HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 99/5/2641    most recent 00028.2008v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mehaffey, W. H. Articles by Turner, R. W. Search for Related Content PubMed PubMed Citation Articles by Mehaffey, W. H. Articles by Turner, R. W.

Intrinsic Frequency Tuning in ELL Pyramidal Cells Varies Across Electrosensory Maps

¹Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta; and ²Center for Neural Dynamics and Department of Cellular and Molecular Medicine, University of Ottawa, Ottawa, Ontario, Canada

Submitted 9 January 2008; accepted in final form 19 March 2008

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The tuning of neuronal responsiveness to specific stimulus frequencies is an important computation across many sensory modalities. The weakly electric fish Apteronotus leptorhynchus detects amplitude modulations of a self-generated quasi-sinusoidal electric organ discharge to sense its environment. These fish have to parse a complicated electrosensory environment with a wide range of possible frequency content. One solution has been to create multiple representations of the sensory input across distinct maps in the electrosensory lateral line lobe (ELL) that participate in distinct behavioral functions. E- and I-type pyramidal cells in the ELL that process sensory input further exhibit a preferred range of stimulus frequencies in relation to the different behaviors and sensory maps. We tested the hypothesis that variations in the intrinsic spiking mechanism of E- and I-type pyramidal cells contribute to map-specific frequency tuning. We find that E-cells exhibit a systematic change in their intrinsic spike characteristics and frequency tuning across sensory maps, whereas I-cells are constant in both spike characteristics and frequency tuning. As frequency tuning becomes more high-pass in E-cells, the refractory variables of spike half-width and afterhyperpolarization magnitude increase, spike threshold increases, adaptation becomes faster, and the gain of the spiking response decreases. These findings indicate that frequency tuning across sensory maps in the ELL is supported by differences in the intrinsic spike characteristics of pyramidal cells, revealing a link between cellular biophysical properties and signal processing in sensory maps with defined behavioral roles.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The complex and multi-dimensional nature of sensory stimuli often contains more information than a single neuron can accurately represent in a spike train. For the nervous system to encode important characteristics of the outside world, sensory neurons must initially respond to specific aspects of the sensory input. One component of natural signals commonly encoded by sensory neurons is frequency (either spatial or temporal), which has been shown to be important for visual (Butts et al. 2007; Lesica and Stanley 2004), auditory (Fritz et al. 2003; Woolley et al. 2005), somatosensory (Luna et al. 2005; Romo et al. 2004), and electrosensory (Chacron et al. 2003; Gabbiani and Metzner 1999; Oswald et al. 2004; Shumway 1989) processing. The ability of a cell to respond to specific frequencies within a broad band signal implies mechanisms for frequency tuning. Such frequency tuning can be implemented by intrinsic conductances and cell dynamics (Ellis et al. 2007b; Izhikevich 2002), by feedforward synaptic dynamics (Zhang et al. 2003), by short-term synaptic plasticity (de la Rocha and Parga 2005), and by feedback from higher brain centers (Chacron et al. 2003, 2005).

The electric fish Apteronotus leptorhynchus generates a weak quasi-sinusoidal electric organ discharge (EOD). Amplitude modulations (AMs) generated by perturbations of the field created by the EOD are used to sense the environment, to establish social hierarchies, and to communicate with conspecifics (Heiligenberg 1991). The electrosensory lateral line lobe (ELL) is the first CNS stage of electrosensory processing in the medulla of gymnotiform fish. It is divided into four topographic maps of the body surface termed the medial (MS), centromedial (CMS), centrolateral (CLS), and lateral (LS) segments (Bell and Maler 2005). Pyramidal neurons of the CMS, CLS, and LS receive primary afferent input from tuberous receptors that detect perturbations in the amplitude of the EOD. On entering the ELL, primary afferent fibers trifurcate and provide identical tuberous receptor input to each of the three maps (Carr et al. 1982; Heiligenberg 1991). The neural circuits in each of the three segments are also morphologically similar and function essentially as repeating topographical maps with no inter-map connections. The formation of distinct maps within a sensory modality is common in sensory systems and has been shown to allow parallel processing of different types of information (Metzner 1999; Schreiner and Winer 2007; Young 1998). In the ELL, pyramidal cells within these three segments show different preferences for the frequency content of input signals. Recordings in vivo reveal that cells in the more medial segments prefer inputs at lower frequencies while the more lateral segments prefer higher frequencies (R. Krahe, personal communication) (Shumway 1989). Furthermore, the differential frequency tuning between maps correlates to specific electrosensory behaviors: more medial segments process low-frequency-related input associated with a jamming avoidance response, and lateral segments process the high-frequency inputs involved in communication (Metzner and Juranek 1997). As ELL pyramidal cells show frequency selectivity to broadband inputs both in vitro (Ellis et al. 2007b) and in vivo (Chacron et al. 2003, 2005), they present an ideal model to identify the cellular and synaptic basis of frequency tuning across multiple sensory maps in the CNS.

Pyramidal cells within the ELL maps can be subdivided into two types based on the presence or absence of a basilar dendrite. Basilar pyramidal cells (E-cells) receive direct excitatory input from electrosensory afferents, whereas nonbasilar pyramidal cells (I-cells) are inhibited by granule cell interneurons (GC2) activated by primary afferents. This architecture allows these two cell classes to respond preferentially to upstrokes (E-cells) and downstrokes (I-cells) of EOD AMs as they are excited or inhibited by increases in EOD amplitude in the intact animal (Heiligenberg 1991). In vivo recordings have established a further frequency preference between these cells in the CLS and LS maps, with E-cells exhibiting tuning to higher frequencies and exhibiting greater regulation of frequency tuning than I-cells (Chacron et al. 2005). These results are also consistent with observations in vitro that intrinsic membrane properties of the E- and I-cell class are biased toward different frequency ranges (Ellis et al. 2007b). However, the contribution of intrinsic cellular mechanisms to differential frequency tuning across ELL maps remains to be determined. One promising line of investigation has examined the detection of input frequency by bursts of spikes. Indeed the segregation of bursts and single spikes to encode distinct frequency ranges has been observed in electric fish (Doiron et al. 2007; Gabbiani et al. 1996; Oswald et al. 2004, 2007) and in the mammalian visual (Lesica and Stanley 2004; Lesica et al. 2006) and auditory systems (Eggermont and Smith 1996). Bursts have also been shown to be more reliably timed to stimulus features, to show superior feature detection properties, and to more reliably activate downstream cells than can single spikes (Gabbiani et al. 1996; Izhikevich et al. 2003; Kepecs and Lisman 2003; Lisman 1997; Metzner et al. 1998). This suggests that bursts may have specific functions in information coding.

Extensive work on ELL pyramidal cells in vitro and in vivo has identified the role and significance of an intrinsic soma-dendritic interaction involving a conditional backpropagation of dendritic spikes in the generation of burst discharge (Turner et al. 1994). Moreover, this work has shown that pyramidal cell bursts are important for feature detection of sensory stimuli and particularly for low-frequency components of a stimulus (Mehaffey et al. 2007; Metzner et al. 1998; Oswald et al. 2004, 2007). The ability of pyramidal cells to encode low-frequency events with bursts is under significant regulation by intrinsic conductances (Ellis et al. 2007b; Rashid et al. 2001b), synaptic input (Mehaffey et al. 2007), and neuromodulatory agents (Ellis et al. 2007a) that can affect the underlying soma-dendritic interaction. Indeed we have previously shown that burst discharge and frequency tuning can be affected by the properties of pyramidal cell spike discharge. For instance, a high-threshold potassium current expressed in pyramidal cells allows high frequencies of spike discharge (Fernandez et al. 2005a). Increases in somatic spike width can alter bursting by reducing the mismatch between somatic and dendritic voltage that underlies a depolarizing afterpotential involved in producing spike bursts (Fernandez et al. 2005b). An apamin-sensitive AHP is involved in creating the frequency tuning profile of pyramidal cells (Ellis et al. 2007b), while a theoretical study has further shown that the adaptation rate of pyramidal cell discharge is capable of regulating low-frequency tuning (Benda and Herz 2003). Inasmuch as these parameters of spike discharge could be regulated in pyramidal cells, they could contribute to establishing differential frequency tuning across sensory maps.

