Intracellular acidification of gastrula ectoderm is important for posterior axial development in Xenopus

There is evidence that neural induction is a two step process. During amphibian gastrulation, an initial activation signal is thought to induce neural tissue with an anterior phenotype, which later differentiates to forebrain and eyes, while a subsequent transformation signal converts anterior neural tissue to a more posterior fate, resulting in the development of hindbrain and spinal cord (Nieuwkoop et al., 1952, 1985).

In 1975, Picard showed that incubating Xenopus gastrula ectoderm in a medium containing high concentrations of bicarbonate and ammonium ions, a treatment which could be expected to elevate intracellular pH (pH_i), induces differentiation of the cement gland, an extremely anterior ectodermal structure. Considering the evidence that cement gland induction precedes forebrain induction (Sive et al., 1989), Picard’ s results raised the possibility that pH_i elevation can cause, and may even be required in vivo, for anterior neural development, namely for the action of the activation signal.

It has been reported, in a wide variety of systems, that pH_i changes are required for the onset of differentiation and cell division (Busa and Nuccitelli, 1984). During development, cytoplasmic alkalization occurs during Xenopus oocyte maturation (Chambard and Pouysségur, 1986) and in response to fertilisation in Xenopus eggs (Webb and Nuccitelli, 1981; Grandin and Charbonneau, 1990) and sea urchin eggs (Shen and Steinhardt, 1979) as well as before prespore development in Dictyostelum (Lookeren Campagne et al., 1989). An acidic pH_i shift has also been reported during the larval development of Caenorhabditis elegans (Wadsworth and Riddle, 1988). Research concerning pH_i changes, in response to growth factors and to the activation of signal transduction pathways, has shown further that pH_i is determined by the synergistic action of differentially regulated ion exchangers (Bierman et al., 1988; Bierman, 1988; Ganz et al., 1989; Gillies and Martinez-Zaguilan, 1991). The ion exchangers in question are the sodium/hydrogen (Na⁺/H⁺) exchanger and the Na⁺-dependent bicarbonate/chloride (HCO₃⁻/Cl⁻) exchanger, mediating both an increase in pH_i, and the Na⁺ independent HCO₃⁻/Cl⁻ exchanger, mediating cytoplasmic acidification (Boron, 1983; Vigne et al., 1988; Ganz et al. 1989, Gillies and Martinez-Zaguilan, 1991). The Na⁺/H⁺ exchanger, in particular, functions as an essential signal transducer during the initiation of differentiation and mitosis (Boonstra et al., 1988; Horvat et al., 1992; Harguindey and Cragoe, 1992). Its exchange activity is an integrated response to the activation of different signal transducers (Sardet et al., 1990, 1991; Wakabayashi et al., 1993) including tyrosine kinases (Moolenaar et al., 1984b; Moolenaar, 1986a, b; Sardet et al., 1991), G-proteins (Huang et al., 1987; Sardet et al., 1990, 1991), protein kinase C (PKC) (Moolenaar et al., 1984a; Vigne et al., 1988; Shen and Buck, 1990; Mitsuka and Berk, 1991; Horie et al., 1992) and cyclic AMP (cAMP) (Kong et al., 1989; Borgese et al., 1992). Considering that PKC and cAMP signalling are also known to mediate neural induction (Davids et al., 1987; Otte et al., 1988, 1989; Durston and Otte, 1991; Otte and Moon, 1992; Collett and Steele, 1993) we were especially interested in determining whether pH_i changes and Na⁺/H⁺ exchange activity participate in the signalling pathways leading to neural induction and we set up pH_i measurements using the BCECF fluorescence ratio technique, to examine these questions.

We found that pH_i decreases continuously in explants of both uninduced ectoderm and neurectoderm during the time course of gastrulation, so that no correlation was found between pH_i changes and neural induction. The observed pH_i decrease is mediated via a shift in the pH_i set point of the Na⁺/H⁺ exchanger to more acidic pH_i values so that the alka-lizing activity of this exchanger is shut off at increasingly acidic pH_i. A role for this pH_i decrease was suggested by the finding that elevating pH_i in dorsal meso/ecto explants affected posterior axial development: posterior Hox gene expression was already reduced during neurula stages and development of posterior structures, namely the tail and the trunk, was suppressed at tadpole stages. Head development was unaffected by treatments leading either to an extensive pH_i decrease or to a pH_i elevation. These data support the idea that the signal transduction pathways mediating head development and neural induction are pH_i tolerant under physiological conditions, while the signals mediating posterior axial development may require a sustained pH_i decrease for their action.

Female and male Xenopus were injected with 375 IU chorionic gonadotropin (Organon) into their dorsal lymphsac to stimulate gamete production. Eggs were artificially fertilized. Embryos were staged according to Nieuwkoop and Faber (1975). During blastula stages, the embryos were dejellied in 2% cysteine pH 7.8, washed thoroughly and cultured in tap water.

Dissections and explant culture were performed in modified Flickinger medium (MFM; NaCl 58 mM, KCl 1 mM, MgSO₄ 1 mM, NaHCO₃ 2.4 mM, Hepes 5 mM, CaCl₂ 0.5 mM) of varying pH (pH 6.8; 7.6; 8.8) or in Flickinger medium (FM, where Hepes is replaced by Na₂HPO₄ 1 mM and KH₂PO₄ 0.82 mM) pH 6.9 or pH 8.3. At the appropriate stages, the vitelline membranes were removed with watchmaker’ s forceps and the embryos were dissected with electrolytically sharpened tungsten needles.

For the morphological assay, dorsal mesoderm, including the dorsal blastore lip, was cut together with the animal cap at stage 10.5 (meso/ecto). Explants were cultured overnight in MFM at varying pH with or without the protonophore carbonylcyanide p-trifluo-romethoxyphenylhydrazone (FCCP) (0.5 μM) (Sigma). At stages 16/17, explants were fixed in MEMPFA for 90 minutes, washed with PBS, stored in methanol and then submitted to whole-mount in situ hybridisation to detect Hox gene expression. For assaying morphological effects, explants were cultured until control embryos reached stage 42, after changing the medium to MFM pH 7.6 at stage 20. These explants were scored for axial defects.

