Development 112, 153-160 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

153

Gap-junctional permeability in early and cleavage-arrested ascidian

embryos

BRIAN DALE*, LUIGIA SANTELLA and ELISABETTA TOSTI

Stauone Zoologica, Villa Comunale, 80121, Naples, Italy

* Author for correspondence

Summary

Using the whole-cell voltage clamp technique, we havestudied jufictional conductance (Gj), and Lucifer Yellow(LY) coupling in 2-cell and 32-cell ascidian embryos. Gjranges from 17.5 to 35.3 nS in the 2-cell embryo wherethere is no passage of LY, and from 3.5 to 12.2 nS in thelater embryo where LY dye spread is extensive. In bothcases, Gj is independent of the transjunctional potential(Vj). Manually apposed 2-cell or 32-cell embryosestablished a junctional conductance of up to 10 nSwithin 30 min of contact. Furthermore, since we did notobserve any significant number of cytoplasmic bridges atthe EM and Gj is sensitive to octanol, it is probable thatblastomeres in the 2-cell and 32-cell embryos are incommunication by gap junctions. In order to compare

Gj in the two stages and to circumvent problems of cellsize, movement and spatial location, we used cytochal-asin B to arrest cleavage. Gj in cleavage-arrested 2-cellembryos ranged from 25.0 to 38.0 nS and remainedconstant over a period of 2.5 h. LY injected into ablastomere of these arrested embryos did not spread tothe neighbour cell until they attained the developmentalage of a 32- to 64-cell control embryo. Our experimentsindicate a change in selectivity of gap junctions at the 32-cell stage that is not reflected by a macroscopic change inionic permeability.

Key words' ascidian embryos, gap junctions, dye coupling,whole cell voltage clamp, electrical conductance.

Introduction

Gap junctions appear to play a role in the regulation ofearly developmental events. First, experimental pertur-bation of junctional conductance using antibodies oranti-sense RNA to connexons can lead to developmen-tal defects (Warner etal. 1984; Warner, 1987; Bevilac-qua etal. 1989). Second, junctional conductance maychange both qualitatively and quantitively at develop-mentally significant times (see reviews, Finbow, 1982;Caveney, 1985; Revel, 1986; Warner, 1988) and, third,cell populations of different developmental fate oftenform distinct communication compartments (Warnerand Lawrence, 1982; Weir and Lo, 1982; Kimmel andLaw, 1985; Serras and van den Biggelar, 1987).

In holoblastic regulative embryos, gap junctionsappear consecutively with initial events of differen-tiation, usually at the 8- to 16-cell stage. For example inthe mouse, dye and electrical coupling occurs at the8-cell stage when compaction first starts (Lo and Gilula,I979a,b; Goodall and Johnson, 1984). Early blasto-meres in the sea urchin embryo are not electricallycoupled (Dale etal. 1982), while electron microscopyshows gap junctions first appear at the 16-cell stage(Andreucetti et al. 1987). In the meroblastic cephalo-pod embryo, blastomeres are extensively coupled by

gap junctions before differentiation of the germ layers(Marthy and Dale, 1989). Mosaic embryos also expressgap junctions early in development. Morphologicalstudies have identified gap junctions in the 2-cellmolluscan embryo Patella (Dorresteijn et al. 1982),although Lucifer Yellow (LY) does not spread betweenblastomeres until the 32-cell stage (Dorresteijn et al.1983).

Intracellular voltage measurements indicate couplingbetween blastomeres in the 2- to 8-cell ascidian embryo(Dale et al 1982), and it has been suggested that ionicpermeability between cells increases at the 32-cellstage, coinciding with the appearance of LY dyecoupling in this embryo (Serras etal. 1988). Complexmulticellular interactions, together with cell size,movement and division, render it difficult to comparejunctional conductance in the 2-cell and 32-cell ascidianembryo. To circumvent this problem, we have usedcleavage-arrested embryos, since elsewhere it has beenshown that cleavage is not necessary for the temporallyprogrammed differentiation of several cytoplasmic andmembrane-located gene products (Whittaker, 1973;Hirano and Takahashi, 1984; Okado and Takahashi,1988; Crowther et al. 1990). Coupling ratios betweencells, as estimated from voltage measurements, givelittle information about the cell-to-cell communication

154 B. Dale, L. Santella and E. Tosti

pathways. By applying the whole-cell voltage clamptechnique to a cell pair, junctional conductance may bemeasured directly (Neyton and Trautmann, 1985;Veenstra and DeHaan, 1988; DeHaan and Chen, 1990).The purpose of the present investigation was tomeasure junctional conductance in early control andcleavage-arrested ascidian embryos, using the whole-cell voltage clamp technique, and to determine whetherearly ascidian blastomeres are coupled by gap junc-tions.

