Eliminating Afferent Impulse Activity Does Not Alter the Dendritic Branching of the Amphibian Mauthner Cell Linda A. Goodman* and Pat G. Model Department of Neuroscience, Rose F. Kennedy Center for Research in Mental Retardation and Human Development, Albert Einstein College of Medicine, Bronx, New York 10461
SUMMARY In the developing amphibian, the formation of extra vestibular contacts on the Mauthner cell (M-cell) enhances dendritic branching, while deprivation reduces it (Goodman and Model, 1988a). The mechanism underlying the interaction between afferent fibers and developing dendritic branches is not known; neural activity may be an essential component of the stimulating effect. We examined the role of afferent impulse activity in the regulation of M-cell dendritic branching in the axolotl (Ambystoma mexicanurn) embryo. M-cells occur as a pair of large, uniquely identifiable neurons in the axolotl medulla. Synapses from the ipsilateral vestibular nerve (nV11I) are restricted to a highly branched region of the M-cell
lateral dendrite. W e varied the amount of nVII1 innervation and eliminated neural activity. First, unilateral transplantation of a vestibular primordium deprived some M-cells of nVIII innervation and superinnervated others. Second, surgical fusion of axolotls to 'ITX-harboring California newt (Taricha torosa) embryos paralyzed the Ambystoma twin: voltage-sensitive Nat channel blockade by 'ITX eliminated action potential propagation. Reconstruction of M-cells in 18 mm larvae revealed that dendritic growth was influenced by ingrowing axons even in the absence of incoming impulses: impulse blockade had no effect on the stimulation of dendritic growth by the afferent fibers.
INTRODUCTION
Covell, and Model, 1988). We have begun to explore the nature of the transsynaptic stimulating effect imposed by afferent fibers on dendritic branching patterns: neural activity could be an essential component of the interaction. Certain aspects of neural development are independent of impulse activity. For example, prolonged immobilization of amphibian embryos by immersion in anesthetic solutions before the onset of motility does not interfere with or delay the development of normal motor function (Harrison, 1904; Carmichael, 1926; Matthews and Detwiler, 1926; Haverkamp and Oppenheim, 1986). In addition, elimination of impulse activity does not disrupt axonal pathfinding: retinal ganglion cells project to and synapse in the correct segment of the tectum (in salamanders, Hams, 1980, 1984; in goldfish, Meyer, 1987). Finally, the membrane properties underlying synaptic function in am-
A neuron's dendritic branching pattern is its most characteristic morphological feature. Ingrowing afferent axons are important in the regulation of dendritic growth on target neurons. For example, in the developing axolotl (Anzhystorna mexicanurn). the formation of extra vestibular (nVIII) contacts on the Mauthner cell (M-cell) enhances dendritic growth in the superinnervated region, while deprivation reduces it (Goodman and Model, 1988a). In older larvae, regenerating nVIII axons also stimulate dendritic growth (Goodman, Received July 27. 1989: accepted October 10. 1989 Journal ofNeurobiology, Vol. 21, No. 2. pp. 283-294 (1990) 0 1990 John Wiley & Sons, Inc. CCC 0022-3034/90/020283- 12$04.00 * To whom correspondence should be addressed.
283
284
Goodman and Model
phibian myotomal muscle (Kullberg, Owens, and Vickers, 1985)and cultured spinal neurons (Bixby and Spitzer, 1984) develop normally in the absence of nerve impulses. On the other hand, various aspects of neural development are activitydependent. Neuromuscular activity blockade rescues motoneurons from natural cell death (Pittman and Oppenheim, 1978), prevents elimination of extra motor nerve terminals (Srihari and Vrbova, 1978), and enhances intramuscular nerve branching (Dahm and Landmesser, 1988). Furthermore. blockade of nerve impulse activity modifies the distribution of acetylcholine receptors on target muscle fibers (L0mo and Kosenthal, 1972). alters the level of biosynthetic enzymes (e.g., Black and Geen, 1973) and neurotransmitters (e.g., Kessler and Black, 1982), precludes preferential loss of retinal ganglion cells with misdirected axons (O’Leary, Fawcett, and Cowan, 1986), and, in culture, affects neuronal survival (Bergey, Fitzgerald, Schrier, and Nelson, 1981) as well as neurotransmitter selection (Walicke, Campenot, and Patterson, 1977). Neuronal morphology is also modified by impulse blockade. For instance, blockade of retinal ganglion cells by application of tetrodotoxin (TTX) prevents the segregation and refinement of axonal arborizations into eye-specific columns in the tectum of goldfish (Meyer, 1 982) and tadpoles (Reh and Constantine-Paton, 1985), and precludes the formation of ocular dominance columns in the lateral geniculate nucleus (Sretavan, Shatz, and Stryker, 1988) and visual coitex (Stryker and Harris, 1986) of cats. There is evidence both for and against changes in dendritic organization due to loss of impulse activity. In mutant crickets, Bentley (1975) observed reduced dendritic branching on an identified interneuron that was deprived of normal synaptic activation. In frog embryos raised in anesthetic solutions, however, the branching of motoneurons was not altered by the absence of impulse activity (Haverkamp, 1986). We have examined the role of neural activity in the regulation of M-cell dendritic branching in the developing axolotl by interrupting synaptic transmission while preserving synaptic contacts. In the medulla of the axolotl, M-cells occur as a pair of large, uniquely identifiable neurons at the level of entry of nV111 (Hemck, 1914). Each cell receives synapses from the ipsilateral nerve; the terminals can be identified morphologically as club terminals in the light (Model, 1978) and electron (Kimme1 and Schabtach, 1974; Model and Wurzel-
mann, 1982) microscopes, and are restricted to a highly branched region of the M-cell lateral dendrite (Fig. I). In the developing embryo, the M-cell is recognized first at Harrison (1969) stage 34, and the earliest nVIII axons contact the growing lateral dendrite at stage 36 (Leber, 1984). To eliminate impulse activity, an experimental preparation devised by Twitty ( 1937) was used: axolotl embryos, stages 28-30, were surgicallyjoined side-by-side to tetrodotoxin (TTX)-harboring California newt (Taricha torosa) embryos in a condition known as parabiosis. In the axolotl, TTX blocks voltage-dependent Na+ channels and thus the propagation of action potentials. The fusion results in complete paralysis of the Ambystoma twin (Twitty, 1937; Harris, 1984; present results); the insensitive newt provides nutritional support for its immobile partner through early larval stages. In the presence of TTX, M-cells in the axolotl develop without incoming impulse activity. If synaptic activity is essential for the development of normal dendritic branching patterns, M-cells exposed to TTX will have reduced branching compared to control. Under normal conditions, the extent of dendritic branching vanes from animal to animal: dendritic loss from both cells in a single larva may be difficult to assess. In contrast, the extent of branching on the M-cells in a single larva is nearly identical, even though the precise pattern of branching may differ. Therefore, the most accurate comparisons are made between M-cells within a single individual. Experimental and control sides were created in each animal by performing unilateral ear transplants. Some experimental M-cells were deprived of nVIlI innervation by removal of a vestibular primordium from donor embryos; others were superinnervated with extra nVIIl afferents by implantation of an extra ear in hosts. The contralateral side served as the intrinsic control. All axolotl embryos were then fused to Taricha embryos. If impulse activity is essential to the stimulation of dendritic growth by ingrowing axons, superinnervated, deprived, and contralatera1 control M-cells in TTX-exposed larvae all should appear deprived. On the other hand, if activity is not required, the dendritic branching pattern of superinnervated M-cells should be enhanced and that of deprived M-cells should be reduced compared to the appropriate contralateral cells. Our data indicate that impulse activity is not essential. Preliminary results have been presented (Goodman and Model, 1988b).
