Early, H⁺-V-ATPase-dependent proton flux is necessary for consistent left-right patterning of non-mammalian vertebrates

Consistent left-right (LR) asymmetries in the vertebrate body plan are of fundamental significance to both basic biology and medicine(Burn, 1991; Kosaki and Casey, 1998; Levin, 1999; Levin et al., 1996). Although the genetic cascades of secreted signals that occur on only one side of the embryo are being elucidated (Burdine and Schier, 2000; Whitman and Mercola, 2001), upstream mechanisms remain poorly understood(Levin, 2005; Yost, 2001). Recent findings shed light on the events which function upstream of asymmetric transcription(Bunney et al., 2003; Fukumoto et al., 2005b; Kramer et al., 2002; Levin and Mercola, 1998; Levin and Mercola, 1999; Tanaka et al., 2005), and it is now known that ion flux is involved in determining the laterality of several vertebrates and invertebrates(Duboc et al., 2005; Hibino et al., 2006; Levin et al., 2002; McGrath et al., 2003; Raya et al., 2004; Shimeld and Levin, 2006). To gain molecular insight into endogenous ion flows that control asymmetry, we performed a loss-of-function drug screen(Adams and Levin, 2006),implicating the H⁺/K⁺-ATPase(Levin et al., 2002). Here, we characterize another crucial component of early embryonic physiology: the H⁺-V-ATPase.

The H⁺-V-ATPase complex (see Fig. S1A in the supplementary material) or V-ATPase (Nishi,2002), is found in the membranes of vacuoles where it acidifies the intravesicular environment (Inoue et al., 2003; Kawasaki-Nishi et al., 2003; Nishi,2002). In many cell types, the H⁺-V-ATPase is also present in the plasma membrane (Klein et al., 1997; Narbaitz et al.,1995; Nishi,2002), where, by pumping protons out of the cell, it affects the pH of the cytoplasm and immediate extracellular environment(Brown and Breton, 2000; Kawasaki-Nishi et al., 2003; Scarborough, 2000). The H⁺-V-ATPase is also strongly electrogenic(Harvey, 1992; Wieczorek, 1999).

Interestingly, besides the `housekeeping' functions of the H⁺-V-ATPase and other electrogenic proteins, their activity has been implicated in the control of cell proliferation, migration and differentiation (Levin, 2003a; Nuccitelli, 2003). One important example of morphogenesis relying upon precisely orchestrated cell behavior is the generation of consistent left-right asymmetry(Cooke, 2004; Levin, 2005), and a number of biophysical mechanisms based on bioelectricity and fluid movement have been implicated (Hamada et al.,2002; Levin,2003c; Tabin and Vogan,2003). Moreover, the conservation of early LR mechanisms is highly controversial (Burdine and Schier,2000; Essner et al.,2002; Levin,2003c). Although ion flux is crucial in early frog and chick embryos (Levin, 2005), ciliary motion has been implicated in rodents(McGrath and Brueckner, 2003; McGrath et al., 2003) and zebrafish (Amack and Yost,2004; Essner et al.,2005; Kramer-Zucker et al.,2005). To understand how ion fluxes participate in asymmetry, it is necessary to characterize the endogenous behavior of the relevant pumps in embryos and to place their function in the context of known LR patterning mechanisms. Here, we explore the properties of H⁺-V-ATPase function in several vertebrate embryos. Through endogenous localization of the H⁺-V-ATPase and gain- and loss-of-function experiments in chick,frog and zebrafish, we identify the H⁺-V-ATPase as a novel,conserved, obligate component of LR patterning upstream of asymmetric gene expression.

Xenopus embryos were collected according to standard protocols(Sive et al., 2000) in 0.1× Modified Marc's Ringers (MMR) pH 7.8 + 0.1% Gentamicin. Xenopus embryos were staged according to Nieuwkoop and Faber(Nieuwkoop and Faber, 1967). Chick embryos from Charles River Laboratories, maintained at 38°C, were staged according to Hamburger and Hamilton(Hamburger and Hamilton,1992). Zebrafish embryos(Westerfield, 1995) were maintained at 28.5°C in water containing 1 drop per gallon Methyl Blue.

Xenopus embryos at stage 45 were analyzed for position(situs) of three organs: the heart, stomach and gallbladder(Levin and Mercola, 1998). Heterotaxia was defined as reversal in one or more organs(Fig. 1K,L). Only embryos with normal dorsoanterior development (DAI=5) and clear left- or right-sided organs were scored; percent heterotaxia was calculated as number heterotaxic divided by the number of scorable embryos, i.e. embryos normal in all other ways, with DAI=5. The proportion of unscorable embryos (number unscorable over total number of embryos), is reported as a measure of treatment toxicity, but has no influence on the heterotaxia score. The autofluorescence of gallbladder and pancreas was used to score situs in zebrafish embryos. A Pearsonχ ² (increased stringency) was used to compare absolute counts of heterotaxic embryos.

Xenopus embryos were incubated from 45 minutes post-fertilization to stage 6-7 in drugs (Fig. 1)at doses reported in Table 1. All of the reagents used were selected on the basis of high specificity for known electrogenic targets (Drose et al.,1993; Shen et al.,2003; Wheatly and Gao,2004), and were titered to ensure that the DAI of the treated embryos was normal, thus avoiding confounding randomization caused by midline defects (Danos and Yost,1996). Embryos were transferred to 0.1× MMR at stage 6-7 and scored at stage 45. For all data shown, normal midline development and DAI were observed. For cytoskeleton disruption experiments(Fig. 3), embryos were treated with 50 nM nocodazole or with 0.7 μM latrunculin, in 0.1× MMR(Qiu et al., 2005).

