Fig. 1. PFKFB4 promotes NB and NC formation.
(A) NC is induced during neural plate stage (st.12.5-14).
Expression of NB specifiers (red) is robustly established
lateral to the neural plate, expression of early NC
specifiers (snail2, foxd3) is faintly initiated (light pink).
(B,C) During neural fold elevation (st.14-17, B), future
NC cells progressively acquire their definitive
specification and specific cellular properties (e.g.
survival, expression of late NC specifiers, cadherin
switch) enabling them to undergo EMT and migration
(st.18-19, C). (D-F,H,I) pfkfb4 is enriched in the NB/NC in
neurulas and tadpoles. (J) pfkfb3 is expressed in the
neural tube. Sense probes are shown inG( pfkfb4) and K
( pfkfb3). Neurula stages: late gastrula/early neural plate
stage (st.12, D); end of neural plate stage (st.14, E);
neural fold stage (st.16, F); neural tube closure/end of
neurulation (st.19, H); tailbud stage (st.22, I).
E-H,J,K: dorsal views, anterior to the right. D,I: side
views. Scale bar: 1 mm. (L-N) Cross-sections through
the mid-neurula anterior neural plate (st.14) show snail2
and pfkfb4 expression in NB/NC. (L) The notochord,
neural plate and paraxial mesoderm are outlined on
Hematoxylin and Eosin-stained sections. (M,N)
Vibratome sections. Arrow indicates the midline.
Arrowheads indicate the st.14 NB (red bar). Scale bars:
200 μm. (O-S) pfkfb4 gain-of-function expanded the
st.14 NB (O,P) and premigratory NC (st.14, Q; st.18, R,
S). Migrating NC streams were expanded (st. 22-24, T,
Z). In contrast, pan- or regional neural plate (sox2, otx2,
hoxb9), ectoderm (ep. ker.) or paraxial mesoderm
(myod) markers were unaffected (U-Y). O-S,U-X: dorsal
views; T,Z: side views; Y: dorsal-posterior view. Scale
bar: 500 μm. co, control side; inj, injected side.
Phenotype scores are shown in Table S6.
Fig. 2. Moderate PFKFB4 depletion results in severe craniofacial defects
in tadpoles. (A) pfkfb4 mRNA levels were quantified on neurula dorsal tissues
by RT-qPCR. Injections at ‘high concentration’ (10 ng/cell in 4-cell-stage
embryos) reduced pfkfb4 levels by 52% and caused developmental arrest. A
‘milder’ concentration (5 ng/cell in 4-cell-stage embryos) decreased pfkfb4
mRNA by 32% (low-level depletion), and allowed development to swimming
tadpoles. (B-E) Craniofacial morphology in tadpoles st.45-46. Low-level
PFKFB4MO depletion reduced the size of craniofacial structures on the
injected side. The branchial arch cartilages area was severely diminished on
the injected side. Lower doses did not significantly reduce cartilage area, or
affect head morphology. n.s., not significant. (C) Head morphology, dorsal
view; arrowhead indicates the injected side. (D) Dissected jaw and branchial
arches cartilages, ventral view. inj, injected side. (E) Schematic of the image
shown in D. Dashed red line indicates midline; light blue, Meckel’s cartilage;
mid blue, ceratohyal cartilage; dark blue, branchial cartilage. Scale bar:
500 μm. Phenotype scores are shown in Table S7.
Fig. 3. PFKFB4 low-level depletion delays NC
early specification, causes retention of NB
character, and impairs NC late specification and
migration. (A) At st.14, snail2 expression was
severely reduced, or abolished, on the injected side.
(B) At st.18, sibling embryos had recovered snail2
expression. (C,D) Whereas sox10 expression was
mainly unaffected, twist1 was severely impaired.
(E-H) In contrast, expression of the immature NC
marker hes4 was expanded, as were some NB
markers, either strongly ( pax3) or moderately (zic1).
