FGF4 and FGF8 Comprise the Wavefront Activity That Controls Somitogenesis

FGF4 and FGF8 comprise the wavefront activity that controls somitogenesis

L. A. Naiche, Nakisha Holder1, and Mark Lewandoski2

Genetics of Vertebrate Development Section, National Cancer Institute, Frederick, MD 21701

Edited by Gail Martin, University of California, San Francisco, CA, and approved January 26, 2011 (received for review May 29, 2010)

Somites form along the embryonic axis by sequential segmenta- required for somitogenesis (7). Furthermore, mice homozygous for tion from the presomitic mesoderm (PSM) and differentiate into null mutations in other FGF ligand genes (Fgf3, Fgf5, Fgf15, Fgf17, the segmented vertebral column as well as other unsegmented and Fgf18) expressed in the PSM also show no early somitogenesis tissues. Somites are thought to form via the intersection of two defects, or die before somitogenesis (Fgf4)(7–9). These results activities known as the clock and the wavefront. Previous work suggest that if FGFs have any function in somitogenesis, loss of has suggested that fibroblast growth factor (FGF) activity may be individual ligands is compensated for via genetic redundancy. the wavefront signal, which maintains the PSM in an undifferen- To circumvent the problem of redundancy, the gene encoding tiated state. However, it is unclear which (if any) of the FGFs FGF receptor 1 (Fgfr1), the only FGF receptor known in the PSM (10, 11), was conditionally deleted (11, 12). These studies dem- expressed in the PSM comprise this activity, as removal of any one fi onstrate a strong effect of FGF signaling on controlling clock gene gene is insuf cient to disrupt early somitogenesis. Here we show expression and in positioning the determination front. However, that when both Fgf4 and Fgf8 are deleted in the PSM, expression FGF signaling was retained during anterior somite formation and of most PSM genes is absent, including cycling genes, WNT path- progressively lost as somitogenesis proceeded. Perhaps due to this fi way genes, and markers of undifferentiated PSM. Signi cantly, gradual loss, premature differentiation of the PSM, a key pre- markers of nascent somite cell fate expand throughout the PSM, dicted outcome of the loss of wavefront activity, was not observed. demonstrating the premature differentiation of this entire tissue, WNT signaling has also been proposed to contribute to a highly unusual phenotype indicative of the loss of wavefront wavefront activity (13, 14). Manipulation of canonical WNT activity. When WNT signaling is restored in mutants, PSM progen- signaling also causes corresponding shifts in the determination BIOLOGY

itor markers are partially restored but premature differentiation of front (13, 15, 16). However, in WNT loss-of-function embryos, DEVELOPMENTAL the PSM still occurs, demonstrating that FGF signaling operates FGF signaling is also affected. Thus, the observed somitogenesis independently of WNT signaling. This study provides genetic evi- defects may be primarily due to loss of WNT signaling or to dence that FGFs are the wavefront signal and identifies the spe- secondary effects on FGF signaling. cific FGF ligands that encode this activity. Furthermore, these data To understand the actual role of FGF signaling in somito- show that FGF action maintains WNT signaling, and that both genesis, we initiated a comprehensive strategy to inactivate all signaling pathways are required in parallel to maintain PSM pro- combinations of FGF ligand genes expressed in the PSM. We genitor tissue. report here that inactivation of both Fgf4 and Fgf8 in the PSM results in dramatic shifts in the determination front and the genetic redundancy | Spry | T-Cre | axis extension | paraxial mesoderm premature differentiation of the PSM. Moreover, the restoration of WNT signaling to these mutants does not restore the determination front. Our data demonstrate that these two FGFs central problem in developmental biology is the identifi- constitute the proposed wavefront activity that maintains the A cation of molecular signals that control the acquisition PSM in an undifferentiated state. of differentiated cell fates. In the vertebrate embryo, a critical differentiation event is the formation of bilaterally paired epi- Results thelialized segments, called somites, from the presomitic meso- We have inactivated Fgf8 pairwise with other Fgfs expressed in derm (PSM). Somites give rise to the vertebral column as well as the PSM. Null alleles were used where possible. To bypass early skeletal muscle and the dermal skin layers (1). Defects in somi- lethality, conditional alleles of Fgf4 and Fgf8 (8, 17) were used in togenesis cause common human developmental disorders such as combination with the pan-mesodermal T-Cre line (7), which abnormal vertebrae segmentation and scoliosis (2). Somite seg- recombines throughout the PSM before the onset of somito- mentation occurs in an anterior-to-posterior progression as the genesis (Fig. S1). Compound mutant embryos lacking Fgf8 and embryo elongates, with a new somite emerging periodically from either Fgf3, Fgf5,orFgf17 showed no obvious synergistic defects the PSM approximately every 2 h in the mouse. in somitogenesis. However, compound Fgf8 and Fgf4 mutants The prevailing hypothesis for somitogenesis posits two inter- have a severe defect in somitogenesis, and we here report our secting activities in the PSM, known as the clock and the wave- analysis in detail. front (3). Clock activity is characterized by oscillating gene expression that cycles in tandem with somite emergence. The Normal Somitogenesis with One Functional Allele of Either Fgf4 or wavefront is an activity gradient, strongest at the caudal end of Fgf8. Embryos in which both Fgf4 and Fgf8 had been ablated in fl fl the embryo, which maintains the undifferentiated PSM; below the PSM (T-Cre; Fgf4 ox/null; Fgf8 ox/null; hereafter “FGF mu- a critical wavefront threshold, called the determination front, tants”) died prenatally (Fig. 1A). Embryos with one or more anterior PSM cells differentiate (1). If cells pass the determination front when clock genes are at a certain point in their cycle, segmentation occurs, resulting in a somite boundary. Whereas numerous genes in the Notch, WNT, and FGF path- Author contributions: L.A.N., N.H., and M.L. designed research; L.A.N., N.H., and M.L. fi performed research; L.A.N., N.H., and M.L. analyzed data; and L.A.N. and M.L. wrote ways have been identi ed as oscillating clock genes, wavefront the paper. identity has been more controversial. Initially, Fgf8 was proposed to fl encode wavefront activity because experimental manipulation of The authors declare no con ict of interest. FGF8 levels caused corresponding shifts in the position of the de- This article is a PNAS Direct Submission. termination front in cultured chick and zebrafish embryos (4, 5). 1Present address: Johns Hopkins Technology Transfer, Baltimore, MD 21201-3894. Moreover, manipulation of downstream components implicated 2To whom correspondence should be addressed. E-mail: [email protected]. FGF signaling in determination front positioning (6). However, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. mouse embryos lacking Fgf8 in the PSM revealed that Fgf8 is not 1073/pnas.1007417108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1007417108 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 functional alleles of either Fgf4 or Fgf8 (hereafter “control em- chyury and Shh on its anterior surface (Fig. S2), and variably also bryos”) were viable to birth (Fig. 1A). Skeletal preparations expressed somite markers (Fig. 2 E–J). This suggests that this showed changes in vertebral numbers and homeotic trans- protrusion is formed by the elongating notochord, but may also formations of vertebral identity in some control genotypes contain some paraxial mesoderm. (Table S1) but otherwise normal vertebral development. Despite Normally, somites form in a linear array (Fig. 2C). By contrast, previous observations showing that chemical inhibition of FGF FGF mutants form misshapen somite-like epithelial conden- signaling reduces PSM length (4, 5), no difference was seen sations that overlap in the dorsal–ventral axis due to abnormal between control genotypes in the length or extension rate of the segmentation planes (Fig. 2D). Markers of anterior (Tbx18) and PSM (Fig. 1 B and C). Collectively, these data demonstrate that posterior (Uncx4.1, Dll1) somite compartments are disorganized – a single copy of either Fgf4 or Fgf8 is sufﬁcient for somitogenesis. in FGF mutants, with no clear stripes observed (Fig. 2 E J). These data suggest that whereas a somitogenic program initiates Loss of Fgf4 and Fgf8 Ablates FGF Signaling in the PSM. We exam- in FGF mutant embryos, the resulting somites are malformed ined the expression of the FGF-responsive genes (18) Etv4, and disorganized. Spry1, Spry2, and Spry4 in FGF mutant embryos. Expression of We investigated whether these disordered somites might form these target genes was lost in the primitive streak of gastrulae at from cells that had not undergone Cre-mediated recombination. bud stages (19) (Fig. 1 D–I) as well as the PSM during early Examination of R26R-lacZ, a reporter of Cre activity, indicated that all cells contributing to FGF mutant somite-like tissue had somitogenesis (Fig. S1). Some FGF signaling presumably oc- indeed undergone Cre recombination and presumably excised curred in streak-stage embryos because it is required for gas- Fgf4 and Fgf8 (Fig. 2 C and D). These data, together with the trulation onset (8), but these results show that FGF signaling was early loss of FGF-responsive genes (Fig. 1 D–I), indicate that ablated substantially before the initiation of somitogenesis. This these structures form in the absence of FGF signaling. demonstrates that FGF4 and FGF8 are the critical ligands for FGF signaling in the PSM. Loss of FGF Signaling Results in Loss of PSM Markers and Cycling Genes. To determine the origin of the somite defects, we in- Loss of Fgf4 and Fgf8 Results in Axis Truncation and Abnormal vestigated the expression of genes known to be important in Somitogenesis. At embryonic day (E)9.5, FGF mutant embryos different phases of PSM development. Genes thought to main- exhibited multiple abnormalities, including a severely truncated tain or mark the progenitor population of the PSM, including body axis (Fig. 2 A and B). Examination of Mox1 expression, Brachyury (T), Tbx6, Wnt3a, and Wnt5a, are severely down- a marker of differentiated somites, revealed that a few small and regulated or absent in the mutant PSM as early as the bud and misshapen somites form in the FGF mutants (Fig. 2 A–B′). At headfold stages and completely lost by early somite stages (Fig. 3 E8.5, FGF mutant embryos exhibited axis truncation and a ven- A–D and G–J and Fig. S3). Conversely, retinoic acid (RA) ac- tral protrusion (Fig. 2 E–J), which consistently expressed Bra- tivity, which is thought to promote PSM differentiation (20), appeared to be up-regulated in the FGF mutants (Fig. 3 E and F). We could not determine whether the RA-responsive domain was also caudally expanded, as predicted by mutual antagonism (20), because at the relevant stages, RA activity is observed throughout the PSM even in control embryos (Fig. S3). To determine whether loss of gene expression was due to tissue loss, we sectioned E7.5 embryos and used TUNEL labeling and anti-phosphohistone H3 staining to visualize cell death and proliferation, respectively. We observed no difference between controls and FGF mutants (Fig. S4). Additionally, we observed ongoing expression of the undeleted portion of the Cre-recombined Fgf8 allele in FGF mutant PSM, demonstrating that the posterior tissue was present and transcriptionally active (Fig. 3 K and L). Because a few somites form in the FGF mutants, we investigated whether the mutant PSM expressed cycling genes. With a few exceptions described in further detail below, we observed no expression of cycling genes Hes7 (n =14/14)orLfng (n =11/13) in FGF mutant embryos (Fig. 3 M–P and Fig. S3).

The Entire PSM Undergoes Somitogenesis in the Absence of FGF Signaling. Although Lfng expression was lost in most mutants, we found that in two out of three late headfold (LHF)-stage mutants it was expanded throughout the PSM immediately before the normal onset of somitogenesis (Fig. 4 A and B). In wild-type embryos, Lfng expression has two separately regulated domains: cycling waves in the posterior PSM, and a stripe in the rostral half of the newly forming somite (21). The absence of other cycling genes inspired us to speculate that this abnormal Lfng expression represented a general expansion of anterior fates in the mutant PSM. Fig. 1. Creation of FGF mutant embryos. (A) Mating scheme for production To test this idea, we examined Fgf18 expression, which is of FGF mutant embryos, which were not found in Mendelian ratios at birth normally observed in the forming somite and anterior PSM (22). (P < 0.001). Δ, null allele; fl, floxed conditional allele. (B) Representative The Fgf18 domain is indeed expanded throughout the FGF embryo from C. PSM length was measured from the last visible somite to the mutant PSM in early headfold (EHF) and LHF stages (Fig. 4 C embryonic posterior end (black line). (C) PSM length in E8.5–E9.5 control and D). No expression of Fgf18 is observed at earlier or later genotypes, as indicated by corresponding color in A. Lines represent linear stages (Fig. S5). These data together suggested that the entire regression. No significant difference was seen in length or rate of growth mutant PSM had acquired anterior PSM fate by the LHF stage. between different control genotypes (P ∼ 0.9, ANCOVA test). (D–I) Expres- We then asked whether the expansion of anterior PSM fates sion of FGF-responsive genes Spry2, Spry4, and Etv4 is lost from bud-stage led to expansion of somitogenic differentiation. We examined (D–G) and headfold-stage (H and I) FGF mutant embryos. Mesp2, which initiates segmentation of the forming somite and

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1007417108 Naiche et al. Downloaded by guest on October 2, 2021 serves as a marker of the anterior of the wavefront (1). Mesp2 expression is expanded throughout the PSM of FGF mutant embryos (Fig. 4 E and F). This Mesp2 expansion is transient, occurring only between LHF and one-somite stages and ceasing thereafter (Fig. S5), consistent with its normal rapid down- regulation via a negative feedback loop (23). We further con- ﬁrmed the somitic differentiation of the mutant PSM by examining the somite marker Mox1. Normally, Mox1 is excluded from the PSM, but in FGF mutants, Mox1 is expressed throughout the posterior of the embryo (Fig. 4 G and H). Thus, in FGF mutants, the entire PSM expresses genes typical of normal somitogenesis, supporting the hypothesis that FGF4 and FGF8 signaling com- prises the wavefront activity, which would normally repress somitic differentiation in the posterior PSM. Interestingly, despite the simultaneous differentiation of the mutant PSM, a marker of somite polarity, Uncx4.1, still appears in a rostral–caudal progression (Fig. S5).

FGF Signaling Acts Independently of WNT Signaling to Maintain the Wavefront. Our results show that ablation of FGF signaling in the PSM results in the loss of WNT ligand expression (Fig. 3 G–J), suggesting that the WNT pathway is downstream of FGF signaling. Conversely, loss of WNT signaling, which similarly halts axis extension and somitogenesis, reciprocally results in the loss of FGF gene expression (13, 16). This reciprocal and comple- mentary down-regulation prevents the determination of which signal is the wavefront. To dissect the requirements for FGF and WNT signaling, we BIOLOGY restored WNT signaling to the PSM of FGF mutant embryos using a conditional gain-of-function (GOF) Ctnnb1 allele (24). DEVELOPMENTAL The recombined allele produces constitutively active β-catenin in the PSM, producing “WNT GOF” embryos (13, 16). By in- troducing this allele into the FGF mutant background (T-Cre; fl fl Fgf4 ox/null; Fgf8 ox/null; Ctnnb1lox(ex3)/+), we maintained canonical WNT signaling in the PSM of FGF mutants, generating “WNT- restored” embryos. We confirmed that WNT signaling had been restored to the mutant PSM before the onset of somitogenesis by examining Sp5, a direct target of canonical WNT signaling (25) (Fig. 5 A–D). As previously reported, WNT GOF embryos display an expanded PSM (13, 16), although this effect is modest at early stages (Fig. 5 F, J, and N). However, in WNT-restored embryos this expansion did not occur, although some markers of PSM progenitor populations were partially rescued. Compared with FGF mutants, WNT-restored embryos show limited up-regulated Brachyury and Tbx6 expression in the primitive streak and PSM (Fig. 5 E–L and Fig. S6). These data suggest that WNT signaling can partially activate the expression of PSM progenitor genes but cannot maintain robust target gene expression without FGF signaling. Furthermore, WNT-restored embryos truncate at a similar axial level as FGF mutants, indicating that WNT signaling is insufficient to rescue PSM elongation (Fig. S6). Consistent with this observation, WNT-restored embryos show the same caudal expansion of Lfng and Mox1 expression as FGF mutants (Fig. 5 M–T), indicating that even when WNT signaling is maintained in the PSM, the loss of FGF signaling results in the lack of wavefront activity and premature PSM differentiation. Therefore, FGF4 and FGF8 signaling acts as the sole mediator of wavefront activity. Discussion Here we show that the PSM-specific inactivation of Fgf4 and Fgf8 before the onset of somitogenesis results in embryonic axis truncation and premature differentiation of the entire PSM. This Fig. 2. Abnormal somites in FGF mutants. Lateral (A and B) and dorsal (A′ and B′) views of E9.5 embryos. FGF mutant embryos are dramatically truncated, but remaining paraxial tissue expresses somite marker Mox1.(A′ and B′) FGF mutant somites are small and irregularly shaped. (C and D) X-gal– the posterior somite compartment Dll1 and Uncx4.1 is disorganized and up- stained sagittal sections of E8.5 embryos carrying the Cre reporter R26R-lacZ. regulated in E8.5 FGF mutant embryos. (I and J) Expression of anterior somite Asterisks indicate the center of an epithelialized somite; lines indicate the compartment marker Tbx18 is disorganized and faint in FGF mutant embryos. plane of segmentation separating somites. (E–H) Expression of markers of Red arrowheads indicate ventral protrusion of extending notochord.

Naiche et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 Fig. 3. Loss of PSM gene expression in somite-stage FGF mutants. (A–D) Expression of T-box genes required for mesoderm formation, Brachyury and Tbx6,is lost in the primitive streak and PSM of FGF mutant embryos. (E and F) RARE-lacZ, a reporter of retinoic acid activity (43), is up-regulated in FGF mutants. (G–J) Wnt3a and Wnt5a are not expressed in FGF embryos. (K and L)Exon1ofFgf8, which is intact in the recombined Fgf8 allele, is expressed in the PSM of FGF mutants. (M–P) Oscillating clock genes Hes7 and Lfng are silent in the PSM of somite-stage FGF mutant embryos. All views are lateral with anterior to the left. Due to lack of visible somites, FGF mutant embryos were staged by comparison of headfold development with littermates.

establishes that signaling via FGF4 or FGF8 is required for the Cre excision at the Fgfr1 locus, extraordinary stability of FGFR1 suppression of differentiation in posterior PSM and is synony- protein after gene excision, or expression of another FGF re- mous with the wavefront. This study therefore lays a foundation ceptor at levels undetectable by in situ hybridization. Regardless that extends earlier data from cultured embryos (4, 5) and pro- of cause, T-Cre; Fgf4; Fgf8 mutant embryos provide a system in vides a basis for genetic investigation of wavefront activity and which loss of FGF signaling before the onset of somitogenesis mechanisms of somitogenesis. has been evaluated. In FGF mutant embryos, markers of somitic differentiation are Interestingly, FGF signaling is lost (as shown by the down- expressed only at the early headfold to one-somite stage, after regulation of Etv4) despite the ongoing expression and even ex- which they disappear. Given that the embryo truncates immedi- pansion of two other FGFs, Fgf3 and Fgf18 (Fig. S7). It is possible ately thereafter, this single wave of differentiation appears to be that in the absence of FGF4 and FGF8, other FGF ligands are the origin of the later abnormal somites. Our observation that insufficient to activate the pathway; this is analogous to the limb Mesp2 expression, which determines segmentation boundaries (1), bud, where Fgf4 and Fgf8 are required despite the expression of is expanded throughout the mutant PSM suggests that multiple other FGF ligand genes (17, 28–30). Alternately, it is possible that cleavage planes form simultaneously, resulting in an abnormal other FGFs in the PSM are acting via noncanonical modes of FGF cluster of somites. Conversely, we also observe that the appear- signaling, such as activation of the PLCγ pathway, which do not ance of a somite polarity marker, Uncx4.1, is progressive in FGF activate classical FGF signaling markers (31–33). mutants, suggesting that complete differentiation does not occur We cannot identify the wavefront activity solely from our FGF simultaneously in the entire PSM. Premature differentiation of the mutants or previously studied WNT loss-of-function mutants (13, PSM is also presumably responsible for axis truncation in the ab- 16) because there is a loss of gene expression of key ligands in one sence of cell death; progenitor cells that normally elongate the axis pathway when the other pathway is genetically ablated. However, instead assume somitic fate. Our previous work showed that Fgf8, widely thought to be in- restoration of canonical WNT signaling in FGF mutant PSM dispensable for somitogenesis (1, 26), is not required for this partially restored expression of progenitor markers Tbx6 and process (7). Here we observe that Fgf4 likewise is not required for Brachyury, a direct WNT target (34), but did not rescue the pre- somitogenesis per se. However, embryos with one functional Fgf8 mature differentiation of the PSM or axis extension. Together, fl allele and lacking Fgf4 (T-Cre; Fgf4flox/null; Fgf8 ox/+) have more these observations suggest a model in which FGF signaling and severe homeotic vertebral transformations than those with one WNT signaling are mutually reinforcing, forming a positive feed- fl fl functional Fgf4 allele but lacking Fgf8 (T-Cre; Fgf4 ox/+;Fgf8 ox/null) back loop in the PSM and functioning in tandem to maintain PSM (Table S1), suggesting that Fgf4 plays the greater role in generating progenitors (Fig. 6). Additionally, FGF signaling has longer-range vertebral position information. functions as a suppressor of somitogenesis (i.e., the wavefront). Fgfr1 is the only FGF receptor gene thought to be expressed in This model predicts that the axial truncations observed in FGF the PSM (11). However, three publications analyzing conditional mutants, WNT-restored mutants, and WNT loss-of-function Fgfr1 inactivation in the mouse describe loss of FGF signaling mutants are due to the premature somitic differentiation of PSM and somitogenesis defects that are milder and occur later than in progenitor tissue that would normally proliferate to allow axis our mutants (11, 12, 27). Importantly, one of these studies also extension. Our data also demonstrate that the expansion of the used T-Cre, making it directly comparable with our study (11). PSM in WNT GOF embryos (13, 16) is FGF-dependent, and This discrepancy between phenotypes may be due to inefficient suggests that the axis truncation occurs slightly later in WNT loss-

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1007417108 Naiche et al. Downloaded by guest on October 2, 2021 BIOLOGY DEVELOPMENTAL

Fig. 5. FGF signaling is required independently of WNT signaling. (A–D) Sp5, a WNT target, is dramatically down-regulated in FGF mutant PSM but restored in WNT-restored late bud-stage embryos. (E–H) Brachyury is expressed in the PSM of control and WNT GOF embryos, lost in FGF mutant embryos, and partially restored in the posterior PSM of WNT-restored embryos. (I–L) Tbx6 expression is lost from FGF mutant PSM but partially restored in the anterior PSM of WNT-restored embryos. (M–P) Lfng is Fig. 4. Expanded differentiation markers in FGF mutant PSM. (A–F) Lateral expressed in a narrow stripe in control one-somite embryos. Two wider views. At LHF to one-somite stages, markers of anterior PSM and forming stripes are observed in WNT GOF embryos. FGF mutant and WNT-restored somite Lfng, Fgf18, and Mesp2 are expanded from a narrow stripe in control embryos both show broad expansion of a single stripe of Lfng throughout embryos to a broad domain covering the PSM in FGF mutant embryos. (G the PSM. (Q–T) Expression of somitic marker Mox1 is excluded from the and H) Posterior views. Somite marker Mox1 is expanded into the PSM of PSM of control and WNT GOF embryos, but expanded into the PSM in FGF FGF mutant embryos. mutant and WNT-restored embryos. Brackets indicate region of expression; asterisks indicate node. A–D and Q–T are posterior views; E–P are lateral views. of-function embryos than in FGF mutants because of the later, secondary, loss of FGF signaling (13, 16). In the normal embryo, somitogenesis initiates at approximately Finally, this study has implications for the role of Fgf4 and Fgf8 the same time that the embryonic axis starts to rapidly elongate. in vertebrate evolution. It is now clear that FGF4 and FGF8 are Prior work in other systems has shown that FGF signaling is suf- the principal FGFs required for both axis extension and limb bud ficient to initiate gene oscillations that propagate to regions of outgrowth (29, 37). Both genes are also expressed in the genital lower FGF expression (35, 36), and our work has shown that FGF tubercle (GT), and whereas GT-specific inactivation of Fgf4 and signaling suppresses PSM differentiation. It is therefore tempting Fgf8 shows that they are dispensable for GT outgrowth, this may to speculate that initiation of somitogenesis is triggered when the be due to increased activity of other ligands (38). Furthermore, axis has elongated sufficiently such that the anterior PSM no in all three systems, FGF activity is regulated by WNT signaling longer receives FGF signals, which permits both the translation of (38, 39), and the coordinate WNT/FGF signaling system may local oscillations into the propagating wave of the clock and the have evolved in early chordates to link the somite clock to body axis extension (40). These data support speculations that pro- differentiation of the anterior PSM as it leaves the wavefront. grams controlling axis and tail extension have been co-opted over However, our evidence does not support this hypothesis. We ob- the course of evolution to control the outgrowth of multiple serve the complete loss of FGF signaling in bud-stage FGF mutant appendages (41, 42). embryos, whereas we do not observe the onset of somitic differentiation until the LHF stage, when somitic differentiation would Materials and Methods ∼ normally initiate. This represents a gap of 6 h of development Mouse Strains and Embryo Harvesting. Previously described mice carrying (19) between the loss of FGF signaling and the onset of somitic conditional mouse mutations (Fgf8, Fgf4, Ctnnb1lox(ex3), R26R-lacZ)or differentiation, suggesting a separate, currently unknown, per- transgenes (T-Cre, RARE-lacZ) were used and genotyped as described (7, 17, missive signal that initiates the somitic program within the PSM. 24, 43–45). Mice were maintained on a mixed genetic background. Embryos

Naiche et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 development were scored by comparing headfold development with littermates.

Marker Analysis. In situ hybridization and lacZ staining were performed via standard techniques (46, 47). At least four embryos of each genotype were observed for each marker.

Morphometric Analysis. Fixed, unstained embryos were photographed and the PSM length was determined by using ImageJ (http://rsbweb.nih.gov/ij)(Fig.1B).

Skeletal Preparations. Embryonic day 18.5 to postnatal day 1 skeletons were Fig. 6. Proposed model for FGF and WNT signaling roles in somitogenesis. prepared using standard techniques (47). FGF signaling (in red) and WNT signaling (in blue) reciprocally up-regulate each other. Both pathways help maintain the PSM progenitor population. FGF signaling separately inhibits somitic differentiation. As the axis extends, ACKNOWLEDGMENTS. We thank C. Elder and E. Truffer for expert technical assistance and A. Liu for cell death/proliferation studies. We thank PSM cells become displaced more anteriorly relative to the sources of FGF M. Fivash and the Histopathology laboratory (NCI-Frederick) for technical signaling, pass the determination front, where FGF signaling is attenuated assistance. We also thank A. Boulet and M. Capecchi for sharing prepublica- below a threshold, and initiate somite segmentation. tion data and T. Yamaguchi, M. Perry, S. Mackem, M. Kuehn, J. Gibson-Brown, and all members of the Lewandoski laboratory for insightful comments on the manuscript. This work was supported by the Center for Cancer Research harvested on E7–E8 were precisely scored using the Downs and Davies of the Intramural Research Program of the National Institutes of Health staging criteria (19) or by somite count. Embryos deﬁcient for somite through the National Cancer Institute.

1. Dubrulle J, Pourquié O (2004) Coupling segmentation to axis formation. Development 24. Harada N, et al. (1999) Intestinal polyposis in mice with a dominant stable mutation of 131:5783–5793. the β-catenin gene. EMBO J 18:5931–5942. 2. Turnpenny PD, et al. (2007) Abnormal vertebral segmentation and the Notch 25. Fujimura N, et al. (2007) Wnt-mediated down-regulation of Sp1 target genes by signaling pathway in man. Dev Dyn 236:1456–1474. a transcriptional repressor Sp5. J Biol Chem 282:1225–1237. 3. Cooke J, Zeeman EC (1976) A clock and wavefront model for control of the number of 26. Wilson V, Olivera-Martinez I, Storey KG (2009) Stem cells, signals and vertebrate body repeated structures during animal morphogenesis. J Theor Biol 58:455–476. axis extension. Development 136:1591–1604. 4. Sawada A, et al. (2001) Fgf/MAPK signalling is a crucial positional cue in somite 27. Niwa Y, et al. (2007) The initiation and propagation of Hes7 oscillation are boundary formation. Development 128:4873–4880. cooperatively regulated by Fgf and Notch signaling in the somite segmentation clock. 5. Dubrulle J, McGrew MJ, Pourquié O (2001) FGF signaling controls somite boundary Dev Cell 13:298–304. position and regulates segmentation clock control of spatiotemporal Hox gene 28. Lewandoski M, Sun X, Martin GR (2000) Fgf8 signalling from the AER is essential for activation. Cell 106:219–232. normal limb development. Nat Genet 26:460–463. 6. Delfini MC, Dubrulle J, Malapert P, Chal J, Pourquié O (2005) Control of the 29. Sun X, Mariani FV, Martin GR (2002) Functions of FGF signalling from the apical segmentation process by graded MAPK/ERK activation in the chick embryo. Proc Natl ectodermal ridge in limb development. Nature 418:501–508. Acad Sci USA 102:11343–11348. 30. Moon AM, Capecchi MR (2000) Fgf8 is required for outgrowth and patterning of the 7. Perantoni AO, et al. (2005) Inactivation of FGF8 in early mesoderm reveals an essential limbs. Nat Genet 26:455–459. role in kidney development. Development 132:3859–3871. 31. Sivak JM, Petersen LF, Amaya E (2005) FGF signal interpretation is directed by Sprouty 8. Sun X, Meyers EN, Lewandoski M, Martin GR (1999) Targeted disruption of Fgf8 causes and Spred proteins during mesoderm formation. Dev Cell 8:689–701. failure of cell migration in the gastrulating mouse embryo. Genes Dev 13:1834–1846. 32. Nutt SL, Dingwell KS, Holt CE, Amaya E (2001) Xenopus Sprouty2 inhibits FGF- 9. Itoh N, Ornitz DM (2008) Functional evolutionary history of the mouse Fgf gene mediated gastrulation movements but does not affect mesoderm induction and family. Dev Dyn 237:18–27. patterning. Genes Dev 15:1152–1166. 10. Yamaguchi TP, Conlon RA, Rossant J (1992) Expression of the fibroblast growth factor 33. Haremaki T, Fraser ST, Kuo YM, Baron MH, Weinstein DC (2007) Vertebrate Ctr1 receptor FGFR-1/flg during gastrulation and segmentation in the mouse embryo. Dev coordinates morphogenesis and progenitor cell fate and regulates embryonic stem Biol 152:75–88. cell differentiation. Proc Natl Acad Sci USA 104:12029–12034. 11. Wahl MB, Deng C, Lewandoski M, Pourquié O (2007) FGF signaling acts upstream of 34. Yamaguchi TP, Takada S, Yoshikawa Y, Wu N, McMahon AP (1999) T (Brachyury) is the NOTCH and WNT signaling pathways to control segmentation clock oscillations in a direct target of Wnt3a during paraxial mesoderm specification. Genes Dev 13: mouse somitogenesis. Development 134:4033–4041. 3185–3190. 12. Oginuma M, Niwa Y, Chapman DL, Saga Y (2008) Mesp2 and Tbx6 cooperatively 35. Nakayama K, Satoh T, Igari A, Kageyama R, Nishida E (2008) FGF induces oscillations create periodic patterns coupled with the clock machinery during mouse of Hes1 expression and Ras/ERK activation. Curr Biol 18:R332–R334. somitogenesis. Development 135:2555–2562. 36. Ishimatsu K, Takamatsu A, Takeda H (2010) Emergence of traveling waves in the 13. Aulehla A, et al. (2008) A β-catenin gradient links the clock and wavefront systems in zebrafish segmentation clock. Development 137:1595–1599. mouse embryo segmentation. Nat Cell Biol 10:186–193. 37. Boulet AM, Moon AM, Arenkiel BR, Capecchi MR (2004) The roles of Fgf4 and Fgf8 in 14. Dequéant ML, Pourquié O (2008) Segmental patterning of the vertebrate embryonic limb bud initiation and outgrowth. Dev Biol 273:361–372. axis. Nat Rev Genet 9:370–382. 38. Miyagawa S, et al. (2009) Dosage-dependent hedgehog signals integrated with Wnt/ 15. Aulehla A, et al. (2003) Wnt3a plays a major role in the segmentation clock β-catenin signaling regulate external genitalia formation as an appendicular program. controlling somitogenesis. Dev Cell 4:395–406. Development 136:3969–3978. 16. Dunty WC, Jr., et al. (2008) Wnt3a/β-catenin signaling controls posterior body de- 39. Soshnikova N, et al. (2003) Genetic interaction between Wnt/β-catenin and BMP velopment by coordinating mesoderm formation and segmentation. Development receptor signaling during formation of the AER and the dorsal-ventral axis in the 135:85–94. limb. Genes Dev 17:1963–1968. 17. Sun X, et al. (2000) Conditional inactivation of Fgf4 reveals complexity of signalling 40. Beaster-Jones L, et al. (2008) Expression of somite segmentation genes in amphioxus: during limb bud development. Nat Genet 25:83–86. A clock without a wavefront? Dev Genes Evol 218:599–611. 18. Turner N, Grose R (2010) Fibroblast growth factor signalling: From development to 41. Kondo T, Zákány J, Innis JW, Duboule D (1997) Of fingers, toes and penises. Nature cancer. Nat Rev Cancer 10:116–129. 390:29. 19. Downs KM, Davies T (1993) Staging of gastrulating mouse embryos by morphological 42. Minelli A (2002) Homology, limbs, and genitalia. Evol Dev 4:127–132. landmarks in the dissecting microscope. Development 118:1255–1266. 43. Rossant J, Zirngibl R, Cado D, Shago M, Giguère V (1991) Expression of a retinoic acid 20. Goldbeter A, Gonze D, Pourquié O (2007) Sharp developmental thresholds defined response element-hsplacZ transgene defines specific domains of transcriptional through bistability by antagonistic gradients of retinoic acid and FGF signaling. Dev activity during mouse embryogenesis. Genes Dev 5:1333–1344. Dyn 236:1495–1508. 44. Soriano P (1999) Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat 21. Cole SE, Levorse JM, Tilghman SM, Vogt TF (2002) Clock regulatory elements control Genet 21:70–71. cyclic expression of Lunatic fringe during somitogenesis. Dev Cell 3:75–84. 45. Meyers EN, Lewandoski M, Martin GR (1998) An Fgf8 mutant allelic series generated 22. Vermot J, et al. (2005) Retinoic acid controls the bilateral symmetry of somite by Cre- and Flp-mediated recombination. Nat Genet 18:136–141. formation in the mouse embryo. Science 308:563–566. 46. Wilkinson DG, Nieto MA (1993) Detection of messenger RNA by in situ hybridization 23. Morimoto M, et al. (2007) The negative regulation of Mesp2 by mouse Ripply2 is to tissue sections and whole mounts. Methods Enzymol 225:361–373. required to establish the rostro-caudal patterning within a somite. Development 134: 47. Nagy A, Gertsenstein M, Vintersten K, Behringer R (2003) Manipulating the Mouse 1561–1569. Embryo (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY).

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1007417108 Naiche et al. Downloaded by guest on October 2, 2021