To apply reagents to chick embryos with minimal disturbance, a small hole was made in each egg (prior to incubation), and 5 ml of light albumin removed. The experimental solution, consisting of pharmacological reagent(Table 1) in 5:1 light albumin and Pannett-Compton (Stern and Holland,1993) was placed into the egg. Eggs were wrapped with Scotch tape,incubated at 37.5°C, then fixed in 4% paraformaldehyde.

Danio rerio embryos were incubated at 28.5°C in 1 μM lobatomide A16 or 250 nM concanamycin during stages indicated.

For self referencing ion selective probe (SERIS) measurements of proton flux (Smith et al., 1999),healthy Xenopus embryos were cultured in 0.1× MMR at pH 7.0. H⁺ flux measurements were made to the left and right, equidistant from the ventral midline, near the AV midline, slightly on the animal side,and approximately in the middle of the left-ventral quadrant and the right-ventral quadrant of the embryo. The displacement of the electrode was along a line that lay 45° from the normal to the surface of the embryo. This was true for both measuring positions. We calculate a correction factor of 105 for the presence of 0.5 mM HEPES buffer and have applied that factor to generate the absolute flux values reported in the Results section(Arif et al., 1995). The ratio of the fluxes across the ventral midline is unaffected by this correction.(Contact the authors for further details.)

Bis-(1,3-dibarbituric-acid)-trimethine-oxanol [DiBAC₄(3),Molecular Probes] accumulates in proportion to membrane voltage; the more depolarized a cell, the greater the accumulation and fluorescence intensity of DiBAC₄(3). Stock DiBAC₄(3) (1 mg/ml in DMSO) was diluted 1:10 in distilled water, then 1:100 in 0.1× MMR (1.9 μM final). Xenopus embryos were soaked for 30 minutes, then imaged submerged in dye, using a Leica TCS SP2 Spectral Confocal Imaging System. The dye was excited at 488 nm and a 20 nm band of emission wavelengths centered at 515 was collected. On any given day, photomultiplier tube gain was kept constant to allow comparisons among images.

Whole mount in situ hybridization (WISH) on Xenopus was performed using a standard protocol (Harland,1991) using XNr1(Lohr et al., 1998; Lowe et al., 1996a), chick Shh (Levin et al.,1995) and chick Nodal(Levin et al., 1995) probes. WISH analysis was performed on zebrafish embryos as described(Albertson and Yelick, 2005). After staining, embryos were refixed in 4% paraformaldehyde, and dehydrated in 100% methanol overnight to remove background staining. Images were manipulated with Adobe Photoshop to increase clarity; data were neither added nor removed.

Immunocytochemistry was performed as described(Levin, 2004). Briefly, Xenopus embryos were fixed overnight at 4°C in MEMFA; chick embryos were fixed overnight in 4% PFA. Some embryos were embedded for 40μm sectioning on a vibratome. Samples were blocked with 20% goat serum+0.2%BSA and incubated overnight at 4°C with primary antibody:anti-subunit A, 1:500 (Kawa et al.,2000); anti-myosin V, 1: 500 (Chemicon #AB5887), anti-RFX3, 1:500(Bonnafe et al., 2004);anti-subunit F, 1:500 (Peng et al.,1996); anti-subunit c, 1:500 (against peptide DAGVRGTAQ,Invitrogen). After washing, samples were incubated overnight at 4°C with alkaline phosphatase-conjugated secondary antibodies. The chromogenic reaction was timed to optimize signal to noise ratio. Standard no-primary, no-secondary and peptide-adsorbed primary controls were used and resulted in no staining. Patterns reported for localization of V-ATPase subunits represent a consensus of data obtained from at least 15 embryos. Images were manipulated with Adobe Photoshop to increase clarity; data were neither added nor removed.

Xenopus embryos were injected with capped, synthetic mRNAs(Sive et al., 2000) into the animal hemisphere of one-cell embryos 45-90 minutes post-fertilization. NHE3 constructs were injected into the vegetal hemisphere within 60 minutes of fertilization. Zebrafish embryo microinjection was performed as described previously (Payne-Ferreira and Yelick,2003). Results of injections are reported as: % of otherwise normal embryos that were heterotaxic; sample size (n); % of injected embryos that died or were abnormal after gastrulation and not scored; andχ ² and P values comparing treated with controls.

To identify molecular candidates for electrogenic proteins involved in LR patterning, we performed a pharmacological loss-of-function screen(Fig. 1A-J, Table 1)(Adams and Levin, 2006; Fukumoto et al., 2005a; Fukumoto et al., 2005b; Levin et al., 2002; Mitchison, 1994). A panel of inhibitors targeting a wide variety of transporters(Fig. 1B-J), like many others reported previously (Levin et al.,2002), induced a very low incidence of laterality phenotypes,ruling out the corresponding targets from important roles in asymmetry and demonstrating the general stability of LR patterning. By contrast, inhibitors of the H⁺-V-ATPase all induced strong heterotaxia (compare Fig. 1K with 1L) in the absence of changes in dorsoanterior character or general toxicity(Danos and Yost, 1995; Danos and Yost, 1996). For example, the potent and highly-specific V-ATPase blockers concanamycin(Huss et al., 2002) and bafilomycin (Bowman et al.,1988) induced heterotaxia in 35% (n=100) and 36%(n=100) of the embryos respectively. Importantly, inhibition of the other two major classes of H⁺ transporters, carbonic anhydrase(Fig. 1B) and the sodium-hydrogen exchanger (NHE, Fig. 1C), did not affect laterality. Specific and dose-dependent(Fig. 1M) randomization of LR asymmetry, induced by loss-of-function reagents targeting the H⁺-V-ATPase, but not other H⁺ pumps or a wide range of other ion transporters, implicate endogenous H⁺-V-ATPase in the LR pathway in Xenopus.