Other NB markers were unperturbed (msx1).
(I-K) Neural plate (sox2), non-neural ectoderm (ep.
ker.) and paraxial mesoderm (myod) seemed to be
unaffected. A-K: dorsal views. (L) Percentage of
embryos with each phenotype, i.e. diminished,
increased or normal expression. Snail2 score at
st.14 is indicated as the first bar, then several gene
scores at st.18 are indicated in the following ten
bars. (M-R) St.24 tailbud embryos exhibited a
severe NC migration defect (M-Q). Sox2 expression
appeared grossly unaffected, despite marginal
reduction of optic vesicle size (R). The injected side
(inj) is compared with the control side (co) in side
views (M,N,P,Q, anterior to the right) or frontal views
(O,R; red arrow on injected side). (S) Co-injection of
pfkfb4 mRNA with PFKFBMO rescued both
sox10/twist1 alterations of expression and NC
migration defects in a significant proportion of the
embryos, compared with PFKFB4MO injections
alone. sox10 and twist1 expression were restored or
increased a majority of the embryos. Scale bar:
500 μm. Phenotype scores are shown in Table S8.
Fig. 4. Inducible depletion demonstrates that PFKFB4 has independent roles in NC specification and migration.We co-injected a UV-cleavable senseMO
(PhMO) to block PFKFB4MO until the desired developmental stage and monitored twist1 expression at st.18. (A-H) Strategy of phMO validation: lacZ control
injections (A), unmasked PFKFB4MO, ‘high’ dose (B), PhMO/MO without UV illumination (C; uninduced, u.i.); UV illumination at early NC specification st.12.5
(D-G). D-F show two examples of diminished phenotype and one example of abolished phenotype at st.14;Gshows an example of diminished phenotype at st.19.
(H) Percentage of embryos with each phenotype, i.e. abolished, diminished, increased or normal twist1 expression. lacZ injections did not alter twist1 pattern.
PFKFB4MO severely blocked twist1 expression. twist1 expression was decreased in PhMO/MO-injected embryos UV illuminated immediately after injection (a.i.)
compared with uninduced PhMO/MO-injected embryos. Pfkfb4 depletion, activated at st.12.5, decreased twist1 expression strongly or moderately. This effect
persisted until EMT and early NC emigration (G; st.20). (A-G): dorsal views. (I-W) To deplete pfkfb4 during EMT/early migration, st.18 neurulas were UV
illuminated and analyzed at tadpole st.24. twist1 expression was normal in embryos injected with lacZ (I,J), or with PhMO/MO but not illuminated (M,N). Cells
injected with PFKFB4MO alone died as expected (K,L). PhMO/MO-injected embryos UV illuminated at st.18 exhibited three phenotypes classes: ‘migration’ (O,
P), ‘staining and migration’ (Q,R) and ‘severe staining’ defects (S,T). We compared the percentage of each phenotype (U), of migration distance (V) and of NC
stream area (W; twist1-expressing area) with the contralateral side. Scale bars: 500 μm. All phenotype differences were statistically significant between
illuminated PhMO/MO-injected embryos, and either lacZ-injected, or uninduced PhMO/MO-injected embryos (*P<0.05; **P<0.01; ***P<0.001; ns, not
significant). Error bars represent s.e.m. Phenotype scores are shown in Table S9.
Fig. 5. AKT signaling mediates the effect of PFKFB4 on premigratory NC
maturation. (A) In embryos, PFKFB4 regulates AKT signaling in addition to
glycolysis. Dissected morphant st.14 NB or st.17 NC displayed decreased AKT
signaling. (B) AKT regulates many aspects of cell homeostasis including cell
proliferation, cell survival, and cell metabolism. However, EdU incorporation
showed that cell proliferation rate was normal after PFKFB4MO injections.
