Cre-mediated gene inactivation
demonstrates that FGF8 is required
for cell survival and patterning of the
first branchial arch
Andreas Trumpp,1,4Michael J. Depew,2John L.R. Rubenstein,2J. Michael Bishop,1
and Gail R. Martin3,5
1G.W. Hooper Foundation, Department of Microbiology, School of Medicine, University of California at San Francisco
(UCSF), San Francisco, California 94143-0552 USA;2Nina Ireland Laboratory of Developmental Neurobiology, Department
of Psychiatry, School of Medicine, UCSF, San Francisco, California 94143-0984 USA;3Department of Anatomy and Program
in Developmental Biology, School of Medicine, UCSF, San Francisco, California 94143-0452 USA
In mammals, the first branchial arch (BA1) develops into a number of craniofacial skeletal elements including
the jaws and teeth. Outgrowth and patterning of BA1 during early embryogenesis is thought to be controlled
by signals from its covering ectoderm. Here we used Cre/loxP technology to inactivate the mouse Fgf8 gene
in this ectoderm and have obtained genetic evidence that FGF8 has a dual function in BA1: it promotes
mesenchymal cell survival and induces a developmental program required for BA1 morphogenesis. Newborn
mutants lack most BA1-derived structures except those that develop from the distal-most region of BA1,
including lower incisors. The data suggest that the BA1 primordium is specified into a large proximal region
that is controlled by FGF8, and a small distal region that depends on other signaling molecules for its
outgrowth and patterning. Because the mutant mice resemble humans with first arch syndromes that include
agnathia, our results raise the possibility that some of these syndromes are caused by mutations that affect
FGF8 signaling in BA1 ectoderm.
[Key Words: Agnathia; Barx1; BMP4; endothelin-1; FGF8; first branchial arch]
Received September 13, 1999; revised version accepted October 19, 1999.
In vertebrates, many structures including parts of the
face develop from small primordia or “buds” consisting
of undifferentiated mesenchymal cells covered by a layer
of epithelium. One of these primordia is the first bran-
chial arch (BA1), which in mammals develops into teeth,
skeletal elements of the jaws, lateral skull wall, and
middle ear, as well as part of the tongue and other soft
tissue derivatives. In the mouse embryo, BA1 first be-
comes apparent at the six- to eight-somite stage [approxi-
mately embryonic day (E) 8.25] as a small swelling on the
side of the head. This bud rapidly increases in size as
cranial neural crest cells migrate into and proliferate
within the arch. This neural crest-derived mesenchyme,
which is termed ectomesenchyme and is localized im-
mediately subadjacent to the covering epithelium, differ-
entiates into cartilagenous (chondrocranial) and osseous
(dermatocranial) structures. The central core of BA1
mesenchyme is derived from somitomeres and forms
craniofacial muscle and vascular tissue. At ∼E9.5 the
outgrowing BA1 on each side of the head develops into
the primordia of the mandibular and maxillary arches,
which grow toward the ventral midline. Subsequently,
multiple fusions involving the paired mandibular arches,
maxillary arches, and the frontonasal process establish
the basic form of the face. Errors in these complex mor-
phogenetic events cause craniofacial anomalies includ-
ing defects of the mandible, which are among the most
common malformations in humans. More than 130 hu-
man syndromes appear to involve incorrect development
of the BA1. The most severe of these syndromes are as-
sociated with agnathia, a condition in which the lower
jaw and other BA1-derived structures are absent (for re-
view, see Bixler et al. 1985; Escobar 1993).
Classic experimental embryological studies have sug-
gested that development of BA1 mesenchyme is con-
trolled by signals from its covering ectoderm, which
regulate cell proliferation, survival, patterning, and dif-
ferentiation. In turn, signals from the mesenchyme may
4Present address: Swiss Institute for Experimental Cancer Research
(ISREG), CH 1066, Epalinges, Lausanne, Switzerland.
E-MAIL firstname.lastname@example.org; FAX (415) 476-3493
3136 GENES & DEVELOPMENT 13:3136–3148 © 1999 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/99 $5.00; www.genesdev.org
influence development of the ectoderm. The molecular
basis of these epithelial–mesenchymal interactions is
not yet well understood. Genetic analysis in mice has
provided evidence that members of several homeobox
gene families are necessary for normal BA1 development
(for review, see Francis-West et al. 1998). However, rela-
tively little is known about the precise role of the epi-
thelium and the signaling molecules it produces in the
regulation of outgrowth and patterning at early stages of
Members of the FGF family, particularly FGF8, have
been implicated as epithelial signals that regulate gene
expression during BA1 development. When beads soaked
in FGF8 (FGF8-beads) were inserted into isolated man-
dibular mesenchymal explants, expression of several
genes could be induced (for review, see Peters and Balling
1999). Although such bead implantation studies provide
valuable clues to the identity of signaling molecules and
potential target genes, the long-term consequences of al-
tering gene expression on skeletal patterning cannot be
addressed in organ culture. Furthermore, firm conclu-
sions about gene function, particularly when several
members of a multigene family are coexpressed, require
loss-of-function studies. However, standard genetic ap-
proaches cannot be used to analyze FGF8 function in
craniofacial development because Fgf8−/−embryos die
during gastrulation (Sun et al. 1999). Here we have used
Cre/loxP technology to circumvent this problem and to
generate mutant embryos in which the Fgf8 gene is in-
activated in BA1 ectoderm. Our analysis identifies FGF8
as an epithelial signal essential for the outgrowth and
patterning of BA1 from the earliest stages of its develop-
Cre-mediated inactivation of Fgf8 in BA1 ectoderm
Inactivation of a particular gene in a specific tissue can
be achieved by mating mice carrying a mutant allele in
which essential regions of the gene are flanked by loxP
sites (the recognition sequence for the site-specific DNA
recombinase, Cre) with mice that express the cre gene in
the tissue of interest (Gu et al. 1994; Tsien et al. 1996;
Kulkarni et al. 1999). To inactivate Fgf8 in BA1, we used
mice carrying Fgf8flox, an allele in which vital coding
exons are flanked by loxP sites. Fgf8floxhas wild-type
Fgf8 activity but can be converted to a null allele
(Fgf8?2,3) by Cre-mediated recombination (Meyers et al.
1998). Using the mating scheme outlined in Figure 1A
we produced animals that were compound heterozygotes
for Fgf8floxand a null allele of Fgf8, and which also car-
ried Nes–cre1, a transgene containing a modified cre
gene (Lewandoski et al. 1997) under the control of the rat
Nestin promoter and intron-2 enhancer (Zimmerman et
al. 1994; A. Trumpp, G.R. Martin, and J.M. Bishop, un-
publ.). In Fgf8flox/Fgf8 null;Nes–cre1 embryos, hereafter
referred to as Fgf8;Nes–cre or mutant embryos, Cre-me-
diated conversion of Fgf8floxto a null allele occurs only
in cells that express Nes–cre1, resulting in complete loss
of Fgf8 gene function in those cells and their descen-
dants. Their Fgf8flox/Fgf8 null littermates that did not
inherit Nes–cre1 are phenotypically normal and served
The stage and tissue specificity of Nes–cre1 activity
was determined by crossing males carrying Nes–cre1 to
females carrying the Z/AP reporter gene, which pro-
duces human alkaline phosphatase (hAP) only after it
has undergone Cre-mediated recombination (Lobe et al.
1999). A detailed description of the results of that analy-
sis will be reported elsewhere. Here we focused on an
assessment of cre activity in the head. In Nes–cre1;Z/AP
embryos assayed at ∼E9.0 (16-somite stage), little hAP
activity was detected in the brain, but strong activity
was detected in surface ectoderm, particularly that cov-
ering BA1 (Fig. 1B,C). The domain of Nes–cre1 activity in
BA1 ectoderm had a sharp caudal limit at the level of the
first pharyngeal groove (red arrowheads in Fig. 1B,C).
