Despite solid evidence for the importance of ion-based signaling during development, only Ca2+-dependent signals have received extensive attention. Regulated ion flux, or bioelectricity, is a fundamental aspect of physiology, and new reporting dyes promise to reveal the roles of other ions the way Fura and Fluo dyes have done for Ca2+. Accumulating evidence shows that patterning and morphogenesis require regulated bioelectrical signals (McCaig et al.,2005; Zhao et al.,2006) because they affect cell shape, proliferation, and growth (Adams et al.,2007; Blackiston et al.,2009; Levin,2007a,b,2009; Morokuma et al.,2008a,b; Sundelacruz et al.,2008,2009). Evidence for the role of ion flux during craniofacial (CF) development includes: (1) K+ ion flux is required for cochlear development in guinea pigs (Jin et al.,2008); (2) development of the zebrafish otic placode requires two different Na+/K+-ATPase genes (Blasiole et al.,2006); (3) Ca2+-ATPase is required for zebrafish semicircular canal and otolith formation (Cruz et al.,2009); and (4) in fish, feedback between eye horizontal cells and retinal neurons may be mediated by a voltage-dependent extracellular proton concentration regulated by H+-V-ATPase activity (Jouhou et al.,2007).
The H+-V-ATPase (Fig. 1A.i) contributes to resting Vmem (Hinton et al.,2009) and regulates pH inside the cell (e.g., vesicle acidification) and outside (e.g., bone breakdown by osteoclasts; Fig. 1A.ii) (Nishi and Forgac,2002). The H+-V-ATPase proton-transporting subunit c, also known as ductin, is highly conserved among eukaryotes (NCBI Reference Sequence NP_001082675.1, see Fig. 1B), indicating a fundamental role in cell physiology. Ductin binds to β1-integrin, a protein critical for cell locomotion (Lee et al.,2004), and is also required for developmentally regulated functions including anterior neural patterning (Cruciat et al.,2010), tadpole tail regeneration (Adams et al.,2007), and zebrafish eye development (Wang et al.,2008). Ductin is also the binding site for Human Papilloma Virus (HPV)16 subunit E5 (Anderson et al.,1995; Disbrow et al.,2005; Finbow et al.,1991; Goldstein et al.,1991; Saito et al.,1998), and is thus a possible factor in cervical cancer (Stoppler et al.,1996). Ductin activity is, therefore, at the hub of three higher order phenomena—development, regeneration, and tumorigenesis—each of which involves regulation of differentiation and morphogenesis.
Ductin participates in bioelectrical signaling, which refers not just to action potentials, but to steady-state conditions like resting potentials that carry patterning information and are instructive (Levin,2009); ion flux directs cell behavior, it is not just a byproduct of cell physiology (Blackiston et al.,2009). Vmem is known to regulate membrane proteins, for example voltage-gated sodium, potassium, and chloride channels; pH affects protein conformation and enzyme activity. It has recently been shown that pH may also regulate osmosis and thus osmotic pressure in cells (Zhao et al.,2009); therefore, regulated pH differences could provide motive force or establish permissive conditions for cell shape changes (Ingber,2006).
Here, we report our discovery of a never-before-seen regionalization of membrane voltage (Vmem) and pH in the Xenopus ectoderm using the voltage-reporting dye pair CC2-DMPE and DiBAC4(3). To examine possible mechanisms, we used ductin inhibitors, both chemical and molecular, as well as reagents that target Vmem and/or pH by ductin-independent mechanisms. Using these reagents, we present evidence that disruption of H+-flux alters craniofacial morphogenesis, expression of craniofacial-relevant mRNAs, and bioelectrical patterns on the developing ectoderm. Finally, we present a model for how these mechanisms might be related to the biochemical pathways known to regulate CF morphogenesis (Brugmann and Moody,2005; Sauka-Spengler and Bronner-Fraser,2008; McCabe and Bronner-Fraser,2009).