In the present study, we tested the hypothesis that differential frequency tuning by pyramidal cells across ELL sensory maps involves the regulation of membrane properties that underlie spike generation and bursting, and their adaptation rates in response to quasi-naturalistic stimuli. In fact, we find that several fundamental aspects of ELL pyramidal cell spike waveform differ across maps in a manner that can contribute to differential frequency tuning.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Preparation of slices for electrophysiology

A. leptorhynchus were obtained from local importers and maintained at 26–28°C in fresh water aquaria in accordance with protocols approved by the University of Calgary Animal Care Committee. All chemicals were obtained from Sigma (St. Louis, MO) unless otherwise noted. Animals were anesthetized in 0.05% phenoxy-ethanol, and ELL tissue slices of 300–400 µm thickness were prepared as previously described (Turner et al. 1994). Slices were maintained by constant perfusion of ACSF (1–2 ml/min) and superfusion of humidified 95% O₂-5% CO₂ gas. ACSF contained (in mM) 124 NaCl, 3 KCl, 25 NaHCO₃, 1.0 CaCl₂, 1.5 MgSO₄, and 25 D-glucose, pH 7.4. HEPES-buffered ACSF for pressure ejection of pharmacological agents contained the same elements with the following differences: 148 mM NaCl, 10 mM HEPES.

Recording procedures

Glass microelectrodes were backfilled with 2 M KAc (pH 7.4; 90–120 M resistance) containing 2% Neurobiotin (Vector Labs, Burlington, ON, Canada). A total of 118 recordings were obtained from pyramidal cells in all three ELL segments receiving tuberous inputs (sample sizes for CMS: 44, CLS: 47, LS: 27). The map in which recordings were made was identifiable as the borders between maps are clearly defined in the in vitro slice. Recordings were digitized at 10–40 kHz using a NI PCI-6030E DAQ board (National Instruments, Austin, TX) and recorded in custom software using the Matlab data-acquisition toolbox (Mathworks, Natick, MA). Cells were held at a level just below firing threshold using negative current injection. Random amplitude modulations (RAMs) consisting of white noise low-pass filtered to 0–60 Hz were provided on top of the bias current and the SD of the waveform was adjusted to give average firing rates of 10–30 Hz, as at these frequencies the estimated coherences were reliable in their frequency preferences. Recordings were rejected if any part of the stimulation protocol could not be performed.

Cell fills

Pyramidal cells were filled with Neurobiotin using a positive current ejection pulse (+1 nA, 2 Hz). Following recording slices were transferred to 4% paraformaldehyde and fixed for several days at 4°C. Slices were then washed in 0.1 M phosphate buffer (PB) for several hours and placed in a solution of PB, Triton X-100 (0.1%), DMSO (0.5%), and streptavidin-Cy3 (1:1500) for 3 days. Slices were slide-mounted for visualization on an Olympus BH-2 research microscope to identify filled cells according to E- or I-cell, position within the ELL pyramidal cell body layer (PCL, superficial, intermediate or deep), and segment (CMS, CLS, or LS). Images were taken using Fluoview software on an Olympus FV300 BX50 confocal microscope. Of 118 cells recorded from the three maps, 43 were successfully filled with Neurobiotin. Superficial pyramidal cells were distinguished by the extent of their dendritic arborization and by location of the soma within the cell layer (Bastian and Nguyenkim 2001). Cells were classified as E- or I-cells depending on the presence or absence of a basilar dendrite or by their frequency preference (Ellis et al. 2007b). Of the filled cells classified following Neurobiotin injection, 8 were from the LS, 11 from the CLS, and 24 from the CMS.

Data analysis

All electrophysiological data were analyzed in Matlab. Data were plotted in Origin (OriginLab, Northhampton, MA) or Igor Pro (Wavemetrics, Lake Oswego, OR). Spike trains were digitized into binary trains in 0.5-ms bins and detrended to obtain a zero mean binary spike train. Coherence estimates were made between the binary spike trains and the original RAM stimulus and given by

where P_ss and P_rr denote the power spectrum of the stimulus and the response respectively, and P_sr denotes the cross-spectrum between the stimulus and response (e.g., the spike train). Pyramidal cells displaying spontaneous slow oscillations (Turner et al. 1996) were excluded from analysis.

Spike characteristics were measured at the weakest magnitude of DC current injection that elicited spiking and analyzed with custom software in MatLab. Afterhyperpolarization (AHP) measurements were taken as the minimum voltage between spikes. Spike threshold was assessed based on the derivative of the voltage trace and defined as the initial value eight times greater than the SD of the subthreshold noise. The level of current injection required to elicit bursting was assessed by visual inspection. Burst threshold was then defined as the amount of current injection required to transition from initial tonic spiking into bursting (Mehaffey et al. 2007; Noonan et al. 2003). Cells were classified as nonbursting if they reached a saturating firing frequency but did not burst or if >0.6 nA of current was injected above spike threshold without eliciting bursting. Gain was determined from the slope of a linear fit to the FI curve (Mehaffey et al. 2005). Adaptation time constants were fit to 100-s RAMs with spike trains binned into 1-s segments. Statistical significance was assessed using a one-way ANOVA with a significance level of P < 0.05 unless otherwise noted, with post hoc analysis using Tukey's HSD. Average values are presented as means ± SE.

Classification of cell types based on coherence

As we recorded primarily from the PCL, we sampled from populations of intermediate and superficial pyramidal cells and were unlikely to have recordings from deep pyramidal cells positioned adjacent to or within the granule cell layer (Bastian and Courtright 1991; Bastian et al. 2004). Further identification of pyramidal cell class on the basis of firing properties to square-wave current injection protocols has traditionally been difficult in vitro, requiring the filling and labeling of cells for histological analysis (Ellis et al. 2007b). In contrast, during in vivo recordings, the response to increases in EOD amplitude can be quickly tested and the cell identified as corresponding to either the E- or I-cell class. Our recent work established that a time-varying stimulus can also be used to distinguish E- and I-type pyramidal cells in vitro in terms of the frequency content of their response (Ellis et al. 2007b). Using this procedure, histologically identified I-cells in all segments were found to be ubiquitously low-pass in their frequency responsiveness, whereas identified E-cells in the CLS and LS were either broadband or high-pass in frequency responsiveness. In the CMS, superficial E-cells could deviate from this pattern in exhibiting low-pass responsiveness (Ellis et al. 2007b). Each of these findings were confirmed in the present study, with low-pass frequency responsiveness defined as coherence ratios [C(30–50 Hz)/C(0–20 Hz)] within 2 SDs above the average for anatomically identified I-cells (upper cut-off of 0.86). No anatomically filled E-cells in the CLS or LS fell into this range. Although this does not guarantee the absence of misclassifications, it suggests that such errors are rare.

The remaining cells were then subdivided into a class of broadband and high-pass cells by a coherence ratio of 0.87 to 1.19 (broadband), or a coherence ratio >1.19 (displaying an apparent preference for higher frequencies, see Fig. 6A and accompanying results). The E-cell group included any basilar pyramidal cells in the CMS that had been successfully labeled regardless of frequency preference, given the similarity in frequency tuning for some CMS E-cells (superficial pyramidal cells) with I-cells. Therefore our group of E-like CMS pyramidal cells is composed of intermediate basilar pyramidal cells along with superficial cells directly identified by intracellular fills. Further, our low-pass cell population may include a small number of superficial E-type CMS pyramidal cells that display low-pass characteristics (Ellis et al. 2007b). Although in the CMS this could lead to misclassification, any such errors would lead to the misclassification of superficial E-cells as I-cells. This would lead to an increase in the number of low-pass cells classified as I-cells, and a corresponding decrease in the number identified as E-cells. As no such differences were found, we are confident in our conclusion that although E- and I-cells of the CMS might show minor differences they are, as a population, indistinguishable.

View larger version (24K): [in this window] [in a new window]   FIG. 6. Frequency selectivity is distributed across maps. A: histogram showing the distribution of low-pass, broadband, and high-pass cells. The lower limit for the broadband cells was 0.86 and the upper limit 1.19 (), cells above this value were considered to display a preference for higher frequencies (e.g., high-pass). Note that for this small sample, the high-pass population occurs as discrete frequency-preferring clusters. The low-pass cells (<0.86) are shown in gray. B: distribution of cell classes across maps. The majority of cells in the CMS are low-pass, with a small number of broadband cells. Cells in the CLS are more equally distributed between low-pass and broadband populations. In the LS, all 3 classes of frequency tuning are observed with low-pass cells making up approximately half of the cells.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

View larger version (69K): [in this window] [in a new window]   FIG. 1. Schematic of electrosensory input to the electrosensory lateral line lobe (ELL) and the distribution of spike bursts across pyramidal cell types and ELL maps. A: the ELL is divided into 4 segmental maps; 3 segments [centromedial (CMS), centrolateral (CLS), and lateral (CMS) segments] receive inputs from peripheral tuberous electroreceptors that code amplitude information. Each primary afferent trifurcates after entering the ELL, resulting in a topographic representation of its receptive field in each of the 3 maps. B: each afferent provides input to both E- (basilar pyramidal) and I-type (nonbasilar) pyramidal cells. E-cells receive excitatory afferent inputs directly onto their basal dendrites and are excited by upstrokes of electric organ discharge (EOD) amplitude modulations (AMs). I-cells receive input through inhibitory granule cells II (GC2) and fire in response to downstrokes of the EOD AMs. C: representative traces of bursts in I-cells in each of the three ELL segments evoked in response to step intracellular current injections. Burst afterhyperpolarizations (AHPs) are indicated by arrows. D: fraction of E- and I-cells exhibiting burst discharge in response to step depolarizations in each segment. Bursts are more common in I- than E-cells within each map (*). E: the amount of current required to evoke the transition from isolated spiking to bursting (e.g., the rheobase for bursting relative to spike threshold) is significantly lower in I-cells of the LS compared with CLS and CMS maps. Note that this was only calculated for I-cells as too few bursting E-cells were observed in the more lateral segments (see D for statistical comparisons). *, statistical significance (P < 0.05).