Analysis at earlier stages suggested that Cre-mediated
recombination first occurred in the cranial surface ecto-
derm at ∼E7.75, and that subsequently, recombination
occurred in cells on the rostral side of BA1 before it oc-
curred in those on the caudal side (data not shown; see
During BA1 development, Fgf8 is expressed in a dy-
namic pattern in the surface ectoderm (Crossley and
Martin 1995; Mahmood et al. 1995; Kettunen and
Thesleff 1998). At E8.5, when the nascent arch is a dis-
crete bud, Fgf8 expression is detected at low levels
throughout the ectoderm, but subsequently becomes
more robust in the rostral- and caudal-most aspects of
BA1 (Fig. 1D,E; data not shown). By E9.5, Fgf8 expression
becomes restricted to the ectoderm on the rostral side.
By E10.5, when the maxillary and mandibular primordia
have expanded, Fgf8 is abundant in the so-called “oral
ectoderm,” which covers the caudal side of the maxillary
and the rostral side of the mandibular arches (see Fig.
In Fgf8;Nes–cre embryos, Fgf8 is inactivated in the sur-
face ectoderm of prospective BA1. By E9.0 (14–18
somites), no Fgf8 RNA was detected in BA1, although it
was detected in all other normal Fgf8 expression do-
mains (n = 5; Fig. 1G,H; data not shown). Significantly,
in two slightly younger embryos a small patch of Fgf8-
expressing cells was detected on the caudal side of BA1
(yellow arrow, Fig. 1F). This patch appeared to be a sub-
set of the normal caudal Fgf8 expression domain. The
presence of this patch confirms that inactivation does
not occur synchronously across the arch, and shows that
some cells on the caudoproximal side transiently express
Mutant embryos survive to birth but lack most
The swelling that heralds BA1 development is evident in
Fgf8;Nes–cre embryos, but by E9.0 it is clearly smaller
than normal (Figs. 1E,H and 2A,B). As the embryos ma-
ture, there is relatively little expansion of either the
mandibular or maxillary primordia, whereas develop-
FGF8 function in first branchial arch development
GENES & DEVELOPMENT3137
ment in neighboring regions, such as the hyoid arch (sec-
ond branchial arch), appears normal (Fig. 2A–D; data not
shown). It is important to note, however, that outgrowth
of the mutant BA1 is not completely arrested, as the
different Fgf8 alleles present in this cross are illustrated in the box on the left. The Fgf8floxallele has wild-type (wt) activity and can
be converted to a null allele (Fgf8?2,3) by Cre-mediated recombination (Meyers et al. 1998). The Nes–cre1 gene used in this cross is
imprinted, and is active in somatic tissues as described (see text) only when it is paternally inherited. However, it is active in the germ
line irrespective of whether it is paternally or maternally inherited (A. Trumpp, B. Bates, R. Jaenisch, G.R. Martin, and J.M. Bishop,
unpubl.). In the cross performed, the parental male has a normal phenotype because he carries a maternally inherited Nes–cre1
transgene (blue filled oval), which has little activity in somatic cells. When this transgene is transmitted to his offspring (unfilled oval),
it is active in somatic tissues. In addition, this male transmits two different Fgf8 null alleles: Fgf8?2,3n(shown in green) and Fgf8?2,3
(shown in red), which is generated by Cre-mediated recombination in his germ line of the Fgf8floxallele he carries (also shown in red).
Thus, all offspring carry a paternally inherited Fgf8 null allele, as well as a maternally inherited Fgf8floxallele (shown in black). In the
50% of offspring that inherit Nes–cre1, the floxed allele (*Fgf8flox) undergoes Cre-mediated recombination, resulting in tissue-specific
inactivation of Fgf8. (B,C) Assay for Cre-mediated recombination using the Z/AP reporter gene (Lobe et al. 1999). Cells in which the
Nes–cre1 gene was active produce human alkaline phosphatase (hAP). (B) Nes–cre1;Z/AP embryo at E9.0 (16 somites) stained for hAP
activity in whole mount. The dashed line indicates the level of the section shown in C. The caudal limit of recombination (marked
by red arrowheads) is located in the first pharyngeal groove. Note that recombination occurs in BA1 ectoderm. (D–H) Fgf8 expression
as detected by whole mount RNA in situ hybridization in D and E control and F to H Fgf8;Nes–cre mutant embryos at E9.0 (14
somites). The dashed boxes in D and G indicate the regions shown at higher magnification in E and H, respectively. In the mutant
embryo, the levels of Fgf8 RNA in most expression domains are similar to those in the normal embryo (D). However, they appear to
be reduced in the front of the face/anterior neural ridge, and, significantly, no Fgf8 RNA is detected in BA1. Black and white arrows
in E point to the normal Fgf8 expression domains on the rostral and caudal side of BA1, respectively. (F) Fgf8 RNA is detected on the
caudal side of BA1 (yellow arrow) in a mutant embryo at a slightly earlier stage. (ANR) Anterior neural ridge; (BA1) first branchial arch;
(BA1*) mutant BA1; (Ca) caudal; (Di) distal; (Ect) ectoderm; (Fb) forebrain; (Hy) hyoid arch (second branchial arch); (Is) isthmic
constriction; (PG) pharyngeal groove; (Pr) proximal; (PS) primitive streak; (Ro) rostral.
Inactivation of the Fgf8 gene in BA1 ectoderm. (A) Breeding scheme used to generate 50% mutant offspring. The three
Trumpp et al.
3138GENES & DEVELOPMENT
embryos (B,D) at E9.5 (A,B) and E11.25 (C,D). The yellow arrow in D points to a small outgrowth in the mutant BA1 primordium
(BA1*) adjacent to the external acoustic meatus (EAM). The white arrowhead points to the region in which the distal ends of the paired
BA1* meet. (E,F) Lateral views of control (E) and mutant (F) embryos at E16.5. (G–N) Skeletal preparations (cartilage stained blue and
bone stained red) of control (G,I,K,M) and mutant (H,J,L,N) embryos. (G,H) Lateral view of E14.5 chondrocranium. Note the absence
of the body of Meckel’s cartilage (bMC) and ala temporalis (AT) in the mutant embryo. The arrow points to the region in which the
AT is normally found. A vestigial rostral process (RP*) is present. (I–L) Newborn skulls. (I,J) Ventral views in which the hyoid and roof
bones have been removed from both control and mutant, and the dentary has been removed from the control skull. (K,L) Lateral views.
Note that in the mutant the dentary (De) is absent, except for the RP* with vestigial incisors (In*) and associated alveolar bone. The
distal maxilla (Mx*) and a vestigial squamosal (Sq*) are present. (M,N) Lateral view of E16.5 embryos showing the region in which the
ear is developing. The arrowhead in N points to ectopic cartilage of unknown identity found in association with the tegmen tympani.
Ossified elements in the region where the gonial and tympanic normally form are indicated by an asterisk. (O) Schematic diagram
depicting BA1-derived craniofacial elements of normal and mutant skulls. Although the origin of the squamosal is controversial, it is
shown here as being BA1 derived. Abbreviations as in Fig. 1. (Al) Alisphenoid; (BS) basisphenoid; (Ey) eye; (Gn) gonial; (Hy) hyoid arch;
(Inc) incus; (Jg) jugal; (Ma) malleus; (MC) Meckel’s cartilage; (Mo) molars; (Mx) maxillary process; (Pl) palatine; (PS) presphenoid; (Pt)
pterygoid; (PX) premaxilla; (SP) styloid process; (Ty) tympanic; (Vg) swelling over trigeminal ganglion.
Phenotypic analysis of Fgf8;Nes–cre embryos. (A–D) Scanning electron microscope images of control (A,C) and mutant
GENES & DEVELOPMENT 3139
mandibular primordia extend distally and meet at the
midline (white arrowhead in Fig. 2D; see also Fig.