Impulse Block Does Not Alter Brunching
MATERIALS AND METHODS Axolotl embryos used in these experiments were provided by the colony of axolotls maintained in our laboratory. California newt embryos were collected in thc wild by Dr. David Good of the Museum of Vertebrate Zoology, Berkley, CA. Operations were performed on midtailbud embryos at stages 28-30 in stcrile NiuTwitty's complex salt solution (pH 7.4-7.6) with neomycin sulfate (50 mg/liter). Ear transplants were performed on the same side of all axolotl embryos. After decapsulation, host embryos were prepared by unilateral removal of a small piece ofectoderm from a site immediately rostra1 to the in situ vestibular primordium. Donor tissue was prepared by unilateral excision of a vestibular primordium (together with its associated nVIII ganglion and overlying ectoderm). The excised vestibular primordium was then implanted in normal orientation in the site previously prepared in each host. Surgery was performed prior to nVIII outgrowth and M-cell differentiation. After the vestibular implants healed and the donor sites began to fill in (about 30 min), Amby.rtoma hosts and donors were parabiotically fused to Tarichu embryos: a patch of ectoderm was removed from the belly region of one side of axolotl embryos and from the opposite side of Turichu embryos. Each pair was positioned with the excision sites apposed and, over several hours. the ectoderm and endoderm became fused. The resulting parabiotic twins shared a blood circulation as well as some gut structures. TTX stored in the Taricha embryos was delivered to the sensitive axolotl partner via the shared circulation. The pairs were reared at 16°C in individual petri dishes through hatching to early feeding stages. Amhystoma larvae, immobilized by TTX, received nutritional sup-
285
port from the Turichu partners. Controls for TTX exposure were prepared by fusing several pairs of normal axolotl embryos. When axolotl larvae reached 18 mm, the parabiotic twins were anesthetized in MS 222, Axolotl heads were removed and immediately immersed in 3%1glutaraldehyde, 2% paraformaldehyde, 1% acrolein, and 2.5% DMSO in 0.1 Mcacodylate buffer (pH 7.4) at room temperature. After 24 h at 4"C, the heads were washed several times in 0.1 M cacodylate buffer (pH 7.4, total time 1.5 h). Heads were postfixed in cold 2% Os04 in cacodykdte buffer (2 h), washed in veronal acetate buffer, and stained en bloc in 4% uranyl acetate ( 1 h). Heads were dehydrated in a graded series of ethanols and embedded in Epon. Serial cross sections, 3 Fm thick, were cut from the heads and photographed in the light microscope. Pairs of M-cells in both TTX-exposed and control larvae were reconstructed from serial 3-pm-thick sections (Goodman and Model, 1988a). The morphology of experimental M-cells was compared to that of the contralatera1 cells. In addition, both M-cells in TTX-exposed larvae were compared to M-cells in unexposed controls. Several 3-gm-thick sections were mounted onto Epon blocks and sectioned for electron microscopy. Thin sections were collected on mesh grids, stained with uranyl acetate and lead citrate, and photographed in a Philips 300 electron microscope. Club endings were identified on M-cells in both TTX-exposed and control larvae. The morphology of the terminals was compared.
RESULTS
The experimental M-cell in 15 donor embryos was deprived of nVIII innervation by removal of a ves-
Figure 1 Light micrograph of an Epon-embedded, 3-pm-thick, toluidine blue-stained cross section through an M-cell in a control 18 mm larva. Vestibular axons (VIII) enter the medulla at the level of the M-cell and terminate on the ventral surface and branches (between arrowheads) of the lateral dendrite (Id). md = medial dendrite; a = axon; G = nVITI ganglion. Scale bar, 50 pm.
286
Goodman and M o d d
tibular primordium. The experimental cell in 15 hosts was superinnervated by implantation of an extra ear. Axolotl embryos were then surgically joined to Taricha embryos [Fig. 2(A,B)]. In addition, normal (unoperated) pairs of axolotl embryos were fused to each other as controls. By stage 34. control animals showed spontaneous slow contractures and responded readily to touch. In contrast, embryos fused to California newts showed no spontaneous movements and did not respond to even the most vigorous stimulation. Complete paralysis of the Ambystoma twin indicates that Na+-dependent action potentials were eliminated in the TTX-sensitive salamander. When the fastest-growing axolotl larva reached 19.5 mm, it began to show slight twitching movements of the head and gills. It was discarded and all of the remaining larvae were killed at 18 mm (about 3.5 weeks posthatching*). None showed any movement. At that stage, ears can be identified by white otoliths of the two maculae. On the experimental side of donors, the maculae were absent [Fig. 2(C)]; on the experimental side of hosts, extra maculae were visible rostra1 to the in situ ones [Fig. 2(D)]. Previously, we showed that vestibular axons from an ectopic ear enter the medulla at the level of nV and, confined to the nVIII tract, course caudad to innervate the appropriate region of the ipsilateral M-cell. We found that experimental M-cells received 30-67% more club endings than controls did, and that dendritic branching in the superinnervated region was enhanced: the supernumerary afferents stimulated extra dendritic growth (Goodman and Model, 1988a). In the present study, in addition to depriving or superinnervating experimental M-cells, afferent impulse activity was blocked by TTX throughout M-cell differentiation. Six donor and 2 host animals died prior to reaching 18 mm. Five of the remaining 9 pairs of M-cells in donor larvae were reconstructed. In the other donors, the experimental M-cell was absent or the medulla was disrupted. Nine pairs of Mcells in hosts were reconstructed. The remaining hosts were discarded because one separated from its Turichu mate, one was stunted, one was poorly fixed, and one (at 19.5 mm) showed movements. The unilateral ear transplants permitted comparisons between an experimental M-cell and the ap*Individual axolotls reach 18 mm about 12 days posthatching. The growth rate of axolotls fused to California newts is slowed, presumably because axolotls in parabiosis receive limited nutrition.
propriate contralateral cell (one that received a normal amount of innervation and was also exposed to TTX). The dendritic surface of deprived M-cells was significantly reduced in the region that normally receives nVIII innervation (Fig. 3). In contrast, the branching of most (6) superinnervated M-cells was much more abundant in the region receiving extra innervation (Fig. 4). In 2 host larvae, dendritic growth on thc experimental cell showed a small increase over the contralateral cell, and in only I larva, both cells appeared about the same. Taken together, the results show that exposure to TTX does not prevent the stimulation of dendritic growth by ingrowing aferent fibers. TTX-exposed cells were compared also to unexposed M-cells in control larvae (axolotls that were fused to other axolotls). If TTX brought about a substantial alteration in M-cell morphology, the change would be detected easily. On the other hand, if TTX brought about only a small alteration, the change might be obscure: exposed cells must be compared to unexposed cells in different individuals, and in different individuals, the extent of dendritic branching varies even under normal conditions. As a group, experimental TTX-exposed cells differed significantly from unexposed cells: deprived M-cells showed a dearth of branching (Figs. 3 , 5), and superinnervated cells showed an excess of branching in the region where nVIIl terminates (Figs. 4, 5). Comparing contralateral TTX-exposed cells (those receiving a normal complement of innervation) to unexposed controls revealed no apparent effect of TTX on M-cell morphology. Seven pairs of Mcells reconstructed from control (unexposed) axo l o t l ~revealed a range in complexity of dendritic branching (Fig. 5 ) . The degree of branching of the M-cells reconstructed from the control side of TTX-exposed axolotls spanned approximately the same range. These results indicate that TTX exposure did not alter dendritic growth or pattern, although due to the normal variability among individuals. subtle changes would not have been detected. The morphology of club endings in 3 TTX-exposed larvae was examined in the electron microscope and compared to unexposed controls. Club endings are easily identified because they are very large, make morphologically mixed (chemical and electrical) synapses, have mitochondria clustered near sites of contact, and contain densely packed. highly organized neurofilaments. TTX-exposed
Impulse Block Does Not iilter Branching
Figure 2 (A and B) Parabiotic twins 30 h after surgery. At the fusion site, some pigmented cells from the axolotl embryos (left) have migrated across to the California newt embryos (right). (A) Donor axolotl embryo with its vestibular primordium removed (arrowhead). (B) Host axolotl embryo with an extra vestibular primordium (double arrowhead) implanted rostra1 to the in silu one (single arrowhead). (C) and (D) Parabiotic twins with axolotl larvae (left) at 18 mm. Maculae are absent in the donor axolotl (C); an extra pair of maculae (double arrowhead) are visible in the host (D). Scale bars, 1 mm.
287
Figure 3 Thrce pairs of M-cells reconstructed from donor larvae. The axon projects toward the center of the page; the lateral dendrite, toward the side of the page. Deprived cells (left) are paired with the appropriate control (right). Dendritic branching of deprived cells is significantly reduced in the region that normally receives nVIII terminals (between arrowheads) compared to the appropriate control. Scale bar, SO pm. c l u b tcrminals were morphologically indistinguishable from controls (Fig. 6). Regions of contact ranged in size from 3 to 10 pm in micrographs from both experimental and control larvae.
DISCUSSION
In the absence of incoming impulse activity, developing M-cells elaborate a normal dendritic
Figure 4 Three pairs of M-cells reconstructed from host larvae. The branching of supennnervated cells (left) is greatly enhanced in the region where nVIII terminates (between arrowheads) compared to the appropriate control (right). Scale bar, 50 wm.