To test unequivocally the requirement for H⁺-V-ATPase activity using specific molecular reagents, we analyzed asymmetry in embryos injected with YCHE78 mRNA, which encodes a well-characterized dominant negative H⁺-V-ATPase subunit E (Lu et al., 2002). Misexpression of this dominant-negative construct specifically induced 20% heterotaxia [(n=191), controls=4%(n=422), χ²=40.6, P≪0.001]. By contrast,injection of mRNAs encoding the Xenopus H,K-ATPase α subunit(Mathews et al., 1995), or the dominant-negative Kir2.2 subunit (Zobel et al., 2003) did not cause heterotaxia (1% heterotaxia each, n=85,93 respectively). These data are consistent with the screen results, demonstrate that embryonic asymmetry is not generally labile with respect to microinjection per se (even when performed during cytoplasmic rotation), and strongly support the hypothesis that H⁺-V-ATPase function is necessary for LR patterning.

To determine the relationship between H⁺-V-ATPase and the asymmetric gene cascade, we analyzed the expression of the earliest known asymmetric gene in Xenopus: Nodal(Hyatt et al., 1996). In embryos treated with the potent and specific heterotaxia-causing H⁺-V-ATPase inhibitor concanamycin(Table 1), the situs of the left-sided marker XNr1 (Fig. 2A) was effectively randomized(Fig. 2B-D), exhibiting aberrant sidedness of expression in 54% of the embryos(Table 2). We conclude that H⁺-V-ATPase functions upstream of the asymmetric transcription of Nodal and feeds into the known asymmetric gene cascade.

Because prior experiments and physiological data (below) indicated that the first cell cleavages are key to establishing ion flux-related asymmetries(Levin, 2003b; Levin et al., 2002), we focused our immunohistochemical analysis on the localization of H⁺-V-ATPase subunits during the first few hours of development in Xenopus. In early embryos, most proteins are symmetrically localized with respect to the LR axis (Fig. 3A,B). Immunohistochemistry revealed that at the two-cell stage,H⁺-V-ATPase subunits (Fig. 3D) exhibited `fingers' extending up from a pool in the vegetal cytoplasm into the animal half. Such patterns are specific to a number of LR-relevant proteins but are not a general feature of all protein localization in the early frog embryo (Qiu et al.,2005) (Fig. 3A,B). In sections perpendicular to the AV axis at the two-cell stage, one cell was often more heavily stained (>80% of the embryos examined, Fig. 3E) (see Fig. S1I-K in the supplementary material), exhibiting both a cytoplasmic pool and localization at the cell cortex. At the four-cell stage, staining was still asymmetrical,and was on the right side in embryos that could be oriented by the cleavage furrows and pigmentation (Fig. 3F, see Fig. S1 in the supplementary material). Consistent with the proposed role of the H⁺-V-ATPase in LR asymmetry, these electrogenic protein subunits are novel markers of LR asymmetry as early as the first two cleavages in Xenopus.

To probe upstream mechanisms (Qiu et al., 2005; Yost,1991), we asked whether the asymmetric localization of H⁺-V-ATPase required organized microtubules/microfilaments. Treatment of one-cell embryos with the actin depolymerizer latrunculin(Ayscough, 1998) caused apparently random localization of subunit A(Fig. 3G). By contrast,treatment with the microtubule disrupting agent nocodazole(De Brabander et al., 1986)had no obvious effect on the LR distribution of subunit A(Fig. 3H). Embryos treated with low levels of these specific cytoskeletal disruptors develop normally to later stages, but, consistent with these data, displayed randomized asymmetry(Qiu et al., 2005).

In light of the loss-of-function and localization data, we asked whether the activity of H⁺-V-ATPase (H⁺ flux) might provide a consistently asymmetric physiological signal to the L and R sides of the body. Using a SERIS probe (Hotary et al.,1992; Smith et al.,1999; Smith and Trimarchi,2001), we detected a large net efflux of protons from the cleavage furrow of the two-cell stage. This efflux averaged 12.7±22 pmole cm^-2s^-1 (n=5) at about the midpoint of cleavage. Importantly, we also found evidence for asymmetry of H⁺-flux. As early as the four-cell stage, a distinct difference in proton efflux was detected across the ventral midline, with larger efflux occurring on the right side, consistent with the immunological localization of H⁺-V-ATPase subunits on one side(Fig. 3F, see Fig. S1F,G,K in the supplementary material). In measurements on 15 embryos made between the four-cell stage and stage 6, the average proton efflux from the middle of the right ventral quadrant was 4.1±0.48 pmole cm^-2s^-1and from the left ventral quadrant was 1.9±0.29 pmole cm^-2s^-1. The average ratio of the efflux from the right side to efflux from the left side was 2.3±0.3.

To confirm that the asymmetric efflux was due to asymmetric H⁺-V-ATPase function, we compared H⁺ efflux in control embryos with embryos treated with ion flux inhibitors. Neither the absolute net proton fluxes nor their right/left ratios were affected by the application of omeprazole (0.27 μM), an inhibitor of H⁺/K⁺-ATPase. However, the H⁺-V-ATPase inhibitor concanamycin (0.28 μM) reduced the proton efflux on the right side to about half of its original value and eliminated the LR asymmetry in the proton fluxes [ratio went from 2.1±0.2 to 1.1±0.05(n=3), t=4.85, P<0.05]. Taken together, these data reveal the existence of consistent H⁺ flux asymmetry in the four-cell embryo, and confirm that, as suggested by the localization data, the asymmetry in the flow of H⁺ ions out of the early blastomeres is due specifically to differential H⁺-V-ATPase activity.