Striking, Dynamic, Regional Vmem Patterns Occur on the Surface of Neurulae
To explore changes in bioelectrical phenomena during normal development of Xenopus embryos, we imaged endogenous Vmem during neurulation using the voltage-reporting dye pair CC2-DMPE and DiBAC4(3) (Blackiston et al.,2011). Still images and time-lapse videos of Vmem in embryos developing from gastrula to tailbud stages revealed remarkable, never-before-seen patterns of relatively hyper- and depolarized subpopulations of visible ectodermal cells (hereafter referred to as compartments or regions; see Supp. Movie S1, which is available online). We distinguished three courses of hyperpolarization. The first was a wave that moved across the entire embryo, apparently coincident with the appearance of cilia at the blastula surface and the beginning of neurulation-related convergence and extension (Fig. 2A). Course II comprised the closure of the neural tube, distinguished by a bright signal coming from the median ectoderm as the folds close over it and a somewhat dimmer signal from the lateral ectoderm; the neural folds stay relatively depolarized (Fig. 2B, green arrows). As neural tube closure ends, distinct spots and lines of hyperpolarization appeared in anterior areas that subsequently invaginated (Fig. 2B). For example, hyperpolarization marked the future stomodeum (Fig. 2B, yellow arrows), olfactory placode (Fig. 2B, lavender arrow), and the first pharyngeal fold (Fig. 2B, brown arrows), as well as the eye field (Fig. 2B, blue arrows), in advance of the invagination of the optic placode (Fig. 2B). Course III was an embryo-wide series of localized hyperpolarizations (Fig. 2C). This course was less orderly than courses I and II, with hyperpolarizations forming and spreading in multiple smaller areas (Fig. 2C, outlined in red). Interestingly, these localized events sometimes overlapped with the regions established during course II without disturbing the previously established patterns (Fig. 2C, eye outlined in blue, stomodeum outlined in yellow). Course III was coincident with the embryonic shape change from spherical to elongated.
Our previous studies revealed a role for the H+-V-ATPase in patterning events during tail regeneration and the generation of left-right asymmetry (Adams et al.,2006,2007). The H+-V-ATPase plays a central role in the regulation of Vmem and pH; therefore, we tested the hypothesis that ductin is required for Course II Vmem patterns and the normal patterning of CF structures. We created a mutant ductin mRNA (xduct-noTM4), which lacks the 4th transmembrane domain (TM4), removing the proton-binding site entirely. We observed that injection of xduct-noTM4 at the 1-, 2-, or 4-cell stage disrupted CF development as assayed by observation at stage 45–48 (Table 1). Importantly, we used dosages that affect CF morphogenesis but were not toxic, as would be expected if we had generally inhibited vesicular acidification; moreover, embryos had normal sizes and behavior, and, with the exception of left-right asymmetry randomization that is caused by ductin loss-of-function (LOF; Adams et al.,2006), tadpoles had no gross malformations such as abnormal dorso-anterior index (Fig. 3).
Table 1. Percent of Injected Embryos With Craniofacial Phenotypes (CFPs) Grouped by Timing of Injectiona
Uninjected controls % CFPs (n)
Injected % CFPs (n)
Toxicity of injected construct (%)
a1/1 = One cell injected of one-cell embryo, 1/4 = one cell of four-cell embryo, etc. Data are also presented to demonstrate the toxicity of each construct; the percentage shown indicates the number of unscorable embryos.
1/1 and 2/1
Affected organs included: jaws and branchial arches (BAs; compare Fig. 3A wildtype morphology to Fig. 3B–G,J,M,Q, red arrowheads, and Supp. Fig. S1); eyes (Fig. 3D,F–P,S, blue arrowheads); otocysts, otoliths, and olfactory pits (Fig. 3Bii,G,L,M,Q,R, orange arrowheads). Unilaterally small heads due to the complete loss or reduction of BAs on one side was the most common phenotype (Fig. 3B,E). Abnormalities of the eyes included: pigment along the optic nerve (Fig. 3H), thickening and pigmentation of the connection between eye and brain (3G,I), ectopic lenses (Fig. 3K), abnormal shapes (Fig. 3D,G,I,K,N–P,S), ectopic pigment (Fig. 3M,P), and connection directly to the brain (Fig. 3I,J,N,P). Reduction of eyes was also seen (Fig. 3J,L). Otocysts and otoliths showed abnormalities including: malformation (Fig. 3Bii), loss (Fig. 3G,L,Q), and in a few cases, ectopic otoliths (Fig. 3R). Finally, the olfactory pits of some injected embryos were misshapen (Fig. 3Dii,I) or fused to eyes and/or brain (Fig. 3J,O,P). Duplication or hypertrophy of the brain was another observed effect (Fig. 3I,M,N,P,Q).
By examining those injected embryos in which the lineage was traced (using the YFP signal from an xduct-YFP fusion construct or the coinjection of β-gal with xduct-noTM4), we found that affected placodal organs (eyes, otic capsule, olfactory pits) were always on the injected side of the embryo. These lineage tracers were also useful to determine that xduct constructs were typically expressed throughout the injected side of the embryo.