Impulse Block Does Not Alter Branching
289
B Figure 5 Two pairs of M-cells reconstructed 1From control larvae never exposed to TTX. The dendritic pattern of control cells varied from moderate (e.g., B) to more generous (e.g., A) amounts of branching. Scale bar, 50 pm.
branching pattern. The elimination of activity in axolotl embryos by surgically joining them to TTX-harboring California newt embryos produced complete paralysis of the Ambystoma twin. Immobilization of the salamander indicates that Na+-dependent action potentials were blocked and that M-cells developed in the absence of afferent impulse activity. Club endings formed on the appropriate dendritic surface of the M-cells and were normal insofar as their light and electron microscopic appearance was indistinguishable from control. Unilateral ear transplants, performed prior to parabiotic fusion, permitted an accurate evaluation of alterations in mature M-cell morphology: superinnervated and deprived cells were compared to the contralateral M-cell in the same individual. Under standard conditions, the usual complement of ingrowing nVIIJ fibers is required for normal development of the M-cell dcndntic branching pattern; the axons stimulate dendritic growth (Goodman and Model, 1988a). That TTX-exposed M-cells receiving a normal amount of nVJTI innervation span a range of branching patterns similar to unexposed controls indicates that the absence of nerve activity does not alter development of M-cell dendritic branching. Since the dendritic surface of superinnervated cells was increased and that of deprived cells, decreased, impulse blockade did not prevent the stimulation of dendritic growth by ingrowing afferents. Thus, neural activity is not an essential component of the stimulating effect. Taricha-supplied TTX produced a transient immobilization of axolotl larvae, i.e., the fastestgrowing salamander became mobile when it reached 19.5 mm. Twitty’s (1937) results were
similar: shortly after the newt’s yolk was used up, the axolotl twin recovered some motor function. The reason for the temporary paralysis is unknown. Twitty (1937) showed that Taricha eggs and embryos are the most potent sources of the toxin; extracts from larvae become progressively less effective with age. Thus, one possibility is that recovery occurs when the maternally-derived toxin becomes depleted. Recovery of impulse activity, however, does not occur in the optic nerve of axolotl eyes transplanted to California newt embryos, even after 6 months have passed (Hams, 1980). Furthermore, in the present work, axolotl larvae that grew more slowly, reaching 18 mm up to 10 days later than the average animal, remained completely immobilized. Another possibility is that the maturing salamander develops the ability to metabolize the toxin, rendering it inactive (Hams, 1984). Studies have shown that the development and maturation of functional synaptic networks is not contingent upon prior electrical impulse activity. For example, in cultured explants of fetal mouse cerebral cortex, synapses that formed in the presence of a local anesthetic were morphologically normal, as observed in the light (Crain, Bornstein, and Peterson, 1968) and electron (Model, Bornstein, Crain, and Pappas, 1971) microscopes. Furthermore, the potential for complex bioelectric activity also developed under conditions of impulse blockade (Crain et al., 1968). In the developing animal, Harris (1980, 1984)demonstrated that retinal ganglion cells formed synaptic contacts in the tectum in the absence of impulse activity. In the present report. blocked nVIIJ axons formed club endings that displayed characteristic ultrastruc-
290
Goodman and Model
Impulse Block Does Not Alter Branching
tural features: the large terminals made both chemical and electrical synaptic specializations, mitochondria clustered near the specialiiations. and neurofilaments were densely packed and organized. Our results agree with the data cited above: synapses that are normal in structure develop in the absence of electrical impulse activity. The presence of morphologically normal synaptic junctions, however, need not imply normal function. This issue was addressed in the classic works of Harrison ( 1 904), Carmichael ( 1926), and Matthews and Detwiler (1926) who showed that neither neuromuscular activity nor proprioceptive input are essential to the development of coordinated swimming behavior in amphibians. Even detailed quantitative measures of behavior do not distinguish normally reared amphibian embryos from those reared in anesthetic solutions (Haverkamp and Oppenheim, 1986). In addition, recordings of ventral root activity in previously immobilized embryos are normal (Haverkamp, 1986). Thus, there is considerable evidence that organized neuronal assemblies form in forward reference to their ultimate use from the standpoint of both morphology and function. It is important to determine whether the TTX circulating in axolotl embryos blocked all action potentials produced by ingrowing nVIII axons. In the developing amphibian, action potentials recorded in Rohon-Beard neurons and dorsal root ganglion neurons in vivo (Spitzer and Baccaglini, 1976; Baccaglini, 1978) and in spinal neurons in vitro (Spitzer and Lamborghini, 1976) show a shift in ionic dependence from Ca2+to Na’. In addition, the ionic basis of action potentials carried by outgrowing processes of individual spinal neurons in vitro follows the same progression (Willard, 1980). Therefore, it is likely that action potentials propagated by nVIII axons in axolotl embryos are initially Ca2+-dependentand insensitive to TTX. We do not know whether early Ca2+-dependent impulses affect developing postsynaptic dendrites. To determine whether they could be important, however, we would have to identify when during nVIII ingrowth the shift to Na+-dependence takes place. Since attempts to study the M-cell electro-
291
physiologically in axolotl embryos and larvae have been unsuccessful, we could not examine the developing system directly. However, due to several lines of evidence. it is probable that the change in ionic dependence takes place before axons reach their targets. For instance, we observed that TTX paralyzed the axolotl embryos through all stages of development: this observation suggests that motoneuron axons make only Na+-dependent (TTXblockable) spikes when they contact muscle. In addition, Kullberg, Lentz, and Cohen (1977) demonstrated that the earliest recorded end-plate potentials at the myotomal neuromuscular junction in Xenopus embryos were blocked completely by TTX. Likewise, retinal ganglion cells in eyes transplanted from Ambystoma into Taricha were silenced by the newt’s TTX, i.e., Ca2+spikes were never recorded in the target tectum (Hams, 1980). Further evidence that the ionic-dependence shift in growing axons takes place early comes from Willard ( I 980) who demonstrated that action potentials propagated by neurites extended from spinal neurons in culture undergo the change from Ca2’- to Na+-dependence considerably before the shift occurs in cell bodies. Moreover, the neuntes lose their capacity to make divalent cation-dependent adion potentials long before the cell bodies. He also showed evidence that neurites sprouted late may produce Na+-dependentaction potentials without ever producing Ca2+ spikes. Finally, the results of several studies in vitro are consistent with the view that axons can convert to Na+-dependent spikes prior to innervating their target: maturational changes in membrane properties occur independently of cellular contacts (Bixby and Spitzer, 1984; Henderson and Spitzer, 1986; Willard, 1980). Thus, it is reasonable to suppose that, in the axolotl, nVllI axons make Na+-dependent, TTX-blockable action potentials by the time the fibers contact the M-cell. In goldfish (and presumably in axolotls) action potentials propagating along nVIII axons excite the M-cell by way of electronic junctions as well as by release of chemical neurotransmitter (Faber and Korn, 1978). Impulse blockade by TTX probably eliminates most, but not all of the transmitter
Figure 6 Club endings in experimental and control larvae. The electron micrograph of the boxed area in each inset shows club endings (*) on a TTX-exposed (A) and on an unexposed control (B) cell. The club terminals are morphologically indistinguishable. Inset scale bar, 50 p m . EM scale bar. 2 pm.
292
Goodman and , o d e 1
release, i.e., it would not block spontaneous quantal release. Spontaneous release in the absence of impulses allows us to separate any theoretical effects of diffusable chemical transmitters from those of electrical activity. The possibility that transmitter may be a regulatory signal between developing nerve terminals and adjacent nerve processes has become more credible since transmitters have been shown to be released from growth cones in culture (Hume, Role, and Fischbach, 1983; Young and Poo, 1983). Several studies have demonstrated that the application of specific transmitters affects neurite growth in vitru. Serotonin, for example, inhibited the elongation of neurites on certain Helisorna neurons (Haydon, McCobb, and Kater, 1987). In addition, glutamate, spontaneously released from axons of entorhinal neurons, reduced dendritic outgrowth from cocultured hippocampal pyramidal neurons (Mattson, Lee, Adams, Guthrie, and Kater, 1988). TTX blocked the inhibitory effect of glutamate by reducing release of the transmitter, but the toxin produced its effect by reducing the number of synaptic contacts rather than by eliminating activity at intact terminals. Since chemical transmitters can influence neurite growth in vitro, perhaps neurotransmitters released from nVIII afferents growing in vivo are responsible for inducing growth of M-cell dendritic branches in the region where nVIII terminates. In the present work, however, TTX probably eliminated nearly all of the transmitter that would have been released by functioning nVIII axons, yet superinnervation still brought about extra dendritic branching. Thus, it appears that the transmitter is not responsible for promoting dendritic growth. On the other hand, since the synaptic endings are intact, miniature postsynaptic potentials are likely to occur in the absence of nerve impulses (as at the neuromuscular junction), i.e., there probably is some spontaneous quanta1 release of transmitter. Therefore, if very small quantities of transmitter are adequate to carry out potential trophic influences, electrical impulse blockade would not eliminate the chemical effect. Once the neurotransmitter in club endings is identified, one may be able to determine whether the compound has a role in inducing dendritic growth by inhibiting its action through the use of receptor blocking agents or, alternatively, enhancing its action through the use of chemical agonists. In summary, parabiotic fusion of axolotl embryos to TTX-harboring Turicha turusa resulted in immobilization of the axolotl twin. Voltage-sensi-
tive Na' channel blockade by TTX interrupted action potential propagation and thus eliminated impulse activity. In the presence of TTX, nVIII afferent fibers formed normal club endings on the appropriate dendritic surface and branches of the M-cell. Eliminating virtually all activity throughout embryonic development did not appear to affect the extent of dendritic branching on M-cells that received a normal complement of nVIII axons. Exposure to TTX did not alter or eliminate the transsynaptic stimulating effect that ingrowing afferent fibers exert on the morphogenesis of the M-cell: superinnervation enhanced dendritic branching while deprivation reduced it. Thus, afferent impulse activity is not required for normal development of postsynaptic dendritic branching patterns. We are indebted to Dr. David Good for providing the Taricha torosa eggs and thus making our study possible. We wish to thank Sarah Wurzelmann for the electron microscopy and for her expert preparation of publication prints. We also thank Kathleen Tinglin for her fine technical assistance. The work reported here was supported in part by a Martin Foundation fellowship awarded to L.A.G.