We found significant differences in the fraction of E- and I-cells that displayed burst firing in each of the tuberous maps (Kruskal-Wallis 1-way ANOVA, P < 0.05). I-cells showed the greatest propensity to burst with the majority of I-cells in each map displaying bursting (CMS: 86%, CLS: 82%, LS: 70%; Fig. 1D). In contrast, bursting was less prevalent in E-cells in all maps with the smallest proportion of bursting cells apparent in the LS (CMS: 22%, CLS: 25%, LS: 7%; Fig. 1D). The incidence of bursting in E-cells proved to be significantly lower than that of I-cells regardless of map (P < 0.05, Mann-Whitney U). We then proceeded to examine the rheobase for burst firing, taken as the difference between the DC current injection required to evoke a burst and that required to evoke initial tonic spiking. Burst rheobase was characterized across maps exclusively for I-cells as the low number of E-cells that displayed bursting precluded any significant statistical power. Burst rheobase was almost identical between I-cells of the CMS and CLS (0.2 ± 0.02 and 0.2 ± 0.03 nA, respectively, P > 0.05) but was significantly lower in the LS (0.09 ± 0.01 nA, P < 0.05) in comparison to the other two maps (Fig. 1E). E-cells as a population were not significantly different from the population of I-cells (E-cells, 0.25 ± 0.07 nA, I-cells, 0.21 ± 0.03, P > 0.05). As ELL pyramidal cells appear to rest at membrane potentials near spike threshold, these results suggest that LS I-cells in vivo will have a lower threshold for bursting relative to I-cells in the other maps, although adaptation, threshold, differences in the convergence of synaptic input, and other processes (see following text) may also have an influence in vivo.

Expression of burst discharge in response to time-varying inputs

We extended our examination of burst firing by characterizing the response to time-varying inputs that more accurately simulate the input pyramidal cells receive in vivo. It has been established previously that pyramidal cell membrane voltage can accurately encode 0- to 60-Hz RAMs presented in vivo (Chacron et al. 2003; Middleton et al. 2006). We have therefore been able to use intracellular RAM current injection in vitro to mimic driving currents that result from external field RAMs in vivo (Ellis et al. 2007b; Mehaffey et al. 2007; Oswald et al. 2004, 2007). Briefly, we adjust the amplitude of a 100-s, 0- to 60-Hz RAM intracellular current stimulus to give an average 10- to 30-Hz firing rate, similar to the firing frequencies recorded in vivo during external RAM stimuli (Chacron et al. 2003, 2005; Doiron et al. 2003a).

Bursts evoked by time-varying inputs require the positive dendro-somatic feedback known to be generated by backpropagating apical dendritic spikes (Oswald et al. 2004). In response to RAMs, bursts are more often truncated to spike doublets, either through dendritic failure or rapid hyperpolarizations induced by high-frequency variations of the signal (Doiron et al. 2007). In fact, the dendrosomatic feedback is present even in cells that do not undergo the slow burst dynamic in response to DC current steps (Mehaffey et al. 2005). As a result of this spike-dependent feedback, time-varying inputs invoke a bimodal ISI histogram that can be divided into burst events and isolated spikes regardless of their ability to burst in response to DC current injections. We therefore parsed the resulting spike trains into bursts (typically spike doublets) and isolated spikes and separately computed the stimulus-response coherence for these events (Fig. 2). No significant difference was noted in the burst fraction between I-cells of the CMS, CLS, and LS and that of E-cells of the CMS (I-CMS: 0.22 ± 0.02, I-CLS: 0.22 ± 0.02, I-LS: 0.24 ± 0.04, E-CMS: 0.19 ± 0.02, P > 0.05; Fig. 2B). However, burst fraction did show a significant reduction in E-cells in both the CLS (0.14 ± 0.02) and LS (0.08 ± 0.02, P < 0.05). This suggests that the mediolateral decrease in the propensity for E-cells to burst in response to step current inputs (Fig. 1D) is also reflected in the response to time-varying inputs (Fig. 2B).

View larger version (27K): [in this window] [in a new window]   FIG. 2. Response of pyramidal cells to time-varying inputs. A: representative recording (top) from an I-cell of the CLS showing the patterns of bursts () and isolated spikes evoked in response to a time-varying intracellular current injection [random AM (RAM) bottom trace]. B: ratio of burst events in comparison to all spike events (burst fraction) evoked in response to time-varying stimuli. The total number of observed bursts decreases from medial to lateral segments in E-cells but not I-cells. C: fraction of the low-frequency (0–20 Hz) coherence encoded by bursts. A large fraction is encoded in all I-cells and CMS E-cells, but a smaller fraction is encoded by CLS and LS E-cells. D and E: representative interspike interval (ISI) densities for I-cells (D) and E-cells (E). Across maps, the timing of the nadir of the bimodal ISI histograms is conserved. Further, the number of events classifiable as burst events (e.g., the left peak) decreases from CMS to LS segments. , the nadir of the histogram used to distinguish short ISIs (e.g., burst events).

Taken together, these data suggest that the degree to which ELL pyramidal cells encode low frequencies with bursts varies across the maps (for E-cells) and cell types (E- vs. I-cells). The cells with the greatest tendency to express the slow burst dynamic (e.g., Fig. 1D) also show the strongest ability to encode low-frequency inputs using bursts. This occurs despite the fact that the positive dendro-somatic feedback required to establish the ISI histogram bimodality (Oswald et al. 2004) occurs even in cells that do not express these slow burst dynamics (Mehaffey et al. 2005). Although LS I-cells show a lower burst threshold in response to brief DC inputs, long-lasting RAMs invoked burst discharge no more often than in I-cells in other segments. This may be due to the adaptation observed over long periods of stimulation (see following text). Such adaptation implies that the lower threshold for bursting in LS I-cells would be observed primarily in response to low-frequency transients larger than the average background fluctuations to which they have adapted.

We plot representative ISI histograms for both I-cells and E-cells in Fig. 2, D and E, using a log scale for time to clearly show the bimodality (Ellis et al. 2007b; Turner et al. 1996). Previous studies have suggested that the nadir of the ISI histogram is highly conserved in the in vitro preparation (Ellis et al. 2007b; Mehaffey et al. 2007; Oswald et al. 2004; Turner et al. 1996). As described in the preceding text, all pyramidal cells, regardless of map or cell class, showed a clear bimodality with a nadir near 10 ms (8–10 ms, P > 0.05 across cell types and all maps), establishing ISIs <10 ms as our burst ISI criterion. Note that the lower density of ISIs in the first peak of the ISI histogram in E-cells of the CLS and LS are indicative of the lower burst fraction observed in these cells. A previous study examined differences between maps and between E- and I-cells, reporting that I-cells were better feature detectors and that the CMS as a whole outperformed the LS at feature detection, particularly when bursts where analyzed (Metzner et al. 1998). As bursting is more common in both I cells, and in the CMS (for E cells), this may be related to these in vivo results, particularly as the frequency cutoff for their RAM stimuli was between 2 and 40 Hz, a frequency range where bursts are expected to play an important role (Doiron et al. 2007).