Because xduct-noTM4 inhibits the H+-V-ATPase, which has structural as well as ion-transporter function, we decided to explore whether ductin's role in craniofacial development was specific to the effects of proton pumping or possibly related to another function. To that end, we altered Vmem and/or pH using ductin-independent reagents that have been characterized previously (see Table 2). NHE3 is a gain-of-function (GOF) reagent that alkalinizes cytoplasm but does not affect Vmem (Praetorius et al.,2000); injection of wildtype nhe3 caused the same CF phenotypes (CFPs) observed with xduct-noTM4, indicating a role for pH regulation in signaling during morphogenesis of the head and face (Table 1 and Supp. Fig. S3). Pma1.2, a single subunit yeast proton pump that localizes to the plasma membrane where it pumps protons out of the cell, acts as a GOF reagent (Adams et al.,2007; Bowman et al.,1997), hyperpolarizing the plasma membrane and alkalinizing cytoplasm. Expression of pma1.2, like nhe3, caused the same CFPs observed with the ductin LOF construct (Table 1 and Supp. Fig. S3). We also examined the effect of overexpressing wildtype ductin (xduct) on CF development. This construct also produced the same CFPs observed with xduct-noTM4 and the ductin-independent reagents (Table 1 and Supp. Fig. S3). Therefore, we conclude that the role of ductin/H+-V-ATPase in normal CF development is related to H+-flux; moreover, because either increasing or decreasing H+-flux causes similar CFPs, we conclude that regulation of H+-flux is required to maintain pH and Vmem within a particular range.
Proton-binding subunit c of H+-V-ATPase; may enhance H+-V-ATPase
Ductin lacking the 4th transmembrane domain; inhibits H+-V-ATPase
YFP tagged xduct; inhibits H+-V-ATPase
In situ hybridization probes
Five mRNAs were used to vary H+-flux. Eight mRNAs were used for in situ hybridization. Column 2 gives information about the protein product. Column three in the top half indicates the known or predicted effect of the protein. LOF, loss-of-function; GOF, gain-of-function.
Ductin Activity Is Upstream of Seven Key Developmentally Regulated Genes
To explore how ductin's function is transduced into cell behaviors required for CF morphogenesis, we injected 1 blastomere of 2-cell embryos with xduct-noTM4 and performed wholemount in situ hybridization (WISH) using probes for gene transcripts known to be critical for CF development (Bowes et al.,2010). We first examined the effect of altered ductin expression by looking at stage 14, when neurulation is beginning and neural crest is being specified (Nieuwkoop and Faber,1967). We examined the expression of two CNC markers (xfz3 and slug; Deardorff et al.,2001; Linker et al.,2000) and a placode marker (sox9). As early as stage 14, we observed abnormal expression of all three markers (Fig. 4A–C), suggesting that ductin influences, either directly or indirectly, gene expression or RNA stability patterns and neural crest specification.
We also examined four CNC markers (xfz3, slug, mitf, sox9) and four placode markers (sox9, pax8, pax6, otx2) in stage 20+ embryos after neural crest migration has begun. We found distorted ISH patterns of slug, sox9, pax8, pax6, and otx2 (Fig. 4D–H), and extra domains of otx2 and mitf (Fig. 4H,I); in embryos co-injected with lineage tracer, the disrupted patterns were invariably on the injected side (Fig. 4G,H). These same patterns were also observed in embryos injected with wildtype xduct (data not shown). Although otherwise normal axial development and metamorphosis of injected tadpoles ruled out nonspecific effects of the reagents, we also confirmed this using a probe for xnr-1, a patterning gene not involved in CF patterning (Fig. 4J). Even in embryos that were positive on the right side for the normally left-sided xnr-1 (the positive control for xduct-noTM4 activity; Adams, et al.,2006), the size and shape of the xnr-1 ISH pattern were unchanged.
Comparing WISH patterns (Fig. 4) to tadpole phenotypes (Fig. 3) revealed that ISH patterns correlated clearly with observed morphological changes, i.e., compare pax6 expression (Fig. 4G) to the eye abnormality in Figure 3J. This correspondence is consistent with the hypothesis that ductin activity is required for establishing the correct ISH patterns of CF regulators.
We examined slug patterns to determine whether ductin disruption affects the ability of neural crest cells to migrate into the pharyngeal arches. At all stages examined, including stages 14 and 20 mentioned above (Fig. 4B,D) and a third stage, stage 25 after neural crest cells have populated the arches, we find evidence of migration including normal population of the arches in addition to ectopic expression of the marker (Fig. 4K). We also examined the same stages for evidence of differences in apoptosis, using both a TUNEL assay and IHC for caspase, and found no difference in the number of apoptotic cells on the control versus the xduct-noTM4 injected sides (data not shown). We conclude that the effect of ductin inhibition is not a general effect on CNC migration or apoptosis, but rather a specific effect on the WISH patterns of CNC and craniofacial placode markers, as early as stage 14.
