INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The synaptic organization of the eyes in Hermissenda has been characterized and described in considerable detail (Alkon 1973a,b; Alkon and Fuortes 1972; Crow et al. 1979; Dennis 1967). The photoreceptors within the eye are mutually inhibitory (Alkon 1973a; Alkon and Fuortes 1972; Crow et al. 1979), receive direct inhibitory input from statocyst hair cells (Alkon 1973b), and inhibit ipsilateral optic ganglion cells (Alkon 1973a). In addition, identified types of photoreceptors excite and inhibit second-order neurons (CP interneurons) located near the terminal photoreceptor processes within the cerebropleural ganglion (Akaike and Alkon 1980; Goh and Alkon 1984). The CP interneurons also receive converging synaptic input from two additional sensory pathways: statocyst hair cells and chemosensory neurons (Akaike and Alkon 1980). In contrast to the convergence observed between different sensory systems and CP interneurons, identified photoreceptors have been shown to directly project to different aggregates of interneurons classified as type I interneurons (Crow and Tian 2000). Monosynaptic connections between identified type A and B photoreceptors and type I interneurons follow a divergent labeled-line principle, in which an individual photoreceptor inhibits and excites two different aggregates of interneurons that do not receive synaptic input from any of the other photoreceptors within the eye. The synaptic connections between identified type A and B photoreceptors and type I interneurons meet accepted electrophysiological criteria for monosynaptic connections (Crow and Tian 2000). While type I interneurons have been examined in some detail (Akaike and Alkon 1980; Crow and Tian 2000; Goh and Alkon 1984), little information is available regarding the synaptic interactions between photoreceptors and other higher-order interneurons that project to pedal motor neurons.

In this report we provide evidence for previously unidentified interneurons that receive indirect or polysynaptic input from identified photoreceptors (here classified as type II interneurons). The indirect (polysynaptic) connections between identified photoreceptors and type II interneurons exhibit significantly longer and more variable synaptic delays between photoreceptor spikes and postsynaptic potentials (PSPs), and longer delays between the generator potential onset and the initiation of PSPs compared with that of type I interneurons receiving direct connections from photoreceptors. Type II interneurons can be further distinguished from type I interneurons based on morphological criteria. Lucifer yellow cell fills of interneurons, which had been classified as receiving direct or indirect synaptic input from photoreceptors based on electrophysiological criteria, showed unique central projections. Type II interneurons that received polysynaptic input from identified photoreceptors projected to either the ipsilateral pedal ganglion or contralateral cerebropleural ganglion. In contrast, type I interneurons that received monosynaptic input from photoreceptors projected to different regions within the same cerebropleural ganglion, depending on whether they received excitatory or inhibitory synaptic input from identified photoreceptors.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals

Adult Hermissenda crassicornis were used in the experiments. The animals were obtained from Sea Life Supply (Sand City, CA) and maintained in closed artificial seawater aquaria at 14 ± 1°C on a 12-h light/dark cycle. All electrophysiological procedures were conducted during the light phase of the light/dark cycle.

Intracellular recordings and cell staining

Intracellular recordings from identified type A or type B photoreceptors and interneurons were collected from isolated nervous systems. Anatomical and electrophysiological criteria were used to identify lateral or medial type A and B photoreceptors within the eyes, as described previously (Alkon and Fuortes 1972; Crow and Tian 2000; Frysztak and Crow 1994). The isolated nervous systems were incubated in a protease solution (Sigma Type VIII; 0.67 mg/ml, 5 min; Sigma, St. Louis, MO) to facilitate microelectrode penetration of photoreceptors. Surgical desheathing of a small area of the cerebropleural ganglion was conducted to expose the cell bodies of interneurons. The criteria for identifying types of interneurons consisted of soma size, cell layer, location in the cerebropleural ganglion, electrophysiological responses to light and extrinsic current stimulation of photoreceptors, and the presence of direct or indirect synaptic input from identified type A or B photoreceptors, as previously reported (Crow and Tian 2000).