Because in many systems the H⁺-V-ATPase contributes significantly to the membrane voltage gradient V_mem(Gluck, 1992), we also used the reporting dye DiBAC₄(3) to determine whether there is a consistent asymmetry to V_mem(Epps et al., 1994). Embryos marked with Alexa 647 dextran in the right ventral cell (RV) at the four-cell stage were monitored using DiBAC₄(3) at 4-32 cells(Fig. 4). Although no pattern was obvious in 4-, 8- or 32-cell embryos, at 16 cells, we observed hyperpolarization of the RV quadrant, relative to the LV quadrant(Fig. 4A, yellow shading). An example of an embryo in which this difference is easily observed is shown in Fig. 4B-D. This pattern and its embryological timing are consistent with the SERIS data, and with the hypothesis that higher net H⁺ efflux on the right side at the four-cell stage results in a difference in membrane potential across the ventral midline by the 16-cell stage.

That this V_mem gradient is detectable in 16-cell embryos implies that H⁺-flux-affecting reagents must be active by this stage in order to affect LR patterning. To test this prediction, we injected YCHE78 mRNA, the dominant-negative H⁺-V-ATPase subunit E, into the LV or RV cell at the four-cell stage, which results in no detectable expression until the ∼64-cell stage. This had no effect on organ situs [controls=0%(n=38, unscorable=4%), LV=0% (n=62, unscorable=10%), RV=1%(n=67, unscorable=12%)], χ²=0.0, P=0.9),although co-injected GFP was expressed normally (DNS). Similarly, we injected YCHE78 DNA at the one-cell stage, which leads to protein expression only after zygotic transcription begins at mid-blastula transition. Co-injectedβ-gal DNA was expressed (and caused higher toxicity than mRNA injections), but injections had no effect on organ situs [β-gal DNA alone=0% (n=38, unscorable=65%), YCHE78 DNA=3% (n=29,unscorable=70%), χ²=2.0, P=0.155]. These data are consistent with a dependence of LR asymmetry on very early, but not later,activity of H⁺-V-ATPase.

We next examined the effects of an excess, symmetrical H⁺-flux across the plasma membrane on both sides of the midline. We induced ectopic H⁺ flux by expressing a well-characterized single-subunit plasma-membrane H⁺-pump, PMA1.2(Masuda and Montero-Lomeli,2000). Use of this construct also allowed us to address whether it is the balance of H⁺ flux at the cell membrane that is important for LR asymmetry, as the H⁺-V-ATPase is also known to occur in vacuoles, while PMA1.2 functions only in the cell membrane. Ubiquitous early misexpression of this H⁺ pump at cell surfaces caused significant heterotaxia [PMA1.2=21% heterotaxic (n=135, unscorable=22%),untreated=2% (n=187, unscorable=3%), χ²=28.3, P≪0.001]. These data are consistent with the SERIS probe measurements and the importance of differential H⁺ flux, and suggest that it is the flow of hydrogen ions across the cell, not vacuole,membrane that is important for LR asymmetry.

To gain insight into the mechanisms by which the H⁺-V-ATPase controls downstream events, we separately tested its two physiological roles -regulation of pH and membrane voltage - by experimentally altering these two parameters independently of direct manipulations of the H⁺-V-ATPase. We first examined the effect of changing the pH of the external environment of the embryo by raising or lowering the pH of the medium during cleavage stages (Table 3). Although pH 7-11 had no effect on asymmetry, pH 5-6 caused a low level of LR patterning defects (6%, P=0.003), and pH 4 caused a significant level of heterotaxia (19%, P≪0.001). This is consistent with inhibition of the activity of H⁺-extruding pumps by high external proton concentrations. To alter embryonic pH without changing V_mem, we used the electroneutral OH^-/Cl^-exchanger tributyltin chloride (TBT, Table 1) to raise internal pH(Matsuya et al., 2000; Simchowitz et al., 1991). Treatment with 10 μM TBT from stages 1 to 13 caused heterotaxia in 17% of embryos [(n=151); untreated 1% (n=361);χ ²=44.7, P≪0.001].

To study the effect of isolated changes in pH using greater molecular specificity, we overexpressed the electroneutral, plasma-membrane Na⁺/H⁺-exchanger NHE3(Praetorius et al., 2000; Sabirov et al., 1999). Injection of this antiporter mRNA within 1 hour of fertilization caused heterotaxia [NHE3=16% (n=77, unscorable=32%), untreated=3%(n=114, unscorable=5%), χ²=8.9, P=0.003]. This confirms the importance of pH for LR asymmetry, and, similar to the PMA1.2 data, indicates that the relevant H⁺ flux occurs at the cell membrane, not in vesicles. Interestingly, NHE3 was active in causing heterotaxia only when injected in the vegetal hemisphere and when injected within 60 minutes of fertilization; later injections had no effect. Assuming that NHE3 mRNA is translated ∼1.5 hours after injection (based on detection of β-gal/GFP translation), the temporal window during which NHE3 can cause heterotaxia ends by 2.0 to 2.5 hours post-fertilization at 18°C. Consistent with expression data, the asymmetric membrane voltages at 16-cell stage, and the lack of effect on asymmetry of late YCHE78 injection or DNA injection, we conclude that the LR-relevant gradients in cell membrane H⁺ flux occur during the first cleavages, and that normal asymmetric pH at the cell membrane is crucial for LR patterning.

The H⁺-V-ATPase is electrogenic(Gluck, 1992; Harvey and Wieczorek, 1997)and can therefore significantly contribute to the steady-state V_mem. We confirmed this in Xenopus blastomeres by imaging V_mem in vivo in control and concanamycin-treated embryos. As predicted, inhibition of H⁺-V-ATPase depolarizes V_mem(Fig. 4E,F). We conclude that H⁺-V-ATPase activity is necessary for the normal pattern of V_mem in Xenopus.

