Cleavage-arrested cell triplets from ascidian embryo differentiate into three cell types depending on cell combination and contact timing

doi:10.1113/jphysiol.2001.013293

Cleavage-arrested cell triplets from ascidian embryo differentiate into three cell types depending on cell combination and contact timing

Motoko Tanaka-Kunishima and Kunitaro Takahashi

Department of Medical Physiology, Meiji Pharmaceutical University, Noshio 2-522-1, Kiyose, Tokyo MZC204-8588, Japan

ABSTRACT

Top
Abstract
Introduction
Methods
Results
Discussion
References

During early ascidian development, which is a prototype of early vertebrate development, anterior neuroectoderm cells (a4.2) from the eight-cell embryo are destined to become anterior neural structures including the brain vesicle, while presumptive notochordal neural cells (A4.1) become larval posterior neural structures including motoneurons. Whereas, an anterior quadrant cell (A3) of the four-cell embryo, from which both anterior neuroectoderm (a4.2) and notochordal neural cells (A4.1) are derived, has both fates. Cleavage-arrested cell triplets were prepared from the anterior quadrant cell and a pair of anterior neuroectoderm cells (A3-aa triplet) or a pair of presumptive notochordal neural cells (A3-AA triplet), and cultured in contact. Differentiation of cells in the triplet was determined electrophysiologically by observing cell type-specific currents. In the A3-aa triplet, when two neuroectoderm cells and an anterior quadrant cell were prepared from the same batch of embryos, all three cells in the triplet developed into neuronal cells in 60 % of cases, but in 40 % of cases all of them differentiated into epidermal cells. However, when the batch of embryos from which neuroectoderm cells were prepared was fertilized 3 h later than that from which the anterior quadrant cell was prepared all three cells in the triplet consistently became neuronal cells. In contrast, when the batch of embryos from which neuroectoderm cells were prepared was fertilized 3 h earlier, all three cells became epidermal. In the A3-AA triplet no switching of differentiation occurred and all three cells in the triplet differentiated into neuronal cells, although the amplitude of inward current was often small. In neuralized A3-aa triplets the spikes in the anterior quadrant cell were characteristically small in amplitude and brief in duration, suggesting the presence of A-currents, which is a characteristic feature of posterior neuronal differentiation. In contrast, the spikes in the anterior neuroectoderm cells were large in amplitude and long in duration, chracteristic to the anterior neuronal type. The majority of single isolated anterior quadrant cells became non-excitable. However, the minority was apparently autonomously neuralized to become the posterior neuronal type. In neuralized A3-AA triplets, the majority of anterior quadrant cells was induced to become the anterior neuronal type. When isolated anterior quadrant cells were neuralized with subtilisin, a protease, they also predominantly became the anterior neuronal type. While, in medium containing a fibroblast growth factor posterior neuralization of isolated anterior quadrant cells was facilitated, but the anterior neuronal type, although minor, appeared anew. These observations indicate that the multiple fates of the anterior quadrant cell expressed in vivo were effectively reproduced in this experimental condition at the single cell level. Interactive differentiation in this triplet system recapitulates not only fundamental neural induction of ascidian neuroectoderm cells, but also functional and positional specificity within the neuronal group.

(Received 19 September 2001; accepted after revision 4 January 2002)
Corresponding author M. Tanaka-Kunishima: Department of Medical Physiology, Meiji Pharmaceutical University, Noshio 2-522-1, Kiyose, Tokyo MZC204-8588, Japan. Email: motokokt{at}my-pharm.ac.jp

INTRODUCTION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Cellular differentiation of an embryonic cell is determined either autonomously by cytoplasmic factors transferred through lineage from the oocyte or inductively by factors which are produced by neighbouring embryonic cells. Neurogenesis in the vertebrate embryo is initiated by the commitment of the dorsal ectoderm, under the inductive influence of the mesoderm, to form the neural plate at the early gastrula stage (Kessler & Melton, 1994; Hemmati-Brivanlou & Melton, 1997; Sasai, 1998), and succeeding regional specification within the CNS, such as anteroposterior or dorsoventral specification, is performed also by various inductive processes (Doniach, 1995; Jessell & Sanes, 2000). Thus, in early vertebrate neurogenesis, inductive differentiation is dominant.

Although an ascidian tadpole larva, a protochordate, has a simplified body organization, its tubular nervous system and notochord are identical to those of vertebrates (Satoh, 1994). Thus, the ascidian embryo has been considered to be a prototype of vertebrate embryos (Meinertzhagen & Okamura, 2001). On the other hand, at the beginning of 20th century, the cell lineage studies by Conklin revealed a fate-map on the ascidian egg and provided a classical example of the mosaic egg, which shows autonomous differentiation (Conklin, 1905b). Whittaker (1973), who provided the evidence of cytoplasmic factors in the ascidian embryo, showed that the cytochemical and micromorphological characteristics of striated muscle fibres in an ascidian tadpole were expressed in presumptive muscular cells, when cleavage of the embryo was arrested in the 8- to 64-cell stages. This report has been further confirmed by examining other characteristics such as various aspects of electrical excitability (Takahashi & Yoshii, 1981). Distinct forms of excitability appear in the respective cleavage-arrested cells as the result of differential expression of Na⁺, K⁺ and Ca²⁺ channels in their membranes (Takahashi & Okamura, 1998).

In the ascidian embryo, Rose (1939) and Reverberi & Minganti (1946) reported that inductive effects of presumptive notochordal regions were required to generate neural tissues, as in vertebrate neurogenesis. Following these classical observations Okado & Takahashi (1990a) first identified autonomous differentiation of cells isolated from cleavage-arrested 1- to 16-cell ascidian embryos. They then demonstrated alteration of their developmental fates by contact with another type of cells (Okado & Takahashi, 1990b). An isolated anterior neuroectoderm (a4.2) cell from the eight-cell embryo autonomously develops long-lasting Ca²⁺-dependent action potentials characteristic of epidermal differentiation. However, when it is cultured in contact with a presumptive notochordal neural (A4.1) cell located in the anterior vegetal region of the eight-cell embryo, the same cell develops a typical Na⁺ spike by expression of Na⁺ and delayed rectifier K⁺ channels characteristic of larval neurons (Okado & Takahashi, 1990b; Okamura et al. 1994; Ono et al. 1999). This unique two-cell system enables us to examine various aspects of inductive differentiation at the cellular level (Takahashi & Okamura, 1998).

However, in contrast to vertebrate embryos, it has recently been shown in the ascidian embryo that even when the inductive influence triggers cell differentiation, some developmental fates are partially autonomous and determined by cytoplasmic factors, as shown in the case of differential neural development between anterior ectoderm (a4.2) cells and posterior ectoderm (b4.2) cells (Hudson & Lemaire, 2001; Ohtsuka et al. 2001) and in the case of differentiation of mesodermal tissues (Kim et al. 2000). Further, while the presumptive anterior neural structures in the head of the tadpole larva are included in a4.2 cells, the posterior neural regions, such as the spinal cord, are included in the A4.1 cells (Okada et al. 1997). In limited cases, A4.1 cells apparently autonomously differentiate into cells which seem to be of the posterior neuronal type, different from the anterior neuronal type expressed in a4.2 cells (Okada et al. 1997).

It is important to ask quantitatively to what extent ascidian neural differentiation is inductive or autonomous for further studies of this system as a prototype for vertebrate neurogenesis. It is noted that an anterior quadrant cell of the four-cell embryo (A3) includes both presumptive anterior and posterior neural regions and the epidermal region for a tadpole larva, as well as the neural inducer region at the early developmental stages, i.e. a4.2 plus A4.1 (Conklin, 1905a; Nishida, 1987; Table 1). In the present experiments, we aimed to use this multi-fated A3 cell and to direct differentiation selectively and inductively under the condition of mixed cytoplasmic factors by cleavage arrest. We prepared a cleavage-arrested A3-aa cell triplet from an A3 cell and a pair of anterior animal a4.2 cells or an A3-AA cell triplet from an A3 cell and a pair of anterior vegetal A4.1 cells (Table 1). And we investigated the neural or epidermal differentiation with electrical markers characteristic of neuronal functions. Parts of the present study were reported elsewhere in abstract form (Tanaka-Kunishima & Takahashi, 1999).

METHODS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Materials

Artificial seawater, Jamarin U (Jamarin Laboratory, Osaka, Japan), was used for both culturing ascidian embryos and experiments. The major ionic composition of Jamarin U is (mM): Na⁺ 409, K⁺ 9, Ca²⁺ 10, Mg²⁺ 47, Cl^- 482.3, SO₄^2- 23.7 and HCO₃^- 2.3. Cytochalasin B (CB), used to arrest cleavage, was purchased from Aldrich and was prepared as a 2 mg ml^-1 stock solution in dimethyl sulphoxide. We added it as a one-thousandth part to seawater to arrest cell-cleavage, and as a ten-thousandth part to maintain the cleavage-arrested condition. We used basic fibroblast growth factor (bFGF) and serine protease, subtilisin, as neuralizing reagents for ascidian embryonic ectoderm cells (Okado & Takahashi, 1993; Inazawa, et al. 1998). Recombinant bovine bFGF and extracted bovine eFGF were purchased from Progen Biotechnik, Heidelberg. Subtilisin was purchased from Boehringer-Mannheim Biochemicals.

Nomenclature of embryonic cells

Ascidian embryonic cells were named following the designation system first adopted by Conklin (1905a). Cells of the animal hemisphere are designated by lower case letters, either a or b, and those in the vegetal hemisphere by capital letters, either A or B. At the four-cell stage, one embryo is composed of two pairs of different cells, A3 and B3. 'A' and 'a' mean anterior cells, and 'B' and 'b' posterior cells. Suffix numbers illustrate the generation and position within a quadrant of a hemisphere.