REFERENCES BACCAGLINI, P. I. (1978). Action potentials of embryonic dorsal root ganglion neurones in Xenoptis tadpoles. J. Physiol. 283585-604. BENTLEY,D. ( 1 975). Single gene cricket mutations: Effects on behavior, sensilla, sensory neurons, and identified interneurons. Scicnce 187:760-764. BERGEY,G. K., FITZGERALD,S. C., SCHRIER, B. K., and NELSON, P. G. (1981). Neuronal maturation in mammalian cell culture is dependent on spontaneous electrical activity. Brain Rex 207:49-58. N. C. (1984). Early differenBIXBY,J. L., and SPITZER, tiation of vertebrate spinal neurons in the absence of voltage-dependent Ca2+ and Na' influx. Dev. Biol. 106539-96. BLACK,I. B., and GEEN,S. C. (1973). Trans-synaptic regulation of adrenergic neuron development: Inhibition by ganglionic blockade. Brain Rex 63:29 1-302. CARMICHAEL, L. ( 1926). The development of behavior in vertebrates experimentally removed from the influence of external stimulation. Psychol. Rev. 335 1-58. CRAIN,S. M., BORNSTEIN, M. B., and PETERSON,E. R. (1968). Maturation of cultured embryonic CNS tissues during chronic exposure to agents which prevent bioelectric activity. Brain Res. 8:363-372.
Impulsc Block Does Not Alter Brunching DAHM, L. M., and LANDMESSER, L. T. (1988). The regulation of intramuscular nerve branching during normal development and following activity blockade. Dev. Biol. 130:62 1-644. FABEK, D. S., and KORN,H. (1978). Elcctrophysiology of the Mauthner cell: Basic properties, synaptic mechanisms, and associated networks. In: Neurobiology of rhe Muuthner Cell. D. S. Faber and H. Korn, Eds., Raven Press, NY., pp. 47-131. GOODMAN,L. A., COVELL,D. A,, and MODEL,P. G. (1988). Regenerating afferent fibers stimulate the recovery of Mauthner cell dendritic branching in the axolotl. J . Neurosci. 8:3025-3034. GOODMAN, L. A,, and MODEL,P. G. (1988a). Superinnervation enhances the dendritic branching pattern of the Mauthner cell in the developing axolotl. J . Neurosci. 8:776-79 1. GOODMAN, L. A., and MODEL,P. G. (1988b). The effect of eliminating afferent impulse activity on the morphogenesis of the amphibian Mauthner cell. Soc. Neurosci. Ahstr. 145 17. HARRIS,W. A. ( I 980). The effects of eliminating impulse activity on the development of the retinotectal projection in salamanders. J . Comp. Neurol. 194:303-317. HARRIS,W. A. (1984). Axonal pathfinding in the absence of normal pathways and impulse activity. J. Neurosci. 4: 1 153- 1 162. HARRISON, R. G. (1904). An experimental study of the relation of the nervous system to the developing musculature in the embryo of the frog. Amer. J. Anat. 3~197-220. HARRISON, R. G. (1969). Organization and Development of'the Embryo. Yale University Press, New Haven, CT. HAVERKAMP, L. J. ( 1986). Anatomical and physiological development of the Xenopus embryonic motor system in the absence of neural activity. .I. Neurosci. 6~1338-1348. HAVERKAMP, L. J., and OPPENHEIM, R . W. (1986). Behavioral development in the absence of neural activity: Effects of chronic immobilization on amphibian embryos. J . Nauosci. 6:1332-1337. HAYDON,P. G., MCCOBB,D. P., and KATER,S. B. ( I 987). The regulation of neuritc outgrowth, growth cone motility, and electrical synaptogenesis by serotonin. J . Neztrobiol. 18:197-215. HENDERSON, L. P., and SPITZER, N. C. ( 1 986). Autonomous early differentiation of neurons and muscle cells in single cell cultures. Dev. B i d . 113:381-387. HERRICK, C. J. (1914). The medulla oblongata oflarval Amblystoma. J. Comp. Neurol. 24:343-427. H U M E ,R. I.. ROLE, L. W.. and FISCHBACH, G. D. (1983). Acetylcholine release from growth cones detected with patches of acetylcholine receptor-rich membranes. Nature 305632-634. KESSLER, J. A., and BLACK,I. B. (1982). Regulation of
293
substance P in adult sympathetic ganglia. Bruin Res. 234:182-187. KIMMEL,C. B., and SCHABTACH, E. (1974). Patterning in synaptic knobs which connect with Mauthner's cell (Ambystoma mexicanum). J. Comp. Neurol. 156:4980. KULLBERG,R., LENTZ,T. L., and COHEN,M. W. (1 977). Development of the myotomal neuromuscular junction in Xenopus laevis: An electrophysiological and fine-structural study. Dev. Biol. 60: 10 1-129. KULLBERG, R., OWENS,J. L., and VICKERS,J. ( 1 985). Development of synaptic currents in immobilized muscle of Xenopus laevis. .I. Physiol. 36457-68. LEBER,S. M. (1984). Effect of precocious and delayed afferent arrival on synapse localization on the amphibian Mauthner cell. Ph.D. Dissertation, Albert Einstein College of Medicine, NY. UMO, T., and ROSENTHAL, J. (1972). Control of Ach sensitivity by muscle activity in the rat. J. Physiol. 221 :493-5 1 3. MATTHEWS,S. A., and DETWILER, S. R. (1926). The reaction of Amhlystoma embryos following prolonged treatment with chloretone. .I. Exp. Zoal. 45279-292. MATTSON,M. P., LEE,R. E., ADAMS,M. E., GUTHRIE, P. B., and KATER,S. B. (1988). Interactions between entorhinal axons and target hippocampal neurons: A role for glutamate in the development of hippocampal circuitry. Neuron 153654376. MEYER,R. L. ( 1 982). Tetrodotoxin blocks the formation of ocular dominance columns in goldfish. Science 218589-591. MEYER,R. L. (1987). Intratectal targeting by optic fibers in goldfish under impulse blockade. Dev. Brain RCS.465:l 15-124. MODEL.P. G. (1978). Aspects of Mauthner cell differentiation in the axolotl, Ambystoma mexicanum. Amer. Zoo/. 18:253-265. MODEL,P. G., BORNSTEIN, M. B., CRAIN,S. M., and PAPPAS,G. D. (1971). An electron microscopic study of the development of synapses in cultured fetal mouse cerebrum continuously exposed to xylocaine. J. Cell Biol. 49:362-37 1. MODEL,P. G., and WURZELMANN, S. (1982). Vestibular axons form synapses on abnormally derived Mauthner cells. Dev. Bruin Res. 3: 123- 1 29. O'LEARY,D. D. M., FAWCETT,J. W., and COWAN, W. M. (1986). Topographic targeting errors in the retinocollicular projection and their elimination by selective ganglion cell death. J . Neurosci. 6:36923705. RTTMAN, R. H., and OPPENHEIM, R. W. (1978). Neuromuscular blockade increases motoneurone survival during normal cell death in the chick embryo. Nature 271~364-366. REH,T. A., and CONSTANTINE-PATON, M. (1985). Eyespecific segregation requires neural activity in threeeyed Runu pipiens. J.Neurosci. 5: 1 132- 1 143. SPITZER,N. C., and BACCAGLINI, P. 1. (1976). Develop-
294
Goodman and Model
ment of the action potential in embryo amphibian neurons in viva Brain Res. 107:610-616. SPITZER,N. C., and LAMBORGHINI, J. E. (1976). The development of the action potential mechanism of amphibian neurons isolated in culture. Proc. Natl. Acad. Sci. (USA) 73: 1641- I 645. SRIHARI, T., and VRBOVA,G. (1 978). The role of muscle activity in the differentiation of neuromuscular junctions in slow and fast chick muscles. J. Neurocyt d . 7~529-540. STRETAVAN, D. W., SHATZ,C. J., and STRYKER, M. P. ( 1 988). Modification of retinal ganglion cell axon morphology by prenatal infusion of tetrodotoxin. Nature 336468-47 1. STRYKER.M. P., and HARRIS,W. A. (1986). Binocular impulse blockade prevents the formation of ocular
dominance columns in cat visual cortex. J. Neurosci. 6:2 1 17-2 133. TWITTY,V. C. (1937). Experiments on the phenomenon of paralysis produced by a toxin occurring in Triturus embryos. J. Exp. Zool. 76:67-104. WALICKE,P. A., CAMPENOT, R. B., and PATTERSON, P. H. (1977). Determination of transmitter function by neuronal activity. Proc. Nutl. Acad. Sci. (USA) 745767-5771. WILLARD,A. L. (1980). Electrical excitability of outgrowing neurites of embryonic neurones in cultures of dissociated neural plate of Xenopus 1aevi.r. J. Pltysiol. 301~115-128. YOUNG,S. H., and Poo, M. (1983). Spontaneous release of transmitter from growth cones of embryonic neurones. Nuture 305634-637.