To address the role of V_mem in the absence of pH changes, we altered V_mem directly by incubating embryos in the Na⁺/K⁺ inhibitor palytoxin (PTX), which converts the Na⁺/K⁺-ATPase into a non-specific ion channel, thus dissipating the voltage gradient and depolarizing cells(Hilgemann, 2003; Tosteson et al., 1997; Tosteson et al., 2003). PTX treatment (2 nM) of embryos from stage 1-6 caused heterotaxia in the absence of DAI change or other defects in 20% of surviving treated embryos[(n=60, unscorable=84%), untreated=2% (n=58, unscorable=3%),χ ²=8.3, P=0.004]. We conclude that normal V_mem, or patterned differences in V_mem, are necessary for normal asymmetry.

To explore the evolutionary conservation of an H⁺-V-ATPase-dependent LR-patterning mechanism, we characterized the expression pattern of H⁺-V-ATPase subunits in chick and zebrafish. In situ hybridization analysis of subunits A, c, and F(Fig. 5A-D), and immunocytochemistry for subunits F and c(Fig. 5E-H) indicated that the H⁺-V-ATPase is expressed in the primitive streak of chick embryos at stages 2-4, and in the node at stage 4 - the time-period in which chick LR sidedness is determined by depolarization in the streak(Levin et al., 2002). Zebrafish embryos (Fig. 7A-F)showed immunohistochemical staining for subunit c in two- to eight-cell stage embryos. By the 32-cell stage, staining appeared stronger in the marginal cells. Early in epiboly, staining was even in all cells, and subunit c was detected surrounding some yolk syncytial nuclei. Staining was apparent in all cells at epiboly.

First, to confirm that H⁺-V-ATPase is upstream of LR patterning events in chick, we examined the effect of H⁺-V-ATPase inhibition on the expression of the early left-sided marker, and correlate of heart looping, sonic hedgehog (Levin et al.,1995). Inhibition of H⁺-V-ATPase by concanamycin(Woo et al., 1996) or DCCD(Finbow et al., 1993)specifically randomized the localization of Shh and Nodal(Fig. 6A-D) Concanamycin induced aberrant sidedness of Shh expression in 67%, and Nodal in 24%, of embryos (Table 4). Thus, as in Xenopus, H⁺-V-ATPase is upstream of the known asymmetric gene cascade, and is required for normal Nodal expression and therefore normal LR patterning in chick.

Next, we assayed whether the pattern of cellular pH in the early chick blastoderm requires H⁺-V-ATPase activity. Using the pH reporting dye SNARF-1-AM (Buckler and Vaughan-Jones,1990; Morley et al.,1996), we found that in control embryos, area pellucida cells maintain higher intracellular pH than primitive streak cells(Fig. 6G,H). And, as predicted by the expression of the H⁺-V-ATPase in the primitive streak, the H⁺-V-ATPase inhibitor concanamycin lowers the pH of streak cells.

The concanamycin-induced decrease in pH in the primitive streak is consistent with H⁺-V-ATPase being the main regulator of pH in these cells. More difficult to interpret is the change in pH of the AP. Neither in situ hybridization nor immunohistochemistry reveals H⁺-V-ATPase subunit expression in the AP; thus, it was surprising that the highly specific H⁺-V-ATPase inhibitor concanamycin affected the AP. One interpretation is that while our imaging studies have only revealed the location of plasma-membrane associated H⁺-V-ATPase, concanamycin treatment also affects the intracellular vesicle-associated H⁺-V-ATPase, and it is that which affects pH in AP cells.

To further examine evolutionary conservation of the V-ATPase, we disrupted its function in Danio rerio embryos. As in Xenopus,microinjection into fertilized zebrafish eggs of mRNAs encoding a dominant-negative subunit E mRNA caused a high level of heterotaxic organ situs [YCHE78=36% (n=42, unscorable=0%), untreated=5% (n=96,unscorable=4%), χ²=19.5, P≪0.001; Fig. 7G-I]. Injections of mRNA for NHE3, to alter cytoplasmic pH only, caused 34% heterotaxia,[(n=99, unscorable=6%), untreated=5% (n=94, unscorable=1%),χ ²=23.4, P≪0.001]. Expression of PMA1.2, to exogenously increase cell membrane H⁺ flux, caused 29% heterotaxia,[(n=52, unscorable=9%), untreated=5% (n=96, unscorable=4%),χ ²=14.2, P≪0.001]. Taken together, these data support the hypothesis that specific levels and/or distributions of H⁺-V-ATPase-dependant H⁺ transport activity at the plasma membrane are necessary for correct LR asymmetry in zebrafish embryos.

Although molecular loss- and gain-of-function experiments demonstrate the importance of H⁺-V-ATPase-dependent H⁺ flux, they do not reveal the endogenous timing of H⁺-V-ATPase function in fish. This is important as no mechanisms upstream of Kupffer's vesicle (KV)(Amack and Yost, 2004) have yet been discovered in zebrafish asymmetry. To determine whether the H⁺-V-ATPase is involved early, as in Xenopus, or later, at stages when cilia might be involved(Essner et al., 2005), we used the specific H⁺-V-ATPase inhibitors lobatamide A16(Shen et al., 2002) and concanamycin (Huss et al.,2002). As in Xenopus and chicks, initiation of H⁺-V-ATPase loss-of-function between fertilization and the two-cell stage causes significant heterotaxia [A16=30% (n=119), untreated=2%(n=138), χ²=37.0, P≪0.001;concanamycin=30% (n=87), χ²=34.1, P≪0.001,phenotypes same as shown in Fig. 7G-I]. By contrast, exposure to lobatamide A16 beginning at three somites (just prior to the formation of KV) did not result in a significant level of heterotaxia [A16=5% (n=56), χ²=0.0, P=0.85]. Thus, in zebrafish embryos, H⁺-V-ATPase-dependent H⁺ flux is important prior to the three-somite stage, i.e. prior to KV formation, and is the earliest known step in LR patterning in zebrafish.

