Requirement of Tbx1 for pouch morphogenesis. (A) A zoomed out view of a 32hpf her5:mCherryCAAX (red) embryo shows the position of pouches relative to the developing eye and ear. (B,C) The first two pouches (p1, p2) form by 20hpf, with the remaining pouches (p3-p6) developing at one per 4h. (D-K) Still images from time-lapse confocal recordings of wild-type and tbx1 mutant pouch development at the times indicated (bottom right of the images) (see supplementary material Movie 1A,B). her5:mCherryCAAX labels endodermal cell membranes. Recordings were started at 26hpf (T0), a stage when the fourth and fifth pouches (p4 and p5) are beginning to develop in the wild-type example. Compared with the typical pouch outpocketing behavior seen in all three wild-type siblings, no endodermal outpocketing was observed in three out of three tbx1 mutants (H-K). (L,M) Fluorescent in situ hybridization for fgf3 (green) and GFP immunohistochemistry to label the her5:GFP-positive pouch endoderm (red). fgf3 was expressed in a segmental pattern in all 39 wild-type siblings and all 14 tbx1 mutants. (N,O) Confocal sections of pre-pouch endoderm in dusp6:dGFP; her5:mCherryCAAX transgenic embryos. In wild type, dusp6:dGFP (green) was expressed segmentally in already formed pouches (arrowhead) and in clusters of presumptive pouch endoderm before outpocketing (arrows); strong expression was also seen in adjacent mesoderm (asterisks) (n=18). In tbx1 mutants, dusp6:dGFP was expressed at lower levels and in fewer cells, yet a segmental pattern in the endoderm was still detected (n=7). Note that we increased the relative gain of the green channel in tbx1 mutants to reveal the weak segmental dusp6:dGFP expression. See supplementary material Fig. S1 for the unaltered images. L′-O′ show the green channel alone. Scale bars: 40μm (B,C,L,M); 20μm (D-K,N,O).

Tbx1 is required in the nkx2.5-positive mesoderm for pouch development. (A-E) Fluorescent in situ hybridization for tbx1 (green) and GFP immunohistochemistry to detect her5:GFP-positive endoderm (red) at 34hpf. In all 58 wild-type siblings, tbx1 expression was observed in her5:GFP-positive endoderm (arrows), adjacent mesoderm (arrowheads) and the ear (asterisk). In all 21 tbx1 mutants, all tbx1 expression was lost. tbx1 expression was selectively restored in the her5:GFP-positive endoderm of all 17 nkx2.3:Tbx1; tbx1^-/- embryos and in the mesoderm of all nine nkx2.5:Tbx1; tbx1^-/- embryos. Inclusion of both transgenes restored tbx1 expression over a broader area (brackets for presumptive endoderm and arrowheads for presumptive mesoderm) in all four nkx2.3:Tbx1; nkx2.5:Tbx1; tbx1^-/- embryos, although for technical reasons her5:GFP was not included. (F-J) In wild-type zebrafish at 34hpf, immunohistochemistry for Alcama (green) labeled five pouches (p1-p5). nkx2.5:Tbx1; tbx1^-/- and nkx2.3:Tbx1; nkx2.5:Tbx1; tbx1^-/- embryos displayed partial rescue of pouches compared with tbx1 mutants, whereas nkx2.3:Tbx1; tbx1^/ embryos did not. Sensory ganglia are indicated with red asterisks. (K-O) Fluorescent in situ hybridization for dlx2a (green) and GFP immunohistochemistry for her5:GFP (red) at 30hpf. In wild-type zebrafish, dlx2a was expressed in the neural-crest-derived mesenchyme of each arch (numbered), with higher expression in posterior arches. In all tbx1^-/- (n=9), nkx2.3:Tbx1; tbx1^/ (n=6), nkx2.5:Tbx1; tbx1^-/- (n=4) and nkx2.3:Tbx1; nkx2.5:Tbx1; tbx1^-/- (n=4) embryos, dlx2a expression was reduced in the second arch and lost from the more posterior arches. For technical reasons, her5:GFP was not included in O. (P-T) Ventral whole-mount views of dissected facial cartilages. Wild-type zebrafish (P) invariantly formed five ceratobranchials (CBs) on each side. CBs were missing in all tbx1 mutants (Q, n=41) and not recovered in nkx2.3:Tbx1; tbx1^-/- (R, n=23), nkx2.5:Tbx1; tbx1^/ (S, n=26) or nkx2.3:Tbx1; nkx2.5:Tbx1; tbx1^/ (T, n=24) embryos. (U) Quantification of pouch defects as assessed by Alcama staining in wild-type zebrafish (n=342), tbx1^-/- (n=63), nkx2.3:Tbx1; tbx1^-/- (n=54), nkx2.5:Tbx1; tbx1^-/- (n=22) and nkx2.3:Tbx1; nkx2.5:Tbx1; tbx1^-/- (n=23). Data represent mean±s.e.m., P values are shown. n.s., not significant. Scale bars: 40μm (A-O).