Ductin Activity Is Required Just Prior to and During Neural Tube Closure
Alterations in gene expression in embryos expressing xduct-noTM4 were observed as early as st. 14, suggesting that H+-flux influences development at or prior to that stage. To explore the critical time period for ductin activity, we exposed dishes of embryos to the highly specific ductin inhibitor concanamycin beginning at different embryonic stages and lasting for different durations. Once embryos had reached st. 45–46, we counted the number of individual tadpoles that displayed one or more CFPs (Fig. 5). We found that concanamycin caused CFPs if embryos were exposed from stage 13 to stage 16, i.e., during the formation of the neural plate and folds. Exposures ending before stage 13 or starting after stage 16 did not induce CFPs, suggesting that concanamycin/H+-flux inhibition affects induction and/or specification of CNC or placodes, but not differentiation of these cell types.
To provide more conclusive evidence that st. 13–16 are the critical ones for the effects of ductin activity on craniofacial morphogenesis, we examined the expression of sox9, slug, and xfz3 in concanamycin-treated embryos using WISH. First, we exposed embryos to concanamycin from st. 13–14 and examined these markers at st. 14–15, at the end of the treatment period. We observed that the expression of all three genes was disrupted by concanamycin treatment (Fig. 5B.i).
We also exposed embryos to concanamycin from st. 14–16, and then examined the expression of the same genes at st. 20+ after neural crest migration has begun. Again, we observed abnormal expression of sox9, slug, and xfz3 in concanamycin-treated embryos (Fig. 5B.ii), similar to the patterns observed in xduct-noTM4-injected embryos. From these data, we conclude that stages 13–16 are the critical ones for the effects of H+-flux on the expression of CNC and craniofacial placode markers.
During Neurulation the Superficial Ectoderm Stains Very Strongly for Ductin Protein
To determine whether ductin is expressed during the period of concanamycin sensitivity, we examined the localization of ductin protein during early neurula stages. As expected for a subunit of the H+-V-ATPase, ductin was expressed throughout the neurulating embryo; the strongest signal localized to the superficial layer of the ectoderm (Fig. 6). Ductin is also expressed in the deep ectodermal cells that will become neural crest and placodes, but localization in this region is much weaker than the expression in the superficial ectoderm. These results suggest that ductin exerts its influence locally, but the effects of ductin activity on developing neural crest and placodes may be non-autonomous.
Ductin Is Necessary for Normal pH and Vmem Patterning
Our imaging of Vmem domains in untreated embryos showed that there was consistent overlapping between the hyperpolarized or depolarized regions of Course II and domains of head patterning genes such as eya1 and sema4A; thus, we focused further analysis on this pattern. To address whether domains of hyperpolarized Vmem would correspond with alkaline pH domains, as predicted by the function of H+-V-ATPase, we imaged embryos in BCECF dye, a pH reporter. We observed that the distribution of iso-pH compartments produced overlapping patterns to those obtained with the Vmem reporter dyes (compare Fig. 7A to B). A correspondence between high pH and hyperpolarization was predicted by the fact that H+-V-ATPase-dependent proton efflux hyperpolarizes membranes and alkalinizes the cytoplasm. Thus these data are consistent with the hypothesis that these electrophysiological conditions are both ductin-dependent.
To test whether alterations in ductin influence these patterns, we imaged pH and Vmem domains in embryos injected with xduct-noTM4. In the majority of injected embryos we examined (9/13), course II hyperpolarization domains and pH patterns were disrupted. In every embryo with an abnormal pH or Vmem pattern, the disrupted side subsequently developed abnormal CF morphology (Fig. 7C,D). This indicates that ductin is a necessary determinant of bioelectrical patterns that influence the ISH patterns of CF related genes and correlate with the morphology of craniofacial tissues.
We conclude that normal craniofacial patterning includes dynamic ectodermal cell compartments defined by specific, ductin-dependent Vmem and pH. Divergence of these characters affects the expression of well-characterized craniofacial markers and correlates strongly with changes to the shapes of CNC- and placode-derived structures. Because hyperpolarizations during the latter half of course II so conspicuously predicted the position of imminent shape changes, our data provide suggestive evidence that these compartments are relevant to morphogenetic movements.
We have: (1) discovered dramatic patterns of pH and Vmem changes in the ectoderm during Xenopus CF development; (2) shown for the first time that alterations in H+ flux can alter development of BAs, jaws, otoliths, otocysts, eyes, and olfactory pits, all derivatives of CNC and placodes; (3) determined that ductin is required for this process; and (4) identified an effect of ductin disruption on the expression patterns of important head-patterning genes.
