| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Biology, University of Virginia, Charlottesville, Virginia 22903
Submitted 28 March 2003; accepted in final form 25 May 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Spruston et al. 1995). Nonetheless, new aspects of dendritic function in even well-studied central neurons continue to be discovered (Magee 2000; Thomas et al. 1998). Furthermore, since only a small number of the many kinds of central neurons has been examined in any detail, we can expect that novel functional properties of dendritic structure will continue to emerge as electrophysiological techniques are brought to bear on previously unexamined kinds of central neurons in both vertebrate and invertebrate preparations.
Parasol cells are multimodal sensory interneurons of the crustacean lateral forebrain (a structure forming the most proximal neuropil within the eyecups of crayfish, lobsters, crabs, and shrimps) (Blaustein et al. 1988; Mellon 2000, 2002). The expanded dendritic trees of parasol cells comprise and define a morphologic feature of the lateral forebrain called the hemiellipsoid body (HEB), which, along with the terminal medulla (MT), is believed to be critical for the initiation of feeding behavior (Maynard and Yager 1968). Parasol cells receive indirect olfactory, tactile, and visual sensory input via midbrain projection neurons and photic input from visual ganglia in the ipsilateral compound eye (Mellon 2000; Sandeman et al. 1995; Sullivan and Beltz 2001; Wachowiak et al. 1996). Parasol cells [which we formerly called lateral protocerebral interneurons (LPIs)] are anatomically and functionally polarized, with a dense apical dendritic arbor mostly confined to the HEB, a soma at the end of a long neurite in the ventromedial MT, and a characteristically recurved axon with branched terminals in the caudal MT (McKinzie et al. 2003; Mellon and Alones 1997).
Parasol cells are of special interest because they exhibit two forms of electrical activity also observed in the central pathways of other animals (e.g., Huerta and Lisman 1995; Kim and McCormick 1998) but not understood in terms of their functional significance in any system. These two forms of activity, population-synchronous oscillations in membrane potential and stimulus-driven impulse bursts, attract interest because 1) they involve the dendrites themselves and therefore can directly influence the integrative functions of the dendritic arbor, and 2) they resemble activities seen in other invertebrate and vertebrate brain cells; thus this crustacean preparation can serve as a relevant and experimentally accessible model for these generally distributed physiological properties.
Intracellular recordings from parasol cell dendrites exhibit periodic excitatory postsynaptic potentials (EPSPs), "background activity," on which sensory input generates additional depolarizations and resultant impulses. Strong sensory stimulation also evokes one or more impulse bursts, occurring at the peaks of the background activity (Mellon 2000; Mellon and Alones 1997; Mellon and Wheeler 1999). Since, aside from occasional single spontaneous bursts, these cells require strong synaptic input to generate bursts, they may be classified as conditional bursters. Periodic EPSPs among individual parasol cells in each of the two major neuropils of the HEB are coherent and phase-locked (McKinzie et al. 2003), presumably because they are common targets for spontaneously active neurons within the MT (Mellon and Wheeler 1999); this coherence also assures that some solitary spikes and impulse bursts generated in individual parasol cells by sensory input will be synchronous, and this synchronous output may be of crucial importance in terms of network properties. However, there may be other advantages to a continuous barrage of depolarizing input that concern the dynamic state of specific membrane conductances in these neurons. Moreover, in parasol cells (and some other arthropod central sensory pathways), impulse bursts depress subsequent activity for periods of 2–5 s (Mellon 2001, 2003; Ogawa et al. 2001; Sobel and Tank 1994). Here I present an electrophysiological description of these highly compartmentalized, functionally complex central neurons, including experimental evidence for a dendritic trunk initiation point for orthodromic impulses and electrical coupling between small groups of parasol cells.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
For electrophysiological experiments, crayfish were chilled in crushed ice for 10 min and decapitated. The isolated head was mounted in a recording chamber with the antennules protruding into and sealed within an olfactometer, and the right-hand eyestalk firmly secured in space by means of heat-shrink tubing glued over the cornea (Mellon and Wheeler 1999). The chamber was then flooded with chilled crayfish saline, and the medial cerebral artery and the left-hand lateral cephalic artery were cannulated and flushed continuously with oxygenated, chilled crayfish saline composed of (in mM) 205 NaCl, 5.4 KCl, 13.6 CaCl2 · 2H2O, 2.7 MgCl2 · 7H2O, and 2.4 NaHCO3, pH adjusted to 7.4 with concentrated HCl. The right eyecup then was subdissected to expose the lateral forebrain. The loosely applied connective tissue sheath surrounding the neural axis was torn with forceps to reveal the HEB, a dorso-medial protuberance on the MT. Hemolymph and loose glial tissue were then gently washed away with a saline-filled syringe, and the major lobe of the HEB (neuropil II of Sullivan and Beltz 2001) was directly penetrated with glass micropipette electrodes filled with 3 M potassium acetate. Dendritic trunks recordings were obtained by rotating an eyestalk 45° clockwise about its long axis medially so that the microelectrode approached the bundle of dendritic trunks at a 90° angle. The electrode was advanced 200–350 µm below the surface of the MT prior to contacting parasol cell trunks. Electrodes were connected via Ag2+-AgCl bridge electrode holders to an Axoclamp 2B amplifier (Axon Instruments) operated in bridge mode, via head stages having a gain of 0.1. For experiments that involved current passage through the