SUMMARY In the developing amphibian, the formation of extra vestibular contacts on the Mauthner cell (M-cell) enhances dendritic branching, while deprivation reduces it (Goodman and Model, 1988a). The mechanism underlying the interaction between afferent fibers and developing dendritic branches is not known; neural activity may be an essential component of the stimulating effect. We examined the role of afferent impulse activity in the regulation of M-cell dendritic branching in the axolotl (Ambystoma mexicanurn) embryo. M-cells occur as a pair of large, uniquely identifiable neurons in the axolotl medulla. Synapses from the ipsilateral vestibular nerve (nV11I) are restricted to a highly branched region of the M-cell
lateral dendrite. W e varied the amount of nVII1 innervation and eliminated neural activity. First, unilateral transplantation of a vestibular primordium deprived some M-cells of nVIII innervation and superinnervated others. Second, surgical fusion of axolotls to 'ITX-harboring California newt (Taricha torosa) embryos paralyzed the Ambystoma twin: voltage-sensitive Nat channel blockade by 'ITX eliminated action potential propagation. Reconstruction of M-cells in 18 mm larvae revealed that dendritic growth was influenced by ingrowing axons even in the absence of incoming impulses: impulse blockade had no effect on the stimulation of dendritic growth by the afferent fibers.
INTRODUCTION
Covell, and Model, 1988). We have begun to explore the nature of the transsynaptic stimulating effect imposed by afferent fibers on dendritic branching patterns: neural activity could be an essential component of the interaction. Certain aspects of neural development are independent of impulse activity. For example, prolonged immobilization of amphibian embryos by immersion in anesthetic solutions before the onset of motility does not interfere with or delay the development of normal motor function (Harrison, 1904; Carmichael, 1926; Matthews and Detwiler, 1926; Haverkamp and Oppenheim, 1986). In addition, elimination of impulse activity does not disrupt axonal pathfinding: retinal ganglion cells project to and synapse in the correct segment of the tectum (in salamanders, Hams, 1980, 1984; in goldfish, Meyer, 1987). Finally, the membrane properties underlying synaptic function in am-
A neuron's dendritic branching pattern is its most characteristic morphological feature. Ingrowing afferent axons are important in the regulation of dendritic growth on target neurons. For example, in the developing axolotl (Anzhystorna mexicanurn). the formation of extra vestibular (nVIII) contacts on the Mauthner cell (M-cell) enhances dendritic growth in the superinnervated region, while deprivation reduces it (Goodman and Model, 1988a). In older larvae, regenerating nVIII axons also stimulate dendritic growth (Goodman, Received July 27. 1989: accepted October 10. 1989 Journal ofNeurobiology, Vol. 21, No. 2. pp. 283-294 (1990) 0 1990 John Wiley & Sons, Inc. CCC 0022-3034/90/020283- 12$04.00 * To whom correspondence should be addressed.
283
284
Goodman and Model
phibian myotomal muscle (Kullberg, Owens, and Vickers, 1985)and cultured spinal neurons (Bixby and Spitzer, 1984) develop normally in the absence of nerve impulses. On the other hand, various aspects of neural development are activitydependent. Neuromuscular activity blockade rescues motoneurons from natural cell death (Pittman and Oppenheim, 1978), prevents elimination of extra motor nerve terminals (Srihari and Vrbova, 1978), and enhances intramuscular nerve branching (Dahm and Landmesser, 1988). Furthermore. blockade of nerve impulse activity modifies the distribution of acetylcholine receptors on target muscle fibers (L0mo and Kosenthal, 1972). alters the level of biosynthetic enzymes (e.g., Black and Geen, 1973) and neurotransmitters (e.g., Kessler and Black, 1982), precludes preferential loss of retinal ganglion cells with misdirected axons (O’Leary, Fawcett, and Cowan, 1986), and, in culture, affects neuronal survival (Bergey, Fitzgerald, Schrier, and Nelson, 1981) as well as neurotransmitter selection (Walicke, Campenot, and Patterson, 1977). Neuronal morphology is also modified by impulse blockade. For instance, blockade of retinal ganglion cells by application of tetrodotoxin (TTX) prevents the segregation and refinement of axonal arborizations into eye-specific columns in the tectum of goldfish (Meyer, 1 982) and tadpoles (Reh and Constantine-Paton, 1985), and precludes the formation of ocular dominance columns in the lateral geniculate nucleus (Sretavan, Shatz, and Stryker, 1988) and visual coitex (Stryker and Harris, 1986) of cats. There is evidence both for and against changes in dendritic organization due to loss of impulse activity. In mutant crickets, Bentley (1975) observed reduced dendritic branching on an identified interneuron that was deprived of normal synaptic activation. In frog embryos raised in anesthetic solutions, however, the branching of motoneurons was not altered by the absence of impulse activity (Haverkamp, 1986). We have examined the role of neural activity in the regulation of M-cell dendritic branching in the developing axolotl by interrupting synaptic transmission while preserving synaptic contacts. In the medulla of the axolotl, M-cells occur as a pair of large, uniquely identifiable neurons at the level of entry of nV111 (Hemck, 1914). Each cell receives synapses from the ipsilateral nerve; the terminals can be identified morphologically as club terminals in the light (Model, 1978) and electron (Kimme1 and Schabtach, 1974; Model and Wurzel-
mann, 1982) microscopes, and are restricted to a highly branched region of the M-cell lateral dendrite (Fig. I). In the developing embryo, the M-cell is recognized first at Harrison (1969) stage 34, and the earliest nVIII axons contact the growing lateral dendrite at stage 36 (Leber, 1984). To eliminate impulse activity, an experimental preparation devised by Twitty ( 1937) was used: axolotl embryos, stages 28-30, were surgicallyjoined side-by-side to tetrodotoxin (TTX)-harboring California newt (Taricha torosa) embryos in a condition known as parabiosis. In the axolotl, TTX blocks voltage-dependent Na+ channels and thus the propagation of action potentials. The fusion results in complete paralysis of the Ambystoma twin (Twitty, 1937; Harris, 1984; present results); the insensitive newt provides nutritional support for its immobile partner through early larval stages. In the presence of TTX, M-cells in the axolotl develop without incoming impulse activity. If synaptic activity is essential for the development of normal dendritic branching patterns, M-cells exposed to TTX will have reduced branching compared to control. Under normal conditions, the extent of dendritic branching vanes from animal to animal: dendritic loss from both cells in a single larva may be difficult to assess. In contrast, the extent of branching on the M-cells in a single larva is nearly identical, even though the precise pattern of branching may differ. Therefore, the most accurate comparisons are made between M-cells within a single individual. Experimental and control sides were created in each animal by performing unilateral ear transplants. Some experimental M-cells were deprived of nVIlI innervation by removal of a vestibular primordium from donor embryos; others were superinnervated with extra nVIIl afferents by implantation of an extra ear in hosts. The contralateral side served as the intrinsic control. All axolotl embryos were then fused to Taricha embryos. If impulse activity is essential to the stimulation of dendritic growth by ingrowing axons, superinnervated, deprived, and contralatera1 control M-cells in TTX-exposed larvae all should appear deprived. On the other hand, if activity is not required, the dendritic branching pattern of superinnervated M-cells should be enhanced and that of deprived M-cells should be reduced compared to the appropriate contralateral cells. Our data indicate that impulse activity is not essential. Preliminary results have been presented (Goodman and Model, 1988b).