Preparation of cell pairs and cell triplets

Embryos of Halocynthia aurantium were used in this study. We selected healthy large adult ascidia to obtain embryos to perform the experiments. Adults were acquired at Wakkanai in northern Japan and maintained in an aquarium with circulating seawater at 3 °C. The animals were immobilized by chilling with seawater at 4 °C. Their oesophageal ganglia, which are parts of the CNS in adult ascidia, were gently crushed with forceps. Then, the oocytes were obtained by dissecting out gonads, and were kept in seawater at 9 °C for 3-10 h for maturation. Eggs were fertilized with sperm obtained from another animal and reared at 9 ± 0.5 °C in filtered Jamarin seawater. Under these conditions, embryos reached the eight-cell stage at 5 h after fertilization. Subsequently the development at 9.0 °C proceeded through blastula between 32- and 76-cell stages at about 10 developmental hours (DHs), the gastrula, which was after the 110-cell stage, at 10-20 DHs, the neurula around 20-30 h, the tailbud at 30 h, and hatching at 60 h.

At the late four-cell or the early eight-cell stage, embryos were transferred into filtered seawater containing cytochalasin B (2 µg ml^-1, Aldrich) to arrest further cleavage and to obtain the cells separated from the four- or eight-cell embryo. The follicular envelope and chorion were removed manually with a pair of sharp tungsten needles, and single cells or cell pairs were separated using a fine glass needle. According to Conklin (1905a), four-cell embryos include two types of cells, two anterior A3 and two posterior B3, and eight-cell embryos consist of four different types of cells, two anterior-animal cells a4.2, two posterior-animal cells b4.2, two anterior-vegetal cells A4.1 and two posterior-vegetal cells B4.1. Embryos after the late four-cell stage showed an anteriorly inclined animal-vegetal axis and symmetry at the midplane. Thus, each type of the cell was easily identified by its size, color and relative position in the embryo.

The A3 cells and a4.2 and A4.1 cell pairs were separated from four- and eight-cell embryos, respectively, with fine glass needles and kept in filtered seawater containing 2.0 µg ml^-1 cytochalasin B for 1 h. And they were transferred into sea water containing 0.2 µg ml^-1 cytochalasin B and flowing over 1.0 % agarose-coated microwells as described previously (Okado & Takahashi, 1990a). Then, various sets of three cells (triplets) were prepared by manipulating a pair of a4.2 cells (A3-aa triplet) or a pair of A4.1 cells (A3-AA triplet) in contact with an A3 cell in a microwell using a fine glass needle after 6-8 DHs, and then continuously cultured at 9.0 ± 0.5 °C. In addition, a pair of a4.2 and A4.1 cells (aA pair) was incubated as a control for neural differentiation, and a cell pair of a4.2 and b4.2 (ab pair) or a cell pair of a4.2 and a4.2 (aa pair) was incubated for epidermal differentiation. Cytochalasin B arrests cleavage of cells by preventing cytokinesis, but does not inhibit nuclear division (Schroeder, 1978). Cytochalasin B treatment allows expression of certain cell differentiation markers, such as tissue-specific enzymes and membrane excitability (Whittaker, 1973; Hirano et al. 1984).

We varied the timing of manipulated contact between an A3 cell and a pair of a4.2 cells to determine conditions under which the A3-aa triplet differentiated into neural or epidermal cells.

Microscopy and imaging

The prepared cell pairs or triplets were cultured in the bath paved with transparent agar containing 0.08 µg ml^-1 cytochalasin B and continuously superfused with circulating Jamarine seawater. The bath was set on the stage of the epifluorescence microscope (Olympus IMT-70) to avoid further cell transfer and to be always ready for observation and recording during development. We monitored the cell conditions and the expressed fluorescence with cCCD camera (Hamamatsu photonics Model c3140) equipped on the microscope throughout experiments. Images were stored on magneto-optical discs. Visualization and analysis of images were later performed with NIH Image or Scion Image software.

Determination of differentiation types

We confirmed the final differentiation types of the cells using electrophysiological techniques or by observing their morphological features.

Differentiation types of a4.2 or A3 cells in aA, A3-aa or A3-AA were identified by their characteristic ionic currents recorded using conventional voltage-clamp procedures (Hirano et al. 1984; Okado & Takahashi, 1990a). Under voltage-clamped conditions Na⁺ and A-type K⁺, or delayed rectifier K⁺ currents represented neurally differentiated cells, and inward rectifier K⁺ and Ca²⁺ currents accompanied with neither Na⁺ nor delayed rectifier K⁺ currents represented epidermally differentiated cells.

Identification of Na⁺ or Ca²⁺ or delayed-K⁺ currents in A3, a4.2, and A4.1 cells has been confirmed by substitution of external medium for those containing tetramethyl ammonium ions or Mn²⁺ and Mg²⁺ ions, as described in previous papers (Okado & Takahashi, 1990a, b; Okamura & Shidara, 1990; Sidara & Okamura, 1991; Okada et al. 1997; Inazawa et al. 1998; Ono et al. 1999). Molecular identity of ascidian neuronal Na⁺ channels, delayed rectifier K⁺ channels and A-type K⁺ channels expressed in cleavage-arrested cells was previously determined as TuNaI, TuKv2 and TuKv1, respectively (Okamura et al. 1994; Ono et al. 1999). It is well established in Halocynthia embryonic cells that neuronal differentiation can be identified when the TuNaI-derived Na channel currents or Na⁺ spikes are observed electrically, because TuNa I is the Halocynthia Na⁺ channel gene which is proved to be specific only for neurons in the tadpole larva with in situ hybridization studies (Okamura et al. 1994; Okada et al. 1997; Takahashi & Okamura, 1998) and electrical and zygotic Na⁺ channel expression in the cleavage-arrested and neuralized cells is blocked by antisense cDNA of the TuNaI gene (Okamura et al. 1994; Inazawa et al. 1998; Ono et al. 1999).

Ion channel currents used for quantitative estimation of channel expression were obtained by subtracting leak currents. All electrophysiological recordings were analysed mainly by pCLAMP 8 software (Axon Instruments Inc, Foster City, CA, USA). Characterization of epidermally differentiated cells could be supported by their electrical properties namely inward rectifier K⁺ and Ca²⁺ channel currents that were recorded under voltage clamp to -120 or +20 mV (Hirano & Takahashi, 1984). However, in fully differentiated epidermal cells the transparent tunic coat on the cell surface prevented the successful electrical recording. Therefore, in most of cases we identified the epidermal type by observing the tunic coat on the cell after 60 DHs (Hirano & Takahashi, 1987).

Measurement of gap junctional communication (GJC)

The gap junctional (GJ) conductance between cells in a cell pair or in a cell triplet was estimated from the current recordings under voltage-clamp conditions.

To quantitatively measure GJC (electrical GJ conductance, G_j), the double voltage-clamp method, as performed by Spray et al. (1981), or fluorescent dye injection into a cell, as used by Saitoe et al. (1996), were employed. We estimated G_j under voltage-clamp conditions by analysing the exponentially rising capacitative current in response to rising ramp potential changes or residual exponentially decaying current in response to a step potential change or after the end of falling ramp potential change (Fig. 4Bb and c). These currents were all derived from the currents charging the neighbouring cell-membrane capacity from the voltage-clamped cell through the GJC. The ramp potential change had a gradient of 2.0 mV ms^-1, and was applied from a holding potential of -75 mV for 10 ms until the end of the slope, keeping the test potential changes within ± 20 mV. Care was taken not to evoke the potential-dependent conductances within the range of the ramp potential change. To precisely estimate the G_j, the extracapacitative current charging the neighbouring cell capacity was preferably saturated within 10 ms. The time constants of the residual current ranged from 2 to 5 ms in the majority of cases. In the case of larger time constants less residual currents were observed. In the case of larger GJ conductances (above 1.0 µS) the time constant was frequently less than 2 ms and the exact estimation of the residual current was difficult. In those cases an additional recording electrode was introduced in the coupled neighbouring cells and electrical connectivity was confirmed by voltage clamping. In the case of smaller GJ conductances (below 0.1 µS) measurement of the residual current was difficult because of the small amplitude. In those cases, the electrical separation was confirmed with the additional recording electrode in the neighbouring cell.

RESULTS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Differentiation of the A3-aa triplet into neural cells

An anterior quadrant (A3 cell) of the ascidian four-cell embryo is by nature a multi-fated precursor for various types of cells of the ascidian tadpole larva (Table 1). However, when isolated, cleavage-arrested and cultured, it occasionally showed neural excitability characteristics with expression of A-type K⁺ channels as a single cell, though it became a non-excitable cell in most cases (Okado & Takahashi, 1990a). A-type K⁺ currents were previously suggested to be a characteristic property of the posterior neuronal type (Okada et al. 1997). Then a question arises whether the multi-fated precursor can be manipulated to differentiate more selectively into various neural or other types of cells by specified interaction with anterior neural ectoderm (a4.2) cells, located in the anterior animal region of the eight-cell embryo (Table 1), as it occurs in the early embryonic environment.

When the cleavage-arrested A3 cell was cultured in contact with a pair of cleavage-arrested a4.2 cells from the eight-cell embryo after 7 developmental hours (DHs) (equivalent to the 16-cell stage of a control embryo), all cells in the triplet became either neuronal or epidermal. Cell contact at earlier than 7 h tended to increase the cases of neuronal differentiation, although about 6 DHs was a lower limit because the eight-cell stage, at which a4.2 cells were prepared, was barely attained at 5 DHs at 9 °C.

When the neuronally differentiated A3-aa triplet was observed after 50 DHs Na⁺ spikes were expressed both in the a4.2 cells and the A3 cell as shown in Fig. 1A and B, respectively. At approximately 50 DHs those cells in a triplet usually showed intercellular gap junctional communication (GJC), as described later. However, as development proceeded, GJC was mostly decoupled, and the excitability type of individual cells could be determined. The spikes in two a4.2 cells showed a larger overshoot of up to 50 mV, a longer duration, and a more negative critical membrane potential than those in A3 cells. Under voltage-clamp conditions only the Na⁺ and delayed K⁺ currents were found in a4.2 cells, while in A3 cells the A-type K⁺ current was also found. The differences in the spike shape characterized the differential developmental fate of respective cells, though all three cells in a triplet became similarly neuronal cells.