The partially desheathed circumesophageal nervous systems were pinned to a SYLGARD (Dow Chemical, Midland, MI) stage in a recording chamber filled with artificial seawater (ASW) of the following composition (in mM): 460 NaCl, 10 KCl, 10 CaCl₂, 55 MgCl₂, buffered with 10 mM HEPES and brought to pH 7.46 with dilute NaOH. Experiments were also conducted in high divalent cation ASW (30 mM CaCl₂ and 165 mM MgCl₂) that raised action potential threshold, thus suppressing spontaneous activity and reducing polysynaptic activation of CP interneurons. The ASW in the recording chamber was monitored by a thermistor and held at 15 ± 0.5°C. Illumination of the eyes was provided by a tungsten halogen incandescent lamp attached to a fiber-optic bundle mounted underneath the recording chamber. Maximum light intensity was attenuated with neutral density filters expressed in negative log units. Photoreceptors were impaled with microelectrodes filled with 4 M KAc and CP interneurons impaled with microelectrodes filled with filtered 4% Lucifier yellow in 0.2 M LiCl. The electrode tips were filled with the Lucifer yellow in LiCl and backfilled from the shank with 0.2 M LiCl. The two electrodes were connected to the two headstages of an Axoclamp 2A (Axon Instruments, Foster City, CA). Standard intracellular recording and stimulation techniques were employed. Electrophysiological data were collected on both videotape (Vetter Instruments, Rebersburg, PA) and a chart recorder (Gould, Cleveland, OH). Single spikes in identified photoreceptors elicited by brief extrinsic current pulses and trains of action potentials elicited by current steps were applied in the dark through a bridge circuit. Depolarizing generator potentials recorded from identified photoreceptors were evoked by light steps following appropriate periods of dark adaptation. Evidence for monosynaptic connections between photoreceptors and interneurons was provided by PSPs with relatively short and constant latencies and a one-for-one relationship between photoreceptor action potentials and PSPs in both normal ASW and in ASW solutions containing high divalent cations (3× Ca²⁺ and 3× Mg²⁺) (see RESULTS).

The connectivity of the CP interneurons was first established with an identified photoreceptor followed by iontophoresis of the dye using a constant negative current (1.0 nA) for 20 min. After an additional 2-3 h to allow for diffusion of the Lucifier yellow to the terminal processes, the ganglion was fixed with 4% paraformaldehyde in 0.2 M cacodylate buffer (pH 7.4) for 3 h, but not >12 h, followed by dehydration in an ascending ETOH series and cleared with methyl salicylate. The stained cells were observed and photographed through a fluorescence microscope (Zeiss, Thornwood, NY) and some preparations were imaged with a confocal laser-scanning microscope (Bio-Rad, MRC 1024 ES). The stack of images was collected through the preparation in 5-µM increments.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Type I excitatory interneurons

Lucifer yellow labeled interneurons were classified as receiving direct monosynaptic input from identified photoreceptors, based on the latency between the peak of the photoreceptor action potential elicited by extrinsic current or initial depolarization of the generator potential produced by stimulation of photoreceptors with light and the initial change in membrane potential underlying the PSP. A total of 31 interneurons were filled with Lucifer yellow after simultaneous electrophysiological recordings were collected from identified photoreceptors and interneurons. Interneurons classified as excitatory type I (n = 12) received direct monosynaptic excitatory synaptic input from lateral type B photoreceptors. In addition, type I excitatory interneurons exhibited distinctive projections to postsynaptic targets within the cerebropleural ganglion and expressed specific branching patterns of the secondary and terminal processes from the main axon.

A representative example of a type I excitatory interneuron is shown in Fig. 1A. The axonal process of the type I excitatory interneurons traveled in the medial region of the cerebropleural ganglion and projected toward the cerebropleural commissure. Along the length of the main axonal process are numerous varicose endings projecting from the primary and secondary processes. The merged optical sections of confocal images of the Lucifer filled type I excitatory interneuron are shown in Fig. 1B. Several regions of the neuron exhibited three distinct clusters of varicosities identified by the filled arrows and one proximal cluster close to the soma (open arrow, Fig. 1B). All type I excitatory interneurons exhibited these morphological characteristics. As indicated, the evidence for a direct monosynaptic connection between identified photoreceptors and type I interneurons was based on electrophysiological criteria examining the latency between photoreceptor spikes and interneuron PSPs and the variability of latency measures.

View larger version (82K): [in this window] [in a new window]   Fig. 1. Morphological and electrophysiological characteristics of cerebropleural interneurons that received monosynaptic excitatory synaptic connections from lateral type B photoreceptors. A: type I excitatory interneuron filled iontophoretically with Lucifer yellow. Type I excitatory interneurons that receive monosynaptic excitatory synaptic input from lateral type B photoreceptors (n = 12) exhibit distinctive projections and branching patterns of the secondary processes. The primary axon traveled in the medial region of the cerebropleural ganglion toward the cerebropleural commissure. Along its length are numerous varicose endings projecting from the primary and secondary processes. Scale bar in A, 125 µM. B: confocal micrograph of Lucifer filled interneuron shown in A representing a reconstructed stack of optical sections (100 µM was scanned at 5-µM intervals), indicating three areas of varicosities (filled arrows) and one region near the soma (open arrow). Scale bar in B, 65 µM. C1: a single spike produced by a 500-ms depolarizing current pulse in the lateral B photoreceptor elicited an EPSP recorded from the Lucifer yellow filled type I interneuron (C2) shown in A. Superimposed recordings of three successive lateral type B spikes (C3) and EPSPs recorded from the type I interneuron (C4). These examples meet the criteria for monosynaptic connections between photoreceptors and interneurons, as previously reported (Crow and Tian 2000).