Materials and methods

Experiments were carried out on embryos of the ascidianCiona wtestinalis collected from the Bay of Naples. Spermand eggs were collected from the gonoducts and the spermkept dry until use. The chonons were removed manually usingfine steel needles. Nude eggs were inseminated in agar-coatedPetri dishes in natural sea water at 20°C and, at theappropriate stage, the embryos were transferred to glassslides for electrical recordings. Cleavage-arrested embryoswere prepared by exposing zygotes or 2-cell embryos to2^gml~1 of cytochalasin B (Sigma, St Louis).

Two standard patch micropipettes were used in the whole-cell clamp configuration to voltage clamp the two blastomeresindependently The electrodes of about lOMohm resistanceand 1-2 /xm in tip diameter were filled with an mtracellular-hke solution composed of: 200 mM KC1, 20 mM NaCl, 250 mMsucrose, 10 mM EGTA, 10 mM Hepes at pH7.4. By usingstandard techniques, we obtained gigaohm seals on the twocells, set the pipette voltage to — 40 mV, and ruptured thepatches Whole cell currents were measured on two List EPC-7 amplifiers. In some experiments, one of the electrodes wasfilled with a 5 % solution of Lucifer Yellow (Sigma, St Louis)in 0 2M LiCl2 and back-filled with 0.2M LiCl2. The lowresistance electrodes favoured rapid diffusion of the dye,which usually filled the cytosol within 10s of patch rupture.Elsewhere using either two patch electrodes (Dale, 1988), or apatch electrode in conjunction with a conventional intracellu-lar electrode (DeFelice et al. 1986), it has been shown that themembrane potential of ascidian eggs, with a diameter of130 /im, is under adequate voltage control in these configur-ations. The changes in resting potential and membraneresistance from the unfertilized egg to the 8-cell stage havebeen reported previously (Dale et al. 1982) Preparations withholding currents greater than 100 pA were discarded

Junctional current (I,), in response to an applied junctionalvoltage (Vj), is inward for the more negative cell and outwardfor its depolarized neighbour. The capacitive currentsrecorded from single blastomeres in response to 10 mV de-polarizing steps, from a holding potential of — 80 mV, displaya single exponential decay. Since the electrode resistance waslow compared to the junctional and non-junctional membraneresistance the potential established across the membrane wasconsidered to be that applied to the electrode Seriesresistance was negligible and therefore compensation was notused in these experiments (see DeFelice et al. 1986). Voltagesand currents were stored on tape and played back on a chartrecorder for subsequent analysis. Junctional conductance (Gj)was calculated from Ij/Vj measured during a series of voltageclamp pulses across the intercellular junction. Cells wereclamped at -40, -60 or -80 mV. One of the pair was thendepolarized or hyperpolanzed in 10 mV steps of 400 msduration to create the junctional voltage difference. Voltage

protocols were applied manually The currents generatedreached a stable state within the step period.

Control 2-cell, 32-cell and cleavage-arrested embryos werefixed in 2.5% glutaraldehyde containing 1% paraformal-dehyde in a buffer composed of 0.2 M sodium cacodylate and20% sea water, pH7.2 for lh and postfixed in 1% osmiumtetroxide. The material was then dehydrated in ethanol,embedded in Epon, cut on a Reichert-Jung ultramicrotomeand examined with a Philips 400 electron microscope.