View larger version
[in this window]
[in a new window]

Figure 1. Na⁺ spikes evoked by constant current stimulation in triplets
Traces were recorded in the a4.2 or A3 cell of A3-aa or A3-AA triplets at ~70 developmental hours (DHs). A, Na⁺ spike recorded in the a4.2 cell of a neuralized A3-aa triplet. B, Na⁺ spike recorded in the A3 cell of a neuralized A3-aa triplet. The spike showed a relatively high threshold and sharp fall. C, long-lasting action potentials recorded in the A3 cell of an epidermally differentiated A3-aa triplet. D, a regenerative response recorded in the A3 cell of an A3-AA triplet. Lower and upper sets of traces illustrate potential and current changes, respectively, in each record. The DHs of the triplet are indicated in respective records.

As described above, in one-third of cases all three cells in an A3-aa triplet became epidermal cells. In those cases, an A3 cell and two a4.2 cells were tightly electrically coupled after 50 DHs and showed a long-lasting Ca²⁺ action potential characteristic of epidermal type differentiation, as shown in Fig. 1C. For comparison, we observed electrical excitability of A3-AA triplets, which included two A4.1 cells, inducer cells from the eight-cell embryo. The triplets were coupled with GJC and showed mostly a type of incomplete regenerative responses. They did not exhibit epidermal characteristics, such as Ca²⁺ action potentials or the tunic coat, as shown in Fig. 1D. The details of this triplet will be described later.

In Fig. 2A, examples of current traces are shown under voltage-clamp of the neuralized cells in the A3-aa triplet. The voltage-clamped A3 cell at 52 DHs expressed relatively small Na⁺ inward currents at -15 mV as shown in Fig. 2A (upper panel) and the recorded currents were the sum of those in an A3 cell and a pair of a4.2 cells due to tight electrical coupling among the three cells. The current traces showed signs of A-type K⁺ channels. However, at later developmental stages, the decoupled A3 cells expressed a significant amount of A-type K⁺ currents and relatively less delayed K⁺ currents when the cell was clamped to +20 mV (Fig. 2A, lower left panel). While, large Na⁺ and delayed K⁺ currents were expressed in decoupled a4.2 cells of the same triplet at the same developmental stages (Fig. 2A, lower right panel). These Na⁺ and K⁺ currents were identical to those previously described in the neuralized a4.2 cells in the neurally induced aA pair from the eight-cell embryo (Okado & Takahashi, 1990b).

View larger version
[in this window]
[in a new window]

Figure 2. Current traces under voltage clamp in the neuralized A3 and a4.2 cells
A, traces from the neuralized A3-aa triplets at 52 and 101 DHs. Upper traces, current traces from a coupled A3-aa triplet at 52 DHs. Lower left traces, current traces from the decoupled A3 cell in an A3-aa triplet at 101 DHs. Lower right traces, current traces from the decoupled a4.2 cell in the same A3-aa triplet as illustrated in the lower left traces at the same time, 101 DHs. For the A3 cell, a depolarizing pulse to -15 mV evoked a Na⁺ inward current with an amplitude of 20 nA, and depolarizing to +20 mV evoked a typical A-type K⁺ (KA) current with an amplitude of 25 nA. For the a4.2 cell, a depolarizing pulse to -15 mV evoked a Na⁺ inward current with an amplitude of 65 nA, and a depolarizing pulse to +20 mV evoked a typical delayed K current with an amplitude of 70 nA. KA current was not observed in the a4.2 cell. B, traces from an isolated A3 cell at 81 DHs. Note that this A3 cell expressed a well-developed A-type K⁺ current (KA). C, inward and outward current expressed in the A3 cell of a coupled A3-AA triplet. For all records, the current traces for depolarizing pulses to -35, -15, and +20 or +25 mV are shown. D, Na⁺ and KA currents and Na⁺ and delayed K⁺ currents were illustrated in A4.1 and a4.2 cells of eight-cell aA pairs, respectively.

To illustrate developmental expression of Na⁺ and K⁺ channels in the neuralized A3-aa triplet, the peak inward current at -15 mV and the steady-state outward current at 200 ms after the onset of a potential step to +20 mV were plotted against DHs in Fig. 3A and B, respectively. Open symbols indicate data from decoupled triplets, and symbols filled with grey represent those from coupled triplets. The decoupling was defined as established when the GJ conductance from the A3 cell to the a4.2 pair became less than 0.1 µS.

View larger version
[in this window]
[in a new window]

Figure 3. Developmental changes in inward and outward currents in a4.2 and A3 cells of A3-aa triplets
A, peak amplitudes of inward currents at -15 mV. In both a4.2 and A3 cells Na⁺ inward currents began to be expressed suddenly around 50 DHs and tended to increase in amplitude during later development. B, developmental change of delayed rectifier K⁺ current at 20 mV expressed in a4.2 and A3 cells of A3-aa triplets. The delayed rectifier K⁺ current amplitude was measured at 200 ms after the onset of a 400 ms depolarizing pulse to +20 mV. In A and B, A3-aa-E represents A3-aa triplets in which all three cells differentiated into epidermal cells and were coupled to each other (n = 9). In the case of epidermalized cells the peak amplitude of inward current (Ca²⁺) at 20 mV was plotted. A3(coupled)-NE: a non-excitable A3 cell (n = 13). A3(coupled)-NA: a neuronally differentiated A3 cell in which the KA current was recorded (n = 6). A3(coupled)-ND: a neurally differentiated A3 cell in which Na⁺ and delayed rectifier K⁺ currents were recorded but the KA current was not (n = 10). A3(decoupled)-NA: an A3 cell in which coupling to the a4.2 pair disappeared and the KA current was recorded (n = 12). a(decoupled)-ND: an a4.2 cell in which Na⁺ and delayed K⁺ currents but not KA currents were recorded and coupling to the A3 cell disappeared (n = 7). Here and in subsequent figures, non-excitable cells were defined when the inward peak current amplitude at -15 mV was less than 1 nA, and the steady outward current component 200 ms after the command step to 20 mV was less than 1 nA. The clearly epidermalized cells judged electrically or with the presence of tunic coat were excluded from NE cells.

Data from the decoupled a4.2 cells in a neuralized triplet are illustated with open circles in Fig. 3A. The peak Na⁺ inward current at -15 mV was larger than 30 nA which exceeded that of about 10 nA observed in the decoupled A3 cell (indicated with open squares in Fig. 3A). The steady-state K⁺ outward current at +20 mV in the a4.2 cells was also larger than that in the A3 cell (Fig. 3B). Na⁺ currents and delayed rectifier K⁺ currents in the neuralized A3-aa triplet appeared after 45-50 DHs, and those ion channels seemed to increase abruptly after initial expression in both decoupled A3 and a4.2 cells (Fig. 3A and B). The developmental time courses in the decoupled a4.2 cells were not significantly different from those in the a4.2 cell of aA pairs from the eight-cell embryo reported previously (Okamura & Shidara, 1990; Shidara & Okamura, 1991).

In addition to decoupled neuralized triplets there was a significant number of coupled and neuralized triplets during the period between 50 and 80 DHs. At the beginning of ion channel expression, that is earlier than 50 DHs, coupled triplets showed smaller inward and outward currents, as shown in the traces of Fig. 2A (upper panel) and with filled grey circles around 50 h in Fig. 3A and B. However, coupled triplets at the later DHs apparently showed larger inward and outward current amplitudes. A large portion of the Na⁺ current recorded in a coupled A3 cell may originate from contacting a4.2 cells, which are coupled to the A3 cell via gap junctions. The enhanced outward currents in coupled triplets cannot be explained solely by GJC. At -15 mV no Na⁺ current was detected in either the a4.2 or A3 cells of the triplets that were epidermally differentiated (Fig. 3A, half-filled circles), as described later.

In summary, the cell-contacted triplet (A3-aa), which consisted of an anterior quadrant cell from the four-cell embryo (A3) and a pair of anterior neuroectoderm cells from the eight-cell embryo (a4.2-a4.2), differentiated into either the neuronal or epidermal cell triplet after culturing under the cell cleavage-arrested condition until the control larva hatched. In the case of the neuralized triplet, the anterior quadrant cell showed briefer and smaller Na⁺ spikes due to A-type K⁺ channel expression, while the anterior ectoderm cells exhibited longer and larger Na⁺ spikes due to absence of A-type K⁺ channels.

The development of A-type K⁺ currents in the neuralized triplet

Since the A-type K⁺ current characterizes the neural differentiation of anterior quadrant (A3)cells, we next studied the development of the A-type K⁺ component quantitatively and used it as a differentiation marker for the neuronal type.

When a decoupled A3 cell of the neuralized A3-aa triplet was voltage clamped at +20 mV, A-type K⁺ currents (KA current) of 20-60 nA were recorded (Fig. 2A, lower left panel). The KA current expressed in the A3 cell showed a peak within 50 ms after the onset of a step pulse above -15 mV. At -15 mV the KA current occurred simultaneously with the Na⁺ current although the delayed-rectifier K⁺ current was hardly activated and must be insignificant within 50 ms because of its relatively slow rise time at this potential level (Shidara & Okamura, 1991). The outward current at -15 mV might also include the Ca²⁺-induced K⁺ current induced by influx of Ca²⁺. However, we think this possibility to be unlikely due to the small Ca²⁺ inward current and the slow rise and high threshold of Ca²⁺-induced K⁺ current in this experiment. The ratio of the amplitude of the inward current at 50 ms after stimulation to the peak amplitude of inward current at -15 mV was calculated for each A3 or a4.2 cell in the A3-aa triplet (Fig. 4A) and plotted against the DHs as shown in Fig. 5A.