Our most exciting observations were that during neurulation, remarkable dynamic ectodermal domains of hyperpolarized cells appear just prior to and throughout important CF morphogenesis events. H+-V-ATPase is implicated in these events because inhibition of ductin changes voltage and pH patterns and causes subsequent, spatiotemporally-correlated CF abnormalities (Fig. 7C,D). Immunohistochemistry shows that ductin is concentrated in all SE cells, not just in those localized to hyperpolarized domains. Further, our injections of xduct-noTM4 did not specifically target ectodermal cells and therefore altered H+-flux is expected throughout the cells on the injected side of the embryo. Yet the effects of xduct-noTM4 are fairly specific to craniofacial structures (Fig. 3), consistent with the higher concentration of ductin in the SE (Fig. 6). As part of the H+-V-ATPase, ductin is required for the acidification of intracellular vesicles such as lysosomes, and, in some cells, such as osteoclasts, for extracellular acidification. Our data show that ductin also has a developmentally regulated role in CF morphogenesis, raising important questions about how ductin, like proteins that control calcium or cAMP signaling, can be required in all cells for essential physiological functions, yet also be required for developmentally regulated, tissue-specific events (Semenov et al.,2007; Ducibella et al.,2006). One hypothesis is that the physiological requirement for ductin varies, and is low during the stages when the developmental requirement for ductin is high. Indeed, developmentally regulated events are much more sensitive to ductin inhibition, as evidenced by the lack of overall toxicity of LOF treatments.
Interestingly, processing of yolk to provide nutrients for differentiating tissues also requires H+-V-ATPase-dependent acidification (Fagotto and Maxfield,1994). Thus, there could be an autonomous effect of H+-V-ATPase. Small decreases in pH cause partial processing of yolk and larger changes cause rapid degradation due to increased activity of proteolytic enzymes (Fagotto,1995). The location of acidified yolk platelets correlates with differentiating ectodermal tissues. Additionally, at tailbud stages, consumption of yolk platelets via acidification occurs in the arch and oral endoderm but not in morphologically static tissues (Jorgensen et al.,2009). Our data are, therefore, consistent with ductin-regulating activity of SE cell surface proteins, and/or increasing the availability of yolk to SE cells. The elaborate pattern of pH regionalization suggests that this mechanism of regulation can be quite intricate.
Other constructs that affected pH and Vmem (pma1.2) or pH alone (nhe3) also caused CFPs. Thus, regulation of pH is required for development of CNC and placode-derived structures. This is consistent with other work investigating effects of pH, particularly alkalinization being required for anterior neural fate (Sater et al.,1994; Uzman et al.,1998). Importantly, that the same CFPs are caused by ductin-independent H+-flux-altering reagents suggests that H+-flux is the important factor regulating craniofacial morphogenesis, and not ductin itself. These results also indicate that the effects of ductin on CF development could not be due to other unidentified ductin-specific factors such as binding partners or nucleotide signaling.
To further explore the relationships among bioelectrical phenomenon and known differentiation and morphogenesis cascades, we compared course II Vmem patterns to published data on the expression patterns of other genes involved in cranial neural crest and placode development. We found that the expression pattern of semaphorin 4A overlaps the stage-19 horseshoe (Fig. 2B, brown arrows) and the dorsolateral domains of hyperpolarization (Fig. 2B, orange arrows) particularly well. Semaphorins are expressed dynamically in areas where morphogenetic cell movements such as neural crest migration occur; Sema4A also stains neurogenic placodes in slightly older embryos (Koestner et al.,2008). We also noted that the neurula-stage expression domains of eya1 and six1, markers of pre-placodal ectoderm, overlap the hyperpolarized regions of the neural tube (Fig. 2B, green arrows) and the anterior semi-circle/pre-placodal area (Fig. 2B, brown arrows; to aid in comparison, the arrows in Fig. 2 are, to a large extent, colored to match the placode maps of Schlosser and Ahrens,2004).
In addition to the overlapping domains of hyperpolarization and CF-relevant genes, we also observed that hyperpolarized regions closely corresponded to the morphological development of specific CF structures. For example, an intense domain of hyperpolarization corresponds to the developing stomodeum, distinguishing these cells from neighboring cells that will not contribute to this structure (Fig. 2B,C). In this way, the bioelectrical patterns we observed demarcate “progenitor fields,” i.e., regions of the embryo comprised of cells whose progeny will produce a specific morphological feature, and can be distinguished from neighboring cells or regions (Davidson,1993). However, the boundary characteristics of these regions may be less restrictive than the classically understood progenitor fields defined by Davidson, as we observed waves of hyperpolarization during course III that crossed, without disturbing, the boundaries of hyperpolarized domains established during course II (Fig. 2C).
Interestingly, most of the course II hyperpolarized areas also undergo epithelial bending, i.e., the closing of the neural tube and the invagination of the olfactory, lens, and otic placodes. While hyperpolarized regions overlap areas where cells are folding and changing shape, we also provide evidence that manipulating H+-flux alters cell shape and size. We observed that cells receiving xduct-noTM4 often appear to be oversized (Fig. 4A.ii), suggesting that H+-V-ATPase inhibition affects cell shape. Differences in pH influence osmotic pressure in cells (Zhao et al.,2009), and are likely to produce conditions where cell shape is affected either in the treated cells or in neighboring cells due to changes in permissive states. Further connections between pH and Vmem via alterations in H+-flux and their effects on cell shape and size should be explored further.