E- and I-cells differ in spike properties

We recently showed that the SK potassium channel mediated AHPs of pyramidal cells differ between E- and I-cells and across ELL maps and differentially regulate the frequency selectivity of cells (Ellis et al. 2007b). We thus tested the hypothesis that differences in burst output and frequency tuning across ELL maps reflect a different complement of ionic currents between E- and I-cells. We first grouped all the identified E-cells (n = 39) and I-cells (n = 79), and examined various parameters of evoked spikes to establish whether there were any significant differences between the two cell classes. We examined the average characteristics of the spike waveform at the minimal current injection sufficient to induce repetitive firing. E- and I-cells proved to be significantly different in all of three key parameters examined (Fig. 3, A–C), including the magnitude of the AHP (I-cells: 6.18 ± 0.33, E-cells: 7.42 ± 0.41 mV, P < 0.05), spike half-width (I-cells: 0.62 ± 0.019 ms, E-cells: 0.81 ± 0.03, P < 0.05), and the voltage threshold for spiking (I-cells: –66.57 ± 074 mV, E-cells: –62.89 ± 0.74 mV, P < 0.05). Interestingly, the differences in voltage threshold for spiking may optimize a pyramidal cell's response to the inputs they receive. E-cells receive direct excitatory input from electroreceptors, whereas I-cells translate the EOD primarily through the removal of disynaptic inhibition. A lower spike threshold in I-cells may then contribute to higher sensitivity to afferent input. The factors determining threshold remain unknown but could involve low-threshold potassium currents known to be expressed in ELL pyramidal cells (Ellis et al. 2007a; Fernandez et al. 2005a; Mathieson and Maler 1988; Mehaffey et al. 2006; Smith et al. 2006). The larger AHP in E-cells may be related to the recently established preferential expression of somatic SK2 potassium channels in this cell class (Ellis et al. 2007b).

View larger version (14K): [in this window] [in a new window]   FIG. 3. E- and I-cell populations are significantly different with respect to spike properties. On average, E-cells exhibit a significantly wider spike (A), a larger magnitude AHP (B), and a higher spike threshold (C) than I-cells. Sample sizes for A–C are shown in brackets in C.

A number of studies have found differences in the expression pattern of specific potassium channels across the tuberous maps of the ELL, generally with an increasing intensity of expression along the mediolateral axis (Deng et al. 2005; Ellis et al. 2007b; Mehaffey et al. 2006; Rashid et al. 2001a; Smith et al. 2006). Gradients of ion channel expression in the auditory system have been shown to underlie differences in the spikes generated, which in turn can contribute to specialized neuronal computations (Brew and Forsythe 2005; Li et al. 2001; Parameshwaran et al. 2001; Rosenblatt et al. 1997; von Hehn et al. 2004). We therefore examined spiking characteristics to see if these gradients of channel expression translate into differences in the spike waveform, and in turn, frequency tuning across maps.

We predicted that spike parameters would vary between cell type and map as suggested by the distribution of potassium channels. Ideally these characteristics would, like coherence, provide an electrophysiological signature to identify cells without histological analysis. We began by examining I-cells (low-pass). Representative examples of spikes recorded from I-cells from each of the three maps are shown in Fig. 4A. When I-cell spikes are superimposed, little variability is detected (Fig. 4B) as supported by the lack of significant differences for various spike parameters between I-cells in each of the three maps. Spike half-width was nearly identical in I-cells between the three maps (CMS: 0.623 ± 0.035 ms, CLS: 0.613 ± 0.028 ms, LS: 0.617 ± 0.018 ms, P > 0.05), suggesting a similar net complement of sodium and potassium channels underlying the spike regardless of map. In support of this, there were no significant differences between I-cells of different maps in terms of the magnitude of the AHP (CMS: 5.70 ± 0.385 mV, CLS: 7.13 ± 0.78 mV, LS: 6.31 ± 1.36 mV, P > 0.05; Fig. 4D) or the spike voltage threshold (CMS: –67.5 ± 1.08 mV, CLS: 65.3 ± 1.15 mV, LS: –67.4 ± 2.02 mV, P > 0.05; Fig. 4C). This suggests that I-cells are highly homogenous in their spiking characteristics across the three maps.

View larger version (25K): [in this window] [in a new window]   FIG. 4. Spike parameters of I- and E-cells across tuberous segments. A: representative traces of individual spikes of I-cells recorded from each of the 3 tuberous segments, CMS (light gray), CLS (dark gray), and LS (black). B: superimposition of spikes shown in A reveals minimal variability in spike shape. C: average values from recordings in each map show no significant differences in I-cell spike half-width, AHP size, or spike threshold. D: representative traces of individual spikes recorded from E-cells within the segments, CMS (light gray), CLS (dark gray), and LS (black). E: superimposition of spikes shown in D reveals larger spike width of LS E-cells. F: spike half-width increases in a medio-lateral fashion and is significantly longer in the LS segment, whereas no significant differences in AHP size or spike threshold are found across maps. Sample sizes for recordings in each segment are shown in brackets.

The expression of SK2 potassium channels varies in a map-specific fashion for E-cells but not for I-cells (Ellis et al. 2007b). Taken together, our results suggest that other currents may follow a similar pattern, preferentially regulating the width and threshold of the spike waveform in the most lateral segment, but in a fashion exclusive to E-cells. However, the lack of a graded difference across maps in such variables as AHP depth does not match the expression pattern observed in immunohistochemical studies where mediolateral gradients of expression were observed for many different channels (Deng et al. 2005; Ellis et al. 2007b; Mehaffey et al. 2006; Rashid et al. 2001a; Smith et al. 2006). We hypothesized that these distinctions could lead to some heterogeneity in the computations performed by E-cells.

Frequency tuning characteristics vary with cell class and across maps

Having examined spike parameters, tendency to burst, and the relationship between bursting and low-frequency tuning in pyramidal cells, we next considered the full spike train, including both bursts and isolated spikes. We calculated the stimulus-response coherence for cells in all three maps and for both E- and I-cells using an intracellular current injected 0- to 60-Hz RAM stimulus. Figure 5 shows Neurobiotin fills of representative cells from each map along with the associated 0–60 Hz coherence. Consistent with a previous study (Ellis et al. 2007b), labeled I-cells were low-pass regardless of map (Fig. 5A), and displayed a clear peak in their stimulus-response coherence at low frequencies (0–20 Hz). In comparison, E-cells identified by their basilar dendrites (Fig. 5B) exhibited peaks in the coherence at low frequencies (e.g., Fig. 5B, CMS), a broadband frequency response (e.g., B, CLS), or a preference for higher frequencies (B, LS). Because our previous work had established that CMS E-cells can contain at least two populations (1 low-pass and 1 without noticeable frequency preference) (Ellis et al. 2007b), we considered whether similar heterogeneities might exist within the more lateral maps. We determined the coherence ratio for all cells (e.g., the ratio of the 30- to 50-Hz coherence to the 0- to 20-Hz coherence). This gave us an index of the frequency tuning characteristics of a cell, where high coherence ratios indicate high-pass characteristics, and numbers near unity indicate broadband characteristics (Ellis et al. 2007b; Mehaffey et al. 2007). Previous work has determined that the low-frequency tuned population of I-cells showed coherence ratios <0.86, and this cut-off was used to separate a low-frequency responsive population (mean coherence ratio, 0.66 ± 0.01). We then considered the possibility that cells may be subdivided into those displaying an opposite frequency preference—e.g., a preference for high frequencies. An examination of the distribution of coherence ratios of the cells (Fig. 6A) allowed a further separation into a population between 0.86 and 1.19 (mean of 0.96 ± 0.01), indicative of approximately broad band tuning. The distribution also displayed a long tail of discretely clustered cells (Fig. 6A, to the right of the arrow) that we used to define cells with a preference for higher frequencies over low frequencies (1.96 ± 0.3). We then examined the distribution of frequency tuning across all three maps.

View larger version (74K): [in this window] [in a new window]   FIG. 5. Frequency selectivity varies with cell type and map. In each case, representative examples of the coherence between a pyramidal cell spike train and the intracellular current RAM stimulus (0–60 Hz) is shown along with a representative Neurobiotin-labeled pyramidal cell. Basilar dendrites of E-cells are denoted in (B) by arrowheads. A: I-cells exhibit low-pass characteristics in preferring low frequency (0–20 Hz) over higher frequency (30–50 Hz) inputs to a similar degree across all ELL maps. B: E-cells show a clear shift in frequency selectivity across maps. Observed frequency tuning for E-cells includes low-pass (CMS), broadband (CLS), and high-pass (LS) selectivities. Scale bars: A: CMS, CLS: 75 µm, LS: 80 µm. B: CMS 60 µm, CLS, LS: 75 µm.