Error bars represent s.e.m. n.s., not significant. (C,D) PFKFB4MO affected late
NC specifier twist1 expression, whereas the NC stem cell marker cmyc was
normally activated. (E-H) Pharmacological treatment during neural fold stage
(st.14-18) showed that blocking MAPK signaling (E,G) did not affect NC
development, but blocking PI3K-AKT signaling (F,H) affected twist1 but not
cmyc, thus phenocopying the PFKFB4MO effect. (I,J) Co-injection of
PFKFB4MO with a constitutively AKT (caAkt; blue arrows) rescued the
morphant twist1 phenotype (red arrowheads): two sibling embryos for each
injection are shown. St.18 pfkfb4 morphants presented diminished twist1. In
contrast, siblings co-injected with PFKFB4MO and caAkt mRNA had normal
twist1 expression in the majority of cases. Scale bars: 500 μm. Phenotype
scores are shown in Tables S8, S13.
Fig. 6. PFKFB4 morphant NC progenitors fail to undergo EMT and
migration. (A,B) When transplanted into a wild-type host embryo, GFPlabeled
wild-type NC efficiently migrates and populates the host branchial
arches, regardless of the size of the grafted tissue, i.e. large graft (A) or small
graft (B). Scale bar: 500 μm. (C,D) In contrast, PFKFB4 morphant NCexhibited
defective healing, resulting in small-size grafts, which yielded few, if any,
migratory NCC into the host craniofacial area. (C) Small graft without migratory
cells; (D) small graft with few cells reaching the branchial arches. In D, gfp was
detected by WISH to enhance individual cell visualization. (E) The effect of
PFKFB4MO on N-cadherin expression (RT-qPCR on dissected explants),
either prior to EMT (NB, st.14), or in premigratory NC (st.17). Error bars
represent s.e.m. *P≤0.05. (F-J) Wild-type (F,G; gfp only, no MO) or morphant
(G,I; MO 10 ng) cells were lineage traced in vivo, on embryos injected in the
prospective neural fold unilaterally. At tailbud stage, control cells (G) efficiently
populated branchial arches and were also found in brain and eye (J). In
contrast, morphant cells failed to populate branchial arches (I) in a dosedependent
manner (J), but were normally distributed in the brain and eye (J).
(K-R) NB tissue was dissected prior to EMT (st.14) and plated onto fibronectin.
Cell behavior was followed by time-lapse videomicroscopy.Wild-type NC (K-N)
adheres efficiently, and undergoes EMT (L), cell scattering and migration (M,
N). Morphant NC (O-R) presented poor adherence, delayed (P) and inefficient
EMT (Q), and few emigrating cells (R). Time after plating is indicated (h:min).
Scale bar: 160 μm. Phenotype scores are indicated in Table S10.
Fig. 7. Glycolysis and PI3K-AKT signaling impact NC migration similarly
to PFKFB4 low-level depletion, and activating AKT signaling rescues
PFKFB4 downregulation. (A-H) When glycolysis (2DG) or PI3K-AKT
pathway (LY294002) were blocked during EMT and migration (st.18-24) both
treatments severely affected NC migration at st.24 (as revealed by twist1 and
sox10 expression). (I-L) At st.45, tadpoles treated during NC migration then
grown in control medium, exhibited general head morphology defects,
including eye defects and smaller branchial cartilages. (I,K) Sibling controls; (J)
2DG; (L) LY294002. A-H show side views; I-L dorsal views. (M-P) At tadpole
st.45, morphant sides were severely affected (M,P), whereas activation of Akt
signaling (N,P) did not affect overall craniofacial morphogenesis. Tadpoles coinjected
with PFKFB4MO and caAkt (O,P) were largely rescued, with 66% of
embryos with injected side symmetrical to contralateral side. (M) Red bar
indicates eye distance from the midline on the morphant side; blue bar
indicates control distance. Both bars are aligned below for comparison. (N,O)
On both sides, the same blue bar measures eye distance from the midline.
Arrowheads indicate the injected side. Scale bar: 500 μm. Phenotype scores
are shown in Tables S11, S12.