As shown in Fig. 1C, 1 and 2, a single spike generated by a 500-ms depolarizing extrinsic current pulse in the photoreceptor elicited an excitatory postsynaptic potential (EPSP) recorded from the interneuron. The EPSPs recorded from the type I interneuron exhibited relatively short and constant latencies between the peak of the B photoreceptor action potential and the initial depolarization underlying the PSP. This is shown by the superimposed recordings of three successive lateral B spikes (C3) and EPSPs recorded from the type I interneuron (C4). The statistical analysis of five successive EPSP latencies for the example shown in Fig. 1C2 yielded a mean latency of 9.7 ms and a  of 0.59 ms. Synaptic connections between identified photoreceptors and type I excitatory interneurons were examined further by exposure to ASW containing high divalent cations (3× Ca²⁺ and Mg²⁺). As shown in the example in Fig. 2, exposing the preparation to a high divalent cation solution did not block the EPSP recorded in the type I interneuron elicited by a single spike in the B photoreceptor. In contrast, type II excitatory interneurons exhibit longer and more variable latencies (see section on type II interneurons below). In addition, type I interneurons exhibited larger amplitude EPSPs compared with that of type II interneurons (Table 1). The mean amplitude of type I interneuron monosynaptic EPSPs (n = 12) was 7.6 ± 0.83 mV compared with a mean of 1.9 ± 0.28 mV for type II indirect or polysynaptic EPSPs (n = 5).

View larger version (11K): [in this window] [in a new window]   Fig. 2. Simultaneous recording from a lateral B photoreceptor (A1) and type I excitatory interneuron (B1) in normal ASW and after exposing the nervous system to a high divalent cation solution (3× Ca2+ and Mg2+) (A2-B2). A spike in the same lateral B photoreceptor (as in A1) bathed in a high divalent cation solution (A2) elicited an EPSP in the same type I interneuron (B2).

The differences between type I and type II EPSP amplitudes were statistically significant (t₁₅ = 4.09; P < 0.001). As previously reported (Crow and Tian 2000), EPSPs in type I interneurons followed photoreceptor spikes one for one. The example shown in Fig. 3, A and B, is a simultaneous recording from the lateral B photoreceptor (A) and the Lucifer filled interneuron (B) shown in Fig. 1. Illumination (4-s light step attenuated 1.0 log unit) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded from the lateral B photoreceptor and a depolarizing complex EPSP, resulting in increased spike frequency of the type I interneuron. The mean amplitude of complex EPSPs recorded from type I interneurons (n = 10) receiving monosynaptic input from photoreceptors was 8.9 ± 0.38 mV. In contrast the mean amplitude of complex EPSPs from type II interneurons receiving polysynaptic or indirect connections (n = 5) was 3.4 ± 0.48 mV. The differences in amplitude of the complex EPSP between type I and type II interneurons were statistically significant (t₁₃ = 2.33; P < 0.03) (see Table 1). As shown in Fig. 3, a one-for-one relationship was detected between light-elicited photoreceptor action potentials and spikes recorded from type I interneurons that received monosynaptic input from identified photoreceptors.

View larger version (48K): [in this window] [in a new window]   Fig. 3. Simultaneous recording from a lateral B photoreceptor (A) and the Lucifer filled type I excitatory interneuron (B) shown in Fig. 1A. Light stimulation (4-s light step attenuated 1.0 log unit) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded from the lateral B photoreceptor and depolarization and increased spike frequency of the type I excitatory interneuron. The maximum discharge rate of the type I interneuron (B) corresponded to the peak of the light-elicited generator potential (A).

Type I inhibitory interneurons

Interneurons that received monosynaptic inhibitory synaptic input from lateral type B photoreceptors based on electrophysiological criteria (n = 9) were classified as type I inhibitory. Type I inhibitory interneurons exhibited distinctive projections within the cerebropleural ganglion and specific branching patterns of the secondary processes. A representative example of a type I inhibitory interneuron is shown in Fig. 4A. The primary axonal process of type I inhibitory interneurons "bends or loops" near the soma and projected to the anterior region of the cerebropleural ganglion. Varicosities and secondary processes were found only in the apparent terminal region of the primary axon in the anterior portion of the ganglion. As previously described (Crow and Tian 2000), inhibitory postsynaptic potentials (IPSPs) in type I interneurons followed spikes in identified photoreceptors with relatively short and constant latencies between the peak of the photoreceptor action potential and the initial hyperpolarization underlying the IPSP.

