Interval-counting neurons in the anuran auditory midbrain: factors underlying diversity of interval tuning
- First Online:
- Cite this article as:
- Rose, G.J., Leary, C.J. & Edwards, C.J. J Comp Physiol A (2011) 197: 97. doi:10.1007/s00359-010-0591-8
- 116 Views
In anurans, the temporal patterning of sound pulses is the primary information used for differentiating between spectrally similar calls. One class of midbrain neurons, referred to as ‘interval-counting’ cells, appears to be particularly important for discriminating among calls that differ in pulse repetition rate (PRR). These cells only respond after several pulses are presented with appropriate interpulse intervals. Here we show that the range of selectivity and sharpness of interval tuning vary considerably across neurons. Whole-cell recordings revealed that neurons showing temporally summating excitatory postsynaptic potentials (EPSPs) with little or no inhibition or activity-dependent enhancement of excitation exhibited low-pass or band-pass tuning to slow PRRs. Neurons that showed inhibition and rate-dependent enhancement of excitation, however, were band-pass or high-pass to intermediate or fast PRRs. Surprisingly, across cells, interval tuning based on membrane depolarization and spike rate measures were not significantly correlated. Neurons that lacked inhibition showed the greatest disparities between these two measures of interval tuning. Cells that showed broad membrane potential-based tuning, for example, varied considerably in their spike rate-based tuning; narrow spike rate-based tuning resulted from ‘thresholding’ processes, whereby only the largest depolarizations triggered spikes. The potential constraints associated with generating interval tuning in this manner are discussed.
KeywordsWhole-cellAuditoryMidbrainTemporal processingAmplitude modulation
The timing of successive sound elements is an important temporal feature of the acoustic communication signals of many animals, including human speech (Ehret 1996). In many anuran species, these and other temporal properties of calls play a pivotal role in reproductive isolation, mate choice, and coordination of reproductive behavior (Gerhardt and Huber 2002; Wells and Schwartz 2007). In particular, the intervals between the onsets of successive pulses and pulse duration constitute the primary temporal acoustic features that enable most frog and toad species to differentiate between conspecific call types and reject heterospecific calls (Gerhardt 1982, 1988). The neural bases of these discriminative abilities are, therefore, of considerable interest.
In the peripheral auditory system, the timing of pulses is represented in the timing of afferent spikes, but selectivity for particular pulse repetition rates (PRRs) is absent (Rose and Capranica 1985). This temporal code is ‘read’ in the central auditory system, however, such that individual neurons in the torus semicircularis (anuran homologue of the inferior colliculus, IC) respond selectively over a narrow range of pulse repetition or amplitude modulation (AM) rates (Rose and Capranica 1983, 1984, 1985; Rose and Gooler 2007). Neurons of one class respond best over a particular range of PRRs, but not to short-duration pulses that are repeated at much slower rates (Alder and Rose 1998, 2000). Perhaps the most interesting property of these neurons is that they respond only after a threshold number of pulses have been presented at the optimal rate. Further, it is the number of consecutive correct intervals (time between successive pulse onsets) that is important for eliciting spikes, not the number of pulses that occur within a particular integration time window (Edwards et al. 2002); that is, responses of these neurons reflect an interval-counting process.
Buonomano (2000) developed a model of how interval selectivity could arise from integration of excitation and inhibition that undergo short-term enhancement or depression, respectively, for a series of short inter-spike intervals. Our initial whole-cell (intracellular) recordings from interval-counting IC neurons, in vivo, have provided some support for a mechanism of this general nature (Edwards et al. 2007). In many cases, pulses presented at slow rates (long interpulse intervals) elicit inhibition and weak excitation. During a series of short interpulse intervals, however, excitation is enhanced and overcomes the concurrent inhibition; the PRR at which enhancement was first observed served as a predictor of best PRR (Edwards et al. 2007). In the present study we further examined how variation in inhibition and activity-dependent excitation contributes to differences in interval selectivity. We predicted that interval-counting neurons that show little, if any, inhibition or rate-dependent enhancement of excitation should be selective for very slow PRRs. We compare the interval tuning of these cells and those that show the more typical pattern of inhibition and rate-dependent excitation.