recording electrode, the bridge was balanced after the recording microelectrode had penetrated the cell membrane and the membrane potential had stabilized. Because the time constant of these extensively branched cells is much longer than that of the microelectrode, we were able to use the following technique. Six hundred-microsecond hyperpolarizing current pulses (0.1–0.2 nAmp) were passed through the recording electrode at 10 Hz, and the bridge was adjusted to provide for the optimum zeroing of the voltage trace following decay of the initial capacitative transient. Nonetheless, in most cases, large capacitative voltage transients remained at the onset and cessation of the experimental current pulses, presumably caused by excessive, uncompensated electrode tip capacitance due to its depth within the surrounding tissue. Extracellular field potentials from parasol cell were recorded using the microelectrodes filled with 1 M NaCl and broken off to tip diameters of 10–20 µm. Parasol cell somata were directly stimulated using a pair of fine (50 µm) silver wire electrodes insulated to their tips, placed adjacent to the parasol cell soma cluster on the ventral surface of the MT. Grass S48 stimulators and isolation units were used to generate brief (0.1–0.5 ms) electrical stimuli to the parasol cell somata or to the antennules. A suction stimulating electrode placed on the contralateral accessory lobe (AL) was used to generate volleys in the AL projection neurons. Electrical signals from parasol cells were filtered at 10 KHz, digitized, and stored in computer files using a Digidata 1320 series A/D converter and Axoscope 8.2 software (Axon Instruments). Some electrical data were analyzed graphically using Microcal Origin software.
In isolated head preparations, parasol cells continued to exhibit background activity and to respond to sensory input for periods lasting up to 7 h after the initial dissection. The antennules were stimulated either by an odorant solution (0.04% tetramin) or by electrical stimuli through paired silver wires imbedded in the walls of the olfactometer, through which a continuous flow of dechlorinated fresh water, or episodically, odorant pulses, were flushed. Odorant pulses and current passed into the recording electrodes were controlled by a Pulsemaster A300 pulse generator manifold (World Precision Instruments). A halogen dissection lamp trained on the ipsilateral retina by means of a fiber optic cable, and a mechanical shutter accomplished photic stimulation of the preparation. All experimental observations were made at a bath temperature of 18°C.
To stain neurons with neurobiotin, individual parasol cell dendrites were penetrated with microelectrodes filled with 2% neurobiotin (Vector Laboratories, Burlingame, CA) in 3 M potassium acetate. Following penetration, 1–2 nAmp of positive current were passed into the cell for about 30 min in 500-ms pulses, recurring at 1 Hz, to iontophorese neurobiotin from the electrode tip. The entire head was fixed overnight in a solution of 2% glutaraldehyde (Fisher Scientific, Pittsburgh, PA) and 2% paraformaldehyde (Fisher Scientific) in 0.1 M phosphate buffer (PB), adjusted to pH 7.4. On the following day, the eyestalk neural centers with the treated cells were dissected from the head, rinsed in 0.1 M PB, and embedded in 14% 100-bloom gelatin (Fisher Scientific). The gelatin blocks were fixed for 1–2 h in 4% paraformaldehyde in 0.1 M PB and cross-sectioned on a vibratome at a thickness of 100 µm. The sections were rinsed in 0.1 M PB for 1 h and then incubated for 72 h at 5–6°C in ABC solution (Vector Labs) in 0.1 M PB/0.3% Triton X-100 (Sigma-Aldrich, St. Louis, MO) solution. After this, sections were rinsed again for 1 h in 0.1 M PB and treated with 3,3'-diaminobenzidine (DAB; Sigma-Aldrich) in a 0.1 M PB/0.5% nickel ammonium sulfate solution at room temperature for 10 min with DAB solution alone, followed by 5 min in DAB solution with 0.2% hydrogen peroxide (Fisher Scientific) for optimal visualization of the peroxidase. The sections were rinsed for 3 x 10 min in cold PB to stop the DAB reaction and were mounted on slides coated in 0.5% pigskin gelatin (Sigma-Aldrich) overnight. The slides were dehydrated in an ethanol series, cleared in xylene, coverslipped using DPX mountant (Fluka Chemicals, Buchs, Switzerland), and examined and photographed on a Zeiss axiophot microscope (Carl Zeiss). Measurements of neuronal processes were made microscopically using an ocular micrometer calibrated with a stage micrometer.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Figures 1 and 2 illustrate the unique morphological properties of parasol cells as a basis for understanding the data that follow. Parasol cell dendrites occupy two separate neuropils that form the two lobes of the crayfish HEB (McKinzie et al. 2003). Parasol cells in neuropil I of the HEB receive synaptic input from the ipsilateral AL; parasol cells in neuropil II receive input from both ipsi- and contralateral ALs, and possibly, from olfactory lobe (OL) projection neurons (Sullivan and Beltz 2001). Synaptic inputs to parasol cells from AL projection neurons are made within microglomeruli of the profusely branched dendritic arbors in the HEB (Mellon 2000; Mellon et al. 1992a). In each neuron, three to four thick (5–6 µm) basal dendritic branches converge on a single dendritic trunk, about 10 µm in diameter, which penetrates the MT, eventually becoming a much thinner neurite that courses through the MT to the neuronal soma. Each basal branch connects to the trunk via a thin (about 1.0 µm diam) isolating segment, having a variable length that ranges from 5 to 40 µm in different branches. The output axon of a parasol cell branches off and curves away from the junction of the trunk, and the neurite and makes synaptic contacts with unidentified neural targets in the lateral MT (Mellon et al. 1992a). Figure 2 shows vibratome sections of portions of four parasol cells that had been filled with biocytin and stained with diaminobenzidine. In both sections, the thin isolating segments connecting the trunk and basal branch compartments are evident, as is the variable nature of their respective lengths.