View larger version
[in this window]
[in a new window]

Figure 4. Schematic illustration of estimation of the A-current and gap junctional conductance
A, as an indicator of the expression of A-current type K⁺ channels we estimated the ratio of current amplitude at 50 ms from the onset of command pulse to -15 mV (red or blue arrow) to the peak amplitude of inward current at -15 mV. As shown in the red trace obtained from the neuralized a4.2 cell of an aA pair, the Na⁺ current without contamination of the KA current showed the ratio of ~0.25, while in the Na⁺ current contaminated with the KA current (blue trace), the ratio was less than 0.25 and mostly became negative. Ba illustrates the electrical equivalent circuit of the triplet in which all cells communicated with each other by gap junctions. The a4.2 pair or A4.1 pair was lumped electrically as cell-2. Assuming that ramp wave as VC command for cell-1ended at t₀ (red arrow) as shown in Bb,
eqt3
By simply solving the differential equations,
$tau$ _Gj, I(t right t₀), I(t₀ ê t) and C₁ can be directly measured at the point of red arrow, knowing $alpha$ = 20 mV/10 ms. Assuming C₂ = C₁ (mostly $gamma$ = 1) because of the geometrical considerations of the cell triplet, G_j can be estimated. I(t) (t > t₀) was obtained by exponential fitting of the extra-capacitative current after the end of the ramp command (blue curve). Fitting was executed by using the data after the point indicated by blue arrow. Similar estimation of Gj could be performed by the extra-capacitative current at the beginning of the step pulse, as shown in Bc.

View larger version
[in this window]
[in a new window]

Figure 5. Developmental change of the KA current in A3-aa triplets
A, in neuralized cells Na⁺ and delayed K⁺ currents were expressed as well as KA currents. However, the delayed K⁺ current was not prominent and showed a slower rise at -15 mV. Thus, the KA current component was approximately estimated by the ratio of inward current amplitude 50 ms after the onset of the depolarizing pulse to -15 mV to Na⁺ inward current peak amplitude at -15 mV. The ratio was plotted against DHs. B, the KA current was expressed during neural differentiation in the A3 cell of A3-aa triplets. The KA peak amplitude measured within 20 ms after the onset of depolarizing step to +20 mV was plotted against DHs. The KA current was recorded in both coupled and decoupled A3 cells. It was also observed in the a4.2 pair coupled to an A3 cell, in which the KA current expressed in the A3 cell was detected through gap junctions. Meanings and plotted data of A3-aa-E, A3(coupled)-NE, A3(coupled)-NA, A3(decoupled)-NA, A3(coupled)-ND, and A3(decoupled)-NA and a(decoupled)-ND are the same as in Fig. 3. In A, A3-NE was omitted because the ratio was undetermined. a4.2(8c-aA) represents neuralized a4.2 cells of aA pairs. The KA current component was predominantly expressed during 48-100 DHs in the A3 cell that was coupled to the a4.2 pair in the A3-aa triplet. It was observed in the A3 cell that was decoupled from the a4.2 pair, but the amplitude was not as large as that in the coupled A3 cell.

The ratio was above 0.25 for the cells that expressed only Na⁺ currents and no KA currents, such as for the decoupled a4.2 cell (open circles). While, the ratio for the A3 cell, either decoupled (open squares) or coupled (filled grey squares), was less than 0.25, if KA currents were expressed, and became significantly negative in most cases, indicating that the outward KA current was already present at the -15 mV level. Thus, the ratio is a simple indicator to illustrate whether the cell expresses the KA current. In Fig. 5A the ratio calculated for the neuralized a4.2 cell in the aA pair from the eight-cell embryo was also plotted, indicating that the channel expression in a4.2 cell was identical between two types of neural induction. In addition, we simply plotted the peak outward current at 20 ms after the onset of command pulse to +20 mV in Fig. 5B. It was aimed to illustrate approximate amounts of KA channel expression. At the 20 mV level the delayed rectifier K⁺ current was not significant up to 20 ms (Shidara & Okamura 1991). And the possible contamination of Ca²⁺-induced K⁺ currents must be negligible because of their slower rise and relatively less contribution (less than one-fifth of the delayed rectifier K⁺ current) to the steady outward current at +20 mV. All cells, which had zero or negative ratios in Fig. 5A, showed significant expression of KA channels in Fig. 5B.

As shown in Fig. 5B, not several A3s that were electrically coupled to a4.2 cells in triplets expressed relatively large KA currents from 50-90 DHs. The expression tended to decrease at later stages, indicating possible maximum expression before 80 DHs. It is also noted that A3 cells, which had already decoupled, expressed less KA currents than coupled A3 cells at the same DHs, and that KA currents were expressed only in A3 cells in the neuralized A3-aa triplet. We suggest that the expression of KA channels in A3 cells is facilitated by coupling and might have a regulatory role in the neuronal development of these cells.

According to previous reports on the cleavage-arrested cells, A-type K⁺ channels are expressed not only in A4.1 cells but also in posterior animal b4.2 cells of the eight-cell embryo, when the cells are neurally induced by cell contact with A4.1 cells or by subtilisin (Okado & Takahashi, 1990b, 1993). It is possible that this expression of A-type K⁺ channels corresponds to the fact that descendants of b4.2 cell-lineage have the fate to become posterior neural structures, such as larval spinal cord or the sensory cells in posterior epidermis (Nishida, 1987; Hudson & Lemaire, 2001; Ohtsuka et al. 2001). Recently it has been shown that the sensory cells are also induced by cell contact with vegetal A4.1 or B4.1 cells (Ohtsuka et al. 2001). And it is certain that anterior neuralized a4.2 cells do not express A-currents in response to any neuralizers, such as cell contact, subtilitisin and bFGF (Okado & Takahashi, 1990b, 1993; Inazawa et al. 1998). Thus, considering these previous reports and the above-described KA expression in A4.1 and A3, it is plausible that A-current spikes or the Na⁺ current plus the A-type K⁺ current are the terminal differentiation marker for posterior neurons in the ascidian tadpole (Table 1).

In summary, we classed A-type spikes, which appeared in anterior quadrant (A3) cells, as the posterior neuronal type and the spikes which appeared in anterior neuro-ectoderm cells (a4.2) as the anterior neuronal type. Correspondingly motoneurons are reported to be major posterior neuronal structures derived from the notochordal and posterior neural precursor (A4.1) cells in the eight-cell embryo by Okada et al. (1997). And the brain vesicle in the head of ascidian tadpoles is known to be the major anterior neural structure derived from anterior ectoderm (a4.2) cells (Nishida, 1987). Here, it should be noted that the A3 cell is the parent cell of both a4.2 and A4.1 cells (Table 1).

Developmental changes of the gap junctional (GJ) conductance in the triplets

In the present experiments, we used electrical measurements of ion channel currents for determination of differentiation types of cells. The ion channels are the most useful markers for functional expression of excitable cells, such as neurons and epidermal cells, in the ascidian tadpole (Takahashi & Okamura, 1998; Takahashi & Tanaka-Kunishima, 1998). However, the electrical measurement of a cell in the contacted cell triplet or pair may be significantly influenced by the presence of GJ conductance within the triplet or pair. Therefore, before determining the ion channel expression of a specific cell, we measured quantitatively the GJ conductance between the concerned cell and the cells that were surrounding it, as described in Methods and in the legend of Fig. 4B.

The GJ conductance between the A3 and a4.2 cell pair in an A3-aa triplet decreased during neural differentiation, as shown in Fig. 6A. GJ conductances of 0.8-1.2 µS were maintained up to 50 DHs. After 50 DHs the population of decoupled triplets increased. Taking measurement errors into account, a gap conductance of less than 0.1 µS was considered to indicate complete decoupling. At around 50 DHs, most A3 cells were coupled to a4.2 cells, and the GJ conductance ranged from 0.5 to 1.0 µS. While, some A3 cells were decoupled (4 of 12 were decoupled at 50-56 DHs). In this situation, the Na⁺ current, which was expected to be expressed in the neuralized a4.2 cells, was recorded in the A3 cell through the gap junction. The GJ conductance of the neuralized A3 tended to decrease in the later development from 70 to 90 DHs, and GJ communication (GJC) almost disappeared at 100 DHs.

View larger version
[in this window]
[in a new window]

Figure 6. Developmental changes in gap junctional conductance
The gap junctional conductance (G_j) between the A3 and a4.2 or A4.1 cell pair was estimated from extracapacitative currents under voltage clamp. A, in A3aa triplets. Meanings and plotted data for A3-aa-E, A3(coupled)-NE, A3(coupled)-NA, A3(decoupled)-NA, A3(coupled)-ND, and A3(decoupled)-NA and a(decoupled)-ND are the same as in Fig. 3. We regarded that the coupling disappeared when the G_j became less than 0.1 µS. B, in A3-AA triplets. G_j more than 0.5 µS was observed in almost all A3 cells in A3-AA triplets. A few exceptionally decoupled A3 cells, of which G_j was less than 0.1 µS, were also illustrated. C, in aA pairs from eight-cell embryos. G_j was mostly less than 0.1 µS. A few exceptionally coupled a4.2 and A4.1pairs were also illustrated. For explanation, see text. a(coupled)-E: an epidermalized a4.2 cell coupled to an A4.1 cell. a(coupled)-NA: a neuralized A4.1 cell coupled to an a4.2 cell. a(coupled)-ND: a neuralized a4.2 cell coupled to a non-excitable A4.1 cell. A(decoupled)-NE: a decoupled and non-excitable A4.1 cell. A(decoupled)-NA: a decoupled and neuralized A4.1 cell with KA currents. a(decoupled)-ND: a decoupled and neuralized a4.2 cell without KA currents.