6D,H,L). Furthermore, at E11.25 a distinct hillock (yel-
low arrow in Fig. 2D) forms in caudoproximal BA1, im-
mediately adjacent to the external acoustic meatus
(EAM). This hillock appears to form in the region in
which Fgf8 expression was detected transiently at E9.0
(yellow arrow in 1F).
Fgf8;Nes–cre mutants die shortly after birth. New-
borns appear grossly normal except for severe craniofa-
cial defects and the presence of an ectodermal covering
over the prospective mouth (Fig. 2E,F; data not shown).
The lungs do not inflate suggesting that lethality is
caused by anoxia. Characterization of the craniofacial
defects in Fgf8;Nes–cre embryos revealed that cartilag-
enous elements thought to develop from BA1 by E14.5,
such as the ala temporalis and incus (maxillary arch-
derived), and body of Meckel’s cartilage (mandibular
arch-derived) fail to develop. However mutant embryos
have a vestigial malleus (Ma*) and rostral process (RP*),
the distal symphyseal end of Meckel’s cartilage (Fig.
Normally, between E14.5 and birth, a number of bones
develop from BA1 neural crest-derived mesenchyme
(summarized in Fig. 2O). In Fgf8;Nes–cre newborns,
most of these bones are absent. A few mandibular arch-
derived elements remain, including alveolar bone asso-
ciated with RP* (Fig. 2L), Ma*, and some small ossified
elements in the general region of the gonial and tym-
panic bones (asterisk in Fig. 2N). Of the maxillary arch-
derived dermal elements, the mutants contain the distal
maxilla (Mx*) and portions of the squamosal (Fig. 2L).
Molars are absent, but vestigial lower incisors are ob-
served in association with RP* (Figs. 2L and 5H). A sum-
mary of skeletal development in Fgf8;Nes–cre mutants is
presented in Figure 2O.
Newborn mutants had other abnormalities consistent
with a failure of BA1 development, including a small
disorganized tongue (microglossia) and a truncated and
misrouted mandibular division of the trigeminal nerve
(data not shown). There was some variation in the mu-
tant phenotype. The majority (28 of 42) displayed the
phenotype described above, whereas the remaining mu-
tants showed a slightly less severe phenotype, with a
small increase in the extent of dermal bone develop-
ment. In many cases, this less severe phenotype was ob-
served on only one side of the head, which is reminiscent
of asymmetries found in a variety of human congenital
syndromes (for review, see Escobar 1993).
FGF8 is necessary for survival of mesenchymal cells
Analysis of Fgf8-null mutant homozygotes has shown
that Fgf8 is required for cell migration at primitive streak
stages of development (Sun et al. 1999). Therefore, we
speculated that the early failure of BA1 outgrowth in
mutant embryos might be due to a lack of neural crest
cell migration into the BA1 primordium. To investigate
this possibility we assayed for Cad6, Crabp1, and Ap2.2
expression, which marks migrating neural crest cells
(Mitchell et al. 1991; Maden et al. 1992; Ruberte et al.
1992; Inoue et al. 1997). Expression of these markers was
similar in mutant and normal embryos (Fig. 3A,B; data
not shown), indicating that despite loss of Fgf8 function
in BA1, the neural crest cell population follows its nor-
mal migration path and appears normal in size.
To determine why the mutant BA1 is hypoplastic, we
then studied cell proliferation by assaying for BrdU in-
corporation. BrdU-labeled cells were detected in the mu-
tant BA1 at E9.5, at least 12 hr after Cre-mediated inac-
tivation of Fgf8, and the ratio of BrdU-labeled to unla-
beled cells appeared roughly the same as in the normal
arch (data not shown). This result suggests that FGF8 is
(C–F) Cell death as assayed by Nile blue sulfate (NBS) staining in whole mount. Lateral (C,D) and frontal (E,F) views of E9.5 embryos.
Black arrowheads point to areas of cell death in the mutant BA1. (G,H) Frontal sections of E9.5 embryos at the level of the maxillary
primordium, stained for the presence of apoptotic cells using TUNEL. Abbreviations as in Figs. 1 and 2. (Mes) Mesencephalon; (NT)
neural tube; (OV) optic vesicle; (r) rhombomere.
Analysis of neural crest cell migration and cell death in Fgf8;Nes–cre mutant embryos. (A,B) Expression of Cad6 at E8.5.
Trumpp et al.
3140GENES & DEVELOPMENT
not required for cell proliferation in BA1. We then as-
sayed for cell survival by staining in whole mount for
Nile blue sulfate (NBS) uptake, which marks dying and
dead cells (Bowen 1981), and by performing TUNEL as-
says on tissue sections to detect apoptosis. In the BA1
region of control embryos at E8.5–E10.0, NBS staining
was observed only proximally (Fig. 3C,E), where trigem-
inal ganglion cells die (Noden 1983; Davies and Lumsden
1984). In Fgf8;Nes–cre embryos, however, extensive cell
death was detected at E8.75–E10.0 (Fig. 3D,F; data not
shown). At E9.5, intense NBS staining was detected
throughout the BA1 primordium (Fig. 3D) except at the
extreme distal tip near the ventral midline (Fig. 3F). The
area in which dying cells were detected stretched proxi-
mally to the trigeminal swelling and included the region
forming the maxillary arch (Fig. 3D,H). The results of the
TUNEL analysis indicated that the observed abnormal
cell death, which was detected exclusively in the mes-
enchyme of BA1, is due to apoptosis (Fig. 3G,H). To-
gether these data suggest that the lack of BA1 develop-
ment in Fgf8;Nes–cre embryos is due, at least in part, to
apoptosis of a substantial proportion of the cells that
normally populate the arch, and therefore, that FGF8
produced in the surface ectoderm is essential for their
Patterned expression of some regulatory genes
is maintained in the hypoplastic BA1
of Fgf8;Nes–cre mutants
Because a proportion of mutant BA1 mesenchyme sur-
vives and forms a hypoplastic arch, we performed an
analysis to identify genes that require FGF8 signaling
and to determine the extent of mesenchymal patterning
in the absence of FGF8. Pitx1, a bicoid-related homeobox
gene (Lanctot et al. 1997; Szeto et al. 1999) is normally
coexpressed with Fgf8 in the oral ectoderm of the man-
dibular and maxillary primordia, and is also expressed in
the underlying mesenchyme (Fig. 4A). The oral mesen-
chyme also expresses Lhx7, a LIM homeodomain encod-
ing gene (Grigoriou et al. 1998; Tucker et al. 1999) (Fig.
4C,E). Expression of both genes was clearly detected in
the mutant BA1, but the signals were less extensive than
normal (Fig. 4A–F), presumably because BA1 mesen-
chyme is hypocellular. These results show that mutant
BA1 is sufficiently healthy to express at least some of the
genes that mark the oral side of BA1, and suggest that
some aspects of oral–aboral patterning are intact in the
absence of Fgf8.
Dlx1, Dlx2, and Dlx5 are three members of the Dlx
multigene family related to the Drosophila distalless
gene, which are thought to regulate morphogenesis along
the P–D axis of BA1 (Qiu et al. 1995, 1997; Depew et al.
1999). Dlx1 and Dlx2 are expressed in the ectomesen-
chyme along most of the P–D length of BA1, including
the maxillary primordium (Fig. 4G; data not shown). In
contrast, Dlx5 expression is not detected in the maxil-
lary primordium and is restricted within the distal two-
thirds of the mandibular primordium (Qiu et al. 1997;
Depew et al. 1999) (Fig. 4I). Expression of all three genes
was detected in BA1 of mutant embryos in roughly their
normal domains (Fig. 4H,J). This suggests that cells in
the mutant arch have sufficient P–D positional informa-
tion to maintain differential expression of these genes.