The CFPs we observed, because they are all abnormalities in derivatives of the anterior ectoderm, including brain, CNC, and cranial placodes (eyes, otoliths and olfactory bulbs), suggest that ductin acts during patterning of the anterior neural plate. Timing experiments are consistent with this hypothesis, and indicate that the critical period for ductin activity is from stage 13 to stage 16; this timing corresponds to the bioelectrical activity of courses I and II, but not III, suggesting that course III may not be related to H+-flux. Finally, our ISH data identified CF markers that are directly or indirectly downstream of H+-flux; these markers, xfz3, slug, sox9, mitf, pax6, and otx2, have all been shown to be early acting members of the CNC and placode development pathways. While some of the ISH patterns we found are consistent with an effect of ductin on neural crest migration, many embryos show slug expression in the branchal arches and in ectopic positions, suggesting that migration is not consistently inhibited. We also found no evidence for an effect on apoptosis resulting from ductin disruption. Taken as a whole, our data indicate that ductin inhibition affects expression of these markers early during the development of anterior ectodermal derivatives as a result of changes to H+-flux.
The wnt/fz pathway is known to participate in early stages of neural crest induction (Yanfeng et al.2006) and, in culture, to be sensitive to bafilomycin, another potent ductin inhibitor (George et al.,2007; Hanken and Gross,2005). Recently, it has been reported that frizzled-dependent planar polarity depends on V-ATPase (Hermle et al.2010) as does wnt signaling (Simons et al.,2009; Cruciat et al.,2010), likely because the secretion of WNT depends on vacuolar acidification (Coombs et al.,2010). Moreover, the CF phenotypes produced by overexpression of xfz3 (see fig. 1 of Rasmussen et al.,2001) are phenocopied with remarkable fidelity by ductin inhibition. Our data are thus consistent with a ductin-dependent hyperpolarization/alkalinization being a mechanism to inhibit the wnt/fz signaling pathway outside of its normal domain. The phenotypes caused by ductin disruption are also very similar to abnormalities caused by disruption of xfz3, sox9, otx2, and pax6 (Gammill and Sive,2001; Kumasaka et al.,2005; Purcell et al.,2005; Saint-Germain et al.,2004; Schlosser and Northcut,2000); this relationship extends to other phenotypes we found including melanocyte and somite disruption and altered neural tube closure (data not shown). Thus, ductin also influences other processes controlled by these gene products, consistent with the idea that H+-flux is an important regulator, either direct or indirect, of these proteins. Also interesting in this context is that while we found both increases and decreases in the normal areas of expression patterns for most of the markers (Fig. 4B–F), we never found a reduction in the amount of xfz3 (Fig. 4A). One hypothesis to explain this difference that merits future attention is that xfz3 transcription requires a particular range of H+-V-ATPase-regulated Vmem/pH, and transcription cannot occur outside that range.
A growing body of literature has investigated the mechanisms by which H+-flux (and other bioelectrical events) can regulate gene expression and cell/tissue morphology. In vitro studies using a wide range of differentiated and stem cells show that changes in Vmem alter proliferation in a cell-type-specific manner (Blackiston et al.,2009). Bioelectrical fields can also dictate cell organization, orientation, growth, differentiation, and migration (Levin,2009). Also, intracellular ion concentrations have been shown to mediate DNA-binding of signaling molecules and can control subcellular localization of transcription factors, providing a direct link between bioelectrical phenomena and changes in gene expression (Levin,2007b,2009). Particularly interesting in the context of our findings is evidence that Disheveled-dependent phosphorylation of xfz3 changes its electrophoretic mobility and downregulates its signaling activity (Yanfeng et al.,2006). Thus there is already evidence consistent with the importance of the charge environment in the regulation of xfz3.
Model of Ductin-Dependent Mechanisms in CNC and Placode Development
Figure 8 summarizes our current model of ductin's role in craniofacial development. Ductin is the proton-translocating subunit of the H+-V-ATPase, and in that role is required for the normal, dynamic compartmentalization of electrophysiological state in the SE during early stages of neurulation (Fig. 8A). Different courses of bioelectrical activity correlate in time and space with important morphogenetic events, such as convergence and extension and the emergence of ciliated cells (course I), in-folding of the deep ectoderm into placodes (course II), and the rapid change in embryonic shape from spherical to laterally compressed (course III). We focused our analysis on the consistent pattern found during neural fold stages, when SE of the neural folds is relatively depolarized/acidic, while SE at the midline and lateral ectoderm is relatively alkaline/hyperpolarized (Fig. 7B). In neural fold SE, we believe the H+-V-ATPase is inactive, while it is active in the other cells, leading to two electrophysiological states.
