO), frozen at 280°C andlater processed for total RNA according to manufactur-er’s instructions. Similar extractions were performedon whole chick embryos (days 4 and 6), embryonic chickhearts (days 8–10), and cultured chick embryo fibro-blasts.

Micromass Culture and Harvesting

Chick limb bud mesenchyme (stages 23–24 or 25) washarvested, enzymatically dissociated and plated in10 µL micromass cultures (20 3 106 cells per milliliter)according to previously published methods (San Anto-nio and Tuan, 1986; Oberlender and Tuan, 1994). Cellsattached for 2 hours at 37°C and 5% CO2 and then werefed 1 ml of Ham’s F-12 medium supplemented with 10%fetal calf serum, 0.2% chick embryo extract, 50 U/mlpenicillin, and 50 µg/ml streptomycin. Culture mediumwas replaced daily. Total RNA was extracted from cellsuspensions immediately before plating (day 0) andfrom micromass cultures at days 1, 2, and 3.

RNAAnalysis and DNase Pretreatment

Isolated RNA was quantified on the basis of A260 andexamined for integrity by agarose gel electrophoresisand ethidium bromide staining. For each sample, 5 µgof total RNA was treated with 1 µL RNase-free DNase(Promega, Madison, WI) in 1 3 PCR buffer (Perkin-Elmer, Foster City, CA) at 37°C for 10 minutes, followedby heat inactivation (70°C, 15 minutes).

Reverse Transcription-Polymerase ChainReaction

Each pretreated RNA sample (5 µg) was processed forRT (SuperScripty cDNA synthesis kit, Gibco, GrandIsland, NY) according to the manufacturer’s protocol.For PCR, an aliquot of premixed reagents (10 3 PCRbuffer II, 25 mM MgCl2, 10 mM dNTPs, 0.5 µl AmpliTaqAS; Perkin-Elmer) and gene-specific primers (Pax-1,

112 LeCLAIR ET AL.

Pax-9, or GAPDH; 25 pM, IDT Technologies) was addedto the cDNA (total volume: 50 µl). Primer sequenceswere the same as used previously (Sanzo and Tuan,1998). Reactions were performed in a Perkin-ElmerDNA Thermal Cycler under the following conditions:94°C for 5 minutes (denaturation), 72°C for 2 minutes(extension), and 54°C for 2 minutes (annealing). After40 cycles, reactions were prolonged at 72°C for 3minutes, terminated at 4°C, and stored at 220°C. PCRproducts were visualized by gel electrophoresis (2%agarose) and ethidium bromide staining. Expected prod-ucts were observed at 206 bp (Pax-1), 233 bp (Pax-9),and 300 bp (GAPDH).

Northern Blot Analysis

Of the total RNA/sample, 20 µg was denatured withglyoxal, run on a 1% agarose gel with recirculatingbuffer, and transferred to neutral nylon (Hybond-N,Amersham, Arlington Heights, IL). Specific probes forchicken Pax-1 (780 bp SmaI fragment) and Pax-9 (800bp SacI fragment) excluded the highly conserved 58paired-box region. Probes were gel-purified and 30–50ng of probe cDNA or chicken GAPDH cDNA were32P-labeled by random priming (Amersham Mega-primey kit). After hybridization (48 hours, 65°C) theblots were washed in 40 mM sodium phosphate/1% SDS(pH 7.0) for 3 3 10 minutes at 70°C, then exposed toautoradiography film (280°C), and developed after2–10 days.

Manufacture of Riboprobes

A 1.1-kb clone of chicken Pax-1 cDNA and a 1.4-kbfragment of chicken Pax-9 cDNA (both in pBluescript)were used to generate riboprobes for in situ hybridiza-tion. Plasmid DNA was linearized by restriction en-zyme digest, purified on Chroma-Spin 100 gel columns(Clontech, Palo Alto, CA), and precipitated with etha-nol. Then 1 µg of template DNA was used to makedigoxigenin-labeled sense or antisense probes (Boeh-ringer Mannheim Geniusy kit; Indianapolis, IN).