To further investigate the extent to which distal BA1
is patterned, we assayed for Msx1 and Bmp4 expression.
Between E9.5 and E10.5, Msx1 expression is normally
restricted to the distal (medial) ectomesenchyme and
Bmp4 expression is detected in distal BA1 ectoderm
overlying the Msx1 expression domain (Tucker et al.
1998a) (Fig. 5A,C). Both Msx1 and Bmp4 expression ap-
peared relatively normal in mutant BA1 (Fig. 5B,D).
Similar results (not shown) were obtained with probes
for Msx2, dHand, and eHand, other genes that are nor-
mally expressed in distal BA1 (Thomas et al. 1998). Pax9,
which is required for tooth development, is normally
detected in the mandibular arches at E11.5 in four spots,
the Fgf8;Nes–cre BA1. (A–J) Gene expression in BA1 at E9.5
(A,B,G–J) and E10.5 (C–F) embryos. (A–D,G–J) Lateral views.
(E,F) Ventral views. Abbreviations as in Fig. 1. (Ab) Aboral; (Md)
mandibular primordium of BA1; (Mx) maxillary primordium of
BA1; (Or) oral.
Patterning genes with normal expression domains in
FGF8 function in first branchial arch development
GENES & DEVELOPMENT 3141
two distal and two proximal, representing the prospec-
tive incisor and molar domains, respectively (Neubuser
et al. 1997; Peters et al. 1998) (Fig. 5E). In the mutant
embryos, Pax9 was detected in only two spots near the
ventral midline (Fig 5F), which presumably mark the re-
gions in which the incisors will form. These early gene
expression patterns are consistent with the observation
that the distal-most portions of BA1 are the least af-
fected: in newborn Fgf8;Nes–cre mutants the distal part
of the maxilla and the distal mandible, including trun-
cated lower incisors, are present (Fig. 5G,H; summarized
in Fig. 2O). These data suggest that FGF8 is not required
for development of distal BA1.
Identification of FGF8-responsive genes
FGF8-bead implantation experiments have identified
Lhx7, and the closely related gene Lhx6, which are co-
expressed in oral mesenchyme of mandibular and max-
illary arches at E10.5 (Figs. 4C,E; 6A,C), as potential tar-
gets of FGF8 signaling in BA1 (Grigoriou et al. 1998).
Interestingly, although we found that Lhx7, is expressed
in its normal domain in Fgf8;Nes–cre mutants (Fig.
4D,F), Lhx6 expression was not detected in mutant BA1
(Fig. 6B,D). Thus, our data suggest that they are indepen-
dently regulated in vivo.
Barx1 is another homeobox gene previously identified
as inducible by FGF8 (Tucker et al. 1998b). Beginning at
E9.5 it is expressed in mesenchyme throughout the
proximal but not in the distal portion of BA1 (Fig. 6E,G;
data not shown). In E9.5–E11.5 Fgf8;Nes–cre embryos,
Barx1 RNA was not detectable, except in a small patch
of cells on the caudal side of the mandibular arch (yellow
arrows in Fig. 6F,H; data not shown). Significantly, this
patch of Barx1 expression was localized in mesenchyme
that appeared to underlie the region in which Fgf8 ex-
pression was detected transiently at E9.0 (cf. Figs. 1F and
6F). At E9.5, the mutant ectoderm that appears to overlie
the patch of Barx1-expressing cells expresses Endothe-
lin-1 (Et1; yellow arrow, Fig. 6N). Et1 RNA was not de-
tected elsewhere in mutant BA1 ectoderm, but was de-
tected in the second and third branchial arch epithelium
(Fig. 6N; data not shown). In contrast, in control em-
bryos, Et1 RNA was detected at low levels throughout
the epithelium of branchial arches 1, 2, and 3 (Clouthier
et al. 1998) (Fig. 6M). By E10.5, Goosecoid (Gsc) expres-
sion, which is normally detected throughout the caudal
half of BA1 (Rivera-Perez et al. 1995; Yamada et al. 1995)
(Fig. 6I,K) is restricted in the mutants to the region in
which Barx1- and Et1-expressing cells are found (yellow
arrow, Fig. 6J,L). These results, summarized in Figure
6O, show that expression of Lhx6, Barx1, Et1, and Gsc in
BA1 are dependent, directly or indirectly, on FGF8 sig-
naling. Furthermore, they suggest that transient expres-
sion of Fgf8 on the caudal side of BA1 at E9.0 is sufficient
to induce local expression of Barx1, Et1, and Gsc, but
that continued expression of Fgf8 is not required to
Inactivation of Fgf8 by Cre-mediated recombination in
the ectoderm of the nascent first branchial arch severely
impairs development of the BA1 primordium. This ap-
pears to be due to both apoptosis of the mesenchyme and
failure to express a set of genes essential for BA1 mor-
phogenesis. Newborn mutants lack most BA1-derived
structures, except those formed from the distal-most re-
gion. These results provide genetic evidence that the ec-
toderm produces factors essential for BA1 outgrowth and
patterning at a very early stage. They further suggest a
model in which BA1 is specified into two domains at
early stages in its development, a proximal region that is
dependent on FGF8 signaling for its outgrowth and pat-
terning, and a distal domain that is controlled by other
signaling molecules (Fig. 7).
FGF8 is an essential signal for cell survival
in proximal BA1
Shortly after Fgf8 is inactivated in the nascent BA1 epi-
thelium, there is a brief period of cell death that peaks at
∼E9.5, during which a large proportion of proximal but
not distal BA1 mesenchyme undergoes apoptosis. This
cell death is presumably responsible for the small size of
in distal BA1 of Fgf8;Nes–cre embryos. (A–F) Gene expression at
E9.5 (A,B) E10.0 (C,D) E11.5 (E,F). Arrows point to expression
domains of (A,B) Msx1 in the mesenchyme; (C,D) Bmp4 in the
ectoderm; and (E,F) Pax9 in the mandibular mesenchyme. Note
that in the mutant embryo (F) Pax9 expression is detected in the
presumptive incisor but not in the presumptive molar domain.
(G,H) Frontal sections through the incisor domain of newborn
Tooth formation and expression of patterning genes
Trumpp et al.
3142 GENES & DEVELOPMENT
BA1 from an early stage of its development in the mu-
tant embryos, and most likely contributes to the final
mutant phenotype. Thus, one important conclusion
from our study is that FGF8 is required, directly or indi-
rectly, for cell survival in proximal BA1. Preliminary
data suggest that reduction in the level of FGF8 also
causes cell death in the developing limb and brain (M.
Lewandoski, E. Storm, and G.R. Martin, unpubl.). In con-
trast, analysis of Fgf8−/−embryos indicates that during
gastrulation FGF8 is not required for cell proliferation or
survival, but instead is necessary for cell migration at the
primitive streak stage (Sun et al. 1999). Thus, it appears
that FGF8 performs different functions in different de-
There is a substantial body of evidence from experi-
mental studies in vitro showing that FGFs, particularly
FGF1 and FGF2, can function as survival factors for a
wide variety of cell types including neurons, glia, endo-
thelial cells, and smooth muscle cells (for review, see
Szebenyi and Fallon 1999). For example, it has been
found that addition of FGF2 to explants of trunk neural
crest enhances cell survival without stimulating mitosis
(Kalcheim 1989), and local application of beads contain-
ing FGF4 prevents apoptosis in dental mesenchyme iso-
lated from the mandibular arch at E13 (Vaahtokari et al.
in BA1 ectoderm. Schematic diagram illustrating effects of
FGF8 signaling in BA1. According to this model, FGF8 signaling
is not required to induce gene expression in the distal region.