Results

In Ciona intestinalis, the plane of first cleavage liesalong the A-V axis and divides the embryo symmetri-cally into left and right halves. The second cleavageplane is perpendicular to the first, forming 4 blasto-meres of almost equal size (Conklin, 1905). The thirdcleavage is equatorial, and subsequent divisions giverise to the three-dimensional holoblastic embryo. At22 °C, cleavage occurs about every 20min. Shortly aftercell division, the blastomeres are round showing littlecontact, whereas after lOmin cells become semicircularwith the apposing parallel membranes in apparent closecontact (Fig. 1A).

Lucifer Yellow injected into a blastomere of a 2-cellembryo within 5 min of cleavage spreads slowly to thesister cell. However, as reported by Serras et al. (1988),we did not detect spread when the dye was injected atlater stages (Fig. 1A, n=15). Using the dual whole-cellvoltage clamp technique and by applying rectangularvoltage pulses across the junction, we found thejunctional conductance of 2-cell embryos at this stage torange from 17.5 to 35.3 nS, with a mean of 25.9±5.6nS(«=13). Hyperpolarizing one cell of the pair from-40 mV to — 80 mV induced an inward steady statecurrent and capacitive transients in the stimulated cell(Si, Sn, Fig. 2A). The non-junctional resistance of thestimulated cell in this experiment is approximately50Mohm. Since the partner cell was independentlyvoltage clamped at -40 mV, these non-junctionalcurrents were not detected by the second electrode.The currents recorded by the second electrode areessentially the consequence of current flowing throughthe cell-cell junction (Ru Ru , Fig. 2A). Note that thecurrent pulses reached a steady state within the 400 msvoltage step excursion. Each blastomere was in turndepolarized or hyperpolanzed in 10 mV steps of 400 msduration from holding potentials of —40, —60 or—80 mV to create a transjunctional potential difference(Vj). In seven experiments, it was shown that Gj wasnot dependent on the transjunctional voltage (Vj), orthe holding voltage (Vm). Similarly the I/V character-istics of the junction were not directional, i.e. did notdepend on which cell of the pair was stimulated. Atypical experiment showing that Gj is independent ofVj, where Vm was -60mV, and one cell wasdepolarized in 10 mV steps is shown in Fig. 2B. Inseveral experiments, we were able to inject LuciferYellow into a blastomere while simultaneouslymeasuring Gr

Fig. IB shows a 32-cell control embryo 10 min after

Fig. 1. Lucifer Yellow and electrical coupling in early embryos of the ascidian Ciona intestinalis. (A) Phase fluorescencephotograph of a control 2-cell embryo 15min after 1st cleavage. The blur to the right is the LY-containing microelectrode.(B) Fluorescence photograph of a 32-cell ascidian embryo, lOmin following injection of LYinto one of the blastomeres.Note the extensive dye spread. (C) Phase-fluorescence photograph of a cleavage-arrested ascidian embryo 2h followingexposure to cytochalasin B showing dye spread between the two blastomeres. (D) Two cleavage-arrested ascidian embryosmechanically apposed l h following exposure to CB. The two outer cells are in the whole-cell clamp configuration. Scalebar, 100 fan.

Gap junctions in ascidian embryos 155

-40

-80500 p A

-40

4O0msec

30

Gj(nS)

20

10

-100 - 5 0Vj (mV)

Fig. 2. .Functional currents and conductance in a control2-cell ascidian embryo. (A) Both cells are whole-cellvoltage clamped by two independent circuits at — 40 mV. I(is current from one cell, I2 is current from its neighbour.Each cell is hyperpolanzed alternatively to — 80 mV for400 ms {Si and Sn) The current generated in theneighbour, of opposite direction and lacking capacitivetransients, is the current flowing through the cell-celljunction (R! and R11? Ij). Note that Ij reaches a steadystate within the 400 ms period. (B) Junctional conductance(G,) vs junctional voltage (V.) in the same embryo,calculated from a series of 10 mV depolarizing stepsapplied to one cell from a holding potential of -60mV,while the second cell was held at — 60 mV.