Secondly, we investigated how ‘thresholding’ properties might contribute to the interval selectivity of interval-counting neurons. Intracellular recordings from IC neurons in bats (Gittelman et al. 2009) and visual cortical cells in cats (Priebe and Ferster 2005) have revealed that the proximity of spike threshold to peak stimulus-driven depolarizations strongly influences the spike-rate-based selectivity for temporal features. We predicted, therefore, that such ‘thresholding’ properties would influence the sharpness of interval tuning in the anuran IC. We compare the interval tuning of membrane depolarization with that derived from spike rate measures and provide evidence that supports this hypothesis.
Materials and methods
Pacific tree frogs (Hyla regilla) and northern leopard frogs (Rana pipiens pipiens) were prepared for recording following the methods of Alder and Rose (2000); the justification for using these two species is associated with the temporal features (PRR) of mating and aggressive calls (see Edwards et al. 2007). Frogs were immersed in 3% urethane and a local anesthetic (Lidocaine HCL) was applied topically to the dorsal surface of the skull where a small opening was made to expose the optic tectum. Individuals were allowed to recover overnight from surgery and were subsequently immobilized with either d-tubocurarine chloride (6 μg/g), Mivacron (1 μg/g) or Atracurium (20 μg/g) for recording. Whole-cell patch intracellular recordings from neurons in the torus semicircularis (also referred to as the ICanuran) were made, in vivo, according to methods described in detail by Rose and Fortune (1996) and Edwards et al. (2007).
Patch pipettes were constructed from borosilicate capillary glass (A-M systems #5960; 1 mm outer diameter, 0.58 mm inner diameter) using a Flaming–Brown type puller (Sutter Instruments, model P-97). These pipettes had outside tip diameters of approximately 1–2 μm and had resistances between 15 and 25 MΩ. Electrode tips were back-filled with a solution (pH 7.4) consisting of (values in mM) 100 potassium gluconate, 2 KCl, 1 MgCl2, 5 EGTA, 10 HEPES, 20 KOH, and biocytin at a concentration to bring the final osmolarity to approximately 285 mOsmol. Biocytin was replaced by mannitol in the solution used to fill pipette shanks.
The pipette was advanced into the brain using an ‘inch-worm’ microdrive (Burleigh Co., Fishers, NY) while applying positive pressure. After reaching the recording location, the pipette was advanced in 1.5 μm increments while maintaining positive pressure and passing −0.1 nA square-wave pulses (500 ms) to monitor resistance. Cell contact was indicated by a small increase (10%) in the voltage change. Negative pressure was then applied to the pipette to increase the seal resistance to Giga Ohm (GΩ) levels. Subsequent to seal formation, negative current (approx. −0.5 nA) was applied to rupture the patch and attain an intracellular recording. Seal resistances were typically greater than 2 GΩ with access resistances of 58 MΩ or less. Resting potentials ranged from −48 to −97 mV (median = −70 mV).
Stimulus generation and delivery
Search stimulus carrier frequencies were systematically varied from 300 to 2,200 Hz with modulation frequencies (sinusoidal amplitude modulation, SAM) ranging from 20 to 100 Hz. In cases where this stimulus regimen was ineffective, slower modulation rates and/or lower frequencies were tested. Intracellular recordings were made in an audiometric chamber that was maintained at 18°C. The average PRR of advertisement calls at this temperature is approximately 15 pulses/s (pps) for R. pipiens and 90 pps for H. regilla. Acoustic stimuli were generated using Tucker Davis Technologies (TDT) System II hardware and custom software (Alder and Rose 2000). Stimuli were presented free field in an audiometric room (Alder and Rose 2000). The speaker was situated 0.5 m from the animal and contralateral to the recording site. Neurons were tested with AM and ‘variable duty cycle’ stimuli; in the latter regimen, pulse shape, duration and number were generally held constant and only PRR was varied. Three or more repetitions of a particular stimulus PRR were delivered before proceeding to the next PRR, e.g., AAA, BBB, CCC, etc.