Whole-Mount In Situ Hybridization

In situ hybridization to whole embryos followed themethod of Riddle et al. (1993) with the followingmodifications to improve staining of larger specimens.The concentration of Proteinase K (Boehringer Mann-heim) was increased to 10–30 µg/ml, and digestion timeranged from 10 to 30 minutes at room temperature,depending on embryo size. The substrate concentrationwas increased to 0.50 µg/ml nitroblue tetrazolium(NBT) and 0.13 µg/ml 5-bromo-4-chloro-3-inodyl phos-phate (BCIP) in detection solution at pH 8.0–8.5. Colordevelopment proceeded for 5–6 hours at room tempera-ture, then overnight at 4°C on a rocking platform (totaltime: 20–24 hours). Finished embryos were thoroughlyrinsed in PBS containing 0.1% Tween-20 and 10 mMEDTA (PBTE). To reduce background, selected embryoswere briefly dehydrated in 100% methanol (2–3 min-utes), rehydrated in several changes of PBTE, and

stored at 4°C. Specimens were photographed in PBTEunder a dissecting microscope using Kodak Ekta-chrome 160T color slide film. Specific staining wasevaluated by noting symmetrical patterns in left andright limbs of the same specimen, and by comparingsense and antisense hybridizations.

Cryosectioning of Stained Embryos

Selected WISH-stained embryos were equilibratedovernight in 30% sucrose/PBTE and freeze-embeddedin O.C.T. Compound (Fisher, Pittsburgh, PA) usingliquid nitrogen-cooled isopentane. 16-mm serial sec-tions were mounted on Colorfrost Plus slides (Fisher)and immediately coverslipped in glycerol for viewingunder Nomarski optics. Other sections were vacuum-dried, rehydrated in a graded ethanol series, and brieflycounterstained (2–3 seconds) in dilute (1:5) alcoholiceosin Y (Sigma). Stained slides were dehydrated to100% ethanol, cleared in Histoclear (Amresco, Solon,OH), and mounted in Clarion (Biomeda, Foster City,CA), a xylene-free mounting medium.

In Situ Hybridization With 35S-Labeled Probes

In situ hybridization to sections was performed ac-cording to the detailed protocol of Wawersik and Ep-stein (1999). 35S-labeled sense and antisense probes forPax-1 and Pax-9 were prepared at approximately 1–2 3106 CPM/µl, and 2–4 3 106 CPM in 50 µl of hybrid-ization buffer was applied to each slide for overnighthybridization at 50°C. After stringent washing, dehy-dration, and preliminary autoradiography, slides weredipped in emulsion and exposed at 4°C for 7–14 days.After photographic development, the slides were coun-terstained with hematoxylin and eosin and viewed withsequential brightfield and darkfield illumination on aJenaval microscope.

In Situ Hybridization to Micromass Cultures

Micromass in situ hybridization was performed withthe same reagents and probes used for whole-mounts,with the following modifications. The hydrogen perox-ide step normally used to bleach whole embryos wasomitted to prevent gas bubbles. Cells were permeabi-lized in 10 µg/ml Proteinase K for 10 minutes at roomtemperature. Hybridization and antibody steps were ofthe same duration (overnight), but intervening washtimes were shortened and color developed after only2–3 hours at room temperature.

ACKNOWLEDGMENTS

We thank Dr. Brian Mariani and Dr. Yefu Li forpatient and expert help in RNA analysis, Dr. JamesSanzo for sharing his in situ protocol for micro-mass cultures, Dr. Jonathan Epstein (University ofPennsylvania) for demonstrating hybridization with35S-labeled probes, Ken Campbell and Melissa Tarbyfor technical assistance in cell culture and histology,and Peter Alexander and Karen Clark for their helpfulreview of our manuscript. This work was supported

113Pax-1 AND Pax-9 IN LIMB DEVELOPMENT

in part by NIH grants HD15822, ES 07005, andDE 11327 to R.S. Tuan. E.E. LeClair is the recipient ofan institutional NIH training grant (AR 07583) and anindividual NIH National Research Service Award(HD 830180).