Spike characteristics correlate with frequency tuning

The heterogeneity observed in the frequency tuning of E-cells within maps suggested that frequency tuning may be a more important characteristic to correlate with the spike waveform characteristics rather than the map within which they are located. We therefore compared the spike parameters calculated previously (AHP, threshold, and spike half-width) but grouped the cells by their frequency tuning characteristics rather than by their map of origin. By doing so we in fact found that spike parameters vary most consistently with frequency tuning of the cell, rather than across maps. Low-pass cells showed the shortest spike half-width of any cell class (0.62 ± 0.019 ms), followed by broadband cells (0.73 ± 0.023 ms) with the widest half-widths in high-pass cells (0.98 ± 0.057 ms; Fig. 7A); all populations were significantly different from all others (P < 0.05). We were able to find similar differences in the AHP, where cells belonging to all three categories of frequency tuning were significantly different from each other (low-pass: 6.06 ± 0.26 mV, broadband: 7.19 ± 0.44 mV, high-pass: 9.04 ± 0.60 mV, P < 0.05; Fig. 7B). The threshold for spike initiation in low-pass cells was significantly lower than in broadband or high-pass cells, which were not different from each other (low-pass: –67.0 ± 0.73 mV, broadband: –63.7 ± 0.86, high-pass: –60.6 ± 1.4 mV, P < 0.05; Fig. 7C). We conclude from this that although spike parameters rarely change significantly across maps when cells are grouped purely by map or cell class (see Fig. 4), they do co-vary with frequency tuning of the cells.

View larger version (13K): [in this window] [in a new window]   FIG. 7. Spike characteristics of ELL pyramidal cells covary with frequency tuning. The mean values of spike parameters for all E- and I-cells in all segments are shown when grouped according to the preferred frequency selectivity. A: spike half-width exhibits a significant increase with higher frequency tuning. B: AHP size increases as the cell selects for higher frequencies. C: spike threshold increases in cells with higher frequency preferences. Sample sizes for recordings in each segment used are shown in brackets in C.

Because of the consistent variation of certain spike waveform characteristics with frequency tuning, we further examined the gain (in Hz/nA) of the firing response as another factor that could influence frequency selectivity. Although not a parameter of the spike waveform itself, gain is influenced by spike refractory variables, such as the rate of repolarization and AHP (Mehaffey et al. 2005; Troyer and Miller 1997).

As with the spike parameters, the gain was homogenous across I-cells of the CMS, CLS, and LS as well as E-cells of the CMS (I-cells CMS: 380 ± 27 Hz/nA. CLS: 391 ± 31 Hz/nA, LS: 376 ± 14 Hz/nA, E-cells CMS: 400 ± 20 Hz/nA, P > 0.05; Fig. 8A). Both CLS and LS E-cells were significantly different from all other cell types (P < 0.05) but not from each other (CLS: 270 ± 20 Hz/nA, LS: 186 ± 17 Hz/nA, P > 0.05; Fig. 8A). These distinctions in gain were further accentuated when the cells were grouped by their frequency tuning characteristics, revealing that all cell classes were significantly different from each other with the high-pass cells having the lowest gain (high-pass: 166 ± 22 Hz/nA, broadband: 316 ± 26 Hz/nA, low-pass: 405 ± 21 Hz/nA, P < 0.05; Fig. 8B). These data suggest that one of the outcomes of the differences in spike waveform across different frequency tuning is a variation of gain. This can be readily understood if it is considered that fast firing requires low-frequency inputs as high-frequency time-varying inputs regulate the firing rate independent of the refractory period. In contrast, the firing rate in response to low-frequency inputs depends on the intrinsic refractory period. Thus higher gains encourage low-frequency responses, while a cell with lower gain would be expected to fire fewer spikes in response to low-frequency stimulus components. The lower gain should not affect the response to higher-frequency inputs as under these conditions the limiting factor for the ISIs might be the statistics of the RAM, not the refractory dynamics of the cell.

View larger version (16K): [in this window] [in a new window]   FIG. 8. Response gain changes with segment, cell type, and frequency selectivity. A: gain for I- and E-cells of different ELL segments. I-cells have a homogenous gain across segments with no significant difference to CMS E-cells (including broadband cells). In comparison, E-cells of the CLS and LS have significantly lower gain. B: when classified by frequency tuning, the differences in gain become more apparent, gain decreases significantly with higher frequency tuning characteristics. Sample sizes for recordings used to calculate means are shown in brackets.

Another factor that can contribute to frequency tuning is adaptation (Benda and Herz 2003; Benda et al. 2005). We therefore examined the distribution of adaptation rates across cell types and maps. Adaptation was characterized in response to a 0- to 60-Hz RAM stimulus (Fig. 9A) and was well-fit by two time constants. The first, and largest amplitude, time constant was a fast decay between 0.5 and 4 s (representative data and fit is shown in Fig. 9B). The second time constant was far more variable and could either consist of a much smaller, slower decay or display a gradual acceleration. This slow acceleration may be due to the persistent sodium currents previously observed in these cells (Berman et al. 2001; Doiron et al. 2003b). This small magnitude slow adaptation or acceleration could appear in any cell class and could not be consistently related to any of our classifications. We focus primarily on the larger magnitude, fast component of adaptation as it showed a more clear relationship to cell class and across different maps.

View larger version (56K): [in this window] [in a new window]   FIG. 9. The rate of spike adaptation covaries across segments with frequency tuning. A: representative trace of the spike response of a pyramidal cell to a 0- to 60-Hz 100-s RAM intracellular current injection. B: adaptation of spike rate in response to the RAM in the example shown in A. Black lines show the instantaneous firing rate averaged in 1-s bins. The gray line is the fit to the data. C: distribution of adaptation time constants across ELL cell types and segments. E-cells of the CLS and LS have significantly shorter time constants for adaptation than other cell types in other segments. D: adaptation time constant correlates with frequency selectivity with a progressive and significant decrease in adaptation time constant with higher frequency tuning. Sample sizes for recordings used to calculate means are shown in brackets.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
We have used an in vitro slice preparation to examine the intrinsic cellular properties of two ELL pyramidal cell classes (E- and I-cells) across the three ELL maps receiving tuberous electrosensory input (CMS, CLS, LS). Our primary findings are that 1) the properties of I-cells are relatively homogenous with no significant differences across the maps for any of the properties examined, including frequency tuning, 2) E-cells are highly variable across the maps with respect to many of the biophysical parameters examined including frequency tuning, 3) the biophysical differences between E-cells are strongly correlated with their intrinsic frequency tuning. These results are significant in that our data with respect to frequency tuning are concordant with results previously obtained though recordings from ELL pyramidal cells in vivo. Thus I-cells in vivo are generally low-pass in all the tuberous maps while frequency tuning of E-cells transitions from low-pass (CMS) to high-pass (LS) (Chacron et al. 2005; Shumway 1989; R. Krahe, personal communication). Note that results obtained on frequency tuning in vivo are under the influence of many factors beyond intrinsic cellular mechanisms, such as feedback from higher brain centers. Recordings from the in vitro preparation here now show that differences in the intrinsic ion channel contributions to spike properties in E-cells account at least in part for frequency tuning within and across ELL maps in vivo.

Differences across cells, maps, and frequency tuning

The existence of different neural maps that are optimized for distinct types of information processing is a common architecture in sensory systems (Metzner 1999; Schreiner and Winer 2007; Young 1998). In the auditory brain stem, for example, a frequency preference is established in repeating maps that are correlated with specific patterns of channel expression (Brew and Forsythe 2005; Li et al. 2001; von Hehn et al. 2004). In the ELL, primary afferents from tuberous receptors disperse equally to three distinct spatial maps with similar architectures, yet frequency tuning differs across the three maps. Key differences have also begun to emerge between the E and I subtypes of pyramidal cells in terms of both frequency tuning and ion channel expression (Ellis et al. 2007b). We have now shown that several properties of spike discharge in pyramidal cells are most clearly correlated to frequency tuning. These cells are contained within maps with specific frequency preferences, but spike properties and frequency preferences in pyramidal cells were not perfectly correlated with these maps. This was largely due to the heterogeneity of E-cell as compared with I-cell spike properties and frequency tuning that establishes a differential responsiveness of these projection cells across ELL maps. We would therefore suggest that E-cells should therefore be examined specifically when candidate currents contributing to frequency selectivity are considered. Our results suggest that the conductances underlying the observed differences in spike waveform parameters (e.g., spike half-width, threshold, and AHP), spike threshold and gain, and factors that control pyramidal cell firing patterns (e.g., adaptation and bursting) all contribute to the frequency selectivity displayed by E-cells.