Five explants of a particular stage (stage 10 ectoderm, stage 11 anterior neuroectoderm or stage 13 neural plates) were prepared in HCO₃⁻ free MFM pH 7.6 or in HCO₃⁻ free FM pH 6.8 or pH 8.3. Dissections were performed on bee’ s wax, to which the edges of the explants will briefly attach after dissection. This allows manual removal of dorsal mesoderm, for clean neurectoderm preparations. In order to minimize the chance that non induced cells are measured during the pH_i recordings, of stage 11 neurectoderm, only a small dorsal piece (about 60° wide) of an animal cap was used, which had been underlain by migrating, dorsal mesoderm. Dorsal ectodermal cells which had no contact with the migrating dorsal mesoderm were removed. The identification of stage 13 neural plate is easier, because its outline is already well developed on the dorsal side of the embryo. In control experiments, uninduced stage 10 ectoderm was isolated, sandwiched and cultured in MFM pH 7.6 until controls reached the appropriate stages. The sandwiches were then dissected again. All types of explants were tested routinely for neural development by culturing them for 2 days in MFM and assaying them morphologically and via immunostaining with the 2G9 antibody (Jones and Woodland, 1989). Explants of uninduced ectoderm (stage 10) developed into atypical epidermis, while stage 11 neurectoderm differentiated anterior neural structures, such as forebrain and eyes. Stage 13 neural plates developed anterior and posterior neural structures (forebrain, hindbrain and spinal cord).

Explants were attached to polylysine (50 μg/ml) coated glass. In order to assure good attachment the inner cells were brought in contact with the glass and the outer pigmented layer was removed during a brief (maximal 3 minutes) exposure to modified Stern’ s solution – calcium- and magnesium-free pH 7.6. This treatment was nescessary because the impermeable pigmented cell layer not only hinders dye loading of the inner cells, but also has the tendency to round up and enclose the inner cells (see also below), thereby loosening their attachment with the glass and leading to extensive cell loss during the pH_i recordings. Importantly, the bulk of the nervous system is induced in the massive deeper layer (Nieuwkoop et al., 1985), and this tissue was thus also used for our pH_i recordings. The medium was then quickly changed back to the specific medium named above. This brief dissociation treatment preserved the multi-layered structure of the explants. Explants were loaded with the fluorescent, pH sensitive dye 2′,7′-bis-(2-carboxyethyl)-5(and-6)carboxyfluorescein, acetoxymethyl ester (BCECF-AM; 20 μM; Molecular Probes) (Rink et al., 1982) for 40 minutes. It proved important to load the cells with the fluorophore immediately before the pH_i measurements, because dextran coupled BCECF (M_r 10×10³), microinjected into zygotes, gave extremely low emission intensities when tested in gastrula ectoderm. Additionally, the microinjected BCECF dextran was unresponsive to pH_i elevation, e.g. after elevating pH_o or during the calibration procedure, indicating that its pH_i response was altered during the long incubation (approx.16 hours) up to gastrula stages. For these reasons we considered the use of dextran coupled BCECF unsuitable for our experiments.

Toxic effects, cell death or atypical development of the explants (see above) in response to the loading procedure as well as to treatment with BCECF or its free ester were tested for by exposing explants to dissociation medium. This treatment reduces the speed of explant closure and the epidermis (which lay under the inner cells, here) therfore presented no barrier to loading the (exposed) inner cells with BCECF-AM, following normal procedures. Afterwards, the explants were washed and cultured in MFM. Explants developed normally when compared to control explants (see above) and unusual cell death was never observed.

All pH_i measurements were performed using a Perkin Elmer LS50B fluorescence spectrometer. The bandwidths for both the excitation and emission wavelengths were set to 3 nm. The pH_i-dependent (Ex. 495 nm) and pH_i-independent (Ex. 438 nm) emission intensities (Em. 532 nm) were recorded alternately for 10 seconds each, with a 4 second interval between them (full cycle = 28 seconds). These signals were corrected for background fluorescence and were used to generate a pH_i-dependent ratio (R), by dividing the fluorescence intensity (I) of the pH-dependent excitation by the pH-independent signal measured at the isobestic point (R = I (495)/I (438)). The stage specific pH_i value was determined 30 minutes after the pH_i measurement had started.

Calibration of the ratio as a function of pH_i was obtained by equilibrating H⁺ using the K⁺/H⁺ ionophore nigericin (10 μM) (Fig.3), which sets [H⁺]_i/[H⁺]_o = [K⁺]_i/[K⁺]_o as described previously (Thomas et al., 1979; Rink et al., 1982; Moolenaar et al., 1983, 1984a), in a special calibration buffer (Hepes 10 mM, KCl 100 mM, NaCl 20 mM, CaCl₂ 1 mM, MgSO₄ 1 mM) in which the potassium concentration is similar to cytoplasmic [K⁺] (Gillespie, 1983).

Calibrations were then performed by using calibration buffers with three different pH values: 7.0, 7.6 and 8.0. The recorded BCECF ratios change linearly with pH_i between pH_i 7.0 and 8.0, which is shown by the factor below. Therefore, all experiments were performed within the linear range of the BCECF response to pH_i.

R is ratio, pH_u is pH units.

Using this factor we were able to calculate pH_i after calibrating individual experiments for pH 7.6 only. In order to avoid errors, we decided not to use a common calibration line but to calibrate each experiment separately.

Images of explants loaded with BCECF-AM (see above) were taken on a Leitz Orthoplan epiluminescence fluorescence microscope (25× magnification, water imersion objective) coupled to a Fluorolog Spex dual wavelength fluorimeter as an excitation source. Images at 438 nm and 495 nm were recorded using an image intensifier system coupled to a video camera (adjustable gain) and stored on video tape. The respective images were digitized (530×576 pixels, 256 gray levels) and corrected for background fluorescence. Ratios of 495 nm/438 nm were calculated for each individual pixel and scaled with the use of calibrated values from images taken in the presence of nigericin (calibration procedure see above) at two different known, clamped pH_i values, by linear interpolation of the known values.