The relatively depolarized/acidic compartment of the neural fold SE affects the expression of xfz3, slug, sox9, mitf, pax6, and otx2 in the deep cells. SE cells in the neural groove and lateral ectoderm are relatively hyperpolarized/alkaline. This signal inhibits the expression of xfz3 and mitf in the deep ectoderm. The wnt/fz pathway is known to be upstream of invagination (Zimmerman et al.,2010) thus H+-flux dependent regulation of fz expression could affect both fz-dependent differentiation and fz-dependent morphogenetic movements. In addition, electrophysiological states could affect mechanical/physiological conditions, such as the actin cytoskeleton, that are prerequisites for morphogenetic movements (Lecuit,2008; Ingber2006). Thus, by linking gene expression to control of cell mechanics, the electrophysiological state of the cell could be the mechanism that coordinates differentiation and morphogenesis in time and space.
We have shown that ductin has an essential role in CF morphogenesis: (1) the location and timing of ductin expression is consistent with a role in regulating the pattern of pH and Vmem in the SE of neurulating embryos; (2) the locations of alkalinized/hyperpolarized cell domains overlap with the locations of CF developmental gene expression and with locations of subsequent tissue shape change; (3) ductin disruption affects the expression patterns of CF-relevant markers; (4) ductin disruption causes CF abnormalities; (5) ductin disruption phenocopies the result of xfz3 overexpression and disruption of xfz3, sox9, otx2, and pax6; (6) loss or gain of H+-flux causes the same suite of phenotypes, consistent with a role for ductin in keeping pH/Vmem within a particular range. We conclude that the pH and Vmem distinguished compartments of the SE of Xenopus neurulae are regulated by the activity of ductin/H+-V-ATPase, and that the pH and Vmem of these cells regulate expression of genes involved in CF development. Because morphogenetic movements occurred in regions that were alkalinized/hyperpolarized relative to the surrounding cells, we suggest that ductin may regulate tissue shape change. Thus, we propose that the electrophysiological state of the cell has the required characteristics of a mechanism for coordinating differentiation and morphogenesis in time and space.
Xenopus embryos were collected according to standard protocols (Sive et al.,2000), maintained in 0.1× Modified Marc's Ringers (MMR) pH 7.8, and staged according to Nieuwkoop and Faber (1967). At stage 45–46, after phenotypes became clearly visible, control and treated tadpoles were anesthetized with 1.5% MS-222 (tricaine) then scored for fourteen CF phenotypes.
Concanamycin A (1.5 mM) (Sigma, St. Louis, MO) stocks (in DMSO) were diluted to 50–150 nM in 0.1× MMR immediately prior to each experiment; at this concentration, DMSO has no apparent effect on development. Dishes of embryos were exposed to concanamycin for different durations and at different times (see Fig. 4). When not in concanamycin, embryos were kept in MMR.
Wildtype XDUCTIN (xduct) and a wildtype ductin with a YFP tag (xduct-YFP) were generated in the laboratory. In the wildtype ductin, protons enter subunit a on the cytoplasmic side and are picked up by a negatively charged glutamic acid residue in TM4 of subunit c as it rotates past; after a complete rotation, subunit c delivers the proton to a different site in subunit a, and it is released to the other side of the membrane. A third ductin construct, a truncated ductin mutant (xduct-no TM4), was also created in the laboratory; elimination of TM4 removes the proton binding site entirely (Nishi and Forgac,2002). Other constructs were: pma1.2, single subunit yeast proton pump (Masuda and Montero-Lomeli,2000); and nhe3, a sodium-hydrogen exchanger (Praetorius et al.,2000; see Table 2).
Constructs used for in situ hybridization probes included mitf, a component of the pigmented retinal epithelium and melanocyte development pathways (Kumasaka et al.,2005); xfz3, part of the neural crest, eye, and inner ear induction pathways (Rasmussen et al.,2001); otx2, required for anterior neural patterning (Gammill and Sive,2001); pax8, required for the development of the otic placode (Saint-Germain et al.,2004); slug, a neural crest marker (Carl et al.,1999); sox9, an early marker of neural crest and the otic placode (Saint-Germain et al.,2004); and pax6, required for olfactory and optic placodes (Purcell et al.,2005; see Table 2).
Xenopus embryos were injected with capped, synthetic mRNAs (Sive et al.,2000) in the animal hemisphere of early cleavage stage embryos. For examination of CFPs, 1-cell embryos, or 1 blastomere of 2- or 4-cell embryos were injected; for ISH experiments, 1 blastomere of 2-cell embryos was injected. Rhodamine-labeled dextran or Alexa647-labeled dextran was used as lineage tracers for all experiments. In some experiments, β-gal was coinjected with xduct-no TM4 as a lineage label.