However, in the proximal region, expression of Barx1, Pax9,
Lhx6, and Gsc in the mesenchyme is directly or indirectly de-
pendent on FGF8, which is also required for Et1 expression in
the ectoderm. Signaling molecules produced in the ectoderm are
shown in red; transcription factors expressed in the mesen-
chyme are shown in black.
Inductive activities of signaling molecules produced
E10.5, and (M,N) Et1 at E9.5. Lateral views (A,B,E,F,I,J,M,N) and ventral views (C,D,G,H,K,L) of embryos assayed for expression of the
genes indicated. The yellow arrows point to gene expression in the region in which Fgf8 was expressed transiently before E9.0. (O)
Schematic diagram summarizing gene expression patterns in the normal and mutant BA1 at E10.5.
Gene expression regulated by FGF8 signaling in BA1. (A–D) Expression of Lhx6 at E10.5, (E–H) Barx1 at E10.0, (I–L) Gsc at
FGF8 function in first branchial arch development
GENES & DEVELOPMENT3143
1996). Despite the wealth of data from such in vitro stud-
ies, previous genetic analysis has provided little evidence
that FGFs are required for cell survival. Null mutations
have been generated in Fgf2 and at least 10 other mouse
FGF genes, as well as all four FGF receptor (FGFR) genes
(Goldfarb 1996; Floss et al. 1997; Dono et al. 1998; Min et
al. 1998; Ortega et al. 1998; Weinstein et al. 1998; Zhou
et al. 1998; Sekine et al. 1999; Sun et al. 1999; D. Ornitz,
pers. comm.; C. Basilico, pers. comm.), but to our knowl-
edge, effects specifically on cell survival have been de-
scribed only in Fgf4−/−(Feldman et al. 1995) and Fgfr2−/−
(Arman et al. 1998) embryos. These mutants display a
similar early postimplantation lethal phenotype in
which the inner cell mass dies.
One interesting question is why a proportion of the
cells in proximal BA1 survive. One possibility is that
their survival is dependent on other FGF family mem-
bers, such as Fgf9, which is detected at low levels in both
wild-type (Kettunen and Thesleff 1998) and mutant BA1
epithelium (data not shown), or on other types of signals.
Another intriguing possibility is based on the premise
that FGFs and other survival factors that signal through
receptor tyrosine kinases (RTKs) function to prevent
apoptosis by stimulating the antiapoptotic activity of
Ras (for review, see Downward 1998), and the observa-
tion that FGF signaling can induce the expression of
Sprouty (Spry) genes, which encode inhibitors of RTK
signaling (Hacohen et al. 1998; Casci et al. 1999; Kramer
et al. 1999; Minowada et al. 1999; Reich et al. 1999).
Thus, the extent of cell survival in a given tissue may be
determined by the balance between factors that stimu-
late and inhibit Ras activity. We have found that the
expression of at least one member of the Sprouty gene
family, Spry2, is greatly diminished in BA1 of Fgf8;Nes–
cre embryos (data not shown), supporting the hypothesis
based on FGF-bead implantation experiments that FGF8
positively regulates Sprouty gene expression in BA1 (Mi-
nowada et al. 1999). The reduced level of Spry2 in mu-
tant BA1 mesenchyme might compensate partially for
the reduced Ras activity caused by the inactivation of
Fgf8 by derepressing other RTK pathways normally in-
hibited by Sprouty, and thereby promote cell survival.
FGF8 induces gene expression necessary for proximal
Fgf8;Nes–cre embryos lack most of the chondrocranial
and dermatocranial elements that form the jaws, the lat-
eral skull wall, and middle ear (Fig. 2O). It is possible
that their failure to develop is due solely to the reduction
in cell number that occurs between E8.75 and E10. How-
ever, it seems likely that the failure of surviving cells to
express genes downstream of FGF8 also contributes to
the final phenotype. One genetic pathway that is affected
in Fgf8;Nes–cre mutants involves Endothelin-1 (ET1)
(Levin 1995). Mice in which the genes encoding ET1 or
its receptor (ETA) have been inactivated display a pleio-
tropic phenotype, including severe effects on the lower
jaw and other BA1-derived structures (Kurihara et al.
1994; Clouthier et al. 1998). Although ET1 signaling is
not required for the early stages of arch outgrowth and
BA1 appears grossly normal until E10.5 in Et1−/−mu-
tants (Kurihara et al. 1994), it is required for expression
of Gsc (Clouthier et al. 1998), which is necessary for
(Rivera-Perez et al. 1995; Yamada et al. 1995). Our re-
sults show that Et1 and Gsc expression are dependent on
FGF8 signaling, indicating that FGF8 is upstream of this
pathway (Fig. 7). Consistent with this hypothesis, the
BA1 structures that are missing or malformed in Et1−/−
bryos. However, loss of Fgf8 function in BA1 results in a
more severe phenotype. This could be explained either
by the hypocellularity of the Fgf8;Nes–cre BA1 or by the
lack of expression of genes that are downstream of FGF8
but not ET1 signaling.
One good candidate for such a gene is Barx1 (Tissier-
Seta et al. 1995). Previous studies have shown that Barx1
expression, which is normally restricted to the region
that gives rise to the same elements that fail to develop
in Fgf8;Nes–cre embryos, can be induced in explants of
mandibular arch mesenchyme by FGF8 (Tucker et al.
1998b). Our data show that FGF8 is required on both the
rostral and caudal side of BA1 to establish the Barx1
expression domain. We found that Lhx6 is also down-
stream of FGF8. Surprisingly, Lhx7 expression does not
require FGF8, despite the fact that Lhx6 and Lhx7, which
appear to be expressed in the same cells on the oral side
of the arch, can both be induced by placing FGF8-beads
in isolated mandibular arch mesenchyme (Grigoriou et
al. 1998). A possible explanation for these findings is that
Lhx7 expression is induced by an FGF family member
other than FGF8.
−/−mice are also affected in Fgf8;Nes–cre em-
FGF8 as a switch that induces but does not maintain
the BA1 developmental program
An important question is whether FGF8 functions solely
to induce gene expression or whether it is also necessary
to maintain it. We were able to address this question
because Fgf8 is expressed transiently in a few epithelial
cells on the caudoproximal side before it is inactivated.
After Fgf8 RNA is no longer detected, Et1 is expressed in
what appears to be the same caudal cell population.
Moreover Barx1 and subsequently Gsc are induced in the
adjacent underlying ectomesenchyme. The finding that
expression of Et1, Barx1, and Gsc in the mutant BA1 is
restricted to the region in which Fgf8 is expressed tran-
siently suggests that FGF8 is necessary and sufficient to
induce expression of these genes in vivo. Furthermore,
because Barx1 and Gsc expression persists in this region
through E11.5, at least 60 hr after Fgf8 has been inacti-
vated, we propose that FGF8 can function as a switch to
induce a gene expression cascade that rapidly becomes
independent of FGF8 signaling. Thus, Barx1 and Gsc ex-
pression may be maintained by other signals produced in
the epithelium, possibly ET1.
Analysis of the skeletons of Fgf8;Nes–cre newborns
indicated that the only proximal BA1-derived skeletal
elements present are a malleus-like middle ear element
Trumpp et al.