Dissociated gastrula cells (stages 10 and 13) were loaded for 40 minutes with either 20 μM BCECF or 20 μM Fura-2 AM in MFM pH 7.6. Subsequently the cells were washed three times and incubated for either 45 or 90 minutes in MFM pH 7.6. They were then lysed by gentle pipetting in 0.5 ml organelle isolation buffer (Chase and Dawid, 1972). Nuclei were removed from the homogenate by centrifugation at 100 g for 10 minutes and the pellet was then dissociated in nuclear buffer (Wolffe, 1989), and diluted to a final volume of 0.5 ml. Under these conditions the isolated nuclei remain intact, as verified by DAPI staining, while the total absence of DAPI fluorescent nuclear fragments in the supernatant confirmed that intact nuclei were indeed isolated in the nuclear fraction. The supernatant was then centrifuged at 50,000 g for 30 minutes, a speed at which smaller organelles are also pelleted (Alberts et al., 1989). The supernatant (containing the cytoplasmic fraction) was collected and the pellet (organelle fraction) was dispersed in organelle buffer and brought to the same volume as the supernatant. The isolation of mitochondria in this high-speed fraction was confirmed by measuring cytochrome C reductase activity (data not shown) which was found exclusively in the high-speed pellet and not in the supernatant.

All measurements were performed using an epiluminescence microscope (see above) using a ×50 water imersion lens. Each fraction was measured 4 times in a 100 μm deep chamber either for its BCECF or Fura-2 fluorescence, and the recorded fluorescence intensities were averaged. Fluorophore fluorescence was calculated after subtracting the background fluorescence measured in unlabeled cell fractions, using the same procedure. Fluorescence intensity at the isobestic point (BCECF ex. 438 nm, em. 532 nm, Fura-2 ex. 365 nm, em. 505 nm) was taken as a measure of dye concentration in the different fractions, and the relative fluorescence contribution of each fraction was calculated.

We also tested whether this cell fractionation procedure could be used effectively to isolate dye located in organelles, by dye loading the cells over a period of 3 hours, conditions which have been reported to result in loading organelles with Fura-2 (Williams et al., 1985; Malgaroli et al., 1987). We found that Fura-2 fluorescence in the organelle fraction now increased by 50% while the relative contribution of BCECF increased by only 10%, when compared to the fluorescence contributions observed in this fraction after our normal dye loading procedure (40 minutes) and subsequent incubation for 45 minutes. Further we observed that the fluorophore acridine orange (5 μg/ml), which is strongly sequestered into organelles, remained detectable in highly fluorescent particles in the dispersed high speed pellet. Taken together, these findings indicate that the procedure used by us can be used effectively to isolate dye loaded organelles in Xenopus cells. It is thus suitable for the separation of different cellular compartments.

An acute cytoplasmic acidification of the cells was introduced by the NH₄Cl prepulse method (Roos and Boron, 1981). Briefly, the exposure of the cells to NH₄⁺ results in a rapid intracellular alkalization due to NH₃ entry. The subsequent passive influx of NH₄⁺ causes pH_i to decay slowly. When external NH₄⁺ is removed, the accumulated NH₄⁺ leaves the cell as NH₃, thereby loading the cells with an excess of intracellular protons. Cells were loaded for 5 minutes with 10 mM NH₄Cl under bicarbonate free conditions and subsequently washed with bicarbonate-free MFM pH 7.6 with or without 0.1 mM dimethyl-amiloride (DMA) (Kleyman and Cragoe, 1988).

The rate constants of the exponential pH_i recoveries were calculated, after correcting for the permanently decreasing pH_i (Fig. 6), by fitting the pH_i time course to the equation (Biermann, 1988):

where k is the first order rate constant, t is time in minutes, pH_i(∞) is pH_i at the new asymptotic steady state, pH(0) is pH_i at the begining of the acid load (AL) recovery and pH_i(t) is pH_i at the time after the start of AL recovery. The rate constant k was then fitted by linear regression analysis.

The procedure for whole-mount in situ hybridisation was as described by Harland (1991) using digoxigenin-labeled probes. The probes were generated by in vitro transcription as described by Godsave et al. (1994). Probe lengths were, Hoxb-3 1.3 kb and Hoxb-9 250 bases. After the in situ hybridisation procedure the pigmented explants were bleached in 5% H₂O₂ in PBS and were then photographed in 1.5% methyl cellulose (Sigma) in PBS.

The mean value ± standard error of means (s.e.m.) are shown; n indicates number.