Fig. 8. Pfkfb4 activation during neural fold stage is regulated by the NCGRN.
(A-X) Unilateral depletion of members of the NC-GRN modulated pfkfb4
activation at the NB (st.14) and NC (st.18). (A-H) Blockade of WNT and FGF
signaling strongly decreased snail2 and pfkfb4 expression. (I-L). Conversely,
BMP signaling downregulation upregulated snail2 and pfkfb4 expression,
laterally to the NB. (M-T) Knockdown of the NC specifiers PAX3 and TFAP2a
resulted in loss of snail2 and pfkfb4. (U-X) Knockdown of the NC specifier
SOX9 depleted snail2 and pfkfb4. Dorsal views, injected side on the right.
Scale bar: 500 μm. Phenotype scores are shown in Table S13.
Fig. 9. Model of PFKFB4-controlled AKT signaling in NC maturation, EMT and migration. During neurulation, in controls (left side), NC is induced at the
neural plate stage with weak snail2 expression. This induction is then strengthened during neural fold stage, under the action of the NC-GRN, which notably
activates pfkfb4 expression in NC progenitors. This second phase allows acquisition of cellular ability to undergo EMT and migration upon neural tube closure.
When AKT signaling is defective (right side), either after PFKFB4 low-level depletion, or using pharmacological inhibitors, delays in NC specification cascade
occur, incomplete maturation is observed, and NC fails to undergo EMT. When PFKFB4 and AKT functions are prevented after EMT, fewer NCCs migrate
resulting into reduced craniofacial skeletal elements. Hence, each step of NCearly development relies upon continuous and elevated AKT signaling, sustained by
PFKFB4, itself triggered by the NC-GRN.
Fig. S1. Schematics of glycolysis. Glucose phosphorylation by hexokinase is the first irreversible step
of glycolysis. The second and rate-limiting step of glycolysis is catalyzed by PFK1. This step is tightly
controlled in cells, by the levels of Fructose-2.6-bisphosphate, potent allosteric regulator of PFK1.
Fructose-2.6-bisphosphate is synthesized by PFKFB1-4 enzymes, which thus control the rate of
glycolysis in the cells. PFKFB3 and PFKFB4, with a strong kinase activity over their phosphatase
activity, promote Fructose-2.6-bisphosphate synthesis and stimulate glycolysis. In addition to their
classic role in glycolysis regulation, recent unconventional roles of the PFKFB enzymes have been
described in cancer or development (see text).
Development 144: doi:10.1242/dev.157644: Supplementary information
Development • Supplementary information
Fig. S2. Pfkfb4 is specifically upregulated in the neural border and neural crest during
neurulation. (A-J). Transverse sections through the anterior neural plate at mid and late neurula stages
(stage 14 and 18, respectively) allow comparing Snail2 and Pfkfb4 expression in the neural crest (Snail2
and Pfkfb4), neural plate (Pfkfb3) and somites (Pfkfb1). (A,F): The notochord, neural plate and paraxial
mesoderm are outlined on hematoxylin-eosin stained paraffin sections. (B-E,G-J) Vibratome sections
of WISH-stained embryos. The arrow indicates the midline and the notochord. The blue arrowheads
indicate in situ hybridization staining position. Red bar positions the neural border at stage 14. Scale
bar = 200μm. (K) A quantitative analysis of Xenopus laevis Pfkfb1-4 expression levels shows that Pfkfb4
is specifically enriched at the neural border of frog neurulas (red), compared to the anterior neural fold
(yellow). All values are normalized to Odc expression and to the average expression in a sibling whole
embryo lysed at late neurula stage 18 (blue, value = 1). Snail2 expression is used to monitor the quality
of dissected tissues and shows high and specific expression at the neural border.
