XB-ART-2371Dev Cell February 1, 2005; 8 (2): 167-78.
Msx1 and Pax3 cooperate to mediate FGF8 and WNT signals during Xenopus neural crest induction.
FGF, WNT, and BMP signaling promote neural crest formation at the neural plate boundary in vertebrate embryos. To understand how these signals are integrated, we have analyzed the role of the transcription factors Msx1 and Pax3. Using a combination of overexpression and morpholino-mediated knockdown strategies in Xenopus, we show that Msx1 and Pax3 are both required for neural crest formation, display overlapping but nonidentical activities, and that Pax3 acts downstream of Msx1. In neuralized ectoderm, Msx1 is sufficient to induce multiple early neural crest genes. Msx1 induces Pax3 and ZicR1 cell autonomously, in turn, Pax3 combined with ZicR1 activates Slug in a WNT-dependent manner. Upstream of this, WNTs initiate Slug induction through Pax3 activity, whereas FGF8 induces neural crest through both Msx1 and Pax3 activities. Thus, WNT and FGF8 signals act in parallel at the neural border and converge on Pax3 activity during neural crest induction.
PubMed ID: 15691759
Article link: Dev Cell
Species referenced: Xenopus
Genes referenced: cyp26a1 egr2 fgf8 foxd3 gsk3b msx1 nog nppa pax3 snai1 snai2 sox2 sox9 tfap2a twist1 wnt7b zic5
Morpholinos: fgf8 MO1 msx1 MO1 pax3 MO1 pax3 MO4
Article Images: [+] show captions
|Figure 1. Msx1, Pax3, and Slug Expression in the Presumptive Neural Crest Domain and Msx1 Neural Crest-Inducing Activity In Vivo and in Explants(A) Msx1 (a–d) and Pax3 (a′–d′) expression in X. laevis at stages 11 (a and a′), 13 (b and b′), and 16 (c–d′) costained with Krox20 (d–d′, indicates third rhombomere). Arrows in a–a′ indicate the blastopore lips at stage 11. a–a′, lateral views; b–d′, dorsal views.(B) Coexpression of Msx1 and Slug by double ISH (Msx1 [blue]; Slug [red]; a; b, dorsal views; a′–b′, transverse sections).(C) Msx1 mRNA was injected in one blastomere at the four cell stage, LacZ is revealed in pink (yellow arrow points at injected side). The embryos were analyzed at early (stage 13, a, c, e, and g) or late (stage 16, b, d, f, and h) neurula stages by ISH for Slug (a and b), Pax3 (c and d), Twist (e and f), and Sox2 (g and h) (red bar, uninjected neural plate half; yellow bar, width of the neural plate on the injected side). Dorsal views.(D–F) RT-PCR analysis of neural, ectoderm and early NC markers: Pax3 and ZicR1 (D); AP-2, Snail, and Slug (E); FoxD3, Pax3, Zic5, and Sox9 (F) after coinjection of Msx1 and Noggin. (D) Lane 4, Noggin (25 pg); lane 5, Msx1 (500 pg); lane 6, Msx1 (500 pg) + Noggin (25pg). (E) Lane 4, Noggin (25 pg); lane 5, Msx1 (100 pg); lane 6, Msx1 (500 pg); lane 7, Msx1 (100 pg) + Noggin (25 pg); lane 8, Msx1 (500 pg) + Noggin (25 pg). (F) Lanes 1–4, as in (E). Lanes 5–7, Msx1 (100–250–500 pg). Lanes 8–10, Noggin (25 pg) + Msx1 mRNA (100–250–500 pg).|
|Figure 2. Antisense Morpholino Nucleotide-Mediated Depletion of Msx1 Activity Blocks Neural Crest Induction(A) Msx1-MO inhibited Slug (a) and FoxD3 (b) (stage 16 neurulae, dorsal view). Sibling embryos traced with fluorescence (c and d) showed decreased Slug (e) and enlarged Pax3 (red arrow, f), corresponding to increased Sox2 expression laterally (g) at the expense of the NC and ectoderm domain (h). Small arrow indicates the midline.(B) Phenotype of the Msx1 knockdown in tailbud and tadpoles. ISH for Twist (stage 22, a, dorsal view) and stage 35 (compare b and c to d). Tadpoles exhibit craniofacial defects and decreased melanocyte development (e and f). b–f, lateral views.(C) Dose response and rescue of Msx1 MO injections. At 20 ng dose, Slug expression is reduced in more than 80% of the embryos (red line). This effect is rescued (16%) by the coinjection of Msx1 MO with an Msx1 construct lacking the MO binding sequence (δXT-Msx1, blue line). The control MO (co-MO) does not affect Slug pattern (black line). ISH for rescue of Slug (a) and FoxD3 (b). Sequential injections of Msx1 MO and δXT-Msx1 also restore Twist expression in the cephalic and trunk NC (c, non injected side; d, injected side).|
|Figure 3. Pax3 Overexpression Increases Neural Crest Formation In Vivo, whereas Depletion of Pax3 Activity In Vivo Prevents Neural Crest Formation(A) Injections of 50–100 pg of Pax3 increase Slug (a), Snail (b), and FoxD3 (c) at the neural border as does ZicR1 (500 pg, d). In contrast, coinjections of Pax3 + ZicR1 expand NC at the neural border (not shown) and induce ectopic ventral Slug expression when targeted ventrally (e, ventral views). a–d, dorsal views.(B) Pax3-MO blocks Slug (a) and FoxD3 induction (b). In contrast, Pax3 mRNA is stabilized by the binding of the MO, resulting in an increased ISH signal (c), and Msx1 is unaffected (d).(C) Tadpole phenotype after Pax3 knockdown. Twist expression is strongly decreased (a and b, frontal views), embryos show fin and craniofacial defects (c), and pigment cells do not differentiate after bilateral injections (d amd e, albinos eggs fertilized with sperm from pigmented male).(D) Pax3-MO injection phenotype is dose dependent. More than 80% of the embryos show decreased NC markers expression at doses above 30 ng (red square). An MO with two nucleotide mismatch (Pax3-MO-mis, open squares) downregulates Slug expression in only 30% of the embryos, and the control-MO has no effect (circles). Mouse Pax3 mRNA rescues the knockdown phenotype (rescue, black circles): ISH for Slug (a) and FoxD3 (b).|