In order to use BCECF as a pH_i indicator we first had to verify that it is taken up by Xenopus ectoderm cells, and that cellular organelles are not loaded extensively with the fluorophore. The loading and distribution of BCECF were therefore examined by recording images of loaded explants at it’ s isobestic point (438 nm), which is a useful indicator of intracellular dye concentration. BCECF-AM was found to be taken up effectively by the cells in the explants and selective dye uptake was never observed at the potentially damaged edges of the explants. The dye appeared to be located within the cytoplasm and within the nucleus (Fig. 3B), which was identified by double staining with the DNA-specific Hoechst fluorophore (H-33258) (not shown). BCECF uptake into other organelles was never evident, even at the end of the pH_i measurements (after approx. 90 minutes; Fig. 3B), either by imaging at the isobestic point, indicating dye uptake, or by ratio image measurements (Tsien and Poenie, 1986; Paradiso et al., 1987), indicating differences in pH. In order to examine the distribution of BCECF more precisely, small organelles, nuclei and cytoplasm were fractionated from BCECF loaded gastrula cells and the fluorescence contributions of these three fractions were measured separately. As a control for the fractionation procedure Fura-2 AM, a calcium fluorophore, which is sequestered by nuclei and organelles in a variety of systems (Williams et al., 1985; Malgaroli et al., 1987; Tsien, 1989), was loaded simultaneously into identical cells. The loading and distribution of each of the fluorophores was then measured at their specific isobestic points, which is a good indicator for fluorophore concentration. After incubating the loaded cells for 45 minutes we found that 63.03% of the BCECF fluorescence was localized in the cytoplasm and 30.89% in the nuclei, while only 6.08% of the fluorescence was located in the organelle fraction (Fig. 1A). After 90 minutes incubation the absolute BCECF fluorescence increased further (Fig. 1C), probably due to hydrolysis of the non-fluorescent BCECF-AM ester, a phenomenon which we also observed during the pH_i recordings. Importantly, the relative fluorescence contributions of the three fractions failed to change significantly (Fig. 1C). The relative BCECF fluorescence of the organelle fraction increased by less than 1%. In contrast, Fura-2 exhibits a different dye distribution (Fig. 1B,D). After incubating the loaded cells for 45 minutes we found 68.48% of the fluorescent Fura-2 in the nuclei, and only 22.26% in the cytoplasm (Fig. 1B). The mechanisms behind this strong nuclear Fura-2 sequestration remain unclear to us, but substantial nuclear Fura-2 localisation has been reported in other systems (Williams et al., 1985). Interestingly, we found 9.26% of the Fura-2 fluorescence in the organelles (Fig. 1B), indicating that these show stronger relative Fura-2 labeling than BCECF labeling. Although it does not appear to be dramatically higher, organelle Fura-2 fluorescence is high when compared to cytoplasmic Fura-2 fluorescence, which is only about 2.5 times higher. On the contrary, cytoplasmic BCECF fluorescence is about 10 times higher than organelle fluorescence. When we consider that these dyes presumably reach organelles via the cytoplasm these findings indicate that BCECF sequestration into organelles is not a prominent feature of this fluorophore in Xenopus gastrula cells.

The measurements reported here were performed using stage 13 cells, and the results from measurements using stage 10 gastrula cells are not significantly different (data not shown). We conclued that the BCECF fluorescence contribution from cellular organelles is small and does not change significantly during gastrulation. It cannot account for the intracellular acidification observed in gastrula (neur)ectoderm.

However, a substantial fraction of BCECF fluorescence was found in the nucleus. We estimated from false colour images that the pH of the nucleus is about 0.1 pH_u lower than the cytoplasmic pH (Fig. 2A, B). The reason for this pH difference is unclear, since nuclear pores should permit proton passage between the nucleus and the cytoplasm (Alberts et al., 1989). Importantly, nuclear pH clearly follows cytoplasmic pH: when pH_i decreases, during gastrulation (see below), it does so both in the cytoplasm and in the nucleus (compare Fig. 2A and 2B). This observation also makes it very unlikely that BCECF-loaded organelles make an important contribution to our pH_i recordings, because we would then have expected non-parallel pH_i courses in the cytoplasm and the nucleus, and this was never observed. Further, we also observed that cytoplasmic pH varies little between different cells within one explant (about 0.2 pH_u). The pH_i values reported by us below, thus mainly represent two important compartments, the nucleus and the cytoplasm, which are coupled by signal transduction pathways.

The effectiveness of the calibration procedure was tested by calculating ratio images of previously calibrated explants. There were now no measurable differences in pH_i, between cells or between the cytoplasm and the nucleus (Fig. 3A), indicating that protons are equilibrated across the plasma and nuclear membranes by the calibration procedure.

From these results we can conclude that the fluorophore ratio technique (Rink et al., 1982; Moolenaar et al., 1983, 1984a,b), using BCECF-AM as the pH_i indicator, is a useful tool for studying intracellular pH in Xenopus gastrula cells.

There is evidence suggesting that anterior axial development can be triggered by elevating intracellular pH (see Introduction; Picard, 1975; Sive et al., 1989). We were therefore interested to see if pH_i is elevated in response to neural induction, especially forebrain induction, in Xenopus embryos. We followed pH_i in ectoderm and in neurectoderm, which were explanted following induction by in vivo signals, at sequential stages during gastrulation.

We found that pH_i decreases substantially during the time course of neural induction (Fig. 4A). In stage 10 ectoderm, cultured in MFM at pH 7.6, a mean pH_i of 7.88±0.03 (mean ± s.e.m.) (n=10) was recorded. In stage 11 neuroectoderm explants, this value decreased to pH_i 7.59±0.04 (n=6), and still lower values were recorded in explants of stage 13 neural plates with pH_i 7.26±0.03 (n=6). This pH_i decrease was also observable as a continuous pH_i decrease during the recordings (Fig. 5). Our observations using ratio image and cell fractionation studies (see above) make it most unlikely that this pH_i decrease is an artifactual consequence of changes in BCECF loading of, or availabilities of cellular organelles. It appears to reflect developmentally regulated acidification of the cytoplasmic and nuclear compartments.

We were interested to investigate whether there is a causal relationship between neural induction and the observed cytoplasmic acidification. Uninduced ectoderm explants (stage 10) were therefore cultured up to late gastrula stages, and pH_i was measured. We recorded a similar pH_i decrease in uninduced ectoderm as in neurectoderm during the course of gastrulation (Fig. 4A). We conclude that the observed intracellular acidification shows no correlation with neural induction. Because of this result, we decided to focus further on measurements in the neurectoderm, rather than exploring differences between neurectoderm and uninduced ectoderm.

The occurrence of pH_i changes in gastrula neurectoderm explants suggests that the activities of ion exchangers, functioning as pH_i regulators, are developmentally regulated in this tissue. In order to approach this question, we first needed to identify these ion exchangers. The activities of two common pH_i regulators, the Na⁺/H⁺ exchanger and the Na⁺-dependent HCO₃⁻/Cl⁻ exchanger (Moolenaar, 1984b; Vigne et al., 1988; Bierman, 1988; Ganz et al., 1989) depend on the presence of extracellular Na⁺ or Na⁺ and HCO₃⁻ ions respectively (Roos and Boron, 1981). If these ions are omitted from the medium, these ion exchangers will be inhibited (Roos and Boron, 1981; Boron, 1983). Additionally, specific inhibitors for these exchangers are available, which are effective even in the presence of Na⁺ and HCO₃⁻. These are the amiloride derivative dimethyl-amiloride (DMA), which inhibits Na⁺/H⁺ exchange (Kleyman and Cragoe, 1988; Counillon et al., 1993; Horvat et al., 1993) and the stilbene derivative SITS, which inhibits HCO₃⁻/Cl⁻ exchange (Roos and Boron, 1981; Boron, 1983; Horvat et al., 1993). The functioning and the inhibition of these acid extruding exchangers can be assayed by acutely acidifying cells, and measuring the subsequent pH_i recovery under conditions of ion omission or inhibitor application (Roos and Boron, 1981; Boron, 1983; Bierman et al., 1988, see also Materials and methods: acid load recovery).