Among the A3-aa triplets examined, typing of neural differentiation must be carefully done considering those cases in which GJC between A3 and a4.2 cells did not disappear even after 70 DHs (Fig. 3A and B, grey-filled circles and squares). Following are three such cases which should be specifically commented on. First, when NA (squares)-type excitability was observed, the A3 cell in a triplet, at least, must have differentiated into the posterior neuronal type. Second, when ND (circles) was observed, it is possible that A3 cells were induced to become the anterior neuronal type by contact with a4.2 cells. In this case, the coupling of the cells of the same type, i.e. a4.2 cells, remained. in both these cases, we assumed that the a4.2 cells were either anterior neuronal type or poorly differentiated because they have never been induced to become the posterior neuronal type either by cell contact or by inducers, as reported previously (Inazawa et al. 1998). Third, if both A3 and a4.2 cells were non-excitable or poorly differentiated, the GJ conductance mostly remained (grey filled triangles).

As mentioned later, the GJ conductance in A3-aa that differentiated into epidermal cells did not diminish during development (Fig. 6A, half-filled circles). During neural differentiation of the triplet, disappearance of GJC between the A3 cell and a pair of a4.2 cells occurred first, as described above, and decoupling between two a4.2 cells seemed to occur subsequently (data not shown). This idea was also supported by the observation that the capacitance, measured in each cell, decreased in parallel with the increase in developmental hours.

In summary, to determine the excitability type of individual cells in the triplet, we always measured the GJ conductance between the concerned cell and neighbours. When the GJ conductance was below 0.1 µS, the cell was considered to be isolated. The better electrical isolation between cells with different neural types in the anterior quadrant cell and neuroectoderm cell triplet (A3-aa) could be obtained with advancement of terminal differentiation, but the decoupling between the cells of the same type seemed to be delayed. In the case of epidermal cells, the coupling remained.

Differentiation of an A3-aa triplet into epidermal cells

As described above, all three cells in the triplet, an anterior quadrant cell and two neuroectoderm cells (A3-aa), could also differentiate into epidermal cells (Fig. 1C). We next examined the characteristics of epidermal cells both by electrical and morphological techniques.

When four-cell embryos, from which A3 cells were separated, and eight-cell embryos, from which a4.2 cells were separated, were obtained from the same batch of embryos, which had been subjected to the same developmental conditions, 60 % of A3-aa triplets differentiated into neural cells and the remaining 40 % developed into epidermal cells. In the epidermalized triplet, the transparent tunic coat first appeared on the surface of the two a4.2 cells at about 50 DHs, and subsequently on the surface of the A3 cells (Fig. 7Ca and Cb, arrowheads). The tunic coats on both a4.2 and A3 cells became thicker during later development (Fig. 7Cb). The development of the tunic coat in cleavage-arrested cells was previously analysed in detail using a monoclonal antibody against the tunic substance (Hirano & Takahashi, 1987; Okado & Takahashi, 1990a).

View larger version
[in this window]
[in a new window]

Figure 7. Epidermally differentiated A3-aa triplets
A, current traces under voltage clamp in the A3-aa triplet, which differentiated epidermally. Left traces, Ca²⁺ current traces were obtained by depolarizing pulses to -35, -15 and +20 mV at 50 DHs. Three cells in the triplet were coupled to each other. Right traces, inward rectifier K⁺ (IRK) current traces were obtained by hyperpolarizing pulses to -100, -135 and -175 mV at 50 DHs in the same A3-aa triplet as shown in the left panel. B and C, bright field photographs of the epidermally differentiated A3-aa triplets at 50, 55 and 100 DHs, respectively. The blue arrows point to the tunic coat formed on the surface of the A3-aa triplets. B, the left photograph shows the same A3-aa triplet as illustrated in A. The right photograph shows the same photograph after the process of edge find. Ca and Cb, development of the tunic coat in another epidermalized A3-aa triplet. The same triplet was observed at different stages (55 and 100 h, respectively). At 100 h the tunic coat was distinct on the A3 as well as an a4.2 cell. Right photographs in the pair show the same photographs on the left after 'edge find' processing. The scale in Cb (100 µm) applies for all photographs.

Ca²⁺ currents and inward-rectifier K⁺ currents were observed in the coupled and epidermalized A3-aa triplet at 50 DHs, as shown in Fig. 7A. However, those currents were difficult to record with microelectrodes due to the tunic coat after 60 DHs (Fig. 7Ca and Cb). Their amplitudes may have increased afterwards compared with those detected at 50 DHs. The Ca²⁺ current showed the maximum peak amplitude around 20 mV, and the delayed K⁺ current was expressed neither in epidermalized A3 nor in a4.2 cells, as shown in Fig. 7A, when they were voltage clamped at +20 mV. Instead, the slowly increasing outward current shown in the trace at 20 mV was most likely to be the Ca²⁺-induced K⁺ current. These characteristics were previously reported in epidermalized cells (Hirano & Takahashi, 1984). The peak amplitudes of Ca²⁺ current and outward current at 20 mV and the gap junctional conductance in epidermalized triplets are illustrated in Figs 3A and B and 6A with half-filled circles, respectively. The low value of GJ conductance was probably due in part to the increased intracellular Ca²⁺ resulting from injury during penetration. The highly efficient intercellular dye coupling between epidermalized cells has been reported previously (Saitoe et al. 1996).

In summary, the characteristics of morphology, such as the tunic coat, and the electrical properties, such as Ca²⁺ channels and inward rectifier K⁺ channels, of the epidermalized anterior quadrant (A3) cell and neuroectoderm (a4.2) cells were identical with those in previously studied ascidian epidermalized cells.

Conditions in which A3-aa triplets develop into epidermal cells

Differentiation fate of cells in triplets consisting of an anterior quadrant cell and two neuroectoderm cells (A3-aa) was somewhat unpredictable, as described above. Therefore, we next determined the conditions, in which A3-aa triplets differentiate into either the neural or epidermal type.

In the following experiments, the type of finally differentiated cells was confirmed by electrophysiological characteristics of the cell membrane for the neural cell and tunic formation for the epidermal cells during the period between 80 and 100 DHs. In the experiment represented by the upper open bars in Fig. 8A, two anterior-animal a4.2 cells and one A3 cell in each triplet were obtained from the same batch of embryos and manipulated to be in contact at 7 DHs. Two-thirds of these triplets differentiated into neural cells, and the remaining one-third became epidermal cells. These control data were obtained from the same batches from which the triplets were prepared for the following developmental time shift experiments. When the batch of embryos from which the A3 cell was prepared was fertilized and subjected to development 3 h ahead of the batch from which the a4.2 cells were prepared, and the A3 cell was manipulated to be in contact with the a4.2 cells at 10 DHs (A3), all cells in the triplet differentiated into neural cells (Fig. 8A, dark shaded bars). Conversely, when the batch of embryos from which the A3 cell was prepared was subjected to development 3 h later than the batch of embryos from which the a4.2 cells were prepared, and they were manipulated to be in contact with the A3 at 7 DHs (A3), all cells in the triplet became epidermal except in one triplet (of 12 examined) which developed into neuronal cells (Fig. 8A, light shaded bars).

View larger version
[in this window]
[in a new window]

Figure 8. Relative percentage differentiation to epidermal or neural cells in A3-aa triplets
A, bars illustrate the percentage differentiation depending upon the developmental stages of the a4.2 cells relative to the A3 cell in a triplet. Dark shaded bars indicate cases in which the batch from which the a4.2 cells were obtained was 3 h behind in development from the batch from which the A3 cell was obtained. Light shaded bars indicate cases in which the batch from which the a4.2 cells were obtained was 3 h ahead in development from the batch from which the A3 cell was obtained. The numbers to the left or right of each bar indicate the sample number of triplets. Open bars indicate control cases, where the a4.2 and A3 cells were obtained from the same batch of embryos. These control data were obtained from the same selected batches from which the triplets were prepared for the above developmental time shift experiments. B, schematic illustration of the experimental procedures, by which cells at different developmental stages were manipulated to be in contact.

From the results mentioned above, it is clear that the timing of cell attachment in cleavage-arrested A3-aa triplets determined the cell fate. A3-aa triplets differentiated into the neural type if the pair of a4.2 cells were at the 16-cell stage when they were manipulated to be in contact with A3. All A3-aa triplets differentiated into the epidermal type, when the a4.2 cell pair at the time of contact had developed further than the 116-cell stage (Fig. 8B) and possibly committed to the epidermal fate, according to the previous results in aA pair induction (Okado & Takahashi, 1990b, 1993; Inazawa et al. 1998).

In summary, the neural fate of the anterior quadrant (A3) cell was altered to the epidermal type by contact with neuroectoderm (a4.2) cells in spite of apparently autonomous neuralization of the isolated anterior quadrant (A3) cell, as described later, when the contacting ectoderm cells were committed to the epidermal fate. The developmental stage of the contacting anterior quadrant cell was apparently not critical for the epidermal commitment of the triplet.

Differentiation of an A3-AA triplet

To determine the conditions under which the anterior quadrant cell commits to fates other than epidermal or posterior neural, we examined another type of cell triplet (A3-AA) which consisted of the anterior quadrant (A3) cell from a four-cell embryo and a pair of neurally competent and presumptive notochordal (A4.1) cells located in the anterior vegetal region of an eight-cell embryo (see Table 1).