Rescue of tbx1^-/-pouch defects with mesodermal Fgf8a and Wnt11r. (A-J) Fluorescent in situ hybridization for fgf8a or wnt11r (green) and GFP immunohistochemistry to detect nkx2.5:GFP-positive mesoderm (red) at 30hpf. In wild type, fgf8a was expressed in ventral nkx2.5:GFP-positive mesodermal cores of arches 3 and 4 (m3 and m4), whereas wnt11r was in more dorsal subsets of these mesodermal cores. In tbx1 mutants, mesodermal expression of fgf8a (n=12 of 12) and wnt11r (n=10 of 11) was lost, although fgf8a and wnt11r expression in the anterior (arrow) and posterior (arrowheads) portions of the otic vesicle was unaffected. An nkx2.3:Tbx1 transgene did not restore mesodermal fgf8a (n=0 of 7) and wnt11r expression (n=0 of 5), whereas an nkx2.5:Tbx1 transgene restored mesodermal fgf8a (n=5 of 6) and wnt11r expression (n=9 of 9). Also, nkx2.5:Fgf8-GFP and nkx2.5:Wnt11r transgenes restored fgf8a (n=27 of 27) and wnt11r (n=21 of 21) expression, respectively. (K-O) Alcama immunohistochemistry (green) showed five pouches (p1-p5) in wild-type fish at 34hpf. tbx1^-/- mutants lost all pouches except for the first (p1), whereas individual nkx2.5:Wnt11r or nkx2.5:Fgf8a-GFP transgenes modestly rescued, and combined nkx2.5:Wnt11r and nkx2.5:Fgf8a-GFP transgenes strongly rescued posterior pouches (p2-p5). Sensory ganglia are indicated with red asterisks. (P-T) Ventral views of dissected facial cartilages. A bilateral set of five CBs formed in wild-type zebrafish, and no CBs formed in tbx1 mutants. No rescue of CB cartilage was seen in nkx2.5:Wnt11r; tbx1^-/- (n=39), nkx2.5:Fgf8a-GFP; tbx1^-/- (n=34), or nkx2.5:Wnt11r; nkx2.5:Fgf8a-GFP; tbx1^-/- larvae (n=31). (U) Quantification of pouch defects based on Alcama staining in wild type (n=51), nkx2.5:Wnt11r (n=49), nkx2.5:Fgf8a-GFP (n=36), tbx1^-/- (n=62), nkx2.5:Wnt11r; tbx1^-/- (n=44), nkx2.5:Fgf8a-GFP; tbx1^-/- (n=37), and nkx2.5:Wnt11r; nkx2.5:Fgf8a-GFP; tbx1^-/- (n=24). Data represent mean±s.e.m., P values are shown. (V) Low magnification view of an embryo at 32hpf showing nkx2.5:GFP-positive mesoderm (green) relative to her5:mCherryCAAX-positive pouch endoderm (red) and the developing eye and ear. The schematic shows expression of fgf8a (green) and wnt11r (yellow) within distinct subsets of nkx2.5-positive mesoderm (red) during the formation of endodermal pouches (blue). Scale bars: 40μm (A-O).

Synergistic pouch and cartilage defects in compound wnt11r; fgf8a mutants. (A-D) Alcama immunohistochemistry (green) labeled five pouches (p1-p5) in wild-type embryos at 34hpf. Compared with the mild reductions of pouches in wnt11r or fgf8a single mutants, compound wnt11r^-/-; fgf8a^-/- mutants had a near complete loss of pouches. Sensory ganglia are indicated with red asterisks. (E-H) Dissections of facial cartilage at 5dpf. Wild type had five CBs on each side, wnt11r and fgf8a single mutants had variable fusions and losses of on average one CB and compound wnt11r^-/-; fgf8a^-/- mutants lost multiple CBs. (I) Quantification of defects. Number of embryos examined (pouches, CBs): wild type (106, 102), wnt11r^-/- (66, 57), fgf8a^-/- (68, 98), and wnt11r^-/-; fgf8a^-/- (24, 20). Data represent mean±s.e.m. ***P<0.001 relative to fgf8a mutants alone. Scale bar: 40μm (A-D).