Impulse Block Does Not Alter Brunching
MATERIALS AND METHODS Axolotl embryos used in these experiments were provided by the colony of axolotls maintained in our laboratory. California newt embryos were collected in thc wild by Dr. David Good of the Museum of Vertebrate Zoology, Berkley, CA. Operations were performed on midtailbud embryos at stages 28-30 in stcrile NiuTwitty's complex salt solution (pH 7.4-7.6) with neomycin sulfate (50 mg/liter). Ear transplants were performed on the same side of all axolotl embryos. After decapsulation, host embryos were prepared by unilateral removal of a small piece ofectoderm from a site immediately rostra1 to the in situ vestibular primordium. Donor tissue was prepared by unilateral excision of a vestibular primordium (together with its associated nVIII ganglion and overlying ectoderm). The excised vestibular primordium was then implanted in normal orientation in the site previously prepared in each host. Surgery was performed prior to nVIII outgrowth and M-cell differentiation. After the vestibular implants healed and the donor sites began to fill in (about 30 min), Amby.rtoma hosts and donors were parabiotically fused to Tarichu embryos: a patch of ectoderm was removed from the belly region of one side of axolotl embryos and from the opposite side of Turichu embryos. Each pair was positioned with the excision sites apposed and, over several hours. the ectoderm and endoderm became fused. The resulting parabiotic twins shared a blood circulation as well as some gut structures. TTX stored in the Taricha embryos was delivered to the sensitive axolotl partner via the shared circulation. The pairs were reared at 16°C in individual petri dishes through hatching to early feeding stages. Amhystoma larvae, immobilized by TTX, received nutritional sup-
285
port from the Turichu partners. Controls for TTX exposure were prepared by fusing several pairs of normal axolotl embryos. When axolotl larvae reached 18 mm, the parabiotic twins were anesthetized in MS 222, Axolotl heads were removed and immediately immersed in 3%1glutaraldehyde, 2% paraformaldehyde, 1% acrolein, and 2.5% DMSO in 0.1 Mcacodylate buffer (pH 7.4) at room temperature. After 24 h at 4"C, the heads were washed several times in 0.1 M cacodylate buffer (pH 7.4, total time 1.5 h). Heads were postfixed in cold 2% Os04 in cacodykdte buffer (2 h), washed in veronal acetate buffer, and stained en bloc in 4% uranyl acetate ( 1 h). Heads were dehydrated in a graded series of ethanols and embedded in Epon. Serial cross sections, 3 Fm thick, were cut from the heads and photographed in the light microscope. Pairs of M-cells in both TTX-exposed and control larvae were reconstructed from serial 3-pm-thick sections (Goodman and Model, 1988a). The morphology of experimental M-cells was compared to that of the contralatera1 cells. In addition, both M-cells in TTX-exposed larvae were compared to M-cells in unexposed controls. Several 3-gm-thick sections were mounted onto Epon blocks and sectioned for electron microscopy. Thin sections were collected on mesh grids, stained with uranyl acetate and lead citrate, and photographed in a Philips 300 electron microscope. Club endings were identified on M-cells in both TTX-exposed and control larvae. The morphology of the terminals was compared.
RESULTS
The experimental M-cell in 15 donor embryos was deprived of nVIII innervation by removal of a ves-
Figure 1 Light micrograph of an Epon-embedded, 3-pm-thick, toluidine blue-stained cross section through an M-cell in a control 18 mm larva. Vestibular axons (VIII) enter the medulla at the level of the M-cell and terminate on the ventral surface and branches (between arrowheads) of the lateral dendrite (Id). md = medial dendrite; a = axon; G = nVITI ganglion. Scale bar, 50 pm.
286
Goodman and M o d d
tibular primordium. The experimental cell in 15 hosts was superinnervated by implantation of an extra ear. Axolotl embryos were then surgically joined to Taricha embryos [Fig. 2(A,B)]. In addition, normal (unoperated) pairs of axolotl embryos were fused to each other as controls. By stage 34. control animals showed spontaneous slow contractures and responded readily to touch. In contrast, embryos fused to California newts showed no spontaneous movements and did not respond to even the most vigorous stimulation. Complete paralysis of the Ambystoma twin indicates that Na+-dependent action potentials were eliminated in the TTX-sensitive salamander. When the fastest-growing axolotl larva reached 19.5 mm, it began to show slight twitching movements of the head and gills. It was discarded and all of the remaining larvae were killed at 18 mm (about 3.5 weeks posthatching*). None showed any movement. At that stage, ears can be identified by white otoliths of the two maculae. On the experimental side of donors, the maculae were absent [Fig. 2(C)]; on the experimental side of hosts, extra maculae were visible rostra1 to the in situ ones [Fig. 2(D)]. Previously, we showed that vestibular axons from an ectopic ear enter the medulla at the level of nV and, confined to the nVIII tract, course caudad to innervate the appropriate region of the ipsilateral M-cell. We found that experimental M-cells received 30-67% more club endings than controls did, and that dendritic branching in the superinnervated region was enhanced: the supernumerary afferents stimulated extra dendritic growth (Goodman and Model, 1988a). In the present study, in addition to depriving or superinnervating experimental M-cells, afferent impulse activity was blocked by TTX throughout M-cell differentiation. Six donor and 2 host animals died prior to reaching 18 mm. Five of the remaining 9 pairs of M-cells in donor larvae were reconstructed. In the other donors, the experimental M-cell was absent or the medulla was disrupted. Nine pairs of Mcells in hosts were reconstructed. The remaining hosts were discarded because one separated from its Turichu mate, one was stunted, one was poorly fixed, and one (at 19.5 mm) showed movements. The unilateral ear transplants permitted comparisons between an experimental M-cell and the ap*Individual axolotls reach 18 mm about 12 days posthatching. The growth rate of axolotls fused to California newts is slowed, presumably because axolotls in parabiosis receive limited nutrition.
propriate contralateral cell (one that received a normal amount of innervation and was also exposed to TTX). The dendritic surface of deprived M-cells was significantly reduced in the region that normally receives nVIII innervation (Fig. 3). In contrast, the branching of most (6) superinnervated M-cells was much more abundant in the region receiving extra innervation (Fig. 4). In 2 host larvae, dendritic growth on thc experimental cell showed a small increase over the contralateral cell, and in only I larva, both cells appeared about the same. Taken together, the results show that exposure to TTX does not prevent the stimulation of dendritic growth by ingrowing aferent fibers. TTX-exposed cells were compared also to unexposed M-cells in control larvae (axolotls that were fused to other axolotls). If TTX brought about a substantial alteration in M-cell morphology, the change would be detected easily. On the other hand, if TTX brought about only a small alteration, the change might be obscure: exposed cells must be compared to unexposed cells in different individuals, and in different individuals, the extent of dendritic branching varies even under normal conditions. As a group, experimental TTX-exposed cells differed significantly from unexposed cells: deprived M-cells showed a dearth of branching (Figs. 3 , 5), and superinnervated cells showed an excess of branching in the region where nVIIl terminates (Figs. 4, 5). Comparing contralateral TTX-exposed cells (those receiving a normal complement of innervation) to unexposed controls revealed no apparent effect of TTX on M-cell morphology. Seven pairs of Mcells reconstructed from control (unexposed) axo l o t l ~revealed a range in complexity of dendritic branching (Fig. 5 ) . The degree of branching of the M-cells reconstructed from the control side of TTX-exposed axolotls spanned approximately the same range. These results indicate that TTX exposure did not alter dendritic growth or pattern, although due to the normal variability among individuals. subtle changes would not have been detected. The morphology of club endings in 3 TTX-exposed larvae was examined in the electron microscope and compared to unexposed controls. Club endings are easily identified because they are very large, make morphologically mixed (chemical and electrical) synapses, have mitochondria clustered near sites of contact, and contain densely packed. highly organized neurofilaments. TTX-exposed
Impulse Block Does Not iilter Branching
Figure 2 (A and B) Parabiotic twins 30 h after surgery. At the fusion site, some pigmented cells from the axolotl embryos (left) have migrated across to the California newt embryos (right). (A) Donor axolotl embryo with its vestibular primordium removed (arrowhead). (B) Host axolotl embryo with an extra vestibular primordium (double arrowhead) implanted rostra1 to the in silu one (single arrowhead). (C) and (D) Parabiotic twins with axolotl larvae (left) at 18 mm. Maculae are absent in the donor axolotl (C); an extra pair of maculae (double arrowhead) are visible in the host (D). Scale bars, 1 mm.
287
Figure 3 Thrce pairs of M-cells reconstructed from donor larvae. The axon projects toward the center of the page; the lateral dendrite, toward the side of the page. Deprived cells (left) are paired with the appropriate control (right). Dendritic branching of deprived cells is significantly reduced in the region that normally receives nVIII terminals (between arrowheads) compared to the appropriate control. Scale bar, SO pm. c l u b tcrminals were morphologically indistinguishable from controls (Fig. 6). Regions of contact ranged in size from 3 to 10 pm in micrographs from both experimental and control larvae.
DISCUSSION
In the absence of incoming impulse activity, developing M-cells elaborate a normal dendritic
Figure 4 Three pairs of M-cells reconstructed from host larvae. The branching of supennnervated cells (left) is greatly enhanced in the region where nVIII terminates (between arrowheads) compared to the appropriate control (right). Scale bar, 50 wm.