3144GENES & DEVELOPMENT
and nearby amorphous cartilagenous and bony fragments
(Fig. 2O). It seems likely that they develop from cells on
the caudal side of the mutant BA1 where outgrowth is
observed at early stages in close proximity to a structure
(the EAM) that will form the outer ear canal. Cells in this
outgrowth apparently express Gsc, which is required for
malleus development (Rivera-Perez et al. 1995; Yamada
et al. 1995). Because this outgrowth develops in the re-
gion that was exposed transiently to FGF8 signaling, the
presence of a vestigial malleus in the Fgf8;Nes–cre new-
borns does not contradict the hypothesis that FGF8 is
required for development of the entire proximal domain
Control of distal BA1 development
In contrast to the lack of development of skeletal ele-
ments derived from proximal BA1, structures derived
from the distal portion of the arch are invariably present
in Fgf8;Nes–cre newborns (Fig. 2O). There is genetic evi-
dence that development of the distal region of BA1 is
dependent on Msx1, as the distal maxilla and mandible
(including the incisors), the very structures that are pre-
sent in Fgf8;Nes–cre pups, fail to form in Msx1 null em-
bryos (Satokata and Maas 1994). As expected, Msx1 ex-
pression was detected in its normal domain in distal BA1
of Fgf8;Nes–cre embryos. Although it is formally pos-
sible that the distal domain develops because Fgf8 is ex-
pressed transiently in distal BA1 ectoderm, we think this
unlikely because no such Fgf8 expression was detected
in the mutant embryos. Instead, we suggest that devel-
opment of distal BA1 is not dependent on FGF8. This
leaves open the question of what controls Msx1 expres-
sion? It is possible that FGF signaling is involved, as
several FGFs have been found to induce Msx1 in dental
mesenchyme (Bei and Maas 1998; Kettunen and Thesleff
1998), at the border of the neural plate (Streit and Stern
1999), and in the limb (Wang and Sassoon 1995). Other
than Fgf8, the only FGF family member presently known
to be expressed in BA1 epithelium before E12.5 is Fgf9
(Kettunen andThesleff 1998;
sinki.fi/toothexp/index.htm). Thus, FGF9 or some other
FGF that has yet to be identified may provide the signal
necessary for outgrowth and patterning of the distal do-
main of BA1. The hypothesis that another FGF in addi-
tion to FGF8 is required for BA1 development could also
explain the pattern of expression we observed for Pax9, a
marker for the prospective tooth-forming domains in the
oral mesenchyme, which can be induced in isolated mes-
enchyme by any of several different FGFs (Neubuser et
al. 1997). In the mutant BA1, Pax9 expression is absent
in the proximal, molar domain because it is dependent
on FGF8, whereas Pax9 expression is detected in the dis-
tal, incisor domain, perhaps because it is induced by an
FGF signal other than FGF8.
Another factor that may play a role in the develop-
ment of the distal domain is BMP4, as it too is capable of
inducing Msx1 expression in BA1 (Vainio et al. 1993;
Tucker et al. 1998a,b), limb (Wang and Sassoon 1995),
and brain (Furuta et al. 1997), and is expressed in distal
BA1 of Fgf8;Nes–cre mutants. In view of the evidence
that BMP signaling plays a role in determining tooth
identity (Tucker et al. 1998b), and that FGF and BMP
signaling pathways function antagonistically to regulate
tooth induction (Neubuser et al. 1997), it seems likely
that interactions between the two types of signals con-
trol development of the distal-most BA1 structures.
The Fgf8;Nes–cre mutants resemble humans with first
arch syndromes that include agnathia. Agnathia alone
occurs very rarely, and is often associated with holopros-
encephaly and sometimes with situs inversus totalis, or
both (Pauli et al. 1983; Bixler et al. 1985; Leech et al.
1988; Escobar 1993). Significantly, embryos that are
compound heterozygous for a null and a hypomorphic
allele of Fgf8 show forebrain defects including holopros-
encephaly (E. Meyers, E. Storm, and G.R. Martin, un-
publ.) and display abnormalities in left-right asymmetry
determination (Meyers and Martin 1999). Therefore, it is
tempting to speculate that mutations in Fgf8 or in genes
directly upstream or downstream of it might cause some
of the human syndromes characterized by agnathia/mi-
Materials and methods
Production and analysis of mutant embryos
Production and full characterization of the Nes–cre1 transgenic
mouse line will be reported elsewhere. The Fgf8floxand Fgf8?2,3n
alleles (Meyers et al. 1998) were maintained on a mixed genetic
background. Fgf8;Nes–cre mutants were produced using the
breeding scheme outlined in Figure 1A and genotyped using
previously described primers (Lewandoski et al. 1997; Meyers et
al. 1998; Sun et al. 1999). Histological analysis of embryos, scan-
ning electron microscopy, and skeletal preparations were car-
ried out as described by Depew et al. (1999). Nes–cre1;Z/AP
double hemizygotes (E7.5–E11.5) were stained for alkaline phos-
phatase activity essentially as described by Lobe et al. (1999).
Cell death analysis
For whole mount NBS staining, E8.0–E10.5 embryos were dis-
sected, washed in PBS, and incubated for 30–45 min at 37°C in
filtered NBS solution [10 mg/ml NBS (Sigma N-5632) in PBS
containing 0.1% Tween 20]. Embryos were then washed several
times in PBS at room temperature and photographed immedi-
ately. TUNEL analysis was performed on paraffin sections using
the In Situ Cell Death Detection kit (Boehringer-Mannheim)
following the manufacturer’s protocol.
RNA in situ hybridization
Whole mount RNA in situ hybridization analysis was carried
out as previously described (Neubuser et al. 1997) using ribo-
probes prepared from plasmids described in references cited for
each gene. Fgf8 RNA was detected using a probe for sequences
in exons 2 and 3, which are deleted in the Fgf8?2,3and the
Fgf8?2,3nmutant alleles. The expression of each gene was ana-
lyzed in at least three Fgf8;Nes–cre embryos at each stage.
FGF8 function in first branchial arch development
GENES & DEVELOPMENT 3145
We are very grateful to Andras Nagy for providing the Z/AP
reporter line, and thank the following for providing plasmids
used to prepare probes used in this study: J.F. Brunet (Barx1); E.
deRobertis (Gsc); J. Drouin (Pitx1); B. Hogan (Bmp4); V. Pachnis
(Lhx6, Lhx7); P. Sharpe (Msx1); M. Takeichi (Cad6); and M.
Yanagisawa (Et1). We are grateful to A. Gannon and D. Trail for
excellent technical assistance. We also thank our colleagues in
the Martin and Bishop laboratories for helpful discussion and
critical readings of the manuscript. A.T. was the recipient of
postdoctoral fellowships from the Deutsche Forschungsgemein-
schaft, Human Frontiers Science Program, and the California
Division of the American Cancer Society. This work was sup-
ported by grants from the March of Dimes and Nina Ireland (to
J.L.R.R.), the Howard Hughes Medical Institute Research Re-
sources Program grant (76296-549901) to the UCSF School of
Medicine, and National Institutes of Health grants KO2
MH01046 (to J.L.R.R.), RO1 CA44338 (to J.M.B.), and RO1
HD34380 (to G.R.M.).
The publication costs of this article were defrayed in part by
payment of page charges. This article must therefore be hereby
marked “advertisement” in accordance with 18 USC section
1734 solely to indicate this fact.
Arman, E., R. Haffner-Krausz, Y. Chen, J.K. Heath, and P. Lonai.
1998. Targeted disruption of fibroblast growth factor (FGF)
receptor 2 suggests a role for FGF signaling in pregastrulation
mammalian development. Proc. Natl. Acad. Sci. 95: 5082–
Bei, M. and R. Maas. 1998. FGFs and BMP4 induce both Msx1-
independent and Msx1-dependent signaling pathways in
early tooth development. Development 125: 4325–4333.
Bixler, D., R. Ward, and D.D. Gale. 1985. Agnathia-holoprosen-
cephaly: A developmental field complex involving face and
brain. Report of 3 cases. J. Craniofac. Genet. Dev. Biol.
(Suppl.) 1: 241–249.
Bowen, I.D. 1981. Techniques for demonstrating cell death. In
Cell death in biology and pathology (ed. I.D. Bowen and R.A.
Lockshin), pp. 379–444. Chapman and Hall, London, UK.
Casci, T., J. Vinos, and M. Freeman. 1999. Sprouty, an intracel-
lular inhibitor of Ras signaling. Cell 96: 655–665.
Clouthier, D.E., K. Hosoda, J.A. Richardson, S.C. Williams, H.