|
|
Recordings from basal branches
Figure 3A shows electrical responses recorded simultaneously with sharp electrodes from the dendritic branches of two crayfish parasol cells in neuropil II. As shown, these neurons typically exhibit periodic EPSPs, or "background depolarizations," which are imposed on the cell at a frequency of about 0.5 Hz, usually generating a solitary action potential at each peak of depolarization (Mellon and Alones 1997). Furthermore, the periodic background input is coherent among the population of parasol cells in neuropil II; thus all of the cells exhibit mostly identical, phase-locked depolarizations (Mellon and Wheeler 1999). Slight variations in background waveform between cells are attributable to recordings sites that are, respectively, at different distances from the terminals of possibly several presynaptic, spontaneously active neurons.
|
The basal branch dendritic action potentials generated by the periodic depolarizations are large and usually overshoot the zero potential level. They have a time course of 6–8 ms at half-amplitude (n = 14, 7.3 ± 2.02), or about an order of magnitude longer than axonal impulses recorded from central or peripheral nerves in the crayfish (e.g., Furshpan and Potter 1959; Mellon and Kennedy 1964). These differences suggest that, as with some other arthropod collecting sensory interneuron dendritic trees (Borst and Egelhaaf 1992; Single and Borst 1998), the underlying ion channel kinetics of the basal branches reflect significant contributions from voltage-gated Ca2+ conductances.
More than one-half (133 of 250 instances) of the records obtained from parasol cell dendrites, as in Fig. 3A (bottom trace), exhibit low-amplitude transients that are also timed to the peaks of the periodic depolarizations. As discussed below evidence suggests that these events are spikes conducted electrotonically from neighboring, electrically coupled parasol cells.
Overshooting impulse bursts
Spontaneous bursts of impulses occasionally are generated near the peak of the depolarizations in parasol cells (Fig. 3A, asterisk in top trace). Their form is identical with impulse bursts generated by sensory input (Mellon 2002; Mellon and Alones 1997; Mellon and Wheeler 1999). The bursts last 200–300 ms and exhibit an approximate parabolic intraburst spike frequency profile. Mean intraburst impulse frequencies are about 40 Hz. Figure 3, B and C, illustrates a consistent feature of impulses and impulse bursts recorded from parasol cell dendrites with sharp electrodes, namely, a difference in voltage threshold between the periodic solitary impulses and those impulses within the burst itself. The dashed line in Fig. 3B denotes the approximate threshold for solitary action potentials as recorded at a site in the basal branches in this neuron. The consistently observed disparity between solitary and burst spike thresholds is easily seen by examining the usual preburst spike occurring at a variable latency (40–70 ms) prior to the burst itself. Figure 3C shows the burst of Fig. 3B at a higher gain and sweep speed and also includes definitions of burst structure that are used throughout this paper. I hypothesize that solitary impulses, including those immediately leading a burst, are initiated at a cellular region or regions in response to the periodic depolarizations, from which point they must either propagate from a more peripheral locus past the recording site to the axon, or back-propagate into the membrane at the recording site on the basal dendritic branch from a locus closer to the axon. Impulse bursts, which also are associated with the periodic depolarizing peaks, must have an initiation point that is more distant from the recording site, however, as they not only arise from a voltage threshold that is 4–10 mV more negative than that of the solitary spikes but do so in the wake of the preceding hyperpolarizing afterpotential. Additionally, when the membrane is depolarized to a more positive level by passing current through the recording electrode (Fig. 4, A and B), solitary spike trains are generated at potential levels that are subthreshold for impulse bursts (n = 5). If, as in some other crustacean central neurons, impulse bursts arise from the activation of voltage-gated plateau potentials, bursts may be initiated at a specific circumscribed region of the cell, situated in a different compartment than the basal branches. If this interpretation is correct, then hyperpolarizing the membrane at the recording site should depress solitary impulse initiation at membrane potential levels less negative than those that are required to block invasion of impulse bursts. The recordings in Fig. 4C indeed show that hyperpolarization of the membrane by passing current through the recording electrode suppressed initiation of solitary impulses while permitting invasion of the recording site by spontaneous impulse bursts (n = 5); light-evoked bursts also were invading at these levels of hyperpolarization (data not shown). These experimental data suggest that bursts arise at a cellular locus that is consistently more distant from the recording site in the basal branches than the point at which solitary impulses are generated.