An anterior quadrant (A3) cell and two presumptive notochordal-neural (A4.1) cells were prepared from the same batch, cleavage arrested and manipulated to be in contact at about 6 DHs. All three cells developed ion channel currents more slowly than cells in A3-aa triplets and differentiated into neural cells, judging from weak but definite expression of Na⁺ inward and delayed-rectifier K⁺ outward currents. The currents were small in amplitude, and the electrical excitability presented as incompletely regenerative responses, which were different from those in an A3-aa triplets at similar DHs, as shown in Fig. 1D. The Na⁺ and delayed-rectifier K⁺ outward currents were first detected at ~50 DHs. The amplitude of Na⁺ current at -15 mV did not further increase and was below ~10 nA throughout the embryonic development, and the development of A-type K⁺ channels was rarely observed (Fig. 2C, 9A and B). These characteristics were different from those in A3-aa triplets. The above results indicated that the neural expression or the development of an A-current in A3-AA triplets was relatively suppressed in comparison to those in A3-aa triplets. Further, in an exceptional batch (9 cases plotted at ~55 DHs in Fig. 9A and B), all of nine A3-AA triplets were found to be completely neuralized, showing 2~39 nA of Na⁺ inward currents at -15 mV. This result was apparently in contrast to the common observation of incomplete neuralization in A3-AA triplets as described above. In this case, however, the differentiation was the anterior neuronal type in seven A3 cells as in the a4.2 cells in a neuralized A3-aa triplet. The A-current was observed only in the remaining two A3 cells (squares). Typical current traces obtained in an A3 cell from an A3-AA triplet are illustrated in Fig. 11B (right).

View larger version
[in this window]
[in a new window]

Figure 9. Developmental changes in Na⁺ inward and K⁺ outward currents expressed in cleavage arrested A3-AA triplets and isolated A3 cells
A, peak Na⁺ inward currents measured in A3-AA triplets. B, peak K⁺ outward currents measured in A3-AA triplets. In A and B, both recording and current pipettes were located in the A3 cell except in AA where A4.1 was voltage clamped. A3(coupled)-NE: non-excitable cells, n = 13. A3(coupled)-ND: no KA current, n = 19. A3(coupled)-NA: with KA current, n = 1. A3(decoupled)-ND: decoupled without KA current, n = 3. A3(decoupled)-NA: decoupled with KA current, n = 2. C and D, from cleavage-arrested isolated A3 cells. A3-NE: non-excitable cells, n = 9. A3-NA: with KA current, n = 5. In D, the peak amplitudes of KA currents at 20 mV were also illustrated for the cells that showed KA currents (KA-peak).

The characteristics of epidermal cells were never expressed in A3-AA triplets. Further, isolated and cleavage-arrested A3 cells, i.e. single A3 cells, frequently developed A-type K⁺ channels at later DHs, i.e. after 70 h (Fig. 2B and 9D).

The GJ conductance between the A3 cell and the A4.1 cell pair was not apparently reduced during incomplete neural differentiation (Fig. 6B). The GJ conductance of 0.5-1.5 µS was recorded between the A3 cell and the A4.1 cell pair at 50 DHs, which indicated that the gap junctional communication (GJC) was robustly established between the cells. At ~100 DHs, a GJ conductance of approximately 0.5 µS was recorded. Only in the exceptional completely differentiated cases as described above, were there decoupled triplets as shown in Fig. 6B. These results indicated that GJC partially decreased but only disappeared in exceptional cases. While in A3-aa triplets, GJC was significantly decoupled during late development from 50 to 100 h.

In summary, we conclude that the differentiation of cells in triplets consisting of an anterior quadrant cell and presumptive notochordal neural cells (A3-AA) was often incomplete in expression of ion channels and decoupling of GJC. Their commitment was not epidermal but neural. Cells in those triplets became predominantly the anterior neuronal in type.

Comparison of differentiation of cells in the a4.2-A4.1 (aA) pair with those in the A3-aa triplet

The cleavage-arrested anterior neuroectoderm (a4.2) cell and the presumptive notochordal neural (A4.1) cell pair separated from the eight-cell embryo have been established as the neural induction of the aA pair (Okado & Takahashi, 1990b). In this pair the neuroectoderm (a4.2) cell differentiates into an anterior neuronal cell that expresses Na⁺ and delayed K⁺ channels at about 40-45 DHs at 9 °C. We re-examined the differentiation types of both the neuroectoderm cell and presumptive notochordal neural cell in the pair to compare with those of the cells in the triplets constituted of an anterior quadrant cell and ectoderm cells. It is known that the isolated cleavage-arrested presumptive notochordal (A4.1) cell occasionally expressed A-type K⁺ currents (Okado & Takahashi, 1990a; Okada et al. 1997).

The A4.1 cell in the aA pair was also occasionally neuralized, just like the isolated A4.1 cells, the characteristics of excitability being different from those of neuralized a4.2 cells. The A4.1 cell expressed KA currents when it was voltage clamped at a depolarized potential of -15 or +20 mV, which was the same as that expressed in the A3 cell of neuralized A3-aa triplets (Fig. 2D). In the a4.2 cell of an aA pair, Na⁺ and delayed K⁺ channels were expressed quite similarly to those expressed in the a4.2 cells of the A3-aa triplet (Fig. 2D, lower right). In Fig. 10A and B, the developmental time courses of Na⁺ and delayed K⁺ channel currents in aA pairs were illustrated. Combined with the results reported previously (Okamura & Shidara, 1990; Shidara & Okamura, 1991; Okamura & Takahashi, 1993; Saitoe et al. 1996; Okada et al. 1997; Inazawa et al. 1998), these results indicate that the developmental time courses of both Na⁺ and K⁺ currents in the a4.2 cells from aA pairs are similar to those in a4.2 cells from A3-aa triplets. While, development of Na⁺ and KA currents in the A4.1 cell was virtually identical to that of the A3 cell in the A3-aa triplet.

F10 View larger version
[in this window]
[in a new window]

Figure 10. Developmental changes of Na⁺ inward and K⁺ outward currents in a4.2 and A4.1 of aA pairs
A and B, a(coupled)-E: epidermalized a4.2 cell coupled to a A4.1 cell, n = 1. a(coupled)-NA: an a4.2 cell coupled to a neuralized A4.1 cell, n = 1. a(coupled)-ND: a neuralized a4.2 cell coupled to a non-excitable A4.1 cell, n = 1. A(decoupled)-NE: a decoupled and a non-excitable A4.1 cell, n = 9. A(decoupled)-NA: a decoupled and a neuralized A4.1 cell exhibiting a KA current, n = 3. a(decoupled)-ND: a decoupled and a neuralized a4.2 cell which did not exhibit a KA current, n = 14.

GJC between a4.2 and A4.1 cells during a-A neural differentiation completely disappeared before 40-50 DHs (Fig. 6C), while Na⁺ and delayed K⁺ channels were expressed later in a4.2 cells as previously reported (Saitoe et al. 1996). Gj values of a4.2 cells which expressed Na⁺ currents became less than 0.1 µS at 50 DHs, save in a few exceptional cases in which cells were epidermalized or failed to differentiate neurally. It seemed that GJC between a4.2 and A4.1 cells was disconnected ahead of the neural expression of the a4.2 cell which is different from the case in neuralized A3-aa triplets.

In summary, the presumptive notochordal neural (A4.1) cell in the neurally induced pair from the eight-cell embryo occasionally showed posterior neural differentiation in which the A-type K⁺ current was identical to that in the neuralized anterior quadrant cell. There was no indication that neuralization of the presumptive notochordal neural (A4.1) cell was facilitated by the neuralized neuroectoderm (a4.2) cell.

The induced differentiation of isolated A3 cells with protease or bFGF

In the triplets consisting of a neuralized anterior quadrant cell and neuroectoderm cells (A3-aa), although both quadrant and ectoderm cells showed spike potentials, spikes in the quadrant (A3) cell were characteristically small in amplitude and brief in duration, suggesting the expression of A-type K⁺ currents characteristic of the posterior neuron type. However, in the anterior quadrant (A3) cell of the triplets constituted of an anterior quadrant cell and presumptive notochordal neural cells (A3-AA), the development of A-currents was considerably suppressed. Further, as described, in an exceptionally well-differentiated batch of A3-AA triplets, the anterior quadrant (A3) cells clearly showed the anterior neuronal type in seven of nine cells, just as in the neuroectoderm (a4.2) cells in the triplet consisting of a neuralized anterior quadrant cell and neuroectoderm cells (A3-aa). We then asked whether the anterior quadrant (A3) cell differentiates into the anterior neuronal type with other inducing agents.

When single A3 cells were neuralized with a protease, subtilisin, they always exhibited Na⁺ inward currents and delayed-rectifier K⁺ outward currents without A-currents with few exceptions, suggesting inductive differentiation of the anterior neuronal type, in contrast to the isolated and untreated A3 cells (Fig. 11A (lower) and B (middle)). While, when single A3 cells were cultured in the medium containing fibroblast growth factor (bFGF) after 7-8 DHs, they were considerably facilitated to differentiate into the neuronal type in comparison to untreated A3 cells (Fig. 11A (upper) and B (left)). The neuralized A3 cells showed an increased population of posterior neuronal type, and the anterior neuronal type differentiation was also observed anew (circles in Fig. 11A, upper). The population of non-excitable cells was reduced (triangles in Figs 11A (upper), 12B and 12D). However, the peak amplitude of Na⁺ currents of anterior neuronal type cells at -15 mV in the bFGF medium was not well developed compared with those treated with subtilisin, while the peak amplitude of KA currents of posterior type cells was well developed compared with that of non-treated isolated A3 cells. Thus, the neuralizing effect of bFGF seemed to be relatively potent for the posterior neuronal type, and was different from that of subtilisin, which was a potent anteriorizing inducer.