Spike half-width, AHP magnitude, and frequency tuning

Spike half-width and AHP magnitude co-varied with frequency tuning in a map-dependent fashion in E-cells. Both features are likely associated with an increased magnitude and/or duration of the refractory state and might be expected to contribute to frequency tuning, as predicted by a theoretical analysis (Benda and Herz 2003). The refractory variable depends on both Na⁺ channel inactivation and K⁺ channel activation In fact, Na⁺ channel inactivation and K⁺ channel activation are often grouped into a single refractory variable for simplicity of analysis in some neural models, including this preparation (Fernandez et al. 2005b; Rinzel 1985). We have confirmed that this is the case for SK2 channels, which increase their expression in a mediolateral fashion and contribute to a larger magnitude AHP in E-cells (Ellis et al. 2007b). Indirect evidence suggests that the high-threshold Kv3.1 channel involved in neuronal spike repolarization might also be related to the differences across maps. Expression of Kv3.1 increases from CMS to LS (Deng et al. 2005) and a high-threshold K⁺ channel (possibly of the Kv3 family) has been shown to stabilize high-frequency firing and maintain a brief spike width in ELL pyramidal cells (Fernandez et al. 2005a; Noonan et al. 2003). This suggests that in contrast to SK currents, high-threshold K⁺ currents may be preferentially expressed in I-cells as these cells have a more narrow spike half-width and, compared with E-cells, are capable of higher-frequency firing within each map. Further, these differences between E- and I-cells increase in the more lateral maps, suggesting that an increased Kv3.1 conductance may be contributing to some of the observed differences between E- and I-cells across the maps.

Spike threshold, gain, and frequency tuning

Spike threshold and gain may not intuitively be expected to represent important factors in controlling frequency tuning, but high-frequency-tuned LS E-cells proved to have a higher spike threshold compared with E-cells of the CMS and CLS (Fig. 7C). In this regard, a recent computational analysis has shown that spike threshold, in combination with purely anatomical map differences (e.g., receptive field size), can contribute to frequency tuning (J. Middleton, personal communication). The receptive field size of E-cells increases from CMS to LS (Shumway 1989 for E. virescens; L. Maler, unpublished observation for A. leptorhynchus), and therefore many more P-units converge onto E-cells of the LS compared with the CMS (the CLS is intermediate in this respect). A higher threshold in LS E-cells implies that they require synchronous input from many P-units to reach threshold (J. Middleton, personal communication). Because P-unit discharge is synchronized by high-frequency signals (Benda et al. 2006; J. Middleton, personal communication), this further implies that high-threshold pyramidal cells should be tuned to high-frequency input. The channels responsible for the differences in spike threshold are not known, but one possibility is a differential expression of Na⁺ channels or associated subunits, which, through a shift in either inactivation or activation rates could both raise threshold and increase spike width. Another candidate is the persistent Na⁺ current because this current is prominent in ELL pyramidal cells and enhances excitatory inputs and spike responses (Berman et al. 2001; Doiron et al. 2003b); direct evidence for a differential expression of persistent Na⁺ current in E-cells across maps is, however, not yet available.

Such in vivo effects of spike threshold are likely accentuated by the differential gain of E-cells across the maps. The mechanisms that underlie gain determine the ISI in response to the input rather than the timing of individual spikes. Therefore higher gains are only meaningful for inputs slow enough to generate two or more spikes. A lower gain would be expected to decrease the response to low frequency inputs because fewer spikes would be generated in response to the signal upstrokes. In contrast, the lower gain should not affect the response to higher-frequency inputs as faster oscillations generally create single spikes for each stimulus upstroke (Oswald et al. 2004, 2007). Under these circumstances, the ISI is largely determined by the statistical properties of the time-varying input signal rather than by the gain dynamics of the cell. The biophysical properties determining the intrinsic gain are not completely known, but both subthreshold Na⁺ and spike-initiated Na⁺ and K⁺ channels are likely to contribute (Fernandez et al. 2005b; Noonan et al. 2003).

We have previously shown that the intrinsic gain of pyramidal cells can also be regulated at the network level: dendritic inhibition invoked by a specific interneuron in the ventral molecular layer (VML cell) induces divisive gain control by reducing the DAP emanating from pyramidal cell apical dendrites (Mehaffey et al. 2005). The VML cell receives its input entirely from feedback to the ELL (Maler 1979; Maler and Mugnaini 1994). It is therefore interesting that the LS has the highest density of VML cells (Maler 1979; Maler and Mugnaini 1994; Shumway 1989). This suggests that, more generally, feedback control of gain and frequency tuning (Bastian 1986a,b; Bastian et al. 2004; Chacron et al. 2003; Mehaffey et al. 2005) might be linked through synaptic regulation of the channels responsible for the differential expression of gain across the ELL tuberous maps.

Spike frequency adaptation and frequency tuning

As discussed by Benda and Herz (2003), temporal summation of adapting currents (e.g., currents that lead to a progressive slowing of spike frequency) also contribute to spike frequency adaptation that can promote high-frequency tuning. We have shown that E-cells tuned to higher frequencies show faster adaptation (Fig. 9). These neurons are also those least likely to generate bursts, both in response to DC steps and to time-varying inputs. A faster rate of adaptation is expected to be recruited rapidly during slow upstrokes in the external stimulus by virtue of their generating faster ISIs than those induced by higher-frequency components of the stimulus. A fast firing rate will recruit adapting currents and therefore downregulate the response to low-frequency stimulus components. However, adaptation is unlikely to accumulate as strongly in response to fast stimulus components as these generate primarily single spikes, permitting a more rapid recovery during stimulus-driven pauses in spiking, which in turn allows high-frequency components to be coded more accurately. The biophysical basis of spike frequency adaptation in pyramidal cells is not currently known, but blockade of SK currents with apamin does not prevent adaptation in E-cells (W. H. Mehaffey, unpublished observation), suggesting that another calcium-activated potassium current might be involved. This is consistent with observations in the lateral amygdala where SK currents do not contribute to spike frequency adaptation (Faber and Sah 2002).

Bursting and low-frequency tuning

Many studies have now shown links between burst discharge and low-frequency events in the electrosensory system (Doiron et al. 2007; Krahe and Gabbiani 2004) and in other sensory systems (Krahe and Gabbiani 2004; Lesica and Stanley 2004; Lesica et al. 2006). This appears to be a useful adaptation for generating distinct spike patterns in response to specific features of a sensory signal. In our examination of intrinsic bursting, we found that the distribution of bursting cells varies, occurring most frequently across ELL maps in association with cells that prefer low-frequency events. This included I-cells in all three maps, whereas bursting was less commonly observed in E-cells, particularly in more lateral maps (Fig. 1D). Further, E-cells show lower stimulus-response coherence at low frequencies (0–20 Hz) than I-cells, particularly in the CLS and LS. This is consistent with our previous observations that conductances that can regulate burst threshold are able to decrease the low-frequency component of stimulus-response coherence (Ellis et al. 2007b). In a recent paper, we examined the contribution of an SK-mediated AHP to frequency tuning in response to broadband inputs (Ellis et al. 2007b). This current may make an important contribution to the relationship between AHP and frequency selectivity observed here, but it is interesting to note that although apamin increased the low-frequency coherence in broadband or high-pass cells, it did not alter their high-pass characteristics. This suggests that the SK2 current reduces the low-frequency coherence rather than amplifying high-frequency coherence. Therefore at least one other conductance is likely required to confer high-pass tuning. One candidate current would be the A-type potassium channels that have been described in ELL pyramidal cells (Ellis et al. 2007a; Mathieson and Maler 1988) and that can decrease a neuron's response to low-frequency input (Ellis et al. 2007a). Further, the A-type channel of pyramidal cells is modulated by muscarinic acetylcholine receptors, and this in turn regulates their frequency tuning (Ellis et al. 2007a). Together this suggests that synaptic inputs can modulate pyramidal cell frequency tuning directly via voltage-gated channels (see preceding text) or via second-messenger regulation of intrinsic conductances.

I-cells and low-frequency tuning

We found that I-cells exhibited low-frequency tuning across all three tuberous maps. The low-frequency tuning observed in vivo (Chacron et al. 2005; R. Krahe, personal communication) is therefore due, at least in part, to the intrinsic properties of I-cells; such as their greater propensity to burst, narrow spike-width and higher gain, allowing a greater response to low-frequency stimulus components. The intrinsic low-frequency tuning of I-cells might also be accentuated by the neuronal architecture of the ELL: because I-cells receive inputs indirectly through the granular cell interneurons, it is possible that this disynaptic relay may contribute to low-pass tuning. It should also be noted that decreases in EOD amplitude (e.g., the stimuli for which I-cells are selective) typically result from the fish swimming past nonconductive objects that generate large low-frequency signals relative to the small amplitude increases in EOD amplitude at the trailing edges of the electric image (Chen et al. 2005; MacIver et al. 2001). Thus the intrinsic and network-dependent tuning properties of I-cells may be well suited to the detection of inanimate nonconductive stimuli such as rocks, although other contexts may lead to large-amplitude decreases in EOD (e.g., beats caused by conspecifics that can contain both high- and low-frequency components).