|
Electrotonic spike events
Dendritic recordings from more than one-half of parasol cells exhibit not only the invading impulse events described above, but also classes of recurring small-amplitude (2–10mV) transients having a constant size and time course (cf., Fig. 3A). The frequency of occurrence of the transients is similar to that of the solitary, overshooting action potentials; they occur near the peaks of the periodic depolarizations, and they are frequently grouped into bursts that, as shown by the bar graphs in Fig. 5, have time course and frequency properties similar to those of bursts that actively invade the recording site. As many as three different amplitude classes of these miniature bursts have been seen in some recordings from individual parasol cells. Figure 5, A–D, shows records in which two amplitude classes (S1 and S2) of miniature bursting transients, in addition to overshooting impulses and bursts, were observed in a parasol cell. Since miniature bursts are generated around the peaks of the periodic background depolarizations, this occasionally leads to a temporal overlap not only with invading bursts, but with each other as well. Figure 5, C and D, shows cases in point, showing instances in which the two distinct classes overlap in time, moving in and out of phase with each other and, in consequence, generating occasional summations, as well as overlapping in time with an overshooting impulse burst.
|
Strong sensory input, such as odorant or electrical stimulation of the antennules, evokes EPSPs that are superimposed on the background activity, generating multiple impulses and delayed impulse bursts (Mellon 2000, 2003; Mellon and Alones 1997). This is also the case for miniature transient bursts recorded from parasol cells. Thus the physiological properties of impulse bursts that invade the recording site are exhibited by the miniature transients as well and are consistent with the identity of the miniature bursts as spike events recorded by the electrode at a distance. These events could arise in the neuron impaled by the recording electrode as noninvading dendritic branch spikes, as has been previously suggested (Mellon 2002) or they could arise in neighboring cells to which the impaled neuron is electrically coupled. To address the question of the origin of the miniature potentials, I designed an experimental approach in an attempt to rule out the branch spike hypothesis. I tested whether individual parasol cells exhibiting miniature events in addition to overshooting spikes revealed unique stimulus thresholds for each class of response, when the soma cluster was stimulated directly, evoking antidromic1 spikes, at different intensities of electrical shock. If the electrotonic spikes represent blocked activity in different dendritic branches of the same parasol cell, then, to a recording electrode within a basal branch, threshold antidromic stimulation ought to generate either an invading impulse or one amplitude class of electrotonically recorded spike, depending on where the invasion block may reside with respect to the recording location on the dendritic tree. If, on the other hand, the electrotonic activity represents spikes recorded from an electrically coupled parasol cell, two distinct thresholds should be detectable for the invading and the electrotonic spikes, respectively. This disparity should be most easily observable when the threshold to electrical stimulation for the cell generating the electrotonic activity is lower than that for the overshooting antidromic impulse.
Accordingly, stimulus intensity series were delivered to the parasol cell soma cluster in the ventral MT while recording from a basal dendritic branch in neuropil II of the HEB. The results of these experiments, shown in Fig. 6, provide unequivocal evidence for the presence of miniature event thresholds that, when fortuitous placement of the stimulus electrode pair in relation to the soma cluster so dictated, were lower than those for overshooting impulses recorded at the same site. Furthermore, in most cases, hyperpolarizing currents passed through the recording electrode could be used to block the invading impulse generated by the higher stimulus intensity. Data from one experiment in a cell exhibiting a single class of miniature transient events are shown. In this case, which is typical, a low-intensity shock to the soma cluster generated an antidromic miniature spike first, which was then obscured by the invading antidromic spike at higher stimulus intensities. Hyperpolarizing the membrane by current injection blocked the invading spike but not the electrotonically recorded spike. Although we cannot know precisely the region where the antidromic spikes were initiated by the stimulus shock and whether or not they were evoked at the soma, the primary neurite, or the junction between the axon, neurite, and dendritic trunk, these results can only be interpreted as two different amplitude classes of antidromic events arriving at the recording site over separate conduction pathways having different stimulus thresholds, that is, in different parasol cells. These results therefore are consistent with the hypothesis that the miniature events originate in separate, electrically coupled parasol cells. They are also consistent with findings in our previous studies that injections of neurobiotin or biocytin into one parasol cell usually results in the staining of four to six neighboring parasol cells (McKinzie et al. 2003; Mellon and Alones 1997; Mellon et al. 1992b).
