Multiplexed temporal coding of electric communication signals in mormyrid fishes

Abstract

The coding of stimulus information into patterns of spike times occurs widely in sensory systems. Determining how temporally coded information is decoded by central neurons is essential to understanding how brains process sensory stimuli. Mormyrid weakly electric fishes are experts at time coding, making them an exemplary organism for addressing this question. Mormyrids generate brief, stereotyped electric pulses. Pulse waveform carries information about sender identity, and it is encoded into submillisecond-to-millisecond differences in spike timing between receptors. Mormyrids vary the time between pulses to communicate behavioral state, and these intervals are encoded into the sequence of interspike intervals within receptors. Thus, the responses of peripheral electroreceptors establish a temporally multiplexed code for communication signals, one consisting of spike timing differences between receptors and a second consisting of interspike intervals within receptors. These signals are processed in a dedicated sensory pathway, and recent studies have shed light on the mechanisms by which central circuits can extract behaviorally relevant information from multiplexed temporal codes. Evolutionary change in the anatomy of this pathway is related to differences in electrosensory perception, which appears to have influenced the diversification of electric signals and species. However, it remains unknown how this evolutionary change relates to differences in sensory coding schemes, neuronal circuitry and central sensory processing. The mormyrid electric communication pathway is a powerful model for integrating mechanistic studies of temporal coding with evolutionary studies of correlated differences in brain and behavior to investigate neural mechanisms for processing temporal codes.

Keywords: anti-coincidence detection; coincidence detection; delay line; duration tuning; electric organ discharge; interval tuning; sub-millisecond timing differences; temporal filter; weakly electric fish.

Fig. 1.

Mormyrid electrocommunication consists of a fixed electric organ discharge (EOD) produced at variable interpulse intervals (IPIs). Top: EOD waveforms recorded from the 21 known species within the Ivindo River of Gabon reveal that the waveforms are species-specific and span a wide range of durations (modified from Carlson et al., 2011). In each case, EOD waveforms from different individuals of the same species are normalized to the same peak-to-peak height, superimposed and aligned to the head-positive peak (except for Paramormyrops sp. ‘TEN’, for which waveforms are aligned to the head-negative peak). Bottom: 10 s of a raw electrical recording from an isolated Brienomyrus brachyistius. The spike-like nature of the EODs is evident, and the amplitude changes throughout the recording as a result of the fish moving with respect to the recording electrode. Below, the raw recording has been converted into a plot that illustrates the sequence of IPIs over time.

Fig. 2.

Neuroanatomy of the knollenorgan electrosensory (red), electromotor (blue) and corollary discharge (purple) pathways. Excitatory connections are indicated by arrows and inhibitory connections by punctate terminals. Knollenorgan electroreceptors project somatotopically to the hindbrain nucleus of the electrosensory lateral line lobe (nELL), which projects to two nuclei of the torus semicircularis: the exterolateral nucleus (EL) and the medioventral nucleus (MV). Each EOD command originates in the command nucleus (CN). The CN projects to the medullary relay nucleus (MRN), which in turn innervates the electromotor neurons in the spinal cord that innervate the electric organ. CN output is influenced by excitation from the precommand nucleus (PCN) and the dorsal posterior thalamic nucleus (DP), both of which receive inhibition from the ventroposterior nucleus (VP). When the CN initiates an EOD, it sends a copy of this signal to the bulbar command-associated nucleus (BCA), which projects to the MRN and to the mesencephalic command-associated nucleus (MCA), which mediates inhibition of the motor pathway through the VP, and inhibition of the KO sensory pathway through the sublemniscal nucleus (slem). ELL, electrosensory lateral line lobe; OB, olfactory bulb; tel, telencephalon; val, valvula of the cerebellum.

Fig. 3.

Evolutionary change in the knollenorgan electrosensory system (modified from Carlson et al., 2011). The cladogram on the left illustrates the transition from a relatively small exterolateral nucleus (EL, green) to an enlarged EL with separate anterior and posterior subdivisions (ELa/ELp, magenta). Normalized EL sizes of each branch are shown to the right (median ± range). A parsimonious reconstruction suggests that ELa/ELp evolved twice, once in the lineage leading to clade A within the subfamily Mormyrinae and once in the lineage leading to Petrocephalus microphthalmus within the subfamily Petrocephalinae. The locations of knollenorgans on the body surface (red dots) and 50 μm horizontal sections of the midbrain highlighting the EL and ELa/ELp from four different species are shown at the far right (these species are underlined in the cladogram). Scale bars are 1 mm for the fish outlines, and 300 μm for the brain sections.

Fig. 4.

Temporal multiplexing of electrocommunication signals by knollenorgans. This schematic representation shows a train of EOD stimuli at the bottom, with the responses of different knollenorgans to these stimuli above. Each row represents a different knollenorgan and each tick mark indicates the timing of an individual spike. The location of each knollenorgan on the body surface with respect to the electric field determines which edge of the EOD waveform it will respond to, and the spikes are color-coded for the edge (arrows) to which the knollenorgan is responding. Spike timing differences between different knollenorgans represent EOD waveform, whereas interspike intervals within each knollenorgan represent EOD intervals.

Fig. 5.