NOTE ADDED IN PROOF

A recent report (Hofmann et al., 1998) demonstratesa similar expression pattern of Pax-1 in the early chicklimb bud and explores its regulation by shh and bonemorphogenetic proteins (BMPs).

REFERENCESAgnish ND, Kochhar DM. 1977. The role of somites in the growth and

early development of mouse limb buds. Dev Biol 56:174–183.Balling R, Deutsch U, Gruss P. 1988. undulated, a mutation affecting

the development of the mouse skeleton, has a point mutation in thepaired box of Pax-1. Cell 55:531–535.

Barnes GL, Hsu CW, Mariani BD, Tuan RS. 1996a. Chicken Pax-1gene: structure and expression during embryonic somite develop-ment. Differentiation 61:13–23.

Barnes GL Jr, Mariani BD, Tuan RS. 1996b. Valproic acid-inducedsomite teratogenesis in the chick embryo: relationship with Pax-1gene expression. Teratology 54:93–102.

Bopp D, Burri M, Baumgartner S, Frigerio G, Noll M. 1986. Conserva-tion of a large protein domain in the segmentation gene paired andin functionally related genes of Drosophila. Cell 47:1033–1040.

Brand-Saberi B. 1993. The ventralizing effect of the notochord onsomite differentiation in chick embryos. Anat Embryol 188:239–245.

Brand-Saberi B, Wilting J, Ebensperger C, Christ B. 1996. Theformation of somite compartments in the avian embryo. Int J DevBiol 40:411–420.

Burri M, Tromvoukis Y, Bopp D, Frigerio G, Noll M. 1989. Conserva-tion of the paired domain in metazoans and its structure in threeisolated human genes. EMBO J 8:1183–1190.

Campbell JT, Kaplan FS. 1992. The role of morphogens in endochon-dral ossification. Calcif Tissue Int 50:283–289.

Chalepakis G, Fritsch R, Fickenscher H, Deutsch U, Goulding M,Gruss P. 1991. The molecular basis of the undulated/Pax-1 muta-tion. Cell 66:873–884.

Chevallier A. 1975. Role du mesoderme somitique dans le developpe-ment de la cage thoracique de l’embryon d’oiseau. I. Origine dusegment sternal et mecanismes de la differentiation des cotes. JEmbryol Exp Morphol 33:291–311.

Chevallier A, Kieny M, Mauger A. 1978. Limb-somite relationship:effect of removal of somitic mesoderm on the wing musculature. JEmbryol Exp Morphol 43:263–278.

Dahl E, Koseki H, Balling R. 1997. Pax genes and organogenesis.BioEssays 19:755–765.

Deutsch U, Gruss P. 1991. Murine paired domain proteins as regula-tory factors of embryonic development. Semin Dev Biol 2:413–424.

Deutsch U, Dressler GR, Gruss P. 1988. Pax-1, a member of a pairedbox homologous murine gene family, is expressed in segmentedstructures during development. Cell 53:617–625.

Dressler GR, Deutsch U, Balling R, Simon D, Guenet J-L, Gruss P.1988. Murine genes with homology to Drosophila segmentationgenes. Development 104(Suppl):181–186.

Ebensperger C, Wilting J, Brand-Saberi J, Mizutani Y, Christ B,Balling R, Koseki H. 1995. Pax-1, a regulator of sclerotome develop-ment is induced by notochord and floor plate signals in avianembryos. Anat Embryol 191:297–310.

Fan C-M, Tessier-Lavigne M. 1994. Patterning of mammlian somitesby surface ectoderm and notochord: evidence for sclerotome induc-tion by a hedgehog homolog. Cell 79:1175–1186.