|
If we accept that small groups of parasol cells may be electrically coupled to one another, a question arises as to whether different sensory inputs to these multimodal interneurons evoke identical responses in each of those cells; in other words, do parasol cells that are electrically coupled to one another receive identical or different strengths of each sensory modality? An instance is shown in Fig. 7 in which different modalities of sensory inputs were used to evoke activity in a parasol cell. In these records, the responses of the impaled neuron to photic and antennular stimulation showed qualitative differences; furthermore, there were quantitative differences in the response categories to the two stimulus modalities in the impaled neuron and a neuron electrically coupled to it. These findings suggest that the efficacy of different kinds of sensory input to the different electrotonically coupled neurons is not identical.
|
Burst effects on activity
In a number of invertebrate central neurons, high-frequency trains of dendritic spikes and bursts have functional consequences that include postburst suppression of EPSPs (Ogawa et al. 2001; Sobel and Tank 1994; Wessell et al. 1999). I have reported previously (Mellon 2003) that parasol cells show similar behavior; following impulse bursts, the parasol cell membrane is hyperpolarized for 2–4 s, suppressing background-generated impulses as well as the EPSPs and spikes evoked by sensory input. More current data confirming these findings are shown in Fig. 8. Figure 8A is a peri-event versus time histogram showing the number of periodically evoked impulses during six-second time periods prior to and following 39 spontaneous bursts that invaded the recording site. The data confirm that the frequency of impulses evoked by the periodic depolarizations is reduced during a 2–4 s period following each burst. If the high-frequency groupings of small transients in fact represent electrotonically recorded impulse bursts arising in neighboring, electrically coupled cells, there should be local functional consequences similar to those following actively invading bursts. The histogram of Fig. 8B shows that in this preparation, the frequency of solitary, transient potentials is indeed depressed following each miniature burst, with a time course that is commensurate with the depression after full-amplitude bursts.
|
Trunk recordings
To determine the compartment of origin of sensory-evoked single spikes, I recorded simultaneously from pairs of parasol cells in, respectively, a dendritic basal branch within the HEB and a dendritic trunk in the MT. The trunk bundle was accessed by advancing a recording electrode 250–350 µm into the MT, just below the center of the base of the HEB. Figure 9A shows a diagram of the recording situation. Typical examples of recorded activity are also shown. Spikes recorded in the trunk region have dynamic properties that differ from those recorded in the basal branches; they have a shorter time course (4.0 ± 0.43 ms, n = 12), and their rise times are more rapid and they have a more pronounced undershoot than those recorded in the basal branches. Furthermore, in most recordings, the amplitude and rate of rise of the background depolarizations measured by a trunk electrode usually are smaller than those observed in the basal branches. Under most circumstances, these differences allowed me to identify the dendritic compartment from which I was recording activity. Characteristically, an intracellular trunk electrode records antidromic impulses 2–3 ms prior to an electrode recording from a basal branch. To determine whether consistent timing differences could also be observed at the two recording electrodes in response to repeated sensory input, we initiated maximal impulse volleys in the olfactory globular tract (OGT) by stimulating the contralateral AL with a constant electrical stimulus (0.2 ms) at a fixed interval of 10 s. Volleys in the OGT massively excite the parasol cells, leading to large EPSPs and resultant spikes (Mellon and Wheeler 1999; Mellon et al. 1992b). On some occasions, orthodromic spikes recorded by the trunk electrode in response to these volleys preceded the corresponding spike recorded in the HEB, but this was not invariably the case, and variations in latency of spike arrival at both recording sites was a confounding factor in some preparations. Therefore because of this timing variability in the records from individual parasol cells, I decided to approach to the question of orthodromic impulse origin using extracellular recording techniques, to emphasize the population-averaged field potential responses evoked by maximal volleys, thereby reducing the variability present in individual neurons, again using arrival times of the orthodromically generated impulses recorded both near the trunk and within the HEB. While field potential records can be difficult to interpret, I was helped by the fact that the HEB and parasol cell dendritic array are symmetrical about a central fascicle of dendritic trunks that course together into the MT. Furthermore, the AL excitatory synaptic inputs are arrayed as a spherical shell around the periphery of the HEB, with radially arrayed basal branches extending from this input zone toward the central trunk fascicle. Therefore compared with a distant indifferent electrode, an active extracellular electrode should record field potentials as a predictable function of depth within the HEB and time following arrival of a presynaptic volley. Figure 9B shows paired intracellular recordings from trunk and basal branch electrodes, while Fig. 9C shows examples of evoked field potentials recorded near the trunk and from an electrode about 150 µm below the surface of the HEB, near the basal branches, in response to volleys in one preparation. The field potential data were more consistent than the paired intracellular records from individual neurons; they showed that the trunk electrode invariably recorded evoked, orthodromic compound action potentials 1–2 ms prior to the electrode within the HEB, in the vicinity of the basal branches. Characteristic frequency histograms of latencies for basal branch and trunk field potentials are shown in Fig. 9D for 32 volleys; t-test comparisons of the paired latencies in this and five other preparations indicated that the arrival times at the respective locations were signifi-cantly different (P < 0.05). Furthermore, the latency differences of orthodromic spikes in any one preparation were either identical with, or less than, those in response to antidromic activation of the parasol cells. These data therefore support a dendritic trunk locus for orthodromic impulse initiation, probably close to the point of emergence of the basal branches at the junction of the MT and HEB, from which point spikes proceed toward the axon and also back-propagate into the dendritic arbor.