Consistently, treatment with the H⁺-V-ATPase blocker lobatamide A16 resulted in 55% right/bilateral expression of the asymmetric (left-sided)marker Southpaw (Long et al.,2003) [Fig. 7J-L(n=38), untreated=7% (n=27), χ²=13.8, P≪0.001). Similar results were obtained using concanamycin[treated=58% (n=26), untreated=8% (n=24),χ ²=11.4, P=0.001; Table 5), demonstrating that H⁺-V-ATPase activity is upstream of the earliest known asymmetric marker in zebrafish.

The movement of cilia in KV of fish is important for normal asymmetry(Essner et al., 2005; Kramer-Zucker et al., 2005)and we analyzed the interaction of H⁺-V-ATPase function with the cilia pathway. Although H⁺-V-ATPase was not detected on the surface of KV cilia (Fig. 8A), we found that treatment of zebrafish embryos with the H⁺-V-ATPase inhibitors lobatamide A16 or concanamycin caused the KV cilia to appear short and/or greatly reduced in number [Fig. 8B,C; untreated=15% (n=41); A16=41% (n=39),χ ²=5.7, P=0.017; concanamycin=52% (n=27),χ ²=9.1, P=0.002]. We conclude that H⁺-V-ATPase activity is upstream of the LR signaling events that are dependent on normal KV cilia.

Finally, we linked the H⁺-V-ATPase to another known early LR patterning mechanism: serotonin, which plays an important early role in frog and chick embryos (Fukumoto et al.,2005a; Fukumoto et al.,2005b). In Xenopus, serotonin is initially present in all blastomeres but is localized to one blastomere by the 16/32 cell stages(Fig. 8D,F). By contrast,embryos exposed to the specific H⁺-V-ATPase blocker concanamycin exhibited an almost complete absence of serotonin content(Fig. 8E,G, n=17). We conclude that H⁺-V-ATPase activity is required for the maintenance of serotonin within early blastomeres and its localization to one specific cell, placing the activity of this pump upstream of serotonin signaling in the LR pathway.

The strong and specific randomization(Fig. 1A) of the heart and viscera by H⁺-V-ATPase blockers (which function by different mechanisms) but not by inhibition of other ion channels/pumps, clearly implicated the H⁺-V-ATPase as a promising novel candidate required for LR asymmetry. Although the H⁺-V-ATPase has housekeeping roles(Nishi, 2002), we were able to use inhibitors at doses that affected LR asymmetry strongly while permitting otherwise normal development (Fig. 1L). Use of specific blockers likewise implicated the H⁺-V-ATPase in the control of laterality in the chick and zebrafish(Figs 6, 7), revealing this versatile pump as the first endogenous asymmetry mechanism conserved in three vertebrates.

Crucially, to unequivocally confirm the requirement for the H⁺-V-ATPase, we augmented drug experiments with specific constructs targeting the H⁺-V-ATPase. Misexpression of the dominant-negative E subunit (Lu et al., 2002) in Xenopus and in zebrafish embryos causes randomization, confirming the requirement for H⁺-V-ATPase for correct asymmetry. Though changing V_mem and pH by overexpression can randomize, only the V-ATPase is implicated endogenously in setting LR-relevant values of these physiological parameters, as specific inhibitors of the Na⁺/K⁺-ATPase(Levin et al., 2002) and other H⁺ pumps (Fig. 1) do not cause heterotaxia. Because genetic loss of H⁺-V-ATPase activity in mice causes early embryonic lethality(Sun-Wada et al., 2000), the connection between this mechanism and rodent nodal cilia is unknown. Intriguingly, however, inhibition of carbonic anhydrase causes a LR-relevant phenotype in mice (Brown et al.,1989) but not in Xenopus(Fig. 1B). Perhaps control of pH is the common denominator, while the mechanism of that control has diverged among phyla.

Loss-of-function of the H⁺-V-ATPase in chick randomizes the early left-sided marker Shh (Fig. 6B). This implies that, as in Xenopus, in chicks the H⁺-V-ATPase functions upstream of early asymmetric gene expression. Future efforts to model physiological events that are integrated at the node must take into account not only electroneutral H⁺/K⁺exchange (Levin et al., 2002)and Ca²⁺ flux (Raya and Belmonte, 2004; Raya et al.,2004), but also pH and/or V_mem regulation by H⁺-V-ATPase. In zebrafish embryos, randomization of visceral situs and of Spaw localization, and disruption of KV cilia, was caused by H⁺-V-ATPase inhibition prior to somite formation, suggesting that H⁺-V-ATPase is the earliest-known component of the LR pathway in fish embryos, and acts prior to, and upstream of, events in KV(Amack and Yost, 2004; Essner et al., 2005; Kramer-Zucker et al., 2005). This implies that ciliary action, at least in zebrafish, is an important, but not initial, step in LR patterning (Okada et al., 2005). Interestingly, although inhibition of H⁺-V-ATPase affects KV cilia, inhibition of the LR-crucial H⁺/K⁺-ATPase does not(Kawakami et al., 2005),suggesting that the zebrafish relies on ion fluxes at more than one LR signaling step. Further study of the interaction between early ion flux and ciliary events in fish will be interesting, as this is the first species in which both membrane voltage/pH and ciliary action are involved.