ELa and ELp microcircuitry overlaid on a horizontal Nissl section through the midbrain of Brienomyrus brachyistius (schematic cells are not to scale). As nELL axons enter the ELa, they synapse directly onto adendritic large cells before following a convoluted path to synapse onto adendritic small cells. Large cells are entirely intrinsic to the ELa, providing inhibitory input onto small cells via a large calyx-like terminal. Small cells are hypothesized to act as time comparators that integrate inhibition from one side of the body with excitation from the opposite side of the body to recode the peripheral spike timing differences that code for EOD waveform. ELa small cells project to multipolar cells in the ELp. The microcircuitry within the ELp is less well understood, although there are extensive excitatory and inhibitory interactions among the multipolar cells. Temporal filtering of small cell outputs by the circuit in the ELp establishes tuning to the peripheral spike timing differences that code for IPIs.

Fig. 6.

Friedman–Hopkins model for small cell duration tuning (modified from Xu-Friedman and Hopkins, 1999). Knollenorgans on one side of the body surface (pink) respond to the upward edge of a square pulse, and this gives rise to an excitatory input to a small cell via a nELL axonal delay line (Δt). Knollenorgans on the other side of the body surface (blue) respond to the downward edge, and this gives rise to an inhibitory input to the small cell via an ELa large cell. This is just one example of many possible receptive field organizations of the excitatory and inhibitory inputs to small cells. The responses of small cells to stimuli of different durations are determined by the length of the excitatory axonal delay (below). For short-duration stimuli, inhibition in response to stimulus offset blocks the delayed excitation in response to stimulus onset. For pulses that are longer than the excitatory delay, however, the delayed excitation arrives before the inhibition, and the small cell responds. Different small cells receive excitatory input with different delays, establishing variation in the minimum pulse duration that can elicit a response. As a result, increasing pulse duration leads to the progressive recruitment of small cells with longer axonal delays, and pulse duration is reflected in the total number of responding cells.

Fig. 7.

Multiplexed temporal codes are converted into distributed population codes in the ELa and the ELp. EOD waveform is represented by the pattern of responsive neurons across the population of small cells in the ELa, and both EOD waveform and EOD interval are represented by the pattern of responsive neurons across the population of multipolar cells in the ELp. Five different EOD waveforms are shown, from top to bottom: Brienomyrus brachyistius EOD, reversed polarity B. brachyistius EOD, elongated B. brachyistius EOD, Stomatorhinus ivindoensis EOD and Paramormyrops sp. ‘VAD’ EOD. Nine model small cells are shown and for each of the five EOD waveforms, pink indicates responding cells and black indicates non-responding cells. Each EOD waveform results in a unique pattern of responsiveness across the population of small cells. This information is sent to the ELp, where multipolar cells respond selectively to EOD waveform as well as EOD interval. Nine model multipolar cells are shown and for the five EOD waveforms and three different EOD intervals (from left to right: long, medium and short), yellow indicates responding cells and black indicates non-responding cells. Each possible combination of EOD waveform and EOD interval results in a unique pattern of responsive neurons across the population of multipolar cells.

Fig. 8.

IPI tuning of ELp neurons in response to electrosensory stimulation in vivo. Data from three different neurons are shown (data from Carlson, 2009). Intracellular recordings show responses to, from left to right, 100, 50 and 10 ms IPIs, with stimulus trains shown underneath each recording trace. On the far right, tuning curves show the normalized average spike rate and postsynaptic potential amplitude as a function of stimulus IPI in response to 10 repetitions of each stimulus [details on stimulus presentation and tuning curve generation can be found in Carlson (Carlson, 2009)].

Fig. 9.

Multiple synaptic mechanisms can establish IPI tuning. Different excitatory (green) and inhibitory (magenta) responses to short-interval stimulation (ticks) can give rise to different summated synaptic responses (black). In this simple model, summed responses are determined by subtracting the inhibitory response from the excitatory response. Temporal summation, in which synaptic responses to successive stimuli overlap in time, can establish IPI tuning. When excitation lasts longer than the stimulation interval, temporal summation leads to an increase in response characteristic of high-pass tuning. Conversely, temporal summation of inhibition occurs when inhibition lasts longer than the stimulation interval, causing a decrease in response characteristic of low-pass tuning. Short-term synaptic depression, or a decrease in synaptic strength with repeated stimulation, of inhibition can produce high-pass tuning, and depression of excitation can produce low-pass tuning. Facilitation, or an increase in synaptic strength with repeated stimulation, of excitation can result in high-pass tuning, whereas facilitation of inhibition can result in low-pass tuning. Finally, the relative timing of excitation and inhibition can establish interval-tuned responses. If excitation and inhibition occur near-coincidently, the response to a single stimulus would be small. Increasing latency of inhibition with repeated stimulation could result in high-pass tuning. In contrast, if the latencies of excitation and inhibition were fixed, and inhibition was delayed with respect to excitation, stimulation at short intervals could cause excitation to coincide with inhibition elicited by previous stimuli, resulting in low-pass tuning.

Fig. 10.

Convergence of high- and low-pass tuning can establish diverse temporal filters. The tuning curves of model neurons (purple) that receive synaptic inputs from one high-pass (magenta) and one low-pass (blue) neuron are shown. Excitatory connections are indicated by ‘+’ and inhibitory connections are indicated by ‘−’, and the tuning of each postsynaptic neuron is determined by a simple addition of ‘+’ tuning curves and subtraction of ‘−’ tuning curves. A variety of different types of interval tuning, including high-pass, low-pass, band-pass, band-stop and all-pass, can be created by different combinations of high-pass and low-pass excitation and inhibition.