|
The locus of burst initiation was not directly investigated; nonetheless, two kinds of observations allow for some speculation concerning burst generation in parasol cells. As was noted above, results suggest that bursts arise at a different point on the parasol cell dendritic complex than do spikes generated by background activity. Additionally, I have on at least one occasion recorded from a position on the trunk of a parasol cell that may have been close to its junction with the primary neurite, the trunk, and the axon. In records obtained from this ventral location, the spikes were very fast (mean duration at half-amplitude = 1.2 ± 0.05 ms, n = 25), and interestingly, the usual threshold disparity between the preburst spikes and those within the burst observed in basal branch recordings was reversed; that is, the preburst spikes arose from a membrane potential more negative than the thresholds for the spikes within the burst. By logic similar to that used in interpreting the threshold differences between single spikes and bursts at basal branch recoding sites, these observations suggest that the individual spikes arrive at the recording electrode from an initiation site now more distant than that for the spikes within the burst. Furthermore, at this recording location, the spike array in the burst appeared to be generated by a smoothly rising and falling depolarization superimposed on the usual background depolarizations, and possibly representing driver potentials. These features can be seen in the records of Fig. 10, A and B. Additionally, when depolarizing current was passed into the parasol cell through the recording electrode, trains of spike bursts were initiated (Fig. 10C) at current strengths less than those required to generate continuous trains of spikes. These data provide circumstantial evidence that the recording electrode was close to a site for burst generation on the parasol cell, which in this case was on the trunk well ventral to the basal branches.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
,b; Sandeman et al. 1995; Sullivan and Beltz 2001), and local oscillatory neurons, the presumed source of the periodic depolarizing potentials (Mellon and Alones 1997; Mellon and Wheeler 1999). Additional inputs are the ipsilateral OL projection neurons, which anatomical evidence indicates may contact parasol cell dendrites below the base of the HEB, within the MT (Sullivan and Beltz 2001), and ipsilateral photic input, usually the most effective stimulus for most parasol cells (Mellon 2000). While synaptic contacts made by AL projection neurons are known to be near the distal tips of the parasol cell dendritic branches (Mellon et al. 1992a), sites of the other synaptic inputs are currently unknown. The present data from recordings obtained simultaneously from sites near the trunk and the basal branches of parasol cells indicate that the initiation zone for postsynaptic impulses is on the trunk, probably close to the point of emergence of the basal branches. From this point, spikes must back-propagate into the basal branches, as well as down the trunk to the axon. Although the major synaptic input from the AL projection neurons is at the peripheral dendritic branches of the parasol cells, the excitatory postsynaptic currents must spread throughout the basal branches to converge at the trunk, where apparently the impulse threshold is lower than in the basal branch compartment itself. My findings do not necessarily preclude a basal branch impulse initiation site following especially strong presynaptic input, as apparently occurs in the mitral cell dendrites of the vertebrate olfactory bulb (Chen et al. 2002); however, from considerations of cellular morphology, as exemplified in Fig. 2, this direction of impulse propagation would be problematical. Invasion of the expanded region of the dendritic trunk (the region from which the basal branches diverge) by a spike propagating from one of the basal branches should be marginal, due to the small diameter of the isolating segments connecting each basal branch with that region of the trunk. The impedance load of the expanded trunk (and other basal branch) membrane experienced by a spike invading from such a small-diameter current source could be severe (Goldstein and Rall 1974), preventing activation of the trunk membrane (unless spikes from several basal branches converged on the trunk simultaneously). During periods of high impulse frequencies, e.g., in bursts, these safety factor problems would be exacerbated. In fact, recent studies using computational simulations of dendritic morphology and ion channel kinetic properties suggest that dendritic morphology is an especially critical factor in determining invasion of back-propagating impulses into branching arbors (Vetter et al. 2001). Thus the isolating segments may provide some compensation in the impedance mismatch at this multiple branch point, which otherwise might severely attenuate impulse back-propagation. Therefore physiological and morphological considerations both suggest a trunk locus as the most likely site for impulse initiation in these neurons.