F11 View larger version
[in this window]
[in a new window]

Figure 11. Neuralization of isolated A3 cells with basic fibroblast growth factor, bFGF, or with a protease, subtilisin
A, developmental changes in peak amplitudes of inward current at -15 mV in isolated and neuralized A3 cells (left panels). Right panels, KA current peak amplitude measured within 20 ms after the beginning of a depolarizing step to +20 mV was plotted against DHs. Upper graphs indicate the responses of A3 cells neuralized by adding 1-30 ng ml^-1 bFGF in the case of low (L) doses and 70-100 ng ml^-1 bFGF in the case of high (H) doses to the medium at 7-8 DHs. Lower graphs indicate the responses of A3 cells neuralized by treating with 1 mg ml^-1 subtilisin for 30 min (9 °C) at 7-8 DHs. With both bFGF and subtilisin Na⁺ currents began to be expressed suddenly around 50 DHs and tended to increase in amplitude during the later period of development. The KA current was recorded in the majority of bFGF-treated A3 cells. ND without KA current in low dose bFGF-treated A3 cells, n = 3 and ND in high dose, n = 5. NA which showed the KA current in low doses, n = 9 and in high doses, n = 8. NE, non-excitable cells, in low dose, n = 8 and in high dose, n = 3. In subtilisin-treated A3 cells: ND, n = 33; NA, n = 6; NE, n = 17. B, current traces (upper traces) under voltage clamp of neuralized A3 cells. Voltage commands were -35, -15 and +25 mV (lower traces). Left, current traces from an A3 cell treated with high doses of bFGF (56 DHs). Middle, current traces from an A3 cell treated with subtilisin (60 DHs). Right, current traces from an A3 cell in an A3-AA triplet which showed fully developed Na⁺ and delayed rectifer K⁺ currents without the KA current, which is characteristic of an anterior type neuron (60 DHs).

The dose of subtilisin used, 1 mg ml^-1, was the saturated level for neuralization of a4.2 cells because half-saturation was obtained at ~0.05 mg ml^-1 (Okado & Takahahsi, 1993). Two levels of the bFGF dose were used: the low (half-saturated) 1-30 ng ml^-1, and the high (saturated) 70-100 ng ml^-1 dose. The half-effective dose for the neuralization of a4.2 cells was 8 ng ml^-1 (Inazawa et al. 1998). The high dose may have a higher neuralizing effect upon A3 cells than the low dose but the difference was not significant.

In summary, the isolated anterior quadrant (A3) cell was induced to differentiate into the anterior neuronal type with subtilisin, while bFGF fascilitated the neuralization of the anterior quadrant cell to become the default posterior neuronal type.

Comparison of differential development between anterior and posterior type neurons in the variously neuralized anterior quadrant cell (A3) population.

Figure 12 illustrates frequency histograms of differentiation into the anterior neuronal type or posterior neuronal type in variously neuralized A3 cells.

F12 View larger version
[in this window]
[in a new window]

Figure 12. Summary histograms of differentiation types in neuralized A3-aa and A3-AA triplets, and in isolated A3 cells neuralized by bFGF or subtilisin
A, data from A3 cell differentiation from non-epidermalized A3-aa triplets. The ND type was seen only in the coupled cases. For the context of illustrated data, see text. B, data from isolated A3 cells induced by bFGF. L indicates the low doses of bFGF (1-30 ng ml^-1) and H indicates the high doses (70-100 ng ml^-1). C, data from decoupled a4.2 cell differentiation from neuralized A3-aa triplets. D, data of isolated and non-treated A3 cells were illustrated as control. E, data from A3 cell differentiation from A3-AA triplets. F, data from isolated A3 cells induced by subtilisin (1 mg ml^-1). Data were all derived from those plotted in preceding figures. In all histograms, the ordinate is the number of cells, in which the type of differentiation was determined. ND: neural differentiation showing Na⁺ inward and delayed rectifier K⁺ outward currents, that is, the anterior neuronal type. NA: neural differentiation showing Na⁺ inward and A-type K⁺ outward currents, that is, the posterior neuronal type. NE: non-excitable cells which showed neither inward peak currents more than 1 nA at -15 mV nor outward currents more than 1 nA at +20 mV, as described in the legend of Fig. 3. The clearly epidermalized triplets were excluded from NE cells.

The data in Fig. 12 are all derived from data in previous figures, but exclude data from triplets examined before 48 HDs. In Fig. 12A, the histogram includes only the data from 37 triplets in the non-epidermalized population of the A3-aa triplets. Other clearly epidermalized populations judged electrically (6 triplets) or by the presence of the tunic coat were excluded. In other histograms in Fig. 12B-F, all examined data are included. Neurons of posterior type were defined by the presence of KA currents (NA in previous figures). Neurons of anterior type were defined by the presence of delayed rectifier K⁺ current with no KA current (ND in previous figures). Non-excitable cells, NE, were defined, when the inward peak amplitude at -15 mV was less than 1 nA and the steady outward current at 200 ms after the onset of potential step at 20 mV was less than 1 nA and when they did not show epidermal characteristics. They were thus mostly unsuccessfully differentiated neural cells, but other types of cell, such as notochordal cells, could be included with them. However, cells which were clearly of epidermal type were excluded from NE. In contrast neuralized a4.2 cells, a significant proportion of A3 cells differentiated into NE cells. This can be partly explained by the fact that A3 cells have multiple fates which are not limited to epidermal and neural types (Table 1)

Decoupled A3 cells in neuralized A3-aa triplets were consistently of posterior neuronal type, while among coupled A3 cells there were some cells presumed to be the anterior neuronal type even after 60 DHs (see also Fig. 3A and B, grey-filled circles and comments in the GJ conductance section). The total distribution of A3 differentiation types in non-epidermalized A3-aa triplets was quite similar to those A3 cells treated with bFGF (Fig. 12A and B). When single A3 cells were neuralized in medium containing bFGF, posterior neuronal type differentiation was relatively facilitated in comparison to that of isolated non-treated A3 cells especially in terms of the amplitude of A-currents (Fig. 9C, 9D, 11A (upper), 12B and 12D). However, it should also be noted that in the bFGF medium some anterior type neurons appeared with relatively small Na⁺ currents (Fig. 11A, upper). In neuralized A3-AA triplets, differentiation in A3 cells was mostly of the anterior neuronal type as in a4.2 cells in neuralized A3-aa triplets (Fig. 11B (left panel) and 12E). Further, when isolated A3 cells were neuralized with subtilisin, they consistently showed anterior neuronal type differentiation with a few exceptions (Fig. 12F).

In summary, the similarity of the anterior-type dominant distribution of the subtilisin-induced anterior quadrant cells to that of the anterior quadrant (A3) cells in triplets containing presumptive notochordal neural cells (A3-AA) was evident (Fig. 12E and F). There was also another similarity, that of the posterior-type dominant distribution of the bFGF-neuralized anterior quadrant cells to that of the anterior quadrant cell in the triplet containing neuroectoderm cells (A3-aa).

DISCUSSION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Binary switching between neuronal and epidermal differentiation in A3-aa triplets

The multi-fated anterior quadrant (A3) cell from a four-cell ascidian embryo was used to direct differentiation of single cells inductively and selectively under the condition of mixed cytoplasmic factors by cleavage arrest. To find inductive events by cell contact, cell triplets were prepared from an A3 cell and a pair of anterior ectoderm cells (a4.2-a4.2) or a pair of presumptive notochordal neural cells (A4.1-A4.1) from an eight-cell embryo. The terminal differentiation was identified electrically (summary in Table 2).

All three cells in an A3-aa triplet differentiated into either neural or epidermal cells depending on the timing of cell contact, i.e. shifting the developmental stages of the A3 cell or the a4.2 cell pair at the time of contact (Fig. 8). Thus, the A3-aa triplet is a unique preparation which can switch differentiation patterns within its own cell group by simply adjusting the developmental timetables of the member cells. These patterns are not usually observed in isolated conditions. Previous reports on cell contact experiments of cleavage-arrested cells from early ascidian embryos, such as two-cell, four-cell and eight-cell embryos, suggest that cell contact between the same type of cells, such as AB-AB, left and right cells from the two-cell embryo, or a4.2-a4.2 or b4.2-b4.2 from the eight-cell embryo, never produces new types of differentiation but rather only intensified the characteristics of the terminal differentiation which is expressed already under isolated conditions, such as epidermal characteristics (Hirano et al. 1984; Okado & Takahashi, 1990a,b, 1993). Heterogeneous cell contact reveals inductive effects as typically represented by the neural induction pair, a4.2-A4.1, from the eight-cell embryo. However, in these cases the interaction seems to be one-directional and not mutually interactive (Okado & Takahashi, 1990a,b, 1993). In the new cell triplet, A3-aa, A3 cells include the components homonymous with contacted a4.2 cells in addition to the heterogeneous A4.1 cell components. It is possible that this situation made the new cell triplet A3-aa mutually interactive and the induction within the triplet regulative and self-changeable. Further, it is inferred that the interaction occurring in the vertebrate ectoderm may be produced within partly homonymous and partly heterogeneous cell population due to gradients of various inducers.

Default status of developmental fate in the ascidian ectoderm

In early Xenopus embryos the dorsal ectoderm is predetermined to become neural cells, and the epidermal cells form only when a growth factor, such as bone morphogenetic protein (BMP4) activates the cells (Hemmati-Brivanlou & Melton, 1997). Thus, according to the recent popular hypothesis, newly found endogenous neural inducers, such as noggin, chordin and follistatin, bind to BMP4 to suppress the epidermalizing capacity (Wilson & Hemmati-Brivanlou, 1995, 1997). However those substances can induce only anterior neural structures, and it has been suggested that other inducers, such as bFGF, must be included for formation of the posterior neural structure or for the anteroposterior patterning (Cox & Hemmati-Brivanlou, 1995; Lamb & Harland, 1995; Pownall et al. 1996). Further, Hongo et al. (1999) have recently shown that the anterior neural structure in the Xenopus embryo also became severely defective when a type of bFGF receptors, XFGFR-4a, was blocked by injecting their dominant negative form.

Here, it is important to ask whether the default differentiation of neurally competent regions in ascidian embryos is neural or epidermal, and whether ascidian neural differentiation is inductive or autonomous. Actually, our ascidian neural induction study indicates that at least a4.2 cells follow autonomously the epidermal fate at the default state (Okado & Takahashi, 1990a,b) and that bFGF alone induces a4.2 cells to become anterior type neurons (Inazawa et al. 1998). Thus, in the ectodermal a4.2 cell, neural differentiation is not the default. However, while the presumptive neural regions specific for anterior structures, such as brain vesicle or visual pigments, are included in the a4.2 cell, the other posterior neural regions, such as motoneurons innervating striated muscles, are included in the A4.1 cell (Okada et al. 1997; Nishida, 1987). In limited cases, the A4.1 apparently autonomously differentiates into a neuronal cell, which is considered to be the posterior neuronal type and different from the anterior neuronal type expressed in the anterior animal a4.2 cell. Therefore, in ascidian embryos, some early neural precursor cells, such as A4.1 cells, may autonomously differentiate into posterior neuronal cells (Okado & Takahashi, 1990a; Okada et al. 1997; Fig. 10).