Fan C-M, Porter J, Chiang C, Chang DT, Beachy PA, Tessier-LavigneM. 1995. Long-range sclerotome induction by sonic hedgehog: directrole of the amino-terminal cleavage product and modulation by thecyclic AMP signaling pathway. Cell 81:457–465.

Fang J, Hall BK. 1997. Chondrogenic cell differentiation from mem-brane bone periostea. Anat Embryol 196:349–362.

Glardon S, Callaerts P, Halder G, Gehring W. 1997. Conservation ofPax-6 in a lower chordate, the ascidian Phallusia mammillata.Development 124:817–825.

Gruss P, Walther C. 1992. Pax in development. Cell 69:719–722.Gumpel-Pinot M, Ede DA, Flint OP. 1984. Myogenic cell movement in

the developing avian limb bud in presence and absence of apicalectodermal ridge (AER). J Embryol Exp Morphol 80:105–125.

Hall BK, Miyake T. 1995. Divide, accumulate, differentiate: cellcondensation in skeletal development revisited. Int J Dev Biol39:881–193.

Hamburger V, Hamilton H. 1951. A series of normal stages in thedevelopment of the chick embryo. J Morphol 88:49–92.

Hill RE, Hanson IM. 1992. Molecular genetics of the Pax gene family.Curr Opin Cell Biol 4:967–972.

Hofmann C, Drossopoulou G, McMahon A, Balling R, Tickle C. 1998.Inhibitory action of BMPs on Pax1 expression and on shoulder girdleformation during limb development. Dev Dyn 213:199–206.

Holland ND, Holland LZ, Zozmik Z. 1995. An amphioxus Pax gene,AmphiPax-1, expressed in embryonic endoderm, but not in meso-derm: implications for the evolution of class I paired-box genes. MolMarine Biol Biotechnol 4:206–214.

Hunt P, Clarke JD, Buxton P, Ferretti P, Thorogood P. 1998. Segmenta-tion, crest prespecification and the control of facial form. Eur J OralSci 106(Suppl):12–18.

Johnson RL, Laufer E, Riddle RD, Tabin C. 1994. Ectopic expression ofsonic hedgehog alters dorso-ventral patterning of somites. Cell79:1165–1173.

Kessel M, Gruss P. 1990. Murine developmental control genes. Science249:374–379.

Koseki H, Wallin J, Wilting J, Mizutani Y, Kispert A, Ebensperger C,Herrmann BG, Christ B, Balling R. 1993. A role for Pax-1 as amediator of notochordal signals during the dorsoventral specifica-tion of vertebrae. Development 119:649–660.

LeDouarin NM, Dupin E, Ziller C. 1994. Genetic and epigeneticcontrol in neural crest development. Curr Opin Genet Dev 4:685–695.

Lopez-Martinez A, Chang DT, Chiang C, Porter JA, Ros MA, SimandlBK, Beachey PA, Fallon JF. 1995. Limb-patterning activity andrestricted posterior localization of the amino-terminal product ofsonic hedgehog cleavage. Curr Biol 5:791–796.

Love JM, Tuan RS. 1993. Pair-rule gene expression in the somiticstage chick embryo: association with somite segmentation andborder formation. Differentiation 54:73–83.

Mansouri A, Hallonet M, Gruss P. 1996. Pax genes and their role in celldifferentiation and development. Curr Opin Cell Biol 8:851–857.

Maulbecker CC. 1993. The oncogenic potential of Pax genes. EMBO J12:2361–2367.

Muller TS, Ebensperger C, Neubuser A, Koseki H, Balling R, Christ B,Wilting J. 1996. Expression of avian Pax-1 and Pax-9 is intrinsicallyregulated in the pharyngeal endoderm, but depends on environmen-tal influences in the paraxial mesoderm. Dev Biol 178:403–417.