The site of burst initiation in parasol cells is more problematical. While our evidence suggests that burst initiation occurs in a cellular compartment that is distant from recording electrode sites within the basal branches, that compartment could either be distal or proximal to the recording locus. Spontaneous bursts normally arise in the wake of a spike generated by the background depolarization, whereas sensory-driven bursts, which also are timed to the peaks of the background activity, are not invariably preceded by an orthodromic spike. These observations suggest that both spontaneous and stimulus-evoked bursts arise as a consequence of the membrane potential exceeding a threshold value at some critical site on the parasol cell. In neurons of the crustacean stomatogastric ganglion, impulse bursts are generated in response to triggering voltage-sensitive conductances that underlie slow, regenerative depolarizing potentials referred to as plateau potentials, and which usually are evoked by strong synaptic input (Miller and Selverston 1985; Russell and Hartline 1978). In neurons of the cardiac ganglia of crabs, where similar prolonged, burst-producing depolarizing events are referred to as driver potentials, voltage-sensitive Ca2+ conductances are involved and are confined to anatomically restricted sites on the neuronal membrane (Tazaki and Cooke 1979a,b). In parasol cells, one possible site for such a burst initiation zone could be a region on the dendritic trunk, possibly close to the point at which it joins the axon and the neuritic segment. Following strong sensory input, summed EPSPs and background depolarizations might be expected to cause transient, maximal depolarizations of this region. Spontaneous bursts might be generated when an especially large amplitude background depolarization occurred. While much additional experimental effort, including simultaneous recordings from trunk and basal branch compartments of the same neuron, will be required to provide convincing evidence for a trunk initiation site for bursts, the occasional observation in trunk recordings of slow depolarizing events underlying bursts and the reversal of the relationship at that recording locus between the preburst spike and burst spike thresholds provide provisional support for this interpretation.
I interpret the miniature solitary transients and transient bursts observed in recordings from parasol cell dendrites as electrotonically recorded events from neighboring, electrically coupled parasol cells. A comparison of properties of miniature bursts with those of overshooting bursts (Fig. 5C) shows a close similarity among those events. Moreover, like spontaneous or simulated bursts that invade the recording site, miniature bursts have a transient depressive effect on the frequency of those solitary events of the same amplitude class.
The functional role of electrical coupling between parasol cells is not clear. Since bursts occur near the peaks of the background activity, cells that respond to the same strong sensory input tend to burst in near synchrony (Mellon and Wheeler 1999). Electrical coupling between similarly targeted parasol cells could enhance burst synchrony if the coupling zone were near the point of burst initiation, just as mutual electrical coupling among cuticular afferents has recently been found to synchronize and amplify afferent signals to the lateral giant fibers of the crayfish abdominal ganglia (Herberholz et al. 2002).
An important question concerns the functional significance of the isolating segments interposed between the trunk and the basal branches of parasol cells, a feature of these neurons that has not been described previously. It was suggested above that this diameter reduction would help to assure impulse propagation from the trunk into the several basal branches by reducing the impedance load on spike propagation in this direction. Another consequence would be the reduction in the amplitude and rate of rise of electrotonically conducted EPSPs from the basal branches into the trunk region, thereby possibly protecting voltage-sensitive ion channels associated with burst generation. Ongoing studies to generate a computational model of parasol cells may help to clarify the role of this unusual morphological property in the normal functioning of these neurons.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
1 The terms antidromic and orthodromic can cause some confusion when referred to in invertebrate neurons, where the soma is usually not involved in the impulse transmission pathway. For purposes of this paper, "orthodromic" will refer to the direction normally taken by propagating spikes in parasol cell dendritic trunks following their initiation by background activity or by EPSPs generated by sensory input. "Antidromic" activity refers to the direction of propagation of spikes generated by electrical stimulation of the soma cluster, i.e., backward along the trunk. 
Address for reprint requests and other correspondence: D. Mellon, Gilmer Hall, Univ. of Virginia, Charlottesville, VA 22903 (E-mail: dm6d{at}virginia.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Borst A and Egelhaaf M. Dendritic processing of synaptic information by sensory interneurons. Trends Neurosci 17: 257–263, 1994.[ISI][Medline]
Chen WR, Shen GY, Shepherd GM, Hines MI, and Midtgaard J. Multiple modes of action potential initiation and propagation in mitral cell primary dendrite. J Neurophysiol 88: 2755–2764, 2002.
Furshpan EJ and Potter DD. Transmission at the giant motor synapses of the crayfish. J Physiol 145: 289–325, 1959.
Goldstein SS and Rall W. Changes of action shape and velocity for changing core conductor geometry. Biophys J 14: 731–757, 1974.
Herberholz J, Antonsen BL, and Edwards DH. A lateral excitatory network in the escape circuit of the crayfish. J Neurosci 22: 9078–9085, 2002.
Huerta PT and Lisman JE. Bidirectional synaptic plasticity induced by a single burst during cholinergic theta oscillation in CA1 in vitro. Neuron 15: 1053–1063, 1995.[ISI][Medline]
Kim U and McCormick DA. The functional influence of burst and tonic firing mode on synaptic interactions in the thalamus. J Neurosci 18: 9500–9516, 1998.
Magee JC. Voltage-gated ion channels in dendrites. In: Dendrites, edited by Stuart G, Spruston G, and Häusser M. Oxford, UK: Oxford University Press, 1999.