Fig. S3. Patterning defects in NC are not due to increased cell death, after low-level PFKFB4
depletion in vivo. In this study, we use a low-level depletion of PFKFB4, with reduced MO
concentrations, in order to bypass the earlier developmental arrest observed in previous study, when
PFKFB4 severely depleted (above 60% mRNA depletion) and prevents gastrulating ectoderm
specification (Pegoraro et al., 2015). Importantly, this severe depletion was rescued by adding back a
MO-resistant pfkfb4 mRNA, assessing both MO specificity and lack of non-specific toxicity (Pegoraro
et al., 2015). (A) Here, we check that the morphant cells do not undergo elevated cell death at later
stages of development by analyzing mid-neurulas at stage 14 and the end of neurulation stage 18. (BC)
We find less than 30% of embryos with caspase-positive cells in the lacZ-injected area (B), in
conditions when over 70% of sibling embryos show deficient twist expression (C, either "strong",
"moderate" or "mild" decrease). The number of caspase-positive cells is slightly more elevated than in
uninjected or lacZ-only-injected embryos, or in ventrally-injected morphants, but it remains limited to
few cells and cannot account for the incidence of the patterning defects. Representative exemples of
the majoritary phenotypes are shown for each condition, i.e. "rare" or "+/-" phenotypes. Embryos
analyzed in this experiments: uninjected, n=49; lacZ-injected, n=46; PFFB4misMO-injected, n=16;
Fig. S4. PFKFB4 is required for neural crest migration. (A,D) WISH analysis of neural crest
migration at tadpole stages. When Pfkfb4 MO is injected, embryos display decreased Twist expression
and neural crest migration defects (B), when compared to control embryos (A). (C) In contrast, the coinjection
of Pfkfb4 mRNA with Pfkfb4 MO restores Twist expression and NC migration, showing that the
neural crest migration defects are due to PFKFB4 knowckdown. A-C are side views, anterior is to the
Fig. S5.PFKFB4 depletion and 2DG treatment block glycolysis, LY294002 blocks AKT
phosphorylation. (A) PFKFB4 low-level depletion affects lactate levels in embryos, indicating efficient
glycolysis blockade. (B, C) Lactate concentration and Akt phosphorylation were decreased following 2-
DG or LY294002 treatment (LY, two doses 60-80 μM), respectively. Error bars represent s.e.m (t-test
p<0.05). (C,D) Blocking glycolysis during neurulation, between st. 14 and st. 18 does not affect twist
expression in vivo. (D) control and (E) 2DG-treated embryos displayed normal twist expression (n=15
Fig. S6. Glycolysis and PI3K-Akt signaling control neural crest migration. Glycolysis blockade (AJ)
and PI3K-Akt blockade (K-T) during EMT and neural crest migration (stage 18 to stage 24) affects
neural crest migration as shown by the expression of Sox9, Sox10 and Twist. Embryos rinsed at stage
24 and grown until stage 45 display craniofacial and eye development defects, albeit the presence of
all individual cartilage elements. (A-C,F-H,K-M,P-R, U-W) Side views, anterior is to the left. (D,I,N,S,X)
dorsal views st 45. (E,J,O,T,Y) Ventral views of alcian blue stained visceral cartilages, st 45. Scale bar=
Fig. S7. Pfkfb4 is regulated by the neural crest gene regulatory network. Unilateral overexpression
of members of the neural crest gene regulatory network modulates Pfkfb4 expression at mid and late
neurula stages (stage 14 and 18, respectively). (A-D) Overexpression of WNT signals increases both
Snail2 and Pfkfb4 expression. (E-H) FGF8 overexpression results in complete loss of pfkfb4, while
snai2 is expanded. (I-L) BMP overexpression with a constitutively active form of BMPR1 results in loss
of Snail2 and Pfkfb4 expression in the NC and ectopic expression in the neural plate. (M-T) PAX3 and
TFAP2a positively regulate Pfkfb4 expression, similarly to Snail2. Dorsal views, anterior is to the top.
Scale bar = 500μm.