Fig. 5A shows an acid load recovery in HCO₃⁻ free medium, under conditions which should inhibit the Na⁺-dependent HCO₃⁻/Cl⁻ exchanger. In order to see whether this pH_i recovery results from Na⁺/H⁺ exchange activity, we inhibited this exchanger by two different means: either by adding its specific inhibitor dimethyl-amiloride (DMA) (Fig. 5B) or by omitting Na⁺ ions from the medium (Fig. 5B insert). The degree and the velocity of the resulting pH_i recovery is reduced and this effect is measurable as a strong reduction of the rate constant k in the presence of DMA (Table 1). Therefore, we deduce that an active Na⁺/H⁺ exchanger is present in gastrula (neur)ectoderm. In the presence of HCO₃⁻, the acid load recovery is enhanced (data not shown) while it is strongly reduced in the absence of Na⁺ (Fig. 5B insert). These findings indicate, that Xenopus embryos have functional Na⁺-dependent HCO₃⁻/Cl⁻ exchangers as well as Na⁺/H⁺ exchangers, and that these operate as acid extruders in gastrula neurectoderm. Fig. 5B (insert) shows further that the acid load recovery is not completely blocked by omitting Na⁺ from the medium. This finding indicates that other alkalizing mechanism(s) besides the Na⁺/H⁺ and the Na⁺-dependent HCO₃⁻/Cl⁻ exchangers also operate in Xenopus ectoderm and neurectoderm. These operate independently of Na⁺ and HCO₃⁻ and are non-responsive to DMA (Fig. 5B). The other characteristics of these acid extrusion mechanism(s) remain unclear.

Taken together we find that at least three different acid extrusion mechanisms are present and active in Xenopus gastrulae. These are the Na⁺/H⁺ exchanger, a Na⁺/HCO₃⁻ independent acid extruding mechanism and the Na⁺-dependent HCO₃⁻/Cl⁻ exchanger. Surprisingly, the availability of at least the first two of these acid extruders failed to hinder the pH_i decrease observed in gastrula (neur)ectoderm cells. This finding led us to wonder whether the activity ranges of these acid extruders are developmentally regulated, being shifted to more acidic pH_i values during the course of gastrulation. We therefore assayed their pH_i set point values.

The pH_i set point of an alkalizing exchanger determines at which pH_i its acid extrusion activity is shut off, preventing further alkalization of the cytoplasm. To determine whether this mechanism itself is developmentally regulated we determined the stage-dependent pH_i set points of the Na⁺/H⁺ exchanger and of the HCO₃⁻/DMA insensitive acid extruder(s) by plotting the velocity of the pH_i recovery after an acid load against pH_i, using conditions chosen to reveal the activities of these exchangers (Fig. 6A,B).

Fig. 6A demonstrates that Na⁺/H⁺ exchange is shut off at a specific pH_i set point value, which indeed shifts to an increasingly acidic pH_i value during the time course of gastrulation. In stage 10 explants the pH_i set point is 7.74±0.07 (n=5), at stage 11 it is 7.68±0.04 (n=10), and at stage 13 the pH_i set point has decreased to 7.20±0.05 (n=3). At the beginning of gastrulation, (stage 10) the pH_i set point is thus slightly more acidic than pH_i, and not significantly different from the pH_i set point at stage 11. From this stage onwards, the pH_i set point of the Na⁺/H⁺ exchanger is equal to pH_i, demonstrating that this exchanger then defines pH_i under HCO₃⁻ free conditions.

The recorded pH_i set point value of the DMA/HCO₃⁻ insensitive acid extruder is shifted even further towards acidic pH_i values, because it’ s set point remains lower than the recorded pH_i throughout (Fig. 6B). In stage 10 explants the pH_i set point of this acid extruder is 7.77±0.07 (n=5), and this then decreases slightly to 7.45±0.03 (n=3) at stage 11. In stage 13 explants this set point has decreased to 6.93±0.07 (n=4), which is about 0.3 pH_u lower than pH_i (7.26), indicating that this exchange system is unlikely to be an active pH_i regulator during gastrulation. Fig. 5A,B also shows that, during an acid load recovery, the two acid extruders are activated sequentially. The Na⁺/H⁺ exchanger is activated first, and responds to small pH_i changes with a strong increase in its proton extrusion activity. This is demonstrated by the slope of the pH_i-dependent proton extrusion rate, which is a parameter for the pH_i sensitivity of this exchanger. The DMA/HCO₃⁻ insensitive acid extruder is only activated after an extensive pH_i decrease and it’ s activation is then slight, as shown by the moderate slope of the decreasing exchange velocity in response to pH_i. Consequently, the DMA/HCO₃⁻ insensitive acid extrusion mechanism is almost inactive in the presence of an active Na⁺/H⁺ exchanger. Taken together, these data show that the alkalizing Na⁺/H⁺ exchanger determines and maintains pH_i under HCO₃⁻ free conditions. The acidic shift of its pH_i set point is an important mechanism permitting cytoplasmic acidification to occur, as revealed in gastrula neurectoderm explants.