Lucifer Yellow had been injected into one of the animalblastomeres. It can be seen that the dye has spread tomany of the other blastomeres, including non-sister

cells. By voltage clamping two non-adjacent cells in the32-cell embryo, we measured the junctional conduc-tance of the composite communication pathway. In 8embryos, the composite junctional conductance rangedfrom 3.5 to 12.2 nS with a mean of 7.7'±3.2 nS. Currentflow in these stages is multi-directional and appears tobe modified by the characteristics of the non-junctionalmembranes (Fig. 3A). The input resistance of thestimulated cell is approximately 200Mohm (Fig. 3A). Atypical experiment in which Vm was held at — 40 mV inboth cells, and one cell was hyperpolarized anddepolarized in 10 mV steps is shown in Fig. 3C. Again itcan be seen that G, is essentially voltage independent.

Owing to the small size of the cells, cellularmovements and the rapid mitotic interval, it is not easyto trace systematically the communication network inthese multicellular stages. In addition, since it isprobable that the various cell lines of the 32-cell embryodisplay graded degrees of coupling and other segTe-gational programs have led to a heterogenous cellpopulation, we exposed 2-cell embryos to cytochalasinB to arrest cleavage and studied junctional conductanceand dye-coupling in these cleavage-arrested embryos.Shortly after cleavage arrest, LY was injected into oneof the blastomeres and the embryo periodicallyobserved at the fluorescence microscope for dye-spread. In 14 experiments, spread occurred at 1.5 to2.5 h following arrest, equivalent to the developmentalage of 32- to 64-cell embryos (Fig. 1C). Junctionalconductance in these cleavage-arrested embryos at thisdevelopmental time was comparable to that of thecontrol 2-cell embryos ranging from 25.0 to 38.0 nS witha mean of 28.7±4.2nS (n=S, Fig. 3B and C).

In electron micrographs of the 2-cell, 32-cell or 2-cellcleavage-arrested embryos cytoplasmic bridges werevirtually absent (Fig. 4). The intercellular spacesranged from 5 to 50 nm and were similar in allembryonic stages studied. Gap junctions were notobserved; however, desmosome-like structures werefrequent. Cytochalasin B had no obvious deleteriouseffect on cellular morphology.

In seven experiments, embryos at various stages weremechanically apposed and one cell in each embryo wasvoltage clamped Within 30min, junctional coupling of3.0 to 10 OnS was established between the embryos. Inone experiment, we manually apposed a cleavage-arrested zygote and a 2-cell embryo. When the embryoreached the 8-cell stage junctional conductance was5.4nS. Lucifer Yellow pre-injected into the zygote didnot spread to the blastomeres until the embryo reachedthe 32-cell stage, whereafter it spread to all theblastomeres. Similar results were observed whencleavage-arrested zygotes were apposed and also withtwo apposed 32-cell embryos. Fig. 5 shows G, measuredacross two apposed cleavage-arrested 2-cell embryos(shown in Fig. ID), 1 h after 1st cleavage. By depolariz-ing one of the blastomeres in 10 mV steps from aholding potential of -60mV, it can be seen thatjunctional conductance is independent of the transjunc-tional voltage (between embryos in this case, Fig. 5B).

Finally, exposing 2-cell, 32-cell and cleavage-arrested

156 B. Dale, L. Santella and E. Tostl

-40

100 p A

-80

500pA

400 msec

-40

Gj (nS)

3 0 "

2 0 -

10..

-50 Vj (mV) 50

Fig. 3. Junctional currents in a 32-cell control embryo (A) and a cleavage-arrested 2-cell embryo of equivalentdevelopmental age (B) Two non-adjacent blastomeres were picked at random in the 32-cell embryo Note the smallercurrents in A. The voltage steps are indicated on the current traces. (C) Junctional conductance (G,) vs junctional voltage(Vj) calculated from a series of lOmV hyperpolarizing and depolarizing steps for A (•**•**) and B( ). Cells werevoltage clamped at — 40 mV.

Gap junctions in ascidian embryos 157

Fig. 4. Electron micrographs of the intercellular junctions of a control 2-cell embryo (A and B), a control 32-cell embryo(C and D) and a cleavage-arrested 2-cell embryo, 2h after continuous exposure to 2/igml"1 of cytochalasin B (E and F)Note the similar morphology in all and the desmosome-hke structures seen at high magnification. Cytoplasmic bridges arerare and CB does not appear to affect gross morphology. The scale in A is for A,C and E, while the scale in B is for B,Dand F.