View larger version (51K): [in this window] [in a new window]   Fig. 4. Morphological and electrophysiological characteristics of cerebropleural interneurons that received monosynaptic inhibitory synaptic connections from lateral type B photoreceptors. A: type I inhibitory interneuron filled iontophoretically with Lucifer yellow. Type I inhibitory interneurons that receive monosynaptic synaptic input from lateral type B photoreceptors (n = 9) exhibit distinctive projections and branching patterns of the secondary processes. The axonal process "bends" close to the soma and projected to the anterior region of the cerebropleural ganglion. Varicosities and secondary processes are found in the apparent terminal region of the primary axon. Scale bar in A, 125 µM. B1: a single spike produced by a 250-ms depolarizing current pulse in the lateral B photoreceptor elicited an IPSP recorded from the Lucifer yellow filled type I inhibitory interneuron (B2) shown in A. Three successive superimposed spikes from the lateral B photoreceptor and IPSPs recorded from the type I inhibitory interneuron are shown in B3 and B4, respectively. Consistent with the results from excitatory synaptic connections described in Fig. 1, these examples meet the criteria for monosynaptic connections between photoreceptors and interneurons.

In the representative example shown in Fig. 4B, 1 and 2, a single spike produced by a 250-ms depolarizing current pulse in a lateral B photoreceptor elicited an IPSP recorded from the Lucifer yellow filled type I interneuron shown in Fig. 4A. Three successive superimposed spikes elicited from the lateral B photoreceptor and IPSPs recorded from the filled type I inhibitory interneuron are shown in Fig. 4B, 3 and 4, respectively. The statistical analysis of six successive IPSP latencies measured between the peak of the action potentials in the B photoreceptor and the initial hyperpolarization of the interneuron shown in Fig. 4B4 yielded a mean latency of 23.5 ms and a  of 1.2 ms.

In contrast, type II inhibitory interneurons exhibit longer and more variable latencies (see section on type II interneurons below). Type I interneurons also exhibit larger amplitude IPSPs compared with that of type II interneurons (Table 1). The mean amplitude of type I interneuron IPSPs (n = 7) was 8.2 ± 0.87 mV compared with a mean of 2.4 ± 0.6 mV for type II polysynaptic IPSPs (n = 3; t₈ = 3.1; P < 0.01). The analysis of complex IPSPs of type I inhibitory interneurons elicited by illumination revealed that spike activity was blocked during the light step and IPSPs followed spikes in the B photoreceptor with a one-for-one relationship.

In the representative example shown in Fig. 5 light stimulation (6-s light step attenuated 2.0 log units) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded in the lateral B photoreceptor and hyperpolarization of the type I interneuron sufficient to block spontaneous spike activity. As shown in Table 1 the mean peak amplitude of complex IPSPs recorded from type I interneurons was 10.1 ± 0.64 mV (n = 5). The mean peak amplitude of complex IPSPs recorded from type II interneurons that received indirect input from photoreceptors was 6.0 ± 1.15 mV (n = 3). The difference in amplitude of the complex IPSP in type I and type II interneurons was statistically significant (t₆ = 3.4; P < 0.01).

View larger version (13K): [in this window] [in a new window]   Fig. 5. Simultaneous recording from a lateral B photoreceptor (A) and the Lucifer filled type I inhibitory interneuron (B) shown in Fig. 4A. Light stimulation (6-s light step attenuated 2.0 log units) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded from the lateral B photoreceptor (A) and hyperpolarization of the type I inhibitory interneuron (B) sufficient to block spike generation. IPSPs recorded from the type I inhibitory interneurons follow B spikes with a one-for-one relationship.

Type II excitatory interneurons

A total of 6 interneurons filled with Lucifer yellow were classified as receiving excitatory polysynaptic input from photoreceptors (type II interneurons) based on electrophysiological criteria. Type II interneurons exhibited central projections that were unique compared with type I interneurons that received direct projections from identified photoreceptors. An example of a type II excitatory interneuron is shown in Fig. 6A. In the example shown in Fig. 6A the type II excitatory interneurons' main axonal process projected to the ipsilateral pedal ganglion. In all examples of Lucifer filled type II interneurons that received polysynaptic input from photoreceptors, the cells projected to postsynaptic targets in either the ipsilateral pedal ganglion (n = 4) or contralateral cerebropleural ganglion (n = 2). As shown in Fig. 6B, 1 and 2, a single spike generated in a lateral B photoreceptor elicited a small EPSP recorded from the Lucifer filled type II interneuron (B2).