Requirement for Fgf8a in the directional persistence of pouch cells. (A-J) Representative confocal sections from time-lapse recordings show the development of pouches p4-p6 in a wild type (n=5) and fgf8a mutant (n=3) (see supplementary material Movie 3A,B). her5:mCherryCAAX labels endodermal cell membranes (red) and dusp6:dGFP shows dynamic Fgf activity (green). T0 is 26 hpf. Merged images are shown in A-J and dusp6:dGFP alone in A′′-J′′. Schematics in A′-J′ show the graded intensity of dusp6:dGFP (green) relative to endodermal cells (red). Individual cell tracks used for the quantification are shown in A′′′-J′′′. (K-M) The velocity, persistence of migration and distribution of angles of tracked cells over a 7-h period. For distribution of angles, each bin represents the number of cells moving in a particular direction relative to the last cell position. We tracked the cells of two embryos for each wild type and fgf8a mutant. Mean±s.e.m. and P values are shown. n.s., not significant. Scale bar: 20μm (A-J).

Ectopic Fgf8a redirects pouch outgrowth. (A-P) Still images from time-lapse recordings monitoring the development of the third pouch in wild-type embryos with mosaic mesodermal mis-expression of a nkx2.5:Fgf8a-GFP transgene (two examples are shown: A-D and E-H) or a control nkx2.5:GFP transgene (I-L), as well as wnt11r mutants with nkx2.5:Fgf8a-GFP mis-expression (M-P) (see supplementary material Movie 4). her5:mCherryCAAX labels endodermal cell membranes in red. Recordings started at 25hpf (T0), and subsequent stills were taken every 2h. nkx2.5:Fgf8a-GFP-expressing clones transiently diverted developing third pouch cells (arrows) in four out of six embryos, whereas nkx2.5:GFP-expressing clones had no effect on adjacent third pouch cells (arrowheads) in all three of the embryos examined. nkx2.5:Fgf8a-GFP-expressing clones failed to attract wnt11r^-/- endodermal cells (arrowheads) in all embryos (n=3). (Q) A two-step model of Tbx1 function in pouch morphogenesis. Tbx1 promotes wnt11r and fgf8a expression in distinct domains of the mesoderm, with Wnt11r initiating pouch morphogenesis through epithelial destabilization and Fgf8a guiding subsequent pouch outgrowth. Scale bar: 20μm (A-P).

Original image of dusp6:dGFP; her5:mCherryCAAX; tbx1-/- embryos
(A) The same image as Figure 1O but without increasing the gain. This image shows that overall dusp6:dGFP expression is very reduced in tbx1 mutants compared to the wild-type control of Figure 1N. Scale bar = 40 μM.