We were interested in investigating the developmental relevance of the observed acidic pH_i shift in explants of gastrula (neur)ectoderm. pH_i recordings performed in cells incubated in media with higher or lower pH_o showed that the corresponding pH_i values were higher and lower respectively at all developmental stages tested here (Fig. 4B). These findings demonstrate that pH_i is influenced by pH_o. Importantly, we also always recorded a sustained pH_i decrease during gastrulation, under all pH_o conditions tested, demonstrating that the observed pH_i decrease is not an artifact of incubating cells at a particular pH_o. The alterations in pH_i occurred only within a limited, physiological pH_i range, which was much smaller than the applied pH_o range (Fig. 4B). These data indicate that we can elevate or lower pH_i by changing pH_o.

pH_i was therefore manipulated by changing the pH of the extracellular fluid. In vivo this is the blastocoelar fluid in which the ion concentration and pH is influenced by the mass of surrounding cells, making the manipulation of blastocoelar pH difficult. pH_i was therefore manipulated in smaller dorsal meso/ecto explants. This preparation also allows extensive exchange of the blastocoelar fluid for medium of a known pH_o.

Explants of stage 10.5 dorsal mesoderm/ectoderm (meso/ecto) were prepared and cultured at three different pH_o: 6.8, 7.6 and 8.8. When control embryos reached stage 42 (tadpole) these explants were scored for development of different axial regions, namely head, trunk and tail (Fig. 7F). Under control conditions, at pH_o 7.6, 73.4% of the dorsal meso/ecto explants developed into small tadpoles with a complete anteroposterior axis (Fig. 7C, Table 2). In a much lower percentage (26.6%), tail development was absent (Fig. 7B). Tail and trunk development was absent in only 1.6% of these control explants, but all developed morphologically normal heads. Similar results were achieved when meso/ecto explants were incubated at pH_o 6.8 (Table 2). The incubation of these explants in a high pH medium (pH_o 8.8), on the other hand, led to extensive suppression of posterior axial development. Only 42.3% of the explants now developed the full range of axial structures (Fig. 7C). The remaining 57.7% all showed suppression of tail development (Fig. 7B). Trunk development was now also inhibited in 21.6% of the explants, while all still developed normal heads (Figs 6A, 7A, Table 2). Obviously, the elevation of pH_o to 8.8, even without any further treatment, leads to a disadvantageous situation for the development of posterior axial structures in meso/ecto explants.

In order to exclude artifacts due to modulation of extracellular events, like receptor ligand interactions, by the high pH_o, pH_i was elevated in a different way. The protonophore FCCP selectively permeabilizes the plasma membrane for protons (Li and Poznansky, 1990). This proton channel was used to elevate pH_i above the low values of late gastrula (neur)ectoderm by adding it to cells in medium with a slightly alkaline pH_o (7.6) (Fig. 5C), to determine whether this treatment can also inhibit posterior development.

In dorsal meso/ecto explants cultured at pH_o 7.6 in the presence of FCCP, the development of the tail and the trunk was strongly suppressed (Fig. 7E,D, Table 2), a feature that was not observed in the absence of FCCP (Table 2). The addition of the protonophore to medium at pH 8.8 only slightly intensified the already striking generation of axial defects (Table 2).

However, when explants were cultured in the presence of FCCP at pH_o 6.8, the type of axial development was similar to that obtained in FCCP free medium (Fig. 7F). The explants developed a full axis with high probability, and the frequencies of their head, trunk and tail development were indistinguishable from those in non FCCP treated controls (compare Fig. 7F and C, Table 2). A second feature of interest was that a subpopulation (19.5%) of explants (Table 2, legend) now failed to develop an obvious axis. Histological sections revealed that these explants contained dorsal axial structures, including notochord, muscle and neural tissue (Fig. 8B), but that these were not organized into a proper anterior-posterior axis. This effect was not enhanced by increasing concentrations of FCCP (up to 4 μm), nor was it observed under the alkaline pH_o conditions. These findings indicate that FCCP treatment alone has no specific effect on axial development, but that the protonophore treatment reveals a second pH_i-dependent effect in low pH medium.

The recordings of pH_i in the presence of FCCP show a biphasic pH_i time course (Fig. 5C): initially pH_i decreases strongly. In alkaline medium, (pH 7.6) it then increases again during mid gastrula and late gastrula stages, rising to above the low pH_i value recorded under control conditions (Fig. 4C). In slightly acidic medium (pH 6.8), pH_i remains low (data not shown). Clearly, the initial intracellular acidification observed under all (pH_o) conditions in the presence of FCCP failed to suppress neural induction and was also permissive for anterior axial development (Fig. 7D-F) in the overwhelming majority of explants (Table 2).

We also investigated whether the suppression of posterior development, which we observed during tadpole stages, could be traced back to pH_i effects on patterning during early axis formation. To approach this question, we applied whole-mount in situ hybridisation (Harland, 1991) to examine the expression patterns of two Xenopus Hox genes, which are already expressed in a position-dependent manner along the antero-posterior axis during neurula stages: Hoxb-9 is expressed in the entire spinal cord, thus, in the trunk and the tail, and Hoxb-3 is expressed in a band, immediately posterior to the otic vesicle in the hindbrain (Godsave et al., 1994). These expression patterns were examined in dorsal meso/ecto explants cultured up to the mid neurula stage 16/17. Under conditions leading to extensive axial development, these neurula stage explants showed an extensive posterior expression domain of Hoxb-9 (Fig. 9A) and an anterior stripe of Hoxb-3 expression in the future hindbrain (Fig. 9E). Under conditions leading to the suppression of posterior structures Hoxb-9 expression was reduced to a small posterior zone (Fig. 9B,C) or is absent (Fig. 9D). However, the anterior stripe of Hoxb-3 expression was never supressed, but shifted towards the posterior end of these explants (Fig. 9F-H). These data indicate that pH_i manipulation in gastrula cells affects early processes mediating axis formation, leading to altered gene expression patterns by the neurula stage.

Taken together, we present evidence that elevating pH_i above its low value during late gastrula stages inhibits the development of posterior structures by interfering with early axial patterning, prior to establishment of the hox code. The pH_i decrease, recorded during the time course of gastrulation may therefore be required for normal posterior axial development. The processes mediating anterior and neural development appear to be more pH_i tolerant.