Impulse Block Does Not Alter Branching
289
B Figure 5 Two pairs of M-cells reconstructed 1From control larvae never exposed to TTX. The dendritic pattern of control cells varied from moderate (e.g., B) to more generous (e.g., A) amounts of branching. Scale bar, 50 pm.
branching pattern. The elimination of activity in axolotl embryos by surgically joining them to TTX-harboring California newt embryos produced complete paralysis of the Ambystoma twin. Immobilization of the salamander indicates that Na+-dependent action potentials were blocked and that M-cells developed in the absence of afferent impulse activity. Club endings formed on the appropriate dendritic surface of the M-cells and were normal insofar as their light and electron microscopic appearance was indistinguishable from control. Unilateral ear transplants, performed prior to parabiotic fusion, permitted an accurate evaluation of alterations in mature M-cell morphology: superinnervated and deprived cells were compared to the contralateral M-cell in the same individual. Under standard conditions, the usual complement of ingrowing nVIIJ fibers is required for normal development of the M-cell dcndntic branching pattern; the axons stimulate dendritic growth (Goodman and Model, 1988a). That TTX-exposed M-cells receiving a normal amount of nVJTI innervation span a range of branching patterns similar to unexposed controls indicates that the absence of nerve activity does not alter development of M-cell dendritic branching. Since the dendritic surface of superinnervated cells was increased and that of deprived cells, decreased, impulse blockade did not prevent the stimulation of dendritic growth by ingrowing afferents. Thus, neural activity is not an essential component of the stimulating effect. Taricha-supplied TTX produced a transient immobilization of axolotl larvae, i.e., the fastestgrowing salamander became mobile when it reached 19.5 mm. Twitty’s (1937) results were
similar: shortly after the newt’s yolk was used up, the axolotl twin recovered some motor function. The reason for the temporary paralysis is unknown. Twitty (1937) showed that Taricha eggs and embryos are the most potent sources of the toxin; extracts from larvae become progressively less effective with age. Thus, one possibility is that recovery occurs when the maternally-derived toxin becomes depleted. Recovery of impulse activity, however, does not occur in the optic nerve of axolotl eyes transplanted to California newt embryos, even after 6 months have passed (Hams, 1980). Furthermore, in the present work, axolotl larvae that grew more slowly, reaching 18 mm up to 10 days later than the average animal, remained completely immobilized. Another possibility is that the maturing salamander develops the ability to metabolize the toxin, rendering it inactive (Hams, 1984). Studies have shown that the development and maturation of functional synaptic networks is not contingent upon prior electrical impulse activity. For example, in cultured explants of fetal mouse cerebral cortex, synapses that formed in the presence of a local anesthetic were morphologically normal, as observed in the light (Crain, Bornstein, and Peterson, 1968) and electron (Model, Bornstein, Crain, and Pappas, 1971) microscopes. Furthermore, the potential for complex bioelectric activity also developed under conditions of impulse blockade (Crain et al., 1968). In the developing animal, Harris (1980, 1984)demonstrated that retinal ganglion cells formed synaptic contacts in the tectum in the absence of impulse activity. In the present report. blocked nVIIJ axons formed club endings that displayed characteristic ultrastruc-
290
Goodman and Model
Impulse Block Does Not Alter Branching
tural features: the large terminals made both chemical and electrical synaptic specializations, mitochondria clustered near the specialiiations. and neurofilaments were densely packed and organized. Our results agree with the data cited above: synapses that are normal in structure develop in the absence of electrical impulse activity. The presence of morphologically normal synaptic junctions, however, need not imply normal function. This issue was addressed in the classic works of Harrison ( 1 904), Carmichael ( 1926), and Matthews and Detwiler (1926) who showed that neither neuromuscular activity nor proprioceptive input are essential to the development of coordinated swimming behavior in amphibians. Even detailed quantitative measures of behavior do not distinguish normally reared amphibian embryos from those reared in anesthetic solutions (Haverkamp and Oppenheim, 1986). In addition, recordings of ventral root activity in previously immobilized embryos are normal (Haverkamp, 1986). Thus, there is considerable evidence that organized neuronal assemblies form in forward reference to their ultimate use from the standpoint of both morphology and function. It is important to determine whether the TTX circulating in axolotl embryos blocked all action potentials produced by ingrowing nVIII axons. In the developing amphibian, action potentials recorded in Rohon-Beard neurons and dorsal root ganglion neurons in vivo (Spitzer and Baccaglini, 1976; Baccaglini, 1978) and in spinal neurons in vitro (Spitzer and Lamborghini, 1976) show a shift in ionic dependence from Ca2+to Na’. In addition, the ionic basis of action potentials carried by outgrowing processes of individual spinal neurons in vitro follows the same progression (Willard, 1980). Therefore, it is likely that action potentials propagated by nVIII axons in axolotl embryos are initially Ca2+-dependentand insensitive to TTX. We do not know whether early Ca2+-dependent impulses affect developing postsynaptic dendrites. To determine whether they could be important, however, we would have to identify when during nVIII ingrowth the shift to Na+-dependence takes place. Since attempts to study the M-cell electro-
291
physiologically in axolotl embryos and larvae have been unsuccessful, we could not examine the developing system directly. However, due to several lines of evidence. it is probable that the change in ionic dependence takes place before axons reach their targets. For instance, we observed that TTX paralyzed the axolotl embryos through all stages of development: this observation suggests that motoneuron axons make only Na+-dependent (TTXblockable) spikes when they contact muscle. In addition, Kullberg, Lentz, and Cohen (1977) demonstrated that the earliest recorded end-plate potentials at the myotomal neuromuscular junction in Xenopus embryos were blocked completely by TTX. Likewise, retinal ganglion cells in eyes transplanted from Ambystoma into Taricha were silenced by the newt’s TTX, i.e., Ca2+spikes were never recorded in the target tectum (Hams, 1980). Further evidence that the ionic-dependence shift in growing axons takes place early comes from Willard ( I 980) who demonstrated that action potentials propagated by neurites extended from spinal neurons in culture undergo the change from Ca2’- to Na+-dependence considerably before the shift occurs in cell bodies. Moreover, the neuntes lose their capacity to make divalent cation-dependent adion potentials long before the cell bodies. He also showed evidence that neurites sprouted late may produce Na+-dependentaction potentials without ever producing Ca2+ spikes. Finally, the results of several studies in vitro are consistent with the view that axons can convert to Na+-dependent spikes prior to innervating their target: maturational changes in membrane properties occur independently of cellular contacts (Bixby and Spitzer, 1984; Henderson and Spitzer, 1986; Willard, 1980). Thus, it is reasonable to suppose that, in the axolotl, nVllI axons make Na+-dependent, TTX-blockable action potentials by the time the fibers contact the M-cell. In goldfish (and presumably in axolotls) action potentials propagating along nVIII axons excite the M-cell by way of electronic junctions as well as by release of chemical neurotransmitter (Faber and Korn, 1978). Impulse blockade by TTX probably eliminates most, but not all of the transmitter
Figure 6 Club endings in experimental and control larvae. The electron micrograph of the boxed area in each inset shows club endings (*) on a TTX-exposed (A) and on an unexposed control (B) cell. The club terminals are morphologically indistinguishable. Inset scale bar, 50 p m . EM scale bar. 2 pm.
292
Goodman and , o d e 1
release, i.e., it would not block spontaneous quantal release. Spontaneous release in the absence of impulses allows us to separate any theoretical effects of diffusable chemical transmitters from those of electrical activity. The possibility that transmitter may be a regulatory signal between developing nerve terminals and adjacent nerve processes has become more credible since transmitters have been shown to be released from growth cones in culture (Hume, Role, and Fischbach, 1983; Young and Poo, 1983). Several studies have demonstrated that the application of specific transmitters affects neurite growth in vitru. Serotonin, for example, inhibited the elongation of neurites on certain Helisorna neurons (Haydon, McCobb, and Kater, 1987). In addition, glutamate, spontaneously released from axons of entorhinal neurons, reduced dendritic outgrowth from cocultured hippocampal pyramidal neurons (Mattson, Lee, Adams, Guthrie, and Kater, 1988). TTX blocked the inhibitory effect of glutamate by reducing release of the transmitter, but the toxin produced its effect by reducing the number of synaptic contacts rather than by eliminating activity at intact terminals. Since chemical transmitters can influence neurite growth in vitro, perhaps neurotransmitters released from nVIII afferents growing in vivo are responsible for inducing growth of M-cell dendritic branches in the region where nVIII terminates. In the present work, however, TTX probably eliminated nearly all of the transmitter that would have been released by functioning nVIII axons, yet superinnervation still brought about extra dendritic branching. Thus, it appears that the transmitter is not responsible for promoting dendritic growth. On the other hand, since the synaptic endings are intact, miniature postsynaptic potentials are likely to occur in the absence of nerve impulses (as at the neuromuscular junction), i.e., there probably is some spontaneous quanta1 release of transmitter. Therefore, if very small quantities of transmitter are adequate to carry out potential trophic influences, electrical impulse blockade would not eliminate the chemical effect. Once the neurotransmitter in club endings is identified, one may be able to determine whether the compound has a role in inducing dendritic growth by inhibiting its action through the use of receptor blocking agents or, alternatively, enhancing its action through the use of chemical agonists. In summary, parabiotic fusion of axolotl embryos to TTX-harboring Turicha turusa resulted in immobilization of the axolotl twin. Voltage-sensi-
tive Na' channel blockade by TTX interrupted action potential propagation and thus eliminated impulse activity. In the presence of TTX, nVIII afferent fibers formed normal club endings on the appropriate dendritic surface and branches of the M-cell. Eliminating virtually all activity throughout embryonic development did not appear to affect the extent of dendritic branching on M-cells that received a normal complement of nVIII axons. Exposure to TTX did not alter or eliminate the transsynaptic stimulating effect that ingrowing afferent fibers exert on the morphogenesis of the M-cell: superinnervation enhanced dendritic branching while deprivation reduced it. Thus, afferent impulse activity is not required for normal development of postsynaptic dendritic branching patterns. We are indebted to Dr. David Good for providing the Taricha torosa eggs and thus making our study possible. We wish to thank Sarah Wurzelmann for the electron microscopy and for her expert preparation of publication prints. We also thank Kathleen Tinglin for her fine technical assistance. The work reported here was supported in part by a Martin Foundation fellowship awarded to L.A.G.