158 B. Dale, L. Santella and E. Tosti

A-40

-80

-40

1 nA500 pA

400 msec

-80

Gj (nS)

10

-100 -50Vj(mV)

Fig. 5. (A) .Functional currents established between tomechanically apposed cleavage-arrested embryos (shown inFig. ID) lh following cleavage arrest and 30min followingcontact. (B) .functional conductance (Gj) vs junctionalvoltage (Vj) in embryo A, when the membrane potential ofone embryo was shifted in a series of 10 mV pulses from itsholding value of — 40mV to +60 mV.

embryos to 1 mM 1-octanol resulted in a gradualdecrease in Gj, starting about 5min after bathapplication. In 5 experiments using the standardintracellular pipette solution, which is buffered withEGTA to maintain intracellular calcium to below10~7 M Gj was reduced by about 50 %. For example, inone experiment using a 2-cell control embryo, Gj wasreduced from 29.0nS to 19.7nS, while in a 32-cellembryo junctional conductance was reduced from 7 6 to5.3 nS. To permit fluctuations in intracellular Ca2+

levels, we earned out a further 3 experiments using anintracellular-like pipette solution without EGTA. Anexample is shown in Fig. 6, in which Gj in a 2-cellembryo was annulled by octanol lOmin after exposure.Since current amplitudes increased in the stimulatedcell, it appears that octanol also has an effect on thenon-junctional membrane. Owing to the fragility of thenude embryos, it was not feasible to wash out theoctanol and therefore we have no information onreversibility in this system.

Discussion

The whole-cell voltage clamp technique has been usedto measure junctional conductance in a variety ofprimary cell cultures, known to express extensive gapjunctional communication, including pairs of rat lacn-mal gland cells (Neyton and Trautmann, 1985, 1986)and chick embryonic ventricle cells (Veenstra and DeHaan, 1988). In the latter case, cells of about 5^mdiameter have a junctional conductance (Gj) rangingfrom 0.15 to 35.0nS (Veenstra and DeHaan, 1988).Similar values of Gj have been measured in isolatedpairs of adult rat ventricular myocytes (White et al.1985). A major difference in physiology of adult andembryonic heart gap junctions is that the former arereported to be voltage insensitive, whereas embryonicgap junctions are voltage sensitive (see DeHaan andChen, 1990 for references).

In the 2-cell ascidian embryo, measurement of Gjusing the same technique gave values of 17.5 to 35.3 nS,which, despite the difference in size, is comparable tothe Gj in embryonic ventricular cells in culture(Veenstra and DeHaan, 1988). To our knowledge, Gjhas not been previously measured in an intact earlyembryo using this technique. Gj in an embryonic cellline of Drosophila, Kc, was shown to be about 7 nS, andcoupling in this system was suggested to be due tocytoplasmic bridges (Spray et al. 1989). Since, in thepresent study, we did not observe at the electronmicroscope any significant number of bridges connect-ing cells, and considering that Gj was both sensitive toOctanol (Spray et al. 1985) and reached values of 10 nSbetween manually apposed embryos, it is probable thatearly ascidian embryos express functional gap junc-tions We have not been able to successfully fix gapjunctions in ascidian embryos, a problem common toother marine invertebrates.

Serras etal. (1988) have shown previously that dyecoupling with LY starts at the 16- to 32-cell stage in theascidian embryo, and we have confirmed this result. Bystudying coupling ratios as calculated from voltageexcursions, these authors suggested that the change incommunication pattern from the 2- to the 32-cell stagereflects an increase in ionic permeability. Directmeasurements show that Gj in the 32-cell embryo issmaller than that in the 2-cell embryo, possibly as aresult of cell size, although position and movement ofthese smaller blastomeres precludes such a conclusion.In addition, Gj in cleavage-arrested 2-cell embryos at

2 h was not significantly different from that of control 2-cell embryos, supporting the idea that there is nosignificant change in ionic permeability. Finally, incontrast to the suggestion of Serras etal. (1988), Gj

-40-30

-20-10

10

Gap junctions in ascidian embryos 159

appears to be independent of Vj in all stages up to the32-cell stage.