The vital role of frequency tuning in the ELL is clearly illustrated by the observed contributions at every level of neuronal processing. From the high-frequency selectivity of synchronized presynaptic activity (Benda et al. 2006; J. Middleton, personal communication) to the regulation of conductances that underlie spike generation and firing patterns (Ellis et al. 2007a,b; Fernandez et al. 2005a; Turner et al. 1994) to feedback regulation of frequency tuning by higher brain centers (Chacron et al. 2003, 2005), many factors contribute to regulating the computations performed by ELL pyramidal cells. As we show here, intrinsic cell properties constitute an integral part of the frequency tuning mechanism across sensory maps in the ELL, identifying potential targets for feedback or feedforward mechanisms to regulate cellular properties and thus frequency tuning according to ongoing network activity.

	GRANTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Funding was provided by Canadian Institutes of Health Resources operating grants to R. W. Turner and L. Maler, and CIHR and Alberta Herirage Foundation for Medical Research studentships to W. H. Mehaffey. R. W. Turner is an AHFMR Scientist.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
We thank M Kruskic for technical assistance and F. R. Fernandez, J. Middleton, and J.D.T. Engbers for helpful conversations.

	FOOTNOTES

 
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Address for reprint requests and other correspondence: R. W. Turner, Hotchkiss Brain Institute, University of Calgary, 3330 Hospital Dr. N.W., Calgary, Alberta T2N 4N1, Canada (E-mail: rwturner{at}ucalgary.ca)

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Bastian J. Gain control in the electrosensory system mediated by descending inputs to the electrosensory lateral line lobe. J Neurosci 6: 553–562, 1986a.[Abstract]

Bastian J. Gain control in the electrosensory system: a role for the descending projections to the electrosensory lateral line lobe. J Comp Physiol [A] 158: 505–515, 1986b.[CrossRef][Medline]

Bastian J, Chacron MJ, Maler L. Plastic and nonplastic pyramidal cells perform unique roles in a network capable of adaptive redundancy reduction. Neuron 41: 767–779, 2004.[CrossRef][Web of Science][Medline]

Bastian J, Courtright J. Morphological correlates of pyramidal cell adaptation rate in the electrosensory lateral line lobe of weakly electric fish. J Comp Physiol [A] 168: 393–407, 1991.[CrossRef][Medline]

Bastian J, Nguyenkim J. Dendritic modulation of burst-like firing in sensory neurons. J Neurophysiol 85: 10–22, 2001.[Abstract/Free Full Text]

Bell CC, Maler L. Central neuroanatomy of electrosensory systems in fish. Electroreception. In: Springer Handbook of Auditory Research, edited by Bullock TH. New York: Springer, 2005, p. xvi.

Benda J, Herz AV. A universal model for spike-frequency adaptation. Neural Comput 15: 2523–2564, 2003.[CrossRef][Web of Science][Medline]

Benda J, Longtin A, Maler L. Spike-frequency adaptation separates transient communication signals from background oscillations. J Neurosci 25: 2312–2321, 2005.[Abstract/Free Full Text]

Benda J, Longtin A, Maler L. A synchronization-desynchronization code for natural communication signals. Neuron 52: 347–358, 2006.[CrossRef][Web of Science][Medline]

Berman N, Dunn RJ, Maler L. Function of NMDA receptors and persistent sodium channels in a feedback pathway of the electrosensory system. J Neurophysiol 86: 1612–1621, 2001.[Abstract/Free Full Text]

Brew HM, Forsythe ID. Systematic variation of potassium current amplitudes across the tonotopic axis of the rat medial nucleus of the trapezoid body. Hear Res 206: 116–132, 2005.[CrossRef][Web of Science][Medline]

Butts DA, Weng C, Jin J, Yeh CI, Lesica NA, Alonso JM, Stanley GB. Temporal precision in the neural code and the time scales of natural vision. Nature 449: 92–95, 2007.[CrossRef][Medline]

Carr CE, Maler L, Sas E. Peripheral organization and central projections of the electrosensory nerves in gymnotiform fish. J Comp Neurol 211: 139–153, 1982.[CrossRef][Web of Science][Medline]

Chacron MJ, Doiron B, Maler L, Longtin A, Bastian J. Non-classical receptive field mediates switch in a sensory neuron's frequency tuning. Nature 423: 77–81, 2003.[CrossRef][Medline]

Chacron MJ, Maler L, Bastian J. Feedback and feedforward control of frequency tuning to naturalistic stimuli. J Neurosci 25: 5521–5532, 2005.[Abstract/Free Full Text]

Chen L, House JL, Krahe R, Nelson ME. Modeling signal and background components of electrosensory scenes. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 191: 331–345, 2005.[CrossRef][Web of Science][Medline]

de la Rocha J, Parga N. Short-term synaptic depression causes a non-monotonic response to correlated stimuli. J Neurosci 25: 8416–8431, 2005.[Abstract/Free Full Text]

Deng Q, Rashid AJ, Fernandez FR, Turner RW, Maler L, Dunn RJ. A C-terminal domain directs Kv3.3 channels to dendrites. J Neurosci 25: 11531–11541, 2005.[Abstract/Free Full Text]

Doiron B, Chacron MJ, Maler L, Longtin A, Bastian J. Inhibitory feedback required for network oscillatory responses to communication but not prey stimuli. Nature 421: 539–543, 2003a.[CrossRef][Medline]

Doiron B, Noonan L, Lemon N, Turner RW. Persistent Na⁺ current modifies burst discharge by regulating conditional backpropagation of dendritic spikes. J Neurophysiol 89: 324–337, 2003b.[Abstract/Free Full Text]

Doiron B, Oswald AM, Maler L. Interval coding. II. Dendrite-dependent mechanisms. J Neurophysiol 97: 2744–2757, 2007.[Abstract/Free Full Text]

Eggermont JJ, Smith GM. Burst-firing sharpens frequency-tuning in primary auditory cortex. Neuroreport 7: 753–757, 1996.[Web of Science][Medline]

Ellis LD, Krahe R, Bourque CW, Dunn RJ, Chacron MJ. Muscarinic receptors control frequency tuning through the downregulation of an a-type potassium current. J Neurophysiol 98: 1526–1537, 2007a.[Abstract/Free Full Text]

Ellis LD, Mehaffey WH, Harvey-Girard E, Turner RW, Maler L, Dunn RJ. SK channels provide a novel mechanism for the control of frequency tuning in electrosensory neurons. J Neurosci 27: 9491–9502, 2007b.[Abstract/Free Full Text]

Faber ES, Sah P. Physiological role of calcium-activated potassium currents in the rat lateral amygdala. J Neurosci 22: 1618–1628, 2002.[Abstract/Free Full Text]

Fernandez FR, Mehaffey WH, Molineux ML, Turner RW. High-threshold K⁺ current increases gain by offsetting a frequency-dependent increase in low-threshold K⁺ current. J Neurosci 25: 363–371, 2005a.[Abstract/Free Full Text]

Fernandez FR, Mehaffey WH, Turner RW. Dendritic Na⁺ current inactivation can increase cell excitability by delaying a somatic depolarizing afterpotential. J Neurophysiol 94: 3836–3848, 2005b.[Abstract/Free Full Text]

Fritz J, Shamma S, Elhilali M, Klein D. Rapid task-related plasticity of spectrotemporal receptive fields in primary auditory cortex. Nat Neurosci 6: 1216–1223, 2003.[CrossRef][Web of Science][Medline]

Gabbiani F, Metzner W. Encoding and processing of sensory information in neuronal spike trains. J Exp Biol 202: 1267–1279, 1999.[Abstract]

Gabbiani F, Metzner W, Wessel R, Koch C. From stimulus encoding to feature extraction in weakly electric fish. Nature 384: 564–567, 1996.[CrossRef][Medline]

Heiligenberg W. Neural Nets in Electric Fish. Cambridge, MA: MIT Press, 1991.