As with the H⁺/K⁺-ATPase(Levin et al., 2002), the earliest asymmetries of H⁺-V-ATPase were observed in Xenopus. Maternal H⁺-V-ATPase proteins are present, and asymmetries in their localization, and in directly detectable H⁺-V-ATPase-dependent H⁺ flow, exist by the 2nd cleavage in some percent of embryos. Alteration of H⁺-V-ATPase function randomized Nodal (Fig. 2), the earliest available Xenopus asymmetric marker(Lowe et al., 1996b), placing it upstream of the conserved cascade in frog. Global upregulation of H⁺ flux using the exogenous NHE3 pump did not affect laterality when injection of the construct took place later than 1.5 hours post-fertilization. Likewise, later inhibition of H⁺-V-ATPase,caused by injections of YCHE78 mRNA at the four-cell stage or injections of YCHE78 DNA, also had no effect on organ situs. DiBAC fluorescence, used to reveal differences in membrane potential show biased asymmetry at the 16-cell stage; it is important to note, however, that because of the magnitude of, and the variations in cell volume, DiBAC is probably not accurate at earlier stages. Thus, a consistent asymmetry in V_mem may be present earlier. Localization of H⁺-V-ATPase subunits to the right ventral cell of four-cell embryos, vibrating probe data showing enhanced H⁺-efflux from the right ventral cells of four- and eight-cell embryos, and DiBAC fluorescence showing the right ventral quadrant to be hyperpolarized at the 16-cell stage, all support a role for H⁺-V-ATPase activity during the first few hours of frog development.

Consistent with the loss-of-function results, expression analysis revealed that H⁺-V-ATPase subunits are present in chick from streak initiation (Fig. 5A) and in zebrafish (Fig. 7A) from the first cleavage. We detected no consistent asymmetries of expression in either species, similar to the situation for H⁺/K⁺-ATPase(Levin et al., 2002). It is likely that protein-level regulators remain to be discovered, the asymmetries of which confer functional asymmetries on the activity of pumps that are present symmetrically.

Asymmetrical localization of maternal components of the H⁺-V-ATPase was observed after the first cleavage in Xenopus (Fig. 3). Although the plane of first cleavage can be experimentally repositioned(Black and Vincent, 1988; Danilchik and Black, 1988), in normal embryos the cleavage furrow usually corresponds to the future midline of the embryo (Klein, 1987; Masho, 1990), and injection of one cell at the two-cell stage is routinely used to target half the embryo,allowing the contralateral half to serve as an internal control(Harvey and Melton, 1988; Vize et al., 1991; Warner et al., 1984). Thus,consistent with the timing data, this early asymmetry of localization is likely to be key to the involvement of the H⁺-V-ATPase in LR patterning of the whole embryo.

In light of early work from the Yost laboratory suggesting that organization of the cytoskeleton prior to 1st cleavage is required for LR asymmetry in Xenopus (Yost,1991), the accepted role of the cytoskeleton and motor proteins in the localization of subcellular components, and data implicating dyneins and kinesins (Nonaka et al., 1998; Supp et al., 1997), we examined the interaction of the cytoskeleton with the H⁺-V-ATPase(Fig. 3F-H). Consistent with earlier observations linking embryo-wide asymmetry with subcellular organization (Bunney et al.,2003), and recent work showing that H⁺-V-ATPase subunits interact directly with the actin cytoskeleton(Chen et al., 2004; Lee et al., 1999; Vitavska et al., 2003), we found that the asymmetric targeting of the H⁺-V-ATPase is dependent on actin, but not microtubule, organization. We propose that some as yet uncharacterized aspect of subcellular targeting(Al-Awqati, 1996; Brown, et al., 1992) [perhaps oriented by a nucleation center which fits the role of Wolpert and Brown's`F-molecule' (Brown and Wolpert,1990)] is upstream of the H⁺-V-ATPase in the LR pathway. The ability of very early latrunculin exposure to randomize LR asymmetry in embryos (Qiu et al.,2005) is consistent with the observed effects on localization,linking cytoskeletal organization with large-scale asymmetry. We are currently investigating cytoskeletal processes that assure asymmetric localization of ion pumps.

As the H⁺-V-ATPase normally energizes both vesicle and plasma membranes, it is important to consider which of these roles controls asymmetry. The SERIS measurements show that significant asymmetric flux can be detected at the cell surface. Moreover, equalizing asymmetries of H⁺ flux at the cell membrane, by global overexpression of gain-of-function constructs (NHE3 and PMA1.2) that are known to induce H⁺ flux specifically from cell surfaces, alters asymmetry in both frog and zebrafish. Thus, efflux of H⁺ ions from the cell surface is one crucial function of the H⁺-V-ATPase complex. However, it cannot conclusively be ruled out that subcellular-compartment pH may be important, and vesicular transport at the node could be such a locus(Tanaka et al., 2005).

What aspect of H⁺-V-ATPase activity is required for asymmetry?We distinguished between the two H⁺-V-ATPase functions, pH and V_mem regulation. Manipulating pH directly in different ways, using changes in medium pH, TBT and ectopic overexpression of the H⁺ pump NHE3 in the blastomere plasma membrane, all disrupted asymmetry, consistent with our conclusion that control of pH by H⁺-V-ATPase is important. In particular, as the NHE3 pump is electroneutral, its ability to randomize suggests that the normal pattern of pH in blastomeres is required for normal LR asymmetry. Because H⁺-V-ATPase is also electrogenic, we explored the effect on LR asymmetry of changing V_mem independently of changes to the H⁺-V-ATPase, using palytoxin, which directly abolishes V_mem. Like manipulation of H⁺-V-ATPase activity and of pH, loss of V_mem causes significant levels of heterotaxia. Thus, the normal pattern of V_mem is required for normal LR patterning. Consistent with this finding, it has already been shown that the normal asymmetric pattern of V_mem in chicks is required for normal LR asymmetry (Levin et al.,2002). Both pH and membrane voltage are important for downstream steps, and we are currently working to identify the mechanisms by which cellular ion fluxes are transduced to stable gene expression programs.