View larger version (41K): [in this window] [in a new window]   Fig. 6. Morphological and electrophysiological characteristics of cerebropleural interneurons that receive indirect (polysynaptic) excitatory synaptic connections from lateral B photoreceptors. A: type II excitatory interneuron filled iontophoretically with Lucifer yellow. Type II interneurons that received polysynaptic excitatory synaptic input from lateral B photoreceptors (n = 5) or medial A photoreceptors (n = 1) projected to the ipsilateral pedal ganglion (n = 4) or contralateral cerebropleural ganglion (n = 2). Scale bar in A, 125 µM. B1: a single spike in a lateral B photoreceptor elicited a small EPSP recorded from a Lucifer yellow filled type II excitatory interneuron (B2). The PSPs also did not exhibit relatively constant latencies following photoreceptor spikes; for example, the variance of the latency between the presynaptic spike and PSP was significantly greater compared with measurements of monosynaptic synaptic connections. A representative example of three superimposed spikes is shown in B3 and variable latency and amplitude spike-elicited EPSPs recorded from the interneuron are indicated by the red, blue, and green superimposed traces in B4.

In addition to the EPSPs elicited by type B spikes, spontaneous EPSPs were also detected in type II interneurons, and in some cases spikes in the type B photoreceptor failed to elicit EPSPs in type II interneurons (see Fig. 8). In contrast to type I interneurons, the EPSPs recorded from type II interneurons did not exhibit relatively short and constant latencies between the peak of the photoreceptor action potential and the initial depolarization underlying the EPSP in the interneuron. The statistical analysis of latencies between successive superimposed spikes from a lateral B photoreceptor and EPSPs recorded from the type II interneuron (Fig. 6B, 3 and 4) yielded a mean latency of 17.4 ms and a  of 2.87 ms (n = 9). A statistical comparison of the variance (²) of the latencies for the type I excitatory interneurons and the type II excitatory interneurons revealed a significant difference (F_max = 23.5; P < 0.001). In addition to the longer and more variable latencies between spikes and PSPs exhibited by type II interneurons, the latencies between the onset of the generator potential and the initial depolarization underlying the EPSPs were significantly longer than those of type I excitatory interneurons (t₁₂ = 5.6; P < 0.001). The mean latency between the onset of the generator potential and EPSP for type I excitatory interneurons (n = 9) was 56.4 ± 2.6 ms compared with a mean of 103.6 ± 10.6 ms for type II excitatory interneurons (n = 5).

In addition to the electrophysiological differences based on latencies, there was not a one-for-one relationship between photoreceptor spikes and type II EPSPs. Shown in Fig. 7 is a simultaneous recording from a lateral B photoreceptor (A1) and a Lucifer filled type II excitatory interneuron (A2). Light stimulation (5-s light step attenuated 2 log units) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded from the lateral B photoreceptor (A1) and a single spike followed by an increase in EPSP frequency in the type II interneuron (A2). The same recordings on a faster time scale (B1 and B2) showed that B photoreceptor spikes did not follow EPSPs recorded from the type II interneuron with a one-for-one relationship. As shown in Fig. 8 stimulation of a single lateral B photoreceptor with an extrinsic current pulse initially elicited a small EPSP in the type II interneuron and several EPSP failures, indicating an absence of a one-for-one relationship between B spikes and type II EPSPs.

View larger version (23K): [in this window] [in a new window]   Fig. 7. Absence of a one-for-one relationship between photoreceptor action potentials elicited by light and EPSPs. Simultaneous recording from a lateral B photoreceptor (A1) and a Lucifer filled type II interneuron (A2). Light stimulation (5-s light step attenuated 2 log units) elicited a stereotyped depolarizing generator potential with superimposed spikes recorded from the lateral B photoreceptor (A1) and a single spike followed by an increase in EPSP frequency in the type II interneuron (A2). B1-B2: same recordings as in A1 and A2 above, on a faster time scale, show that spikes in the B photoreceptor did not follow the EPSPs recorded from the type II interneuron (B2) with a one-for-one relationship, consistent with other electrophysiological evidence showing that synaptic connections between B photoreceptors and type II interneurons are indirect (polysynaptic).