Yanagisawa, T. Kuwaki, M. Kumada, R.E. Hammer, and M.
Yanagisawa. 1998. Cranial and cardiac neural crest defects in
endothelin-A receptor-deficient mice. Development 125:
Crossley, P.H. and G.R. Martin. 1995. The mouse Fgf8 gene
encodes a family of polypeptides and is expressed in regions
that direct outgrowth and patterning in the developing em-
bryo. Development 121: 439–451.
Davies, A. and A. Lumsden. 1984. Relation of target encounter
and neuronal death to nerve growth factor responsiveness in
the developing mouse trigeminal ganglion. J. Comp. Neurol.
Depew, M.J., J.K. Liu, J.E. Long, R. Presley, J.J. Meneses, R.A.
Pedersen, and J.L.R. Rubenstein. 1999. Dlx5 regulates re-
gional development of the branchial arches and sensory cap-
sules. Development 126: 3831–3846.
Dono, R., G. Texido, R. Dussel, H. Ehmke, and R. Zeller. 1998.
Impaired cerebral cortex development and blood pressure
regulation in FGF-2-deficient mice. EMBO J. 17: 4213–4225.
Downward, J. 1998. Ras signalling and apoptosis. Curr. Opin.
Genet. Dev. 8: 49–54.
Escobar, L.F. 1993. Facial bones. In Human malformations and
related anomalies (ed. R.E. Stevenson, J.G. Hall, and R.M.
Goodman), II, pp. 629–653. Oxford University Press, New
Feldman, B., W. Poueymirou, V.E. Papaioannou, T.M. DeChi-
ara, and M. Goldfarb. 1995. Requirement of FGF-4 for
postimplantation mouse development. Science 267: 246–
Floss, T., H.H. Arnold, and T. Braun. 1997. A role for FGF-6 in
skeletal muscle regeneration. Genes & Dev. 11: 2040–2051.
Francis-West, P., R. Ladher, A. Barlow, and A. Graveson. 1998.
Signalling interactions during facial development. Mech.
Dev. 75: 3–28.
Furuta, Y., D.W. Piston, and B.L. Hogan. 1997. Bone morphoge-
netic proteins (BMPs) as regulators of dorsal forebrain devel-
opment. Development 124: 2203–2212.
Goldfarb, M. 1996. Functions of fibroblast growth factors in
vertebrate development. Cytokine Growth Factor Rev.
Grigoriou, M., A.S. Tucker, P.T. Sharpe, and V. Pachnis. 1998.
Expression and regulation of Lhx6 and Lhx7, a novel sub-
family of LIM homeodomain encoding genes, suggests a role
in mammalian head development. Development 125: 2063–
Gu, H., J.D. Marth, P.C. Orban, H. Mossmann, and K. Rajewsky.
1994. Deletion of a DNA polymerase beta gene segment in T
cells using cell type-specific gene targeting. Science 265:
Hacohen, N., S. Kramer, D. Sutherland, Y. Hiromi, and M.A.
Krasnow. 1998. sprouty encodes a novel antagonist of FGF
signaling that patterns apical branching of the Drosophila
airways. Cell 92: 253–263.
Inoue, T., O. Chisaka, H. Matsunami, and M. Takeichi. 1997.
Cadherin-6 expression transiently delineates specific rhom-
bomeres, other neural tube subdivisions, and neural crest
subpopulations in mouse embryos. Dev. Biol. 183: 183–194.
Kalcheim, C. 1989. Basic fibroblast growth factor stimulates
survival of nonneuronal cells developing from trunk neural
crest. Dev. Biol. 134: 1–10.
Kettunen, P. and I. Thesleff. 1998. Expression and function of
FGFs-4, -8, and -9 suggest functional redundancy and repeti-
tive use as epithelial signals during tooth morphogenesis.
Dev. Dynamics 211: 256–268.
Kramer, S., N. Hacohen, M. Okabe, M.A. Krasnow, and Y. Hi-
romi. 1999. Sprouty: A common antagonist of FGF and EGF
signaling pathways in Drosophila. Development 126: 2515–
Kulkarni, R.N., J.C. Bruning, J.N. Winnay, C. Postic, M.A. Mag-
nuson, and C.R. Kahn. 1999. Tissue-specific knockout of the
insulin receptor in pancreatic beta cells creates an insulin
secretory defect similar to that in type 2 diabetes. Cell
Kurihara, Y., H. Kurihara, H. Suzuki, T. Kodama, K. Maemura,
R. Nagai, H. Oda, T. Kuwaki, W.H. Cao, N. Kamada et al.
1994. Elevated blood pressure and craniofacial abnormalities
in mice deficient in endothelin-1. Nature 368: 703–710.
Lanctot, C., B. Lamolet, and J. Drouin. 1997. The bicoid-related
homeoprotein Ptx1 defines the most anterior domain of the
embryo and differentiates posterior from anterior lateral me-
soderm. Development 124: 2807–2817.
Leech, R.W., L.S. Bowlby, R.A. Brumback, and G.B. Schaefer Jr.
1988. Agnathia, holoprosencephaly, and situs inversus: Re-
port of a case. Am. J. Med. Genet. 29: 483–490.
Levin, E.R. 1995. Endothelins. N. Engl. J. Med. 333: 356–363.
Lewandoski, M., K.M. Wassarman, and G.R. Martin. 1997. Zp3-
cre, a transgenic mouse line for the activation or inactivation
Trumpp et al.
3146GENES & DEVELOPMENT
of loxP-flanked target genes specifically in the female germ
line. Curr. Biol. 7: 148–151.
Lobe, C.G., K.E. Koop, W. Kreppner, H. Lomeli, M. Gertsen-
stein, and A. Nagy. 1999. Z/AP, a double reporter for Cre-
mediated recombination. Dev. Biol. 208: 281–292.
Maden, M., C. Horton, A. Graham, L. Leonard, J. Pizzey, G.
Siegenthaler, A. Lumsden, and U. Eriksson. 1992. Domains
of cellular retinoic acid-binding protein I (CRABP I) expres-
sion in the hindbrain and neural crest of the mouse embryo.
Mech. Dev. 37: 13–23.
Mahmood, R., J. Bresnick, A. Hornbruch, C. Mahony, N. Mor-
ton, K. Colquhoun, P. Martin, A. Lumsden, C. Dickson, and
I. Mason. 1995. A role for FGF-8 in the initiation and main-
tenance of vertebrate limb bud outgrowth. Curr. Biol.
Meyers, E.N., M. Lewandoski, and G.R. Martin. 1998. An Fgf8
mutant allelic series generated by Cre- and Flp-mediated re-
combination. Nature Genet. 18: 136–141.
Meyers, E.N. and G.R. Martin. 1999. Differences in left-right
axis pathways in mouse and chick: Functions of FGF8 and
SHH. Science 285: 403–406.
Min, H., D.M. Danilenko, S.A. Scully, B. Bolan, B.D. Ring, J.E.
Tarpley, M. DeRose, and W.S. Simonet. 1998. Fgf-10 is re-
quired for both limb and lung development and exhibits
striking functional similarity to Drosophila branchless.
Genes & Dev. 12: 3156–3161.
Minowada, G., L.A. Jarvis, C.L. Chi, A. Neubu ¨ser, X. Sun, N.
Hacohen, M.A. Krasnow, and G.R. Martin. 1999. Vertebrate
Sprouty genes are induced by FGF signaling and can cause
chondrodysplasia when overexpressed. Development 126:
Mitchell, P.J., P.M. Timmons, J.M. Hebert, P.W. Rigby, and R.
Tjian. 1991. Transcription factor AP-2 is expressed in neural
crest cell lineages during mouse embryogenesis. Genes &
Dev. 5: 105–119.