Since cytoplasmic bridges are absent in cleavage-arrested embryos, and Gj does not change in time, what

2030

4050

60

500 p A

2sec

-40

nnnnnnu U

' • ^

Gj(nS)

-100 - 5 0Vj(mV)

Fig. 6. Junctional currents in a control 2-cell embryo where one cell is depolarized by a senes of 400ms steps (A), andlOmin after exposure to 1 mM octanol showing the loss of coupling (B). (C) Gj vs V, before (••••) and after (*•**) exposureto octanol. Holding currents in both cells increased from less than 100pA to about 300pA following exposure to octanol.

160 B Dale, L. Santella and E. Tosti

is the molecular basis for the change in dye couplingproperties at the 32-cell stage? The experiments withcytochalasin B, which functionally blocks cell move-ment and cleavage, rules out spatial organization of theapposing junctional membranes as an important factor.In the early mouse embryo, it has been shown usingcytochalasin B that gap junction assembly is indepen-dent of cell flattening and cytokinesis (Kidder et al1987). One possibility is that the gap junctions in the 32-cell stage are products of the embryonic genome, whilethe gap junctions in the 2-cell stage are the result ofmaternal gene expression. Although macroscopic Gjdoes not change in time, the embryonic junctions maycontain fewer functional channels of larger singlechannel conductance. An alternative possibility is thatjunctional properties are modulated by removal orsynthesis of an accessory protein that controls per-meability. Studies with antibodies to gap junctionproteins and single channel measurements may resolvethis problem.

This work was supported by Nato grant No (30)0255/88.We thank Gianni Gragnaniello for help with the electronmicroscopy

References

ANDREUCCETTI, P , BARONE LUMAGA, R L , CAFIERO, G , FILOSA,A AND PARISI, E (1987) Cell junctions during the earlydevelopment of the sea urchin embryo (Paracentrotus lividus).Cell Differ 20, 137-146

BEVILACQUA, A , LOCH-CARUSO, R AND ERICKSON, R P (1989)Abnormal development and dye-coupling produced by antisenseRNA to gap junction protein in mouse preimplantationembryos Proc natn Acad Sci USA 86, 5444-5448.

CAVENEY, S (1985) The role of gap junctions in developmentA Rev Physiol 47, 319-335

CONKLIN, E. G (1905) The organization and cell lineage of theascidian egg / Acad natn Sci (Philadelphia) 13, 1-119

CROWTHER, R J , MEEDEL, T H AND WHITTAKER, J R (1990)Differentiation of tropomyosin-containing myofibnls in cleavagearrested ascidian zygotes expressing acetylchohnesteraseDevelopment 109, 953-959

DALE, B (1988) Primary and secondary messengers in theactivation of ascidian eggs Expl Cell Res 177, 205-211

DALE, B , D E SANTIS, A , ORTOLANI, G , RASOTTO, M. ANDSANTELLA, L (1982) Electrical coupling of blastomeres in earlyembryos of ascidians and sea urchins. Expl Cell Res 140,457-461.

DEFELICE, L , DALE, B AND TALEVI, R. (1986) Distribution offertilization channels in ascidian oocyte membranes Proc RSoc Lond B 229, 209-214

D E HAAN, R L AND CHEN, Y H (1990) Development of gapjunctions In Embryonic Origins of Defective HearthDevelopment Vol 588 of the Annals of the New York AcademyScience

DORRESTEUN, A W C , BlLINSKI, S M , VAN DEN BlGCELAAR, JA M AND BLUEMINK, J G. (1982) The presence of gapjunctions during early Patella embryogenesis an electronmicroscopical study Devi Biol 91, 397-401

DORRESTEUN, A W C , WAGEMAKER, H A , DELAAT, S. W. ANDVAN DEN BIGGELAAR, J A M (1983). Dye-coupling betweenblastomeres in early embryos of Patella vulgata (MolluscaGasteropoda) Its relevance for cell determination WilhelmRoux Arch dev Biol 192, 262-269.