Neubuser A, Koseki H, Balling R. 1995. Characterization and develop-mental expression of Pax-9, a paired-box-containing gene related toPax-1. Dev Biol 170:701–716.

Oberlender SA, Tuan RS. 1994. Expression and functional involve-ment of N-cadherin in embryonic limb chondrogenesis. Develop-ment 120:177–187.

Peters H, Doll U, Niessing J. 1995. Differential expression of thechicken Pax-1 and Pax-9 gene: in situ hybridization and immunohis-tochemical analysis. Dev Dyn 203:1–16.

Pourquie O, Coltey M, Teillet MA, Ordahl C, Le Douarin NM. 1993.Control of dorsovantral patterning of somitic derivatives by noto-chord and floor plate. Proc Natl Acad Sci USA 90:5242–5246.

Riddle R, Johnson RL, Laufer E, Tabin C. 1993. Sonic hedgehogmediates the polarizing activity of the ZPA. Cell 75:1401–1416.

San Antonio JD, Tuan RS. 1986. Chondrogenesis of limb bud mesen-chyme in vitro: stimulation by cations. Dev Biol 115:313–324.

Sanzo JF, Tuan RS. 1998. High sensitivity analysis of gene expressionin single embryonic somites using coupled reverse transcription-polymerase chain reaction. Mol Biotechnol 9:7–15.

114 LeCLAIR ET AL.

Searls RL. 1986. A description of caudal migration during growthleading to the formation of the pericardial and pleural coeloms, tocaudal movement of the aortic arches, and to development of theshoulder. Am J Anat 177:271–283.

Searls RL. 1990. Development of the embryonic chick wing bud fromstage 24 to stage 32. Am J Anat 188:307–318.

Smith CA, Tuan RS. 1994. Human PAX gene expression and develop-ment of the vertebral column. Clin Orthop Relat Res 302:241–250.

Smith CA, Tuan RS. 1995. Functional involvement of Pax-1 in somitedevelopment: somite dysmorphogenesis in chick embryos treatedwith Pax-1 paired-box antisense oligonucleotide. Teratology 52:333–345.

Stott NS, Chuong CM. 1997. Dual action of sonic hedgehog onchondrocyte hypertrophy: retrovirus mediated ectopic sonic hedge-hog expression in limb bud micromass culture induces novel carti-lage nodules that are positive for alkaline phosphatase and type Xcollagen. J Cell Sci 110:2691–2701.

Stott NS, Lee YS, Chuong CM. 1998. Retroviral gene transfer inchondrogenic limb bud micromass cultures. Biotechniques 24:660–666.

Stuart ET, Kioussi C, Gruss P. 1993. Mammalian Pax genes. Annu RevGen 27:219–236.

Timmons PM, Wallin J, Rigby PWJ, Balling R. 1994. Expression andfunction of Pax-1 during development of the pectoral girdle. Develop-ment 120:2773–2785.

Wallin J, Mizutani Y, Imai K, Miyashita N, Moriwaki K, Taniguchi M,Koseki H, Balling R. 1993. A new Pax gene, Pax-9, maps to mousechromosome 12. Mammal Genome 4:354–358.

Wallin J, Wilting J, Koseki H, Fritsch R, Christ B, Balling R. 1994. Therole of Pax-1 in axial skeletal development. Development 120:1109–1121.

Walther C, Guenet J-L, Simon D, Deutsch U, Jostes B, Goulding MD,Plachov D, Balling R, Gruss P. 1991. Pax: a murine multigene familyof paired box-containing genes. Genomics 11:424–434.

Wawersik S, Epstein JA. 1999. Gene expression analysis by in situhybridization: radioactive probes. In: Tuan RS, Lo CW, editors.Developmental Biology Protocols. Totowa, NJ: Humana Press; Inpress.

Yander G, Searls RL. 1980. An experimental analysis of shoulder andlateral body wall development in the chick embryo. J Exp Zool214:79–92.

115Pax-1 AND Pax-9 IN LIMB DEVELOPMENT