Mesodermal tbx1 expression during pouch formation and a requirement for Tbx1 in ectomesenchyme development
(A-C) Fluorescent in situ hybridization shows expression of tbx1 (green) in nkx2.5:GFP+ mesoderm labeled by GFP immunohistochemistry (red), as well as in nkx2.5:GFP- endoderm. Anterior is to the left and ventral is down. Scale bar represents 20 μM.
(D-F) Alcama immunohistochemistry on 34 hpf embryos. Transgenic expression of Tbx1 in the endoderm or mesoderm of nkx2.3:Tbx1 or nkx2.5:Tbx1 embryos, respectively, does not affect pouch development.
(G-I) In situ hybridization for tbx1 (dark blue) in 34 hpf embryos. In 31/31 wild-type siblings, tbx1 expression is observed in the pharyngeal region ventral to the ear (asterisk). In 8/8 tbx1 mutant siblings, all tbx1 expression is lost. In 11/11 nkx2.3:Tbx1; tbx1-/- embryos, tbx1 expression is partially restored to the pharyngeal region, particularly posterior to the ear.
(J and K) Fluorescent in situ hybridization for sox10 (green) at 12 hpf. Dorsal views with anterior to the left show sox10 expression in two bilateral fields of neural crest cells in both wild types and tbx1 mutants. We also note that sox10 expression was consistently higher in tbx1 mutants.
(L-O) Fluorescent in situ hybridization for dlx2a (green) to detect neural-crest-derived ectomesenchyme. While dlx2a is expressed in three distinct neural crest streams in 11/11 wild-type siblings at 16.5 hpf, its expression is reduced in the first and second streams and nearly lost in the third stream in 3/5 tbx1 mutants. Compared to wild-type siblings at 18 hpf (n=21), the arch expression of dlx2a was greatly reduced in 4/4 tbx1 mutants. her5:GFP labeling of endoderm at 18 hpf (red) shows that pouches have yet to form at this stage. Scale bar represents 40 μM.
(P and Q) BrdU staining (red) labels mitotic cells, and sox10:GFP (green) labels neural-crest-derived ectomesenchyme cells of the pharyngeal arches. Comparable numbers of BrdU+ cells were observed in wild-type siblings (n=4) and tbx1 mutants (n=4) at 30 hpf. Scale bar represents 40 μM. (R and S) Lysotracker Red staining (red) labels dying cells relative to sox10:GFP+ neural-crest-derived cells (green) in wild-type siblings (n=6) and tbx1 mutants (n=6). Arrows show dying cells in the posterior-most pharyngeal arches of a tbx1 mutant.
(T) Quantification of cell death in tbx1 mutants. Dying cells were counted in the entire head (from the posterior-most arch to the anterior limit of the embryo, including the brain, arches, and other tissues), as well as in sox10:GFP+ cells of the arches. A one-tailed Student’s t test with unequal variance was utilized to quantify the number of dying cells, and p values are shown.

wnt4a and fzd8a expression in tbx1 mutants and analysis of nkx2.5:Wnt11r and nkx2.5:Fgf8a-GFP transgenes
(A-H) Fluorescent in situ hybridization for wnt11r, tbx1, or fgf8a mRNA (green). In red, GFP immunohistochemistry detects nkx2.5:GFP+ mesoderm (A-F) or her5:GFP+ endoderm (G,H). At 18 hpf, wnt11r is expressed in the second arch mesodermal core (m2) (A, n=17), and fgf8a is not yet expressed in the nkx2.5:GFP+ mesoderm (E, n=21). By 24 hpf, nkx2.5:GFP labels the third arch mesodermal core (m3) and wnt11r is expressed in both m2 and m3 (B, n=15) and fgf8a just in m2 (F, n=20). fgf8a mRNA expression is also observed at the midbrain-hindbrain boundary (arrowheads in E and F) and the anterior region of the otic vesicle (arrow in F). At 30 hpf, tbx1 expression is unaffected in 8/8 wnt11r mutants (D) or 11/11 fgf8a mutants (H) compared to wild-type siblings (C and G).
(I-N) Fluorescent in situ hybridization for wnt4a or fzd8a mRNA (green) at 30 hpf. In red, GFP immunohistochemistry detects her5:GFP+ endoderm (I,K,L,N) or fli1a:GFP+ neural-crest-derived ectomesenchyme (J,M). Compared to wild types (n=61), ectodermal wnt4a expression (brackets in L) persists but is disorganized in 18/18 tbx1 mutants. In contrast, 9/9 sox32-/- embryos displayed near complete loss of wnt4a ectodermal expression, with the lack of posterior arch segmentation indicative of the absence of pouch endoderm. Compared to normal expression in wild types (n=38), fzd8a expression within the her5:GFP+ endoderm is largely unaffected in 10/14 tbx1 mutants.
(O-Q) Alcama immunohistochemistry shows a series of five pouches at 34 hpf. Compared to wild types, we observed normal pouches in 49/49 nkx2.5:Wnt11r and 27/36 nkx2.5:Fgf8a-GFP embryos. In 9/36 nkx2.5:Fgf8a-GFP embryos, we observed delayed and/or disorganized formation of the last pouch. Sensory ganglia are indicated with red asterisks. Anterior is to the left and ventral is down. Scale bar represents 40 μm.
(R-T) Ventral whole-mount views of Alcian-stained facial cartilage. Compared to wild types, 28/28 nkx2.5:Wnt11r and 25/31 nkx2.5:Fgf8a-GFP embryos formed normal cartilages. In 6/31 nkx2.5:Fgf8a-GFP embryos, we observed partial reductions of one ceratobranchial (arrow in T).