FINBOW, M E (1982) A review of junctional mediatedintercellular communication. In The Functional Integration ofCells in Animal Tissues (ed J D Pitts and M. Finbow), pp1-37 Cambridge U K Cambridge Umv Press

GOODALL, H. AND JOHNSON, M H (1984) The nature ofintercellular coupling within the preimplantation mouse embryo/ Embryol exp Morph 79, 53-76

HIRANO, R AND TAKAHASHI, K. (1984) Comparison of propertiesof calcium channels between the differentiated 1-cell embryoand the egg cell of ascidians J Physiol. 347, 327-344

KIDDER, G M , RAINS, J AND MCKEON, J (1987) Gap junction

assembly in the preimplantation mouse conceptus is independentof microtubules, microfilaments, cell flattening and cytokinesisProc natn Acad Sci USA 84, 3718-3722

KJMMEL, C. AND LAW, R (1985) Cell lineage of zebrafishblastomeres II Formation of the yolk syncytial layer Devi Biol108, 86-93

Lo, C W AND GILULA, N B (1979a) Gap junctionalcommunication in the preimplantation mouse embryo. Cell 18,399-409.

Lo, C W AND GILULA, N B (19796) Gap junctionalcommunication in the postimplantation mouse embryo Cell 18,411-422

MARTHY, H J AND DALE, B (1989) Dye-coupling in the earlysquid embryo Roux's Arch devl Biol 198, 211-218

NEYTON, J AND TRAUTMANN, A (1985) Single channel currents ofan intercellular junction Nature 317, 331-335

NEYTON, J AND TRAUTMANN, A (1986) Physiological modulationof gap junction permeability J exp Biol 124, 93-114

OKADO, H AND TAKAHASHI, K (1988) A simple 'neural induction'model with two interacting cleavage arrested ascidianblastomeres Proc natn Acad Sci USA 85, 6197-6201

REVEL, J. P. (1986) Gap junctions in development Am Zool 26,529-533

SERRAS, F AND VAN DEN BIGGELAAR, J. A. M (1987). Is a mosaicembryo a mosaic of communication9 Devi Biol 120, 132-138

SERRAS, F , BAUD, C , MOREAU, M , GUERRIER, P AND VAN DEN

BIGGELAAR, J A M (1988) Intercellular communication in theearly embryo of the ascidian Ciona intestinalis Development102, 55-63

SPR/IY, D. C , WHITE, R L , VERSELIS, V AND BENNETT, M V L

(1985) General and comparative physiology of gap junctionschannels In Gap Junctions (ed M V L Bennett and D CSpray), pp ,139-153 New York Cold Spring Harbor

SPRAY, D C., CHERBAS, L., CHERBAS, P., MORALES, E A AND

CARROW, G M (1989) Ionic coupling and mitotic synchrony ofsiblings in a Drosophila cell line Expl Cell Res 184, 509-517

VEENSTRA, R D. AND DEHAAN, R L (1988) Cardiac gapjunction channel activity in embryonic chick ventncular cells.Am J Physiol 254, H170-H180.

WARNER, A E (1987) The use of antibodies to gap junctionsprotein to explore the role of gap junctional communicationduring development In Junctional complexes of Epithelial Cells(ed G Bock and S Clark), pp 154-167 Ciba FoundationSymposium 125. Chichester, U K. Wiley

WARNER, A (1988) The gap junctions J Cell Sci 89, 1-7WARNER, A AND LAWRENCE, P. (1982). Permeability of gap

junctions at the segmental border in insects epidermis Cell 28,243-252

WARNER, A E-, GUTHRIE, S C AND GILULA, N B (1984)

Antibodies to gap-junctional protein selectively disruptjunctional communication in the early amphibian embryoNature 311, 127-131

WEIR, M. P. AND LO, C W (1982) Gap junctionalcommunication compartments in Drosophila wing imaginal disk

' Proc natn. Acad Set U S A 79, 3232-3235WHITE, R L , SPRAY, D C , CAMPOS DE CARVALHO, A ,

WITTEMBERG, B A AND BENNETT, M V L. (1985) Someelectrical properties of gap junctions between adult ventncularmyocytes Am. J Physiol 249, C447-C455

WHITTAKER, J R. (1973) Segregation during ascidianembryogenesis of egg cytoplasmic information for tissue specificenzyme development Proc. natn Acad. Sci USA 70,2096-2100.

(Accepted 21 January 1991)