Izhikevich EM. Resonance and selective communication via bursts in neurons having subthreshold oscillations. Biosystems 67: 95–102, 2002.[CrossRef][Web of Science][Medline]

Izhikevich EM, Desai NS, Walcott EC, Hoppensteadt FC. Bursts as a unit of neural information: selective communication via resonance. Trends Neurosci 26: 161–167, 2003.[CrossRef][Web of Science][Medline]

Kepecs A, Lisman J. Information encoding and computation with spikes and bursts. Network 14: 103–118, 2003.[Web of Science][Medline]

Krahe R, Gabbiani F. Burst firing in sensory systems. Nat Rev Neurosci 5: 13–23, 2004.[CrossRef][Web of Science][Medline]

Lemon N, Turner RW. Conditional spike backpropagation generates burst discharge in a sensory neuron. J Neurophysiol 84: 1519–1530, 2000.[Abstract/Free Full Text]

Lesica NA, Stanley GB. Encoding of natural scene movies by tonic and burst spikes in the lateral geniculate nucleus. J Neurosci 24: 10731–10740, 2004.[Abstract/Free Full Text]

Lesica NA, Weng C, Jin J, Yeh CI, Alonso JM, Stanley GB. Dynamic encoding of natural luminance sequences by LGN bursts. PLoS Biol 4: e209, 2006.[CrossRef][Medline]

Li W, Kaczmarek LK, Perney TM. Localization of two high-threshold potassium channel subunits in the rat central auditory system. J Comp Neurol 437: 196–218, 2001.[CrossRef][Web of Science][Medline]

Lisman JE. Bursts as a unit of neural information: making unreliable synapses reliable. Trends Neurosci 20: 38–43, 1997.[CrossRef][Web of Science][Medline]

Luna R, Hernandez A, Brody CD, Romo R. Neural codes for perceptual discrimination in primary somatosensory cortex. Nat Neurosci 8: 1210–1219, 2005.[CrossRef][Web of Science][Medline]

MacIver MA, Sharabash NM, Nelson ME. Prey-capture behavior in gymnotid electric fish: motion analysis and effects of water conductivity. J Exp Biol 204: 543–557, 2001.[Abstract]

Maler L. The posterior lateral line lobe of certain gymnotoid fish: quantitative light microscopy. J Comp Neurol 183: 323–363, 1979.[CrossRef][Web of Science][Medline]

Maler L, Mugnaini E. Correlating gamma-aminobutyric acidergic circuits and sensory function in the electrosensory lateral line lobe of a gymnotiform fish. J Comp Neurol 345: 224–252, 1994.[CrossRef][Web of Science][Medline]

Mathieson WB, Maler L. Morphological and electrophysiological properties of a novel in vitro preparation: the electrosensory lateral line lobe brain slice. J Comp Physiol [A] 163: 489–506, 1988.[CrossRef][Medline]

Mehaffey WH, Doiron B, Maler L, Turner RW. Deterministic multiplicative gain control with active dendrites. J Neurosci 25: 9968–9977, 2005.[Abstract/Free Full Text]

Mehaffey WH, Fernandez FR, Maler L, Turner RW. Regulation of burst dynamics improves differential encoding of stimulus frequency by spike train segregation. J Neurophysiol 98: 939–951, 2007.[Abstract/Free Full Text]

Mehaffey WH, Fernandez FR, Rashid AJ, Dunn RJ, Turner RW. Distribution and function of potassium channels in the electrosensory lateral line lobe of weakly electric apteronotid fish. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 192: 637–648, 2006.[CrossRef][Web of Science][Medline]

Metzner W. Why are there so many sensory brain maps? Cell Mol Life Sci 56: 1–4, 1999.[CrossRef][Web of Science][Medline]

Metzner W, Juranek J. A sensory brain map for each behavior? Proc Natl Acad Sci USA 94: 14798–14803, 1997.[Abstract/Free Full Text]

Metzner W, Koch C, Wessel R, Gabbiani F. Feature extraction by burst-like spike patterns in multiple sensory maps. J Neurosci 18: 2283–2300, 1998.[Abstract/Free Full Text]

Middleton JW, Longtin A, Benda J, Maler L. The cellular basis for parallel neural transmission of a high-frequency stimulus and its low-frequency envelope. Proc Natl Acad Sci USA 103: 14596–14601, 2006.[Abstract/Free Full Text]

Noonan L, Doiron B, Laing C, Longtin A, Turner RW. A dynamic dendritic refractory period regulates burst discharge in the electrosensory lobe of weakly electric fish. J Neurosci 23: 1524–1534, 2003.[Abstract/Free Full Text]

Oswald AM, Chacron MJ, Doiron B, Bastian J, Maler L. Parallel processing of sensory input by bursts and isolated spikes. J Neurosci 24: 4351–4362, 2004.[Abstract/Free Full Text]

Oswald AM, Doiron B, Maler L. Interval coding. I. Burst interspike intervals as indicators of stimulus intensity. J Neurophysiol 97: 2731–2743, 2007.[Abstract/Free Full Text]

Parameshwaran S, Carr CE, Perney TM. Expression of the Kv3.1 potassium channel in the avian auditory brain stem. J Neurosci 21: 485–494, 2001.[Abstract/Free Full Text]

Rashid AJ, Dunn RJ, Turner RW. A prominent soma-dendritic distribution of Kv3.3 K⁺ channels in electrosensory and cerebellar neurons. J Comp Neurol 441: 234–247, 2001a.[CrossRef][Web of Science][Medline]

Rashid AJ, Morales E, Turner RW, Dunn RJ. The contribution of dendritic Kv3 K⁺ channels to burst threshold in a sensory neuron. J Neurosci 21: 125–135, 2001b.[Abstract/Free Full Text]

Rinzel J. Excitation dynamics: insights from simplified membrane models. Fed Proc 44: 2944–2946, 1985.[Web of Science][Medline]

Romo R, Hernandez A, Zainos A. Neuronal correlates of a perceptual decision in ventral premotor cortex. Neuron 41: 165–173, 2004.[CrossRef][Web of Science][Medline]

Rosenblatt KP, Sun ZP, Heller S, Hudspeth AJ. Distribution of Ca²⁺-activated K⁺ channel isoforms along the tonotopic gradient of the chicken's cochlea. Neuron 19: 1061–1075, 1997.[CrossRef][Web of Science][Medline]

Schreiner CE, Winer JA. Auditory cortex mapmaking: principles, projections, and plasticity. Neuron 56: 356–365, 2007.[CrossRef][Web of Science][Medline]

Shumway CA. Multiple electrosensory maps in the medulla of weakly electric gymnotiform fish. I. Physiological differences. J Neurosci 9: 4388–4399, 1989.[Abstract]

Smith GT, Unguez GA, Weber CM. Distribution of Kv1-like potassium channels in the electromotor and electrosensory systems of the weakly electric fish Apteronotus leptorhynchus. J Neurobiol 66: 1011–1031, 2006.[CrossRef][Web of Science][Medline]

Troyer TW, Miller KD. Physiological gain leads to high ISI variability in a simple model of a cortical regular spiking cell. Neural Comput 9: 971–983, 1997.[CrossRef][Web of Science][Medline]

Turner RW, Maler L, Deerinck T, Levinson SR, Ellisman MH. TTX-sensitive dendritic sodium channels underlie oscillatory discharge in a vertebrate sensory neuron. J Neurosci 14: 6453–6471, 1994.[Abstract]

Turner RW, Plant JR, Maler L. Oscillatory and burst discharge across electrosensory topographic maps. J Neurophysiol 76: 2364–2382, 1996.[Abstract/Free Full Text]

von Hehn CA, Bhattacharjee A, Kaczmarek LK. Loss of Kv3.1 tonotopicity and alterations in cAMP response element-binding protein signaling in central auditory neurons of hearing impaired mice. J Neurosci 24: 1936–1940, 2004.[Abstract/Free Full Text]

Woolley SM, Fremouw TE, Hsu A, Theunissen FE. Tuning for spectro-temporal modulations as a mechanism for auditory discrimination of natural sounds. Nat Neurosci 8: 1371–1379, 2005.[CrossRef][Web of Science][Medline]

Young ED. Parallel processing in the nervous system: evidence from sensory maps. Proc Natl Acad Sci USA 95: 933–934, 1998.[Free Full Text]

Zhang LI, Tan AY, Schreiner CE, Merzenich MM. Topography and synaptic shaping of direction selectivity in primary auditory cortex. Nature 424: 201–205, 2003.[CrossRef][Medline]

This article has been cited by other articles:

				J. W. Middleton, A. Longtin, J. Benda, and L. Maler Postsynaptic Receptive Field Size and Spike Threshold Determine Encoding of High-Frequency Information Via Sensitivity to Synchronous Presynaptic Activity J Neurophysiol, March 1, 2009; 101(3): 1160 - 1170. [Abstract] [Full Text] [PDF]

				R. Krahe, J. Bastian, and M. J. Chacron Temporal Processing Across Multiple Topographic Maps in the Electrosensory System J Neurophysiol, August 1, 2008; 100(2): 852 - 867. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 99/5/2641    most recent 00028.2008v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mehaffey, W. H. Articles by Turner, R. W. Search for Related Content PubMed PubMed Citation Articles by Mehaffey, W. H. Articles by Turner, R. W.