REFERENCES BACCAGLINI, P. I. (1978). Action potentials of embryonic dorsal root ganglion neurones in Xenoptis tadpoles. J. Physiol. 283585-604. BENTLEY,D. ( 1 975). Single gene cricket mutations: Effects on behavior, sensilla, sensory neurons, and identified interneurons. Scicnce 187:760-764. BERGEY,G. K., FITZGERALD,S. C., SCHRIER, B. K., and NELSON, P. G. (1981). Neuronal maturation in mammalian cell culture is dependent on spontaneous electrical activity. Brain Rex 207:49-58. N. C. (1984). Early differenBIXBY,J. L., and SPITZER, tiation of vertebrate spinal neurons in the absence of voltage-dependent Ca2+ and Na' influx. Dev. Biol. 106539-96. BLACK,I. B., and GEEN,S. C. (1973). Trans-synaptic regulation of adrenergic neuron development: Inhibition by ganglionic blockade. Brain Rex 63:29 1-302. CARMICHAEL, L. ( 1926). The development of behavior in vertebrates experimentally removed from the influence of external stimulation. Psychol. Rev. 335 1-58. CRAIN,S. M., BORNSTEIN, M. B., and PETERSON,E. R. (1968). Maturation of cultured embryonic CNS tissues during chronic exposure to agents which prevent bioelectric activity. Brain Res. 8:363-372.
Impulsc Block Does Not Alter Brunching DAHM, L. M., and LANDMESSER, L. T. (1988). The regulation of intramuscular nerve branching during normal development and following activity blockade. Dev. Biol. 130:62 1-644. FABEK, D. S., and KORN,H. (1978). Elcctrophysiology of the Mauthner cell: Basic properties, synaptic mechanisms, and associated networks. In: Neurobiology of rhe Muuthner Cell. D. S. Faber and H. Korn, Eds., Raven Press, NY., pp. 47-131. GOODMAN,L. A., COVELL,D. A,, and MODEL,P. G. (1988). Regenerating afferent fibers stimulate the recovery of Mauthner cell dendritic branching in the axolotl. J . Neurosci. 8:3025-3034. GOODMAN, L. A,, and MODEL,P. G. (1988a). Superinnervation enhances the dendritic branching pattern of the Mauthner cell in the developing axolotl. J . Neurosci. 8:776-79 1. GOODMAN, L. A., and MODEL,P. G. (1988b). The effect of eliminating afferent impulse activity on the morphogenesis of the amphibian Mauthner cell. Soc. Neurosci. Ahstr. 145 17. HARRIS,W. A. ( I 980). The effects of eliminating impulse activity on the development of the retinotectal projection in salamanders. J . Comp. Neurol. 194:303-317. HARRIS,W. A. (1984). Axonal pathfinding in the absence of normal pathways and impulse activity. J. Neurosci. 4: 1 153- 1 162. HARRISON, R. G. (1904). An experimental study of the relation of the nervous system to the developing musculature in the embryo of the frog. Amer. J. Anat. 3~197-220. HARRISON, R. G. (1969). Organization and Development of'the Embryo. Yale University Press, New Haven, CT. HAVERKAMP, L. J. ( 1986). Anatomical and physiological development of the Xenopus embryonic motor system in the absence of neural activity. .I. Neurosci. 6~1338-1348. HAVERKAMP, L. J., and OPPENHEIM, R . W. (1986). Behavioral development in the absence of neural activity: Effects of chronic immobilization on amphibian embryos. J . Nauosci. 6:1332-1337. HAYDON,P. G., MCCOBB,D. P., and KATER,S. B. ( I 987). The regulation of neuritc outgrowth, growth cone motility, and electrical synaptogenesis by serotonin. J . Neztrobiol. 18:197-215. HENDERSON, L. P., and SPITZER, N. C. ( 1 986). Autonomous early differentiation of neurons and muscle cells in single cell cultures. Dev. B i d . 113:381-387. HERRICK, C. J. (1914). The medulla oblongata oflarval Amblystoma. J. Comp. Neurol. 24:343-427. H U M E ,R. I.. ROLE, L. W.. and FISCHBACH, G. D. (1983). Acetylcholine release from growth cones detected with patches of acetylcholine receptor-rich membranes. Nature 305632-634. KESSLER, J. A., and BLACK,I. B. (1982). Regulation of
293
substance P in adult sympathetic ganglia. Bruin Res. 234:182-187. KIMMEL,C. B., and SCHABTACH, E. (1974). Patterning in synaptic knobs which connect with Mauthner's cell (Ambystoma mexicanum). J. Comp. Neurol. 156:4980. KULLBERG,R., LENTZ,T. L., and COHEN,M. W. (1 977). Development of the myotomal neuromuscular junction in Xenopus laevis: An electrophysiological and fine-structural study. Dev. Biol. 60: 10 1-129. KULLBERG, R., OWENS,J. L., and VICKERS,J. ( 1 985). Development of synaptic currents in immobilized muscle of Xenopus laevis. .I. Physiol. 36457-68. LEBER,S. M. (1984). Effect of precocious and delayed afferent arrival on synapse localization on the amphibian Mauthner cell. Ph.D. Dissertation, Albert Einstein College of Medicine, NY. UMO, T., and ROSENTHAL, J. (1972). Control of Ach sensitivity by muscle activity in the rat. J. Physiol. 221 :493-5 1 3. MATTHEWS,S. A., and DETWILER, S. R. (1926). The reaction of Amhlystoma embryos following prolonged treatment with chloretone. .I. Exp. Zoal. 45279-292. MATTSON,M. P., LEE,R. E., ADAMS,M. E., GUTHRIE, P. B., and KATER,S. B. (1988). Interactions between entorhinal axons and target hippocampal neurons: A role for glutamate in the development of hippocampal circuitry. Neuron 153654376. MEYER,R. L. ( 1 982). Tetrodotoxin blocks the formation of ocular dominance columns in goldfish. Science 218589-591. MEYER,R. L. (1987). Intratectal targeting by optic fibers in goldfish under impulse blockade. Dev. Brain RCS.465:l 15-124. MODEL.P. G. (1978). Aspects of Mauthner cell differentiation in the axolotl, Ambystoma mexicanum. Amer. Zoo/. 18:253-265. MODEL,P. G., BORNSTEIN, M. B., CRAIN,S. M., and PAPPAS,G. D. (1971). An electron microscopic study of the development of synapses in cultured fetal mouse cerebrum continuously exposed to xylocaine. J. Cell Biol. 49:362-37 1. MODEL,P. G., and WURZELMANN, S. (1982). Vestibular axons form synapses on abnormally derived Mauthner cells. Dev. Bruin Res. 3: 123- 1 29. O'LEARY,D. D. M., FAWCETT,J. W., and COWAN, W. M. (1986). Topographic targeting errors in the retinocollicular projection and their elimination by selective ganglion cell death. J . Neurosci. 6:36923705. RTTMAN, R. H., and OPPENHEIM, R. W. (1978). Neuromuscular blockade increases motoneurone survival during normal cell death in the chick embryo. Nature 271~364-366. REH,T. A., and CONSTANTINE-PATON, M. (1985). Eyespecific segregation requires neural activity in threeeyed Runu pipiens. J.Neurosci. 5: 1 132- 1 143. SPITZER,N. C., and BACCAGLINI, P. 1. (1976). Develop-
294
Goodman and Model
ment of the action potential in embryo amphibian neurons in viva Brain Res. 107:610-616. SPITZER,N. C., and LAMBORGHINI, J. E. (1976). The development of the action potential mechanism of amphibian neurons isolated in culture. Proc. Natl. Acad. Sci. (USA) 73: 1641- I 645. SRIHARI, T., and VRBOVA,G. (1 978). The role of muscle activity in the differentiation of neuromuscular junctions in slow and fast chick muscles. J. Neurocyt d . 7~529-540. STRETAVAN, D. W., SHATZ,C. J., and STRYKER, M. P. ( 1 988). Modification of retinal ganglion cell axon morphology by prenatal infusion of tetrodotoxin. Nature 336468-47 1. STRYKER.M. P., and HARRIS,W. A. (1986). Binocular impulse blockade prevents the formation of ocular
dominance columns in cat visual cortex. J. Neurosci. 6:2 1 17-2 133. TWITTY,V. C. (1937). Experiments on the phenomenon of paralysis produced by a toxin occurring in Triturus embryos. J. Exp. Zool. 76:67-104. WALICKE,P. A., CAMPENOT, R. B., and PATTERSON, P. H. (1977). Determination of transmitter function by neuronal activity. Proc. Nutl. Acad. Sci. (USA) 745767-5771. WILLARD,A. L. (1980). Electrical excitability of outgrowing neurites of embryonic neurones in cultures of dissociated neural plate of Xenopus 1aevi.r. J. Pltysiol. 301~115-128. YOUNG,S. H., and Poo, M. (1983). Spontaneous release of transmitter from growth cones of embryonic neurones. Nuture 305634-637.