Embryos were fixed in MEMFA fixative, embedded in paraffin, and sectioned on a Leica M2255 microtome. After sections were deparaffinized, endogenous peroxidases were quenched and the samples were blocked with 5% Goat serum and 1.5% milk powder, then incubated with anti-ductin antibody (Invitrogen, Carlsbad, CA). After washing, samples were incubated with horseradish peroxidase-conjugated secondary antibodies and the signal was amplified using Tyramide Amplification (Invitrogen). Controls included: primary only and secondary only treatments, and antibody pre-adsorbed against the peptide. Controls and experimentals were incubated for the same durations in all reagents.
In Situ Hybridization
One blastomere of 2-cell embryos was injected with xduct-noTM4 and whole mount in situ hybridization (WISH) was performed (Harland,1991). Probes were generated in vitro using DIG labeling mix (Roche, Indianapolis, IN). Controls included no-probe and sense probes. Chromogenic reaction times optimized the signal:background ratio of treated embryos; controls and treated embryos were incubated for the same duration. In situ hybridization (ISH) patterns were considered abnormal if the pattern on the injected side of the embryo differed obviously from the pattern on the uninjected side. Comparisons to untreated controls were also made.
Imaging Vmem and pH
CC2-DMPE (N-(6-chloro-7-hydroxycoumarin-3-carbonyl)-dimyristoylphosphatidyl ethanolamine) (CC2; Invitrogen) and DiBAC4(3) (bis-(1,3-dibutylbarbituric acid) trimethine oxonol) (DiBAC; Biotium) were used as a ratio pair. A 5-mM stock of CC2 was diluted 1:1,000 directly into 0.1× MMR, pH 7.5 (MMR). Just prior to each experiment, DiBAC stock (1.9 mM), was diluted 1:8,000 in MMR. Embryos were incubated in CC2 solution for 45–60 min. After washing with MMR, they were put into DiBAC solution for at least 45 min before imaging in the DiBAC solution. At the final concentrations (≤1:1,000), DMSO has no effect on fluorescence or development of the phenotype. An Olympus BX-61 with a Hamamatsu ORCA AG CCD camera, controlled by IPLabs or MetaMorph software, was used for imaging. CC2 filters were: EX 405/20; BS 425; EM 460/50 (Chroma filter set 31036). DiBAC filters were: EX 470/20; BS 485; EM 517/23 (Chroma filter set 41001). After corrections to remove camera noise and uneven illumination, the software calculated the ratio of intensities Iem460/Iem517. The result is a picture of Vmem; the brighter the pixel, the more polarized the region it represents. Controls included: artificially changing Vmem in a known direction; using alternative dyes; leaving out one or both dyes (Adams,2008; Adams et al.,2007; Adams et al.,2006; Blackiston et al.,2009; Morokuma et al.,2008a). To image pH, fresh BCECF-AM (stock 5 mM in DMSO, stored at −20°C) was dissolved in MMR to a final concentration of 5 μM (BCECF solution). Embryos were soaked in BCECF solution for 45 min, rinsed, then imaged in MMR. Filter sets used were EX=450/20, BS=460, EM=535/30 and EX=500/20, BS=515, EM=535/30. Metamorph was used to calculate the ratio of Iex490/Iex440. The result was an image in which brighter pixels represent areas that are alkaline relative to darker areas. Except for resizing during figure preparation, no changes were made to these images. Thus, pixel intensity reports relative levels of Vmem or pH (Blackiston et al., 2010).
Tadpoles with one or more CF abnormality were deemed affected, regardless of which structure(s) were malformed; numbers of normal and affected experimental embryos were compared to numbers of normal and affected controls using a χ2 test with Pearson correction for increased stringency (α =0.01, minimum meaningful difference between treated and control [mmd] = 0.1). Next, a χ2 test of heterogeneity was run to determine if experiments could be pooled. If justified (α<0.05), numbers from all similar experiments were pooled and compared to controls by χ2 with Pearson correction (α =0.05, mmd = 0.1).
Photoshop™ was used to orient, scale, and improve clarity of images (not including Vmem or pH images). Data were neither added nor subtracted. Original images are available on request.
The authors thank Doug Blackiston for assistance with WISH, Dayong Qiu and Shing-Ming Cheng for help making mRNA constructs, Punita Koustubhan and Amber Currier for animal husbandry, Bill Baga for administrative aid, J-P Saint-Jeannet for sox9 and pax8, M. Montero-Lomeli for pma1.2, M. Musch for nhe3, C. Wright for xnr-1, C. Labonne for mitf, V. Schneider for otx2, P. Klein for xfz3, Michael Levin for constructive suggestions with experiments and the manuscript, and three anonymous reviewers. This work was supported by NIH grants K22-DE016633 (to D.S.A.), The Russell L. Carpenter Summer Internship (to R.D.M.), and 1F32GM087107-01 (to L.N.V.).