Magee JC. Somatic EPSP amplitude is independent of synapse location in hippocampal pyramidal neurons. Nat Neurosci 3: 895–903, 2000.[ISI][Medline]
Maynard DM and Yager JG. Function of an eyestalk ganglion, the medulla terminalis, in olfactory integration in the lobster, Panulirus argus. Zeit Vergleich Physiol 59: 241–249, 1968.
McKinzie ME, Benton JL, Beltz BS, and Mellon DeF. Parasol cells of the hemiellipsoid body in the crayfish Procambarus clarkii: dendritic branching patterns and functional implications. J Comp Neurol 462: 168–179, 2003.[ISI][Medline]
Mellon DeF. Convergence of multimodal sensory input onto higher-level neurons of the crayfish olfactory pathway. J Neurophysiol 84: 3043–3055, 2000.
Mellon DeF. Active dendritic properties constrain the response of crayfish higher order olfactory neurons to weak sensory stimuli. Soc Neurosci Abstr 27, 726.1, 2001.
Mellon DeF. Crayfish parasol cells exhibit multiple burst initiation sites. Chem Senses 26: A63, 2002.
Mellon DeF. Active dendritic properties constrain input-output relationships in neurons of the central olfactory pathway in the crayfish forebrain. Microsc Res Tech 60: 278–290, 2003.[ISI][Medline]
Mellon DeF and Alones V. Response properties of higher level neurons in the central olfactory pathway of the crayfish. J Comp Physiol [A] 181: 205–216, 1997.
Mellon DeF, Alones V, and Lawrence MD. Antomy and fine structure of neurons in the deutocerebral projection pathway of the crayfish olfactory system. J Comp Neurol 321: 93–111, 1992a.[ISI][Medline]
Mellon DeF and Kennedy D. Impulse origin and propagation in a bipolar sensory neuron. J Gen Physiol 47: 487–499, 1964.
Mellon DeF, Sandeman DC, and Sandeman RE. Characterization of oscillatory olfactory interneurons in the protocerebrum of the crayfish. J Exp Biol 167: 15–38, 1992b.
Mellon DeF and Wheeler CJ. Coherent oscillations in membrane potential synchronize impulse bursts in central olfactory neurons of the crayfish. J Neurophysiol 81: 1231–1241, 1999.
Miller JP and Selverston A. Neural mechanisms for the production of the lobster pyloric motor pattern. In: Model Neural Networks and Behavior, edited by Selverston AI. New York: Plenum, 1985.
Ogawa H, Baba Y, and Oka K. Dendritic calcium accumulation regulates wind sensitivity via short-term depression at cercal sensory-to-giant interneurons synapses in the cricket. J Neurobiol 46: 301–313, 2001.[ISI][Medline]
Russell DF and Hartline DK. Bursting neural networks: a reexamination. Science 200: 453–456, 1978.
Sandeman DC, Beltz B, and Sandeman R. Crayfish brain interneurons that converge with serotonin giant cells in accessory lobe glomeruli. J Comp Neurol 352: 263–279, 1995.[ISI][Medline]
Single S and Borst A. Dendritic integration and its role in computing image velocity. Science 281: 1848–1850, 1998.
Sobel EC and Tank DW. In vivo Ca2+ dynamics in a cricket auditory neuron: an example of chemical computation. Science 263: 823–826, 1994.
Spruston N, Schiller Y, Stuart G, and Sakmann B. Activity-dependent action potential invasion and calcium influx into hippocampal CA1 dendrites. Science 268: 297–300, 1995.
Sullivan JM and Beltz BS. Neural pathways connecting the deutocerebrum and lateral protocerebrum in the brains of decapod crustaceans. J Comp Neurol 441: 9–22, 2001.[ISI][Medline]
Tazaki K and Cooke IM. Isolation and characterization of slow, depolarizing responses of cardiac ganglion neurons in the crab, Portunus sanguinolentus. J Neurophysiol 42: 1000–1021, 1979a.
Tazaki K and Cooke IM. Ionic bases of slow, depolarizing responses of cardiac ganglion neurons in the crab, Portunus sanguinolentus. J Neurophysiol 42: 1022–1047, 1979b.
Thomas MJ, Watabe AM, Moody TD, Makhinson M, and O'Dell TJ. Postsynaptic complex spike bursting enables the induction of LTP by theta frequency synaptic stimulation. J Neurosci 18: 7118–7126, 1998.
Vetter P, Roth A, and Häusser M. Propagation of action potentials in dendrites depends on dendritic morphology. J Neurophysiol 85: 926–937, 2001.
Wachowiak M, Diebel CE, and Ache BW. Functional organization of olfactory processing in the accessory lobe of the spiny lobster. J Comp Physiol [A] 178: 211–226, 1996.
Wessell R, Kristan WB, and Kleinfeld D. Dendritic Ca2+-activated K+ conductances regulate electrical signal propagation in an invertebrate neuron. J Neurosci 19: 8319–8326, 1999.
| ||||||||||||||||||||||||||||||
20 measurements from each of 6 parasol cells combined. Overshooting bursts, black; electrotonic bursts, gray.