Interactions between Wnt4a, Wnt11r, and Fgf8a during pouch formation
(A-F) Alcama immunohistochemistry at 34 hpf shows a series of five pouches. Compound wnt11r+/-; fgf8a-/- but not wnt4a-/-; fgf8a-/- mutants show enhancement of pouch defects compared to fgf8a-/- embryos alone. Sensory ganglia are indicated with red asterisks. (G-L) Facial cartilages stained with Alcian. Compound wnt11r+/-; fgf8a-/- mutants but not wnt4a-/-; fgf8a-/- mutants show enhancement of ceratobranchial cartilage defects compared to fgf8a-/- mutants alone. (M) Quantification of pouch and ceratobranchial cartilage (CB) defects. Number of pouches and CBs examined for each genotype: wild type (106, 102), wnt11r+/- (51, 47), wnt4a-/- (45, 144), fgf8a-/- (68, 98), wnt11r+/-; fgf8a-/- (38, 80), and wnt4a-/-; fgf8a-/- (20, 30). Data represent mean ± SEM. ***, p<0.001. n.s., not significant. The data for wild type and fgf8a-/- are repeated from Figure 4I. (N-Q) Fluorescent in situ hybridization for fgf8a or wnt11r mRNA (green) at 30 hpf. GFP immunohistochemistry labels the nkx2.5:GFP+ mesoderm in red. fgf8a expression is unaffected in 10/13 and slightly reduced in 3/13 wnt11r mutants. wnt4a expression is unaffected in 15/19 and slightly reduced in 4/19 fgf8a mutants. Scale bar represents 40 μM

Preferential contribution of dusp6:dGFP+ cells to pouches
(A-E) Representative sections from a time-lapse recording of a wild-type dusp6:dGFP; her5:mCherryCAAX embryo (see Movie S2) in which individual cells were tracked (colored dots). Orange and blue arrows show two cells that were dusp6: dGFP- in the posterior-most domain of pharyngeal endoderm at the start of the recording (T0) and then turned on dusp6:dGFP and contributed to the sixth pouch (p6) over the next 8 hours. White and red arrowheads show two adjacent cells that never turned on dusp6:dGFP and were excluded from the pouch. (F) Quantification of the contribution of cells to the pouches based on whether they were initially dusp6:dGFP- (left bar) or dusp6:dGFP+ (right bar). For initially dusp6:dGFP- cells, we then quantified the number of cells that stayed GFP- or turned on dusp6:dGFP (GFP+). For initially dusp6:dGFP+ cells, we quantified the number of cells that maintained dusp6: dGFP (GFP+) or extinguished GFP (GFP-). For both, we also scored the contribution of each category to pouches. Both cells that turned on dusp6:dGFP (p<0.001) and maintained dusp6:dGFP (p=0.002) contributed to pouches at a high frequency than cells that remained or turned off dusp6:dGFP. y-axis represents % of cells of each category. (G-I) High magnification confocal sections show her5:mCherryCAAX labeling of endodermal cell membranes within the fourth pouch (p4) and presumptive fifth pouch (p5) at 28 hpf. The epithelium of nascent pouch p5 was multilayered in 8/8 wild-type siblings and 5/5 fgf8a mutants but only in 1/4 wnt11r mutants. Instead, pre-pouch cells in 3/4 wnt11r mutants retained a columnar epithelial morphology. Scale bars represent 20 μM.

Fgf activity in the endoderm of wnt11r mutants
(A and B) Confocal projections show Fgf activity as marked by dusp6:dGFP (green) relative to the her5:mCherryCAAX+ pouch endoderm (red) at 32 hpf. In 8/8 wnt11r mutants, dusp6:GFP expression is grossly unaffected compared to wildtype siblings (n=34). Scale bar represents 40 μM. (C and D) Higher magnification and gain-enhanced sections focused on the posterior pouches seen in (A) and (B) show dusp6:dGFP fluorescence within alternating clusters of endodermal epithelial cells (arrowheads), as well as in clusters of adjacent mesodermal cells (asterisks). While pouch outpocketing is delayed in wnt11r mutants, segmental dusp6:dGFP fluorescence is still detectable within both the mesoderm and endoderm. Scale bar represents 20 μM.

Criteria for scoring pouch rescue
Several examples of 34 hpf Alcama staining are shown that illustrate lack of pouch rescue (score = 0) or partial/full pouch rescue (score = 1). Pouches were scored as “partial rescue” when they were greater than 50% the length of a corresponding wild-type pouch at that position. These criteria were applied consistently across all experimental groups.