View larger version (11K): [in this window] [in a new window]   Fig. 8. Indirect (polysynaptic) connection between a B photoreceptor and a Lucifer filled type II interneuron. A 2-s depolarizing extrinsic current pulse elicited several action potentials recorded from a lateral B photoreceptor (A) and a complex EPSP recorded in the type II interneuron (B). The initial lateral B spike elicited a small EPSP followed by a second EPSP that occurred in the absence of a second B spike in the photoreceptor. Several B spikes generated during the current pulse failed to elicit detectable EPSPs in the type II interneuron. PSPs in type II interneurons do not follow photoreceptor spikes one-for-one, in contrast to type I interneurons, as described previously (Crow and Tian 2000) and in Figs. 1-5 of this study.

Type II inhibitory interneurons

Interneurons that received indirect IPSPs from identified photoreceptors were classified as type II inhibitory (n = 4) based on electrophysiological and morphological criteria. Type II inhibitory interneurons filled with Lucifer yellow were found to project to the contralateral cerebropleural ganglion. An example of a type II inhibitory interneuron is shown in Fig. 9A. The polysynaptic IPSPs recorded from type II interneurons were small and often were not observed following single spikes elicited from identified photoreceptors. In the example shown in Fig. 9B, 1 and 2, a single spike generated in a lateral B photoreceptor elicited a small IPSP recorded from the Lucifer yellow filled type II interneuron (B2) shown in Fig. 9A. In addition, the elicited IPSPs recorded from type II interneurons did not exhibit relatively constant latencies between the peak of the photoreceptor spikes and the initial hyperpolarization underlying the IPSP (mean = 37.6 ms;  = 20.5 ms). This is supported by the example in Fig. 9B, 3 and 4, showing three superimposed B spikes and type II interneuron IPSPs. A statistical comparison of the variance ² of the latencies for the type I inhibitory interneurons and the type II inhibitory interneurons showed a significant difference in variability for the two types of interneurons (F_max = 63.68; P < 0.01). The synaptic connection between an identified photoreceptor and type II inhibitory interneuron was further examined with exposure to ASW containing high divalent cations (3× Ca²⁺ and Mg²⁺). As shown in Fig. 10, exposure of the nervous system to a high divalent cation solution blocked the polysynaptic IPSP elicited by a spike in the lateral B photoreceptors and recorded from the type II interneuron. The statistical analysis of IPSP latencies also showed a significant difference between type I and type II inhibitory interneurons with respect to the latency between the onset of the generator potential and the initial hyperpolarization underlying the IPSPs (t₅ = 3.4; P < 0.02). The mean latency for the type I inhibitory interneurons (n = 5) was 74.4 ± 3.5 ms compared with 103.5 ± 12.5 ms for the type II inhibitory interneurons (n = 2).

View larger version (44K): [in this window] [in a new window]   Fig. 9. Morphological and electrophysiological characteristics of cerebropleural interneurons that received indirect (polysynaptic) inhibitory synaptic connections from lateral B photoreceptors. A: type II inhibitory interneuron filled iontophoretically with Lucifer yellow. Type II inhibitory interneurons that received polysynaptic inhibitory synaptic input from lateral B photoreceptors (n = 4) projected to the contralateral cerebropleural ganglion. Scale bar in A, 125 µM. B1: a single spike in a lateral B photoreceptor elicited a small IPSP recorded from the Lucifer yellow filled type II inhibitory interneuron (B2) shown in A. Characteristic of type II interneurons, the spike-elicited PSPs did not exhibit relatively constant latencies following photoreceptor spikes, as shown by the three superimposed spikes (B3) and the variable latency and amplitude IPSPs recorded from the type II interneuron indicated by the red, blue, and green superimposed traces in B4.

View larger version (9K): [in this window] [in a new window]   Fig. 10. Simultaneous recording from a lateral B photoreceptor (A1) and type II inhibitory interneuron (B1) in normal ASW and after exposing the nervous system to a high divalent cation solution (A2-B2). A spike in the same lateral B photoreceptor (as in A1) bathed in a high divalent cation solution (A2) did not elicit an IPSP in the type II inhibitory interneuron (B2).

An example of a complex light-elicited IPSP is shown in Fig. 11. The example is from a simultaneous recording of a lateral B photoreceptor (Fig. 11A) and the same Lucifer filled interneuron (Fig. 11B) as shown in Fig. 9A. Light stimulation (10-s light step attenuated 1.0 log unit) elicited a stereotyped response characteristic of B photoreceptors (Fig. 11A) and a small prolonged hyperpolarization of the type II inhibitory interneuron (Fig. 11B). The transient increase in B photoreceptor spike frequency after light offset did not inhibit spikes recorded in the type II interneuron (Fig. 11B).