Disruption of consistent asymmetry by global manipulation of pH and/or V_mem suggests that the spatial distribution of the levels of these physiological parameters is important for asymmetry. How are the biophysical asymmetries in pH and/or V_mem coupled to downstream biochemical and genetic asymmetries? Active regulation of pH has been described during different stages of development (Gillespie and Greenwell, 1988; Grandin and Charbonneau, 1989). pH acts synergistically with many other important biochemical pathways (Busa and Nuccitelli, 1984), including Ca²⁺-calmodulin binding and regulation of cAMP levels, which are important in LR patterning(Kawakami and Nakanishi,2001). For example, pH gates gap junctions(Greenfield et al., 1990; Turin and Warner, 1980),which are essential for normal LR axis development(Levin and Mercola, 1999). pH regulation also has a role in gene expression and differentiation(Sater et al., 1994; Uzman et al., 1998) and in other important developmental processes such as proliferation, apoptosis and growth factor release (Bidani et al.,1998; Lee and Steinhardt,1981; Putney and Barber,2003; Shrode et al.,1997). pH also has other physiological effects that are likely to be important during development, although a direct connection has not been shown. For example, the G-protein-coupled receptors OGR1 and GPR4 are both inactive at pH 7.8. At pH 6.8, however, OGR1 leads to inositol phosphate formation, while GPR4 causes cAMP formation(Ludwig et al., 2003). Thus,pH changes can be coupled to two important second messengers. It is also known that certain ion channels are pH sensitive(Akopian et al., 2000; Berdiev et al., 2003), and H⁺-V-ATPase could exert its effect by controlling another electrogenic component of LR patterning. There is also the possibility that activity of the asymmetrically localized H⁺-V-ATPase could cause redistribution of determinants (which may include the H⁺ ion itself) by virtue of electrophoretic forces driven by the strong voltage gradients that the H⁺-V-ATPase can produce. One such plausible determinant is serotonin (5-HT). The normal localization of 5-HT to one blastomere at the 16-cell stage is required for asymmetry(Fukumoto et al., 2005b), and we observed that 5-HT cannot be detected by that stage in embryos exposed to H⁺-V-ATPase inhibitors (Fig. 8D-G). One possibility, based on the requirement for vesicular monoamine transporters (VMAT) in asymmetry(Fukumoto et al., 2005a) and on the well-known role of the H⁺-V-ATPase in powering 5-HT import into vesicles via the proton gradient-dependent transporter VMAT(Hayashi et al., 1999), is that 5-HT is degraded by monoamine oxidase when its import into protective vesicles is inhibited by downregulation of the H⁺-V-ATPase. Thus,one likely link between H⁺-V-ATPase function and downstream events is 5-HT signaling. However, the ability of asymmetry to be perturbed by cell-membrane modulation of pH and voltage suggests strongly that other linking mechanisms remain to be discovered. Although many of the downstream details remain to be worked out, we have formulated a model that accounts for the available data.

We propose that a small charged morphogen, here hypothesized to be a positively charged inhibitor of left-side specific gene products (`inhibitor of leftness' or IOL), is evenly distributed in the fertilized egg, but becomes asymmetrically localized during early cleavage stages(Fig. 9). For this morphogen to be active, two requirements must be met: there must be a threshold concentration and there must be a high pH. The proposed V_memgradient, established by the activity of the right-sided H⁺-V-ATPase, and possibly the H⁺/K⁺-ATPase explored by Levin et al. (Levin et al.,2002), provides the unidirectional motive force that moves the morphogen, allowing the threshold concentration to be reached only on one side(the right). The asymmetric activity of the H⁺-V-ATPase also raises the pH high enough to activate the morphogen, again, only on the right. Under these conditions the morphogen is active and affects the downstream genetic cascade only on the right side.

Because of the dual requirement for both threshold concentration and pH,manipulations that affect either of these conditions will cause heterotaxia. Heterotaxia I results when V_mem is disrupted. In that situation,although pH may be high enough to activate the morphogen, the threshold concentration has not been reached, and heterotaxia results. This may happen in embryos treated with H⁺-V-ATPase inhibitors, expressing the electrogenic H⁺-pump PMA1.2, or treated with the V_memdisrupting agent PTX. Heterotaxia II results when V_mem is normal,but pH is not high enough. Under those conditions, the threshold concentration may be reached, but pH is too low, and again, the morphogen remains inactive and heterotaxia results. This describes the situation in embryos treated with H⁺-V-ATPase inhibitors, low pH medium or the electroneutral ionophore TBT, and in those embryos expressing the electroneutral NHE3. Testing this model and applying it to embryos of other species represents the next exciting challenge for future efforts to understand the integration of bioelectric and genetic information in left-right signaling.

We gratefully acknowledge the following for their contributions to this work: Debra Sorocco and Punita Koustubhan for animal and laboratory care;Shing-Ming Cheng for histology; and Dayong Qiu for molecular biology. For their kind gifts of reagents, we thank R. Shen for lobatomide analogs, M. Lu for YCHE78, C. Masuda and M. Montero-Lomeli for PMA1.2, R. Sabirov for NHE3,Y. Moriyama for anti-subunit A antibody, P. Backx for dominant negative Kir2.2, Kaethi Geering for the H,K-ATPase α subunit, W. Reith for anti-RFX3 antibody, M. Tosteson for palytoxin, and D. Stone and X. S. Xie for anti-A and anti-F subunit antibodies. We also thank Mark Mercola for important early discussions and the encouragement of this work at the beginning. This project was supported by NIH Training Grant 1-T32-DE-08327 and NIH Career Transition Award 1-K22-DE016633-01A1 to D.S.A; NSF grant IOB-0087517 to K.R.R;and NIH Research grant 1-R01-GM-0622, American Cancer Society Research Scholar Grant RSG-02-046-01-DDC and US Department of Transportation Grant to the FRCDB to M.L. This investigation was conducted in a Forsyth Institute facility renovated with support from Research Facilities Improvement Grant Number CO6RR11244 from the National Center for Research Resources, National Institutes of Health.