In this study, we measured intracellular pH (pH_i) in Xenopus gastrula cells via the fluorescence ratio technique using the pH sensitive fluorophore BCECF. Ectoderm and neurectoderm explants from different developmental stages were loaded with the membrane permeant ester BCECF-AM and fluorescence imaging then revealed that nearly all the detectable fluorescence was in the cytoplasmic and nuclear compartment in these cells. Cell fractionation studies, performed as a further control for the contribution of different compartments to total fluorescence confirmed that nearly all of the fluorescence (almost 94%) is localized in the cytoplasmic and nuclear compartments, while only 6% of the fluorescence was found in the organelle fractions, at all stages tested. Importantly, fluorescence imaging also revealed that pH_i values in the nucleus and the cytoplasm are quite similar, and also that the pH_i changes observed during development occur in parallel in both compartments. Taken together, these observations make it very unlikely that changes in BCECF loading of, availabilities of, or pH_i values in, cellular organelles, such as acidic lysozomes, contribute significantly to total cellular fluorescence signal, and thereby yield artifactual pH_i courses. The total pH_i measurements made in this study thus are made up of an average of the pH_i values in these two important cellular compartments, the cytoplasm and the nucleus, which are known to be coupled by signal transduction pathways.

In Xenopus, pH_i measurements have previously been performed in intact embryos from fertilisation up to the mid blastula stage, using microelectrode measurements (Turin and Warner, 1980; Lee and Steinhardt, 1981; Webb and Nuccitelli, 1981; Grandin and Charbonneau, 1990, 1991). These authors reported pH_i values of about 7.7 at early blastula stages. The pH_i values of gastrula ectoderm and neurectoderm reported here (pH 7.9-7.3) and very recently by others (Sater et al., 1994; pH 7.7-7.5), using BCECF ratio measurements, are within a similar pH range. The latter authors also recorded a slight, but continuous pH_i decrease during their measurements in stage 10.5 cells in ectoderm explant, using microinjected BCECF dextran. Since this large molecule is localized in the cytoplasm in their experiments, this finding using a different method than our own emphasizes that decreasing cytoplasmic pH is an endogenous feature of Xenopus gastrula ectoderm cells and not an artifactual result of the method used. Taken together, these findings indicate that BCECF ratio measurements are a useful tool for measuring pH_i in Xenopus cells.

During gastrulation, we recorded a pH_i decrease of about 0.7 pH_u in explants of both neurectoderm and uninduced ectoderm. Surprisingly, the pH_i values and pH_i time courses were similar in both tissues, indicating that the observed pH_i decrease failed to corrrelate with neural induction. We demonstrated further that one mechanism underlying this pH_i decrease is a developmentally regulated, acidic shift in pH_i set points of at least two alkalizing exchangers. This acidic shift thus prevents maintenance of a high pH_i (see below).

In contradiction to our findings here, Sater et al. (1994) recently reported a pH_i elevation of 0.1 pH_u in Xenopus gastrula ectoderm, in response to neural induction by planar signalling. The apparent discrepancy between their observation and ours may arise for several reasons. One obvious difference from our own approach is that Sater et al. (1994) specifically assayed cellular responses to planar signalling. Vertical signalling, which normally occurs during neural induction, was excluded by a special explant preparation technique. Consequently planar signalling was examined in an in vitro situation. On the other hand, we were specifically interested in assaying cellular responses to neural induction in vivo, and we therefore isolated cells from gastrulating embryos following in vivo signalling. It is possible that planar signalling alone leads to a pH_i elevation, as reported by Sater et al. (1994), but that neural induction in vivo may also lead to other cellular responses, which overrule this. Although we found an intracellular acidification following in vivo neural induction, this possibility seems highly unlikely to us, because we would still have expected to record differences in pH_i between uninduced ectoderm and induced neurectoderm explants, following in vivo induction, in our own measurements. This was never observed.

The second and probably most important difference is that their method of recording pH_i differs substantially from our own, notably with respect to the number of cells in which pH_i was recorded. While we measured the average pH_i of a large group of cells (several thousands), Sater et al. (1994) recorded pH_i in single cells. In order to record a pH_i elevation in a large cell population a large fraction of the cells must elevate pH_i. Consequently, if in vivo neural induction leads to a general sustained pH_i increase in the induced cells we should have recorded a more alkaline pH_i value in stage 11 neurectoderm and possibly in stage 13 neural plates (although this stage was not measured by Sater et al., 1994). This was never observed. However, if the pH_i elevation occurs only in a small subpopulation of cells, e.g. in a position-dependent manner, the acidification in the remaining cells could very well have prevented the recording of a pH_i elevation in our measurements. Importantly, other findings made in both studies are very consistent with each other, making it very unlikely that the two different methods yield different results due to general experimental artifacts. We note that Sater et al. (1994) observed a slight intracellular acidification in stage 10.5 uninduced gastrula ectoderm explants, which is comparable to observations made by us. Another consistent feature is that the pH_i images taken of our explants show that pH_i varies by about 0.2 pH_u between different cells, and a similar variation was also observed by Sater et al. (1994). Finally, Sater et al. (1994) also reported that the addition of TPA to ectodermal explants, a treatment known to mediate anterior neural induction (Durston and Otte, 1991; Davids et al., 1987; Otte et al. 1988, 1989; Otte and Moon, 1992; Collett and Steele, 1993), was also followed by a pH_i decrease, a finding which we can confirm (unpublished results). They concluded that TPA may cause neural induction via aspecific interactions, rather than via the activation of the proper signal transduction pathways. Taking into account that we observe an intracellular acidification after in vivo neural induction, the acidification following TPA addition supports the idea that activating signal transduction pathways mediating neural induction is permissive for an intracellular acidification, while it does not neccessarily mediate a pH_i elevation. Further, our manipulations of pH_i also revealed that neural induction is not suppressed by strong intracellular acidification (see below). In conclusion, we think that the data presented by Sater et al. (1994) and by ourselves support the idea that intracellular acidification is a major feature of pH_i regulation in Xenopus gastrula ectoderm. Although we cannot rule out that pH_i is elevated in a small subpopulation of cells during neural induction, our pH_i recordings in large cell populations indicate that pH_i decreases dominate both in neurectoderm and ectoderm during gastrulation. These data already suggest that neural induction is not critically controlled by pH_i. pH_i manipulations of dorsal explants point to another role of pH_i regulation during development, namely that it is critical for the processes leading to axial specification.