View larger version (19K): [in this window] [in a new window]   Fig. 11. Light-elicited complex IPSP recorded from a type II inhibitory interneuron. Simultaneous recordings from lateral B photoreceptor (A) and the Lucifer filled type II inhibitory interneuron (B) shown in Fig. 9. Light stimulation (10-s light step attenuated 1.0 log unit) elicited a stereotyped response characteristic of B photoreceptors as described (A) and a small prolonged hyperpolarization of the type II inhibitory interneuron (B). Note that the increase in the B photoreceptor spike frequency after light offset did not inhibit spikes recorded in the type II interneuron (B).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Characteristics of type I and type II interneurons

We have shown that type I interneurons that receive either excitatory or inhibitory monosynaptic input from identified photoreceptors in the Hermissenda eye exhibit distinct morphologies and central projections to postsynaptic targets within the cerebropleural ganglion. Type I interneurons that received monosynaptic excitatory input from identified B photoreceptors sent an axonal process that traveled in the medial region of the cerebropleural ganglion and terminated near the cerebropleural commissure. Type I excitatory interneurons also have two distinct regions of varicosities. The first is near the soma and is expressed by a fan-shape cluster of short processes that terminated in a series of varicosities. The second region consisted of numerous varicose endings that projected from primary and secondary processes along the length of the main axonal process. In contrast, type I inhibitory interneurons had a primary axonal process that formed a loop near the soma and projected to a specific region in the anterior part of the cerebropleural ganglion. Varicose endings of type I inhibitory interneurons were localized to the terminal region of the primary axonal process. The site of termination was characterized by a small cluster of varicosities in contrast to the numerous varicosities found along the entire length of the primary axonal process of type I excitatory interneurons.

Here, for the first time, we characterize in Hermissenda type II interneurons that express distinct interganglionic projections and synaptic input from identified photoreceptors. All type II interneurons projected to either the ipsilateral pedal ganglion or to the contralateral cerebropleural ganglion. The majority of the type II interneurons that projected to the pedal ganglion were from neurons that received polysynaptic excitatory input from identified photoreceptors. In contrast, all type II interneurons that received polysynaptic inhibitory input from identified photoreceptors projected to the contralateral cerebropleural ganglion. Electrophysiological criteria were used to establish monosynaptic and polysynaptic connections with interneurons. Type II interneurons exhibited significantly longer latencies between the peak of photoreceptor action potentials and the foot of the PSPs compared with those of type I interneurons that received monosynaptic connections from photoreceptors. In addition, the variability of the latency between the photoreceptor action potential peak and the foot of the PSP was significantly larger for type II interneurons compared with that of type I interneurons. In addition the latency between light-elicited generator potentials and the foot of the complex PSP was significantly longer for type II interneurons than for type I interneurons.

Visually mediated behaviors

A number of investigations have revealed a high degree of behavioral complexity in Hermissenda's response to illumination. Visually mediated behaviors that have been examined include the effects of light gradients on locomotion (Lederhendler et al. 1980), positive phototactic responses (Alkon 1974), Pavlovian conditioning of inhibition of normal positive phototactic behavior (Crow and Alkon 1978), and foot contraction (Lederhendler et al. 1986). To date, all of the examples of behavior and its modification by associative learning in Hermissenda involve components affecting locomotion, including the potential interaction between the initiation of locomotion and contraction of different regions of the foot. While considerable progress has been made in identifying conditioning correlates in identified photoreceptors (Crow 1985; Crow and Alkon 1980; Farley and Alkon 1982; Frysztak and Crow 1993, 1994, 1997; West et al. 1982) and in characterizing photoreceptor conductances and their modulation by conditioning (Acosta-Urquidi and Crow 1995; Alkon 1989; Alkon et al. 1982, 1985; Crow 1988; Farley and Han 1997; Farley et al. 1990; Gandhi and Matzel 2000; Yamoah and Crow 1994, 1996; Yamoah et al. 1998), little is known regarding plasticity in other neurons of the conditioned stimulus pathway. Interneurons have been identified as sites of plasticity in a number of examples of learning in invertebrates (for recent examples and discussions, see Burrell et al. 2000; Cleary et al. 1995; Sahley and Crow 1998).

To understand how learning is expressed in behavior of Hermissenda requires the identification of the neural circuit that supports light-elicited mucociliary locomotion and foot contraction elicited by stimulation of statocyst hair cells. As a first step in this analysis the light responses of several pedal neurons have been described and progress has been made on the identification of putative motor neurons in the pedal ganglia that receive synaptic input from the visual system and whose axons project in identified pedal nerves (Goh and Alkon 1984; Hodgson and Crow 1991, 1992; Jerussi and Alkon 1981; Richards and Farley 1987).