Neurophysiological data acquisition and analyses
Recordings were digitized at 10 kHz (power 1401, Cambridge Electronic Design, Cambridge, UK) and stored as data files using Spike-2 software, also from the same supplier. Analyses were performed using acquired and custom Spike-2 programs. Peak membrane depolarizations were measured from averaged traces derived from a minimum of three repetitions of each stimulus condition, with stimulus amplitude held at approximately 10 dB above threshold (Alder and Rose 2000). In some cases, acquired recordings were median filtered (software feature of Spike-2) to remove spikes before averaging; averages taken from filtered and raw traces were compared to determine whether moderate spike activity influenced measurements of membrane depolarization. Neurons were classified as low-pass, high-pass or band-pass based on whether spike rate levels decreased to at least 50% of maximal at PRRs above, below, or above and below the best PRR. To compare interval tuning based on spike rate versus membrane depolarization measures, the difference between the best PRR and the rate above (low-pass cells) or below (band-pass and high-pass neurons) that value at which the response was half-maximal was calculated from the equation, Diffoctaves = log(2) PRRmax/PRR50%.
Histological and anatomical procedures
Following a recording session, each frog was deeply anesthetized by immersion in 5% urethane and perfused through the heart with a physiological saline/heparin solution followed by a 1:1 mixture of 5% glutaraldehyde and 0.2M phosphate buffer (pH 7.4). The brain was then removed, fixed overnight in the glutaraldehyde solution, and sliced into 100 μm sections on a Vibratome. The sections were incubated overnight in a 10 ml solution of 0.3% Triton X-100 in phosphate-buffered saline (PBS) and the A and B reagents of the Vectastain Elite kit (Vector Labs). The sections were then washed 3 × 10 min in 0.01M PBS and processed using the Vector Peroxidase Substrate kit (SK-4700); the slices were allowed to incubate in a solution of 10 ml 0.01M PBS and six drops each of chromogen and hydrogen peroxide until they began to turn a light gray. The reaction was stopped by washing in 0.01 M PBS (3 × 10 min). Sections were then placed on slides, dried overnight, counterstained with Neutral Red (0.5%), dehydrated, cleared in xylene and cover-slipped. The locations of labeled neurons were then determined using an Olympus BH-2 microscope.
Whole-cell recordings: subthreshold correlates of PRR selectivity
To investigate the mechanistic bases of this diversity in PRR tuning, we made whole-cell recordings, in vivo, from interval-counting neurons in the IC. PRR tuning information was obtained for 20 cells (11 in R. pipiens, 9 in H. regilla).
Neurons without inhibition are selective for slow PRRs
Neurons with inhibition and selective for intermediate or fast PRRs
Neurons at one end of this range in postsynaptic potential (PSP) profiles showed prominent (>5 mV) depolarizations to individual pulses (Fig. 5a, b). These two cells responded best at PRRs of 60 (a) and 15 pps (b), and spikes were occasionally elicited at slow PRRs (5–10 pps) in both cases. For the neuron shown in Fig. 5a, hyperpolarizations were observed for PRRs of 10 pps and greater, but not at 5 pps; the amplitude of hyperpolarizations increased with successive pulses, suggesting a rate-dependent enhancement of inhibition. Cells with these physiological properties were either low-pass (n = 1), band-pass (n = 3) or high-pass (n = 1) for PRR (gray traces, Fig. 3).
Neurons that showed hyperpolarizations and relatively small depolarizations (<5 mV, Fig. 2) to individual pulses showed either band-pass tuning to mid or fast PRRs or were high-pass (solid black lines, Fig. 3c, d, respectively). The examples presented in Fig. 6 represent the range of response profiles that were observed. The cell shown in Fig. 6a exhibited small depolarizations (2–5 mV) and hyperpolarizations in response to pulses repeated at 10 pps, and responded maximally to a PRR of 60 pps; the neuron shown in Fig. 6b was similar, but showed more prominent hyperpolarizations at slow PRRs. For neurons at the other end of this range in response profiles, pulses repeated at slow rates, e.g., 5 pps, elicited primarily hyperpolarizations (Fig. 6c). Responses at faster PRRs appeared to result from rate-dependent enhancement of excitation (Edwards et al. 2007). The prominent after-hyperpolarization at fast PRRs appeared to be due to inhibition, rather than intrinsic membrane properties; these PSPs could be reversed by hyperpolarizing cells and were not observed following depolarization from current injection alone (data not shown).
Spike rate versus PSP measures of interval selectivity
It should be noted that many of the PRR tuning functions of neurons with prominent depolarizations to individual pulses (shown as gray or dashed lines in Fig. 3) are as sharp as those that show small depolarizations (shown in black in Fig. 3). We therefore asked to what extent the interval selectivity based on spike rate measures could be understood