In the present experiment, we carried out the analysis of neural competence in the A3 cell from the anterior region of a four-cell embryo. When the A3 cells were disaggregated and cultured in the cleavage-arrested condition, some of them autonomously differentiated into neural cells that expressed posterior neuronal electrical properties, as A4.1 cells (Fig. 12D). Although the A3 cell, the parent cell of both a4.2 and A4.1 cells, includes both presumptive anterior and posterior neural regions, the epidermal region, and the neural inducer region at early developmental stages, the default state of differentiation was apparently posteriorly polarized and neuralized. The apparent neural default in isolated A3 cells might come from autocrine mechanisms of neural inducers, as suggested in the case of A4.1 cells (Okada et al. 1997), because the A3 cells include the inducer region.

Anterior-posterior polarization in neuralized A3 cells

In contrast to vertebrate embryos, in ascidian embryos, even when the inductive influence triggers cell differentiation, some developmental fates, such as antero-posterior polarization, are partially predetermined by cytoplasmic factors inherited from parent cells, as shown in the case of differential neural development of posterior-animal b4.2 cells (Hudson & Lemaire, 2001) and in the case of notochord and mesenchymal cells (Kim et al. 2000). In vertebrate embryos, however, the eggs are apparently regulative due to imperfect segregation of the cytoplasmic factors. In this study we purposely used A3 cells, which have multiple presumptive fates almost for a quadrant of the whole tadpole larva, and mixed their cytoplasmic factors in the cleavage-arrested condition, as a model for regulative precursor cells in vertebrate embryos. The electrophysiological properties of the neurally differentiated A3 cells in triplets or in the isolated condition demonstrated that A3 cells could be either anteriorly or posteriorly polarized by cellular interactions or by inducers.

The neuralization of A3 cells was greatly facilitated in the A3-aa triplets in comparison with isolated A3 cells, as indicated by the population of total differentiated neurons of both ND and NA type in the histograms of Fig. 12A and D, when all three cells in the triplet were switched off from epidermal differentiation. In this triplet, the A3 cell must be an inducer for a4.2 cells to develop neurally at the first step, since a4.2 cells take epidermal fate as default. To explain the enhanced neuralization of the A3 cells, we have to assume another permissive, rather than instructive, interaction of neurally committed a4.2 cells upon A3 cells at the second step (Fig. 13Aa). The A3 cell in A3-aa triplets developed predominantly into the posterior neuronal type possibly because the default state was polarized posteriorly.

F13 View larger version
[in this window]
[in a new window]

Figure 13. Schematic summary of experimental results and the proposed hypothesis of neural cell differentiation in the cell-contact triplet system
The comparison with results in the previously reported aA neural induction pair is also included. A, for A3-aa triplets. Aa, when three cells in the triplets were neuralized, the A3 cell was an inducer secreting both subtilisin-like and FGF-like substances, and the neuralization of the A3 cell was secondarily intensified by the neuralized a4.2 with possible secretion of bFGF-like substance. Ab, when all three cells in the triplets became epidermal cells, the neuron-inducing activity of A3 was blocked by a4.2 cells which are committed to become epidermal cells in advance, rather than A3 cells which are facilitated to become epidermal cells with an epidermalizing factor, like BMP, from a4.2 cells. B, A3-AA triplets. Here, the neural inducer was A4.1 cells, and A3 cells were induced to become anterior neurons. Secondary effects of the neuralized A3 cell upon A4.1 cells were possibly weak as in the case of the aA pair in C. C and D, previously reported neural induction pair aA and homologous a4.2 cell pair from the eight-cell embryo. In the aA pair the A4.1 cell differentiated into a posterior neuron only in limited cases and frequently became non-excitable cells, whereas the a4.2 cell was consistently induced to become an anterior neuron. The secondary neuralizing effects of a4.2 cells on A4.1 cells were weak possibly because of the unresponsiveness of the A4.1 cell.

On the other hand, the A3 cells in A3-AA triplets predominantly developed into anterior type neurons, when the triplets were fully neurally developed and well differentiated. In this A3-AA system, the A4.1 cells must be the inducer at the first step for A3 cells to become anterior neuronal cells (Fig. 13B), since the anteriorly polarized state was not the default of the A3 cells. This may be similar to the situation in which the anterior type differentiation of a4.2 cells in aA pairs from an eight-cell embryo is not the default but requires the A4.1 cells as an inducer (Okado & Takahashi, 1990b). In many cases of A3-AA triplets the differentiation was immature judged from small amplitudes of expressed Na⁺ inward or delayed K⁺ outward currents. This result suggest that the potency of A4.1 cells in A3-AA triplets as an inducer is weaker than that of the A3 cell in A3-aa triplets. A more likely alternative is that induction of the A3 cell into anterior type differentiation requires a more potent inducer. A3 cells in A3-AA triplets sometimes escaped from the inductive effect of A4.1 cells and developed into posterior-type neurons which expressed KA currents (Fig. 9A and B, squares). Altogether in the A3-AA triplets, the A3 cells may be more anteriorly polarized by instructive interaction than their default state, but they are not so polarized as in the case with the a4.2 cells which were consistently anteriorly polarized whenever neuralized (Table 2). The second step interaction from A3 cells to A4.1 cells could be postulated with the analogy of the A3-aa triplet, however, the interaction was not apparent possibly because of the relative lack of responsiveness of A4.1 cells to the neuralizing effects (Fig. 13B). In the aA induction pair there is no significant positive neuralizing interaction from a4.2 cells to A4.1 cells in comparison to isolated A4.1 cells (Fig. 10 and 13C; Okado & Takahashi, 1990b; Okada et al. 1997). It is highly possible that A3 cells are more susceptible to the homologous neuralizing interaction than A4.1 cells, because the A3 cell incorporates a larger fraction of presumptive neural regions than A4.1 cells, as suggested by cell-lineage studies (Nishida, 1987). The proposed schematic models of the differentiation mechanism in newly prepared cell triplets under cleavage-arrested conditions are summarized in Fig. 13.

Comparison of the interactive processes in triplets with those in vertebrate neurogenesis

To compare the mechanism of neural differentiation in the cell-cell contacted A3-aa and A3-AA triplets in this study with previously reported neurogenesis in vertebrate embryos, some suggestions as to the acting inducers in the triplets were apparent from the literature. According to the previous neural induction studies in isolated and cleavage-arrested cells from ascidian eight-cell embryos, bFGF and the protease subtilisin are two major neuralizing agents (Inazawa et al. 1998). Here, in this study, we examined the effects of these two agents upon isolated A3 cells, as shown in Fig. 11 and Fig. 12. To our surprise, bFGF and subtilisin showed different neuralizing effects in contrast to the case of a4.2 cells from eight-cell embryos. bFGF facilitated the differentiation of isolated A3 cells into neuronal cells with a tendency to express more posterior characteristics. Thus, the effects may be equivalent to those of the neuralized a4.2 cells in A3-aa triplets, as shown in Fig. 12A and B. It is suggested that the neuralized a4.2 cells responding to the signalling from A3 cells may secrete bFGF or similar substances to affect A3 cells as a recipient. Here, the interaction may be permissive to intensify the default posterior polarization and/or neuralization in general, being different from the instructive signals of A3 cells to a4.2 cells. While, in Xenopus embryos bFGF is essential for anterior-posterior patterning of neurally induced ectoderm (Cox & Hemmati-Brivanlou, 1995; Lamb & Harland, 1995; Pownall et al. 1996).

On the other hand, subtilisin induced A3 cells to differentiate almost exclusively into anterior-type neurons, more effectively than A4.1 cells in A3-AA triplets (Fig. 12E and F). Subtilisin in the ascidian embryos might be acting as anti-epidermalizing factors and inducing anterior neural structures, such as chordin, noggin and follistatin in vertebrate embryos, by destroying BMP4-like substances. The ascidian BMP homologues are also suggested as neural inhibitors and epidermalizing inducers, although they are not similarly distributed within ectoderms to those in vertebrate embryos (Miya et al. 1996, 1997). However, it should also be noted that bFGF alone can induce a4.2 cells into anterior-type neurons, as previously reported (Inazawa et al. 1998). Thus, the inductive or instructive interactions from A3 to a4.2 in the A3-aa triplet and the interaction from A4.1 to A3 in the A3-AA triplet (Fig. 13Aa and B), that are the first step cell-cell interactions for neurogenesis, are most probably mediated by both subtilisin-like and bFGF-like substances, which are working co-operatively and in a mutually facilitating way. Here, it should be noted that bFGF could also induce A3 cells to differentiate into anterior-type neurons (Fig. 11A, upper graphs, circles). The basic neuralizing effects of bFGF beside its posteriorizing effects in Xenopus ectoderms have been established by Okamoto and colleagues (Kengaku & Okamoto, 1995; Hongo et al. 1999).

Two more comments on the similarity between the ascidian simple cell induction system and vertebrate neurogenesis are worthy of mention. Firstly, the antero-posterior patterning may be predetermined at the stage of neural induction because the neuralized a4.2 cells always expressed the anterior-type differentiation and the neuralized A4.1 cells always expressed the posterior type. However, once the cytoplasmic factors for both anterior and posterior structures were mixed in the A3 cells, the specified cell-cell interaction or inducers could selectively differentiate the anterior or posterior neurons, indicating the similarity of the A3 cell to the vertebrate neuroectoderms. Secondly, a