Neubuser, A., H. Peters, R. Balling, and G.R. Martin. 1997. An-
tagonistic interactions between FGF and BMP signaling
pathways: A mechanism for positioning the sites of tooth
formation. Cell 90: 247–255.
Noden, D.M. 1983. The role of the neural crest in patterning of
avian cranial skeletal, connective, and muscle tissues. Dev.
Biol. 96: 144–165.
Ortega, S., M. Ittmann, S.H. Tsang, M. Ehrlich, and C. Basilico.
1998. Neuronal defects and delayed wound healing in mice
lacking FGF2. Proc. Natl. Acad. Sci. 95: 5672–5677.
Pauli, R.M., J.C. Pettersen, S. Arya, and E.F. Gilbert. 1983. Fa-
milial agnathia-holoprosencephaly. Am. J. Med. Genet.
Peters, H. and R. Balling. 1999. Teeth. Where and how to make
them. Trends Genet. 15: 59–65.
Peters, H., A. Neubuser, K. Kratochwil, and R. Balling. 1998.
Pax9-deficient mice lack pharyngeal pouch derivatives and
teeth and exhibit craniofacial and limb abnormalities. Genes
& Dev. 12: 2735–2747.
Qiu, M., A. Bulfone, S. Martinez, J.J. Meneses, K. Shimamura,
R.A. Pedersen, and J.L. Rubenstein. 1995. Null mutation of
Dlx-2 results in abnormal morphogenesis of proximal first
and second branchial arch derivatives and abnormal differ-
entiation in the forebrain. Genes & Dev. 9: 2523–2538.
Qiu, M., A. Bulfone, I. Ghattas, J.J. Meneses, L. Christensen,
P.T. Sharpe, R. Presley, R.A. Pedersen, and J.L. Rubenstein.
1997. Role of the Dlx homeobox genes in proximodistal pat-
terning of the branchial arches: Mutations of Dlx-1, Dlx-2,
and Dlx-1 and -2 alter morphogenesis of proximal skeletal
and soft tissue structures derived from the first and second
arches. Dev. Biol. 185: 165–184.
Reich, A., A. Sapir, and B. Shilo. 1999. Sprouty is a general
inhibitor of receptor tyrosine kinase signaling. Development
Rivera-Perez, J.A., M. Mallo, M. Gendron-Maguire, T. Gridley,
and R.R. Behringer. 1995. Goosecoid is not an essential com-
ponent of the mouse gastrula organizer but is required for
craniofacial and rib development. Development 121: 3005–
Ruberte, E., V. Friederich, G. Morriss-Kay, and P. Chambon.
1992. Differential distribution patterns of CRABP I and
CRABP II transcripts during mouse embryogenesis. Devel-
opment 115: 973–987.
Satokata, I. and R. Maas. 1994. Msx1 deficient mice exhibit cleft
palate and abnormalities of craniofacial and tooth develop-
ment. Nature Genet. 6: 348–356.
Sekine, K., H. Ohuchi, M. Fujiwara, M. Yamasaki, T.
Yoshizawa, T. Sato, N. Yagishita, D. Matsui, Y. Koga, N.
Itoh, and S. Kato. 1999. Fgf10 is essential for limb and lung
formation. Nature Genet. 21: 138–141.
Streit, A. and C.D. Stern. 1999. Establishment and maintenance
of the border of the neural plate in the chick: involvement of
FGF and BMP activity. Mech. Dev. 82: 51–66.
Sun, X., E.N. Meyers, M. Lewandoski, and G.R. Martin. 1999.
Targeted disruption of Fgf8 causes failure of cell migration in
the gastrulating mouse embryo. Genes & Dev. 13: 1834–
Szebenyi, G. and J.F. Fallon. 1999. Fibroblast growth factors as
multifunctional signaling factors. Int. Rev. Cytol. 185: 45–
Szeto, D.P., C. Rodriguez-Esteban, A.K. Ryan, S.M. O’Connell,
F. Liu, C. Kioussi, A.S. Gleiberman, J.C. Izpisua-Belmonte,
and M.G. Rosenfeld. 1999. Role of the Bicoid-related ho-
meodomain factor Pitx1 in specifying hindlimb morphogen-
esis and pituitary development. Genes & Dev. 13: 484–494.
Thomas, T., H. Kurihara, H. Yamagishi, Y. Kurihara, Y. Yazaki,
E.N. Olson, and D. Srivastava. 1998. A signaling cascade
involving endothelin-1, dHAND and msx1 regulates devel-
opment of neural-crest-derived branchial arch mesenchyme.
Development 125: 3005–3014.
Tissier-Seta, J.P., M.L. Mucchielli, M. Mark, M.G. Mattei, C.
Goridis, and J.F. Brunet. 1995. Barx1, a new mouse homeodo-
main transcription factor expressed in cranio-facial ectomes-
enchyme and the stomach. Mech. Dev. 51: 3–15.
Tsien, J.Z., D.F. Chen, D. Gerber, C. Tom, E.H. Mercer, D.J.
Anderson, M. Mayford, E.R. Kandel, and S. Tonegawa. 1996.
Subregion- and cell type-restricted gene knockout in mouse
brain. Cell 87: 1317–1326.
Tucker, A.S., A. Al Khamis, and P.T. Sharpe. 1998a. Interactions
between Bmp-4 and Msx-1 act to restrict gene expression to
odontogenic mesenchyme. Dev. Dynamics 212: 533–539.
Tucker, A.S., K.L. Matthews, and P.T. Sharpe. 1998b. Transfor-
mation of tooth type induced by inhibition of BMP signaling.
Science 282: 1136–1138.
Tucker, A.S., G. Yamada, M. Grigoriou, V. Pachnis, and P.T.
Sharpe. 1999. Fgf-8 determines rostral-caudal polarity in the
first branchial arch. Development 126: 51–61.
Vaahtokari, A., T. Aberg, and I. Thesleff. 1996. Apoptosis in the
developing tooth: Association with an embryonic signaling
center and suppression by EGF and FGF-4. Development
Vainio, S., I. Karavanova, A. Jowett, and I. Thesleff. 1993. Iden-
tification of BMP-4 as a signal mediating secondary induc-
tion between epithelial and mesenchymal tissues during
early tooth development. Cell 75: 45–58.
Wang, Y. and D. Sassoon. 1995. Ectoderm–mesenchyme and
mesenchyme–mesenchyme interactions regulate Msx-1 ex-
FGF8 function in first branchial arch development
GENES & DEVELOPMENT3147
pression and cellular differentiation in the murine limb bud.
Dev. Biol. 168: 374–382.
Weinstein, M., X. Xu, K. Ohyama, and C.-X. Deng. 1998.
FGFR-3 and FGFR-4 function cooperatively to direct alveo-
genesis in the murine lung. Development 125: 3615–3623.
Yamada, G., A. Mansouri, M. Torres, E.T. Stuart, M. Blum, M.
Schultz, E.M. De Robertis, and P. Gruss. 1995. Targeted mu-
tation of the murine goosecoid gene results in craniofacial
defects and neonatal death. Development 121: 2917–2922.
Zhou, M., R.L. Sutliff, R.J. Paul, J.N. Lorenz, J.B. Hoying, C.C.
Haudenschild, M. Yin, J.D. Coffin, L. Kong, E.G. Kranias, W.
Luo, G.P. Boivin, J.J. Duffy, S.A. Pawlowski, and R.
Doetschman. 1998. Fibroblast growth factor 2 control of vas-
cular tone. Nature Med. 4: 201–207.
Zimmerman, L., B. Parr, U. Lendahl, M. Cunningham, R.
McKay, B. Gavin, J. Mann, G. Vassileva, and A. McMahon.
1994. Independent regulatory elements in the nestin gene
direct transgene expression to neural stem cells or muscle
precursors. Neuron 12: 11–24.
Trumpp et al.
3148 GENES & DEVELOPMENT