Indeed, a neural circuit between identified photoreceptors and a putative motor neuron has been proposed to account for turning movements in Hermissenda (Goh and Alkon 1984). The circuit consists of type B photoreceptor projections to the medial type A photoreceptor that projects to interneurons whose proposed synaptic target is a pedal motor neuron (see Goh and Alkon 1984). The interneurons in this proposed circuit show a striking similarity to -and -interneurons initially described by Akaike and Alkon (1980) and the type I interneurons examined by Crow and Tian (2000) and in the present report. However, it is clear that the interneurons that project to putative motor neurons in the Goh and Alkon (1984) circuit are not of the type I class of interneurons. First, Goh and Alkon (1984) did not have interneuron dye fills for all of their examples where electrophysiological data were collected. Indeed, they provided evidence that the putative motor neuron in their circuit must have received synaptic input from other interneurons in addition to putative type I interneurons, since EPSPs recorded in the putative motor neuron during type A photoreceptor activity were not preceded by spikes in the interneurons that they recorded from. In addition, since cell fills were not provided for all of their examples it was not clear where the interneuron projected to.

Based on our results, where cell fills were always carried out in conjunction with the electrophysiology, the interneuron described by Goh and Alkon (1984) that projected to the pedal ganglia would be classified as type II, receiving indirect projections from photoreceptors. However, the previously reported examples of interneurons that received direct input from photoreceptors (Akaike and Alkon 1980; Goh and Alkon 1984) would be classified as type I excitatory. As shown in the present study, type I interneurons never projected to postsynaptic targets outside of the cerebropleural ganglion. Therefore our observations would indicate that modifying the number of interneurons in the previously proposed circuit for behavioral turning would be necessary. Taken collectively, the recent evidence suggests that the circuitry involving interneuronal contributions to visually guided behavior is more complex than initially envisioned.

Divergence and convergence of photoreceptor synaptic connections

We previously reported that different aggregates of type I excitatory and inhibitory interneurons received synaptic input from different identified photoreceptors, thus maintaining a divergent labeled-line. Divergence of synaptic input from a single photoreceptor to type I interneurons is extensive. A single identified A or B photoreceptor has been shown to synapse with as many as four type I interneurons in addition to the multiple inhibitory synaptic connections to different photoreceptors within the Hermissenda eye (Crow and Tian 2000). The light-elicited activity recorded from a single type I interneuron can be accounted for by the activity of the identified A or B photoreceptor that has a direct connection with the interneuron. The functional significance of maintaining labeled-lines to type I interneurons has not been elucidated. However, the divergence of identified photoreceptor synaptic projections to specific interneuronal aggregates may provide for the maintenance of differential expression of conditioning correlates in different A and B photoreceptors at the level of second-order neurons. This specific organization would provide for plasticity intrinsic to specific neurons of the visual system to be processed at postsynaptic targets that are separate from and parallel to neurons responsible for normal light intensity discriminations. In contrast, type II excitatory and inhibitory interneurons express activity during illumination of the eye that cannot be accounted for exclusively by the light-elicited activity of a single identified photoreceptor. This suggests that the first site of convergence in the visual pathway is at the connections with type II interneurons. Since type II interneurons project to postsynaptic targets in either the pedal or contralateral cerebropleural ganglia, they would be good candidates for the first site of synaptic convergence of visual information in the nervous system. However, synaptic connections between type I and type II interneurons have not been established.

Circuitry supporting visually influenced locomotion

The specific motor neurons mediating mucociliary locomotion, the generation of pedal waves, and/or foot contraction have not been identified. However, it is clear that locomotion is controlled by motor neurons whose axons project to their postsynaptic targets through pedal nerves P1 and P2. Bilateral lesions of either pedal nerve P1 or P2 interfered with the ability of animals to express phototactic locomotion (Richards and Farley 1987). In addition, unilateral axotomy of P1 and P2 eliminated ipsilateral ciliary movement on the ventral surface of the foot, although contralateral activity remained intact (Crow 1981). Previous studies, using morphological criteria, have identified several putative motor neurons whose axons exit the pedal ganglia in nerves P1 or P2 (Goh and Alkon 1984; Hodgson and Crow 1991). The understanding of how mucociliary locomotion is generated and modulated by illumination is a necessary prerequisite for any adequate explanation of how Pavlovian conditioning is expressed in the behavior of Hermissenda. The physiological and anatomical identification of type I and type II interneurons and the description of their central projections presented here is a necessary step in the analysis of how light and Pavlovian conditioning regulates motor behavior in Hermissenda.