Our pH_i recordings prompted us to test whether the intracellular acidification observed in explanted gastrula cells is important for embryonic development, namely for neural induction and/or axis formation. We therefore manipulated pH_i in dorsal mesoderm/ectoderm (meso/ecto) explants, using mild treatments which altered pH_i within the physiological range, while they did not clamp pH_i to a specific pH_i value, permitting intracellular pH_i regulation. We found that these treatments did, indeed, affect development. Treatments causing a mild pH_i elevation inhibited early posterior development, as revealed by the suppression of posterior Hoxb-9 expression and by the increasingly posterior location of hindbrain specific Hoxb-3 expression in mid neurulae stage explants. This indicates that early processes mediating axis formation and posteriorisation are affected by these pH_i changes. The early posterior reduction is not a transient effect but leads later to suppression of tail and/or trunk development during tadpole stages, while head development remains unaffected. Treatments that permitted or enhanced the observed intracellular acidification had no detectable effects on development at all: Hoxb-9 was expressed at normal (control) levels, Hoxb-3 expression was found anteriorly (see also Godsave et al., 1994) and head, trunk and tail structures developed normally. These results suggest that the early mechanisms leading to anterior development are rather pH_i tolerant, while regulatory processes mediating posterior development are pH_i sensitive. They are inhibited by a modest increase in pH_i and may depend on the endogenous acidification which we observed. Our data thus indicate that pH_i regulation is an important feature of axial development.

The conclusions drawn by us contrast with those drawn recently by Sater et al. (1994), who suggested that an endogenous pH_i elevation is important for neural development in vivo. We found no evidence for this. Even extreme acidification of explants, which affected explant morphology, failed to block neural differentiation, indicating that this is not critically controlled by physiological variations in pH_i. Other findings in the literature, namely that NH₄Cl treatment, which possibly elevates pH_i, can induce very anterior cement gland differentiation (Sive et al., 1989) and that the inhibition of an alkalizing ion exchanger using H₂ DIDS can suppress expression of the midbrain-specific marker engrailed-2 (en-2) (Sater et al., 1994) are also fully compatible with the idea that pH_i regulation is important for axial specification and they raise the possibility that the situation is more complex than indicated by our own findings. One should bear in mind, however, that harsh treatments, like immersion in 10 mM NH₄Cl may act as aspecific triggers, suppressing epidermal development and thereby switching differentiation to another developmental pathway (namely to anterior and possibly to neural). This default principle has been suggested recently by others (Sive et al., 1989; Hemmati-Brivanlou and Melton, 1994). Our own results here emphasize however, that, under physiological conditions, cytoplasmic acidification is of great interest as a potential regulator of posterior development. In this context it is interesting that Hox genes are expressed in parallel at the appropriate anteroposterior levels in different ectodermal derivatives, namely in neurectoderm and mesoderm (DeRobertis et al., 1989; McGinnis and Krumlauf, 1992), but that the mechanisms coordinating this alignment remain unclear. We note that the observed cytoplasmic acidification is also a general feature of different ectodermal derivatives, namely at least of explants of neurectoderm and uninduced ectoderm of Xenopus gastrulae. Cytoplasmic acidification may thus fit one of the requirements for a posteriorising regulator, that this should coordinate anteroposterior positional information in different tissue types in the embryo.

Our data presented above suggest that the developmental regulation of intracellular pH_i is an important developmental mechanism regulating posterior development during Xenopus gastrulation. Our investigations also provide insights into the regulatory mechanisms mediating the observed intracellular acidification. For the first time, we identified ion exchangers that function as active pH_i regulators in Xenopus gastrula cells. Interestingly, the presence of active acid extruders, namely the Na⁺/H⁺ exchanger and a Na⁺/HCO₃⁻independent mechanism, failed to prevent the intracellular acidification. This apparent contradiction was resolved by the finding that the pH_i set points of these two exchangers are developmentally regulated and shifted to increasingly acidic pH_i values, permitting intracellular acidification. In this context it is not surprising that inhibition of the activity of the Na⁺/H⁺ exchanger by DMA failed to affect development (unpublished results; Sater et al., 1994). This inhibition will facilitate the endogenous acidification process rather than counteracting it. These data indicate that an acidic shift in the pH_i set points of these alkalizing ion exchangers passively permits the acidification as recorded in Xenopus gastrula ectoderm explants.

The question remains: which cellular processes are involved in actively acidifying ectoderm cells during gastrulation. One possibility is that the metabolic rate increases leading to increased acid production (Busa and Nuccitelli, 1984). Another possibility is that activation of the Na⁺ independent HCO₃⁻/Cl⁻ exchanger leads to a strong acidification of the cytoplasm, even overruling the activities of the alkalizing ion exchangers (Bierman et al., 1988; Ganz et al., 1989). The question of whether either or both of these processes are involved in actively acidifying gastrula tissues will be the subject of further investigation. Clearly, in explants of Xenopus gastrula cells, developmental regulation of the operating range of alkalizing mechanisms, especially the Na⁺/H⁺ exchanger, by an acidic shift in their set points is an important permissive mechanism, which allows acidification to occur.

These investigations were supported by the Netherlands Foundation of Life Sciences (SLW Grant 417.441) which is subsidized by the Netherlands Organisation for Scientific Research (NWO). A.J.D. also acknowledges support from the EEC BIOTECH Program: contract no. PL 920060. The authors with to thank A. P. Otte for critically discussing the manuscript, S. Godsave and J. Hendriks for their indispensable help with whole-mount in situ hybridisation, R. Ameerun, for helping with taking the video images on the SPEX microscope. Many thanks also to W. Hage whose help was indispensable for preparing the false colour pH_i images. The authors wish to thank the referees for valuable comments on the manuscript.