|Figure 4. Relationships and Mechanisms of NC Induction by Msx1 and Pax3(A) Epistasis between Msx1 and Pax3 (stage 17 neurulae, dorsal views). When Msx1-MO, which downregulates Slug expression at the neural border (a), is coinjected with Pax3 mRNA (b), Slug expression is restored. The decreased Slug expression observed after injections of Pax3-MO (c) is not rescued by the coinjections of Pax3-MO and Msx1 mRNA (d).(B) RT-PCR analysis on animal caps injected with Msx1 + Noggin in the presence of control (lane 3), Pax3 (lane 4), or β-catenin (lane 5) MOs.(C) In animal cap dissociation assay, Sox2 is induced (lanes 4–6). NC induction observed in nondissociated sibling caps (lanes 5), but not obtained after dissociation (lanes 4 and 6). However, the dissociated cells induce Pax3 and ZicR1 expression (lane 6). Intact animal caps (+), dissociated AC (−).(D) Pax3 + ZicR1 coinjections induce Slug and FoxD3 when the animal caps are intact (lane 5), but not after dissociation (lane6). Intact animal caps (+), dissociated animal caps (–).(E) Pax3 + ZicR1 activation of Slug is blocked by NFz8, but not by cyp26 or a DNA binding mutant of Su(H) (SDBM).(F) Model part 1: Msx1 combined to BMP signaling attenuation activates Pax3 and ZicR1 in explants. Msx1 induction of Slug requires Pax3 activity in vivo and in explants. In turn, Pax3 + ZicR1 are sufficient to activate Slug in vivo and in explants in cooperation with a WNT signal.(G) Phenotype of the double knockdown, Msx1-MO + Pax3-MO. Twist expression is largely depleted in cephalic NC (a, frontal view) and trunk NC (b, white arrows, normal side; red arrows and dots, injected side; dorsal view of the spinal cord). Branchial arch cartilages do not differentiate or are strongly reduced (c and d, ventral views and dissected cartilage). Fin mesenchyme does not form when NC is fluorescently labeled on the injected side (e and f, see Supplemental Data).|
|Figure 5. Neural Crest Induction by the WNT Pathway Requires Pax3, but Not Msx1 Activity(A) In vivo overexpression of Wnt7b in the neural fold results in Slug and Pax3 in the presence of control or Msx1 MOs. In contrast, Pax3-MO coinjections block Wnt7b activity and result in downregulation of Slug in the neural fold. (MOs are traced with fluorescence, and RNAs are traced with LacZ).(B) RT-PCR analysis of animal caps injected with Wnt7b + Noggin in the presence of control (lane 4), Msx1 (lanes 5 and 7), and Pax3 (lanes 6 and 7) MOs. Msx1-MO does not prevent Pax3 nor Slug induction, whereas Pax3-MO reduces Slug induction.(C) The activity of the canonical WNT pathway is required for activation of Msx1 and Pax3 in the neural folds: β-catenin MO (Supplemental Data) or of GSK3 injections in one cell at the 16 cell stage.(D) Model part 2: activation of Msx1 and Pax3 expression in the embryo requires an active WNT signaling, but WNT pathway depends on Pax3 function only for Slug induction.|
|Figure 6. FGF8 Neural Crest Inducing Properties Requires Msx1 Activity(A) Msx1 exhibits a strong response to FGF8 overexpression (50 pg) and is broadly activated on the injected side, including lateral and ventral areas of the embryo (a). The FGF8-AMO (40 ng) abolishes Msx1 expression (b). FGF8 overexpression does not increase the level of Pax3 expression but, rather, expands the Pax3 domain (c). FGF8-MO decreases Pax3 expression partially on the injected side (d).(B) Fgf8 mRNA injections (125–250 pg) either alone or with the control MO expanded greatly Slug expression around the neural plate (c). At this dose, Pax3 expression also encompassed the ANF (a). In the presence of Msx1 MO, FGF8 induction of Slug is abolished (d), whereas Pax3 remains as with control (b). Pax3-MO also blocks Slug activation by FGF8 in vivo.(C) FGF8MO does not affect global patterning of the neural plate as shown by the mild shift in Krox20 expression by rhombomeres three and five but, rather, prevents NC induction as seen by the loss of rhombomere five NC (arrows).(D) Msx1 overexpression rescued the loss of FGF8 activity, showing that Msx1 mediates FGF8 activity. Both Slug (stage 19 tailbud) and Twist (stage 35 tadpoles) are efficiently rescued both in cephalic NC (individual streams are rescued) and trunk NC.(E) Model part 3: FGF8 controls Msx1 expression in vivo independently of WNTs, but FGF8 depends on both Msx1 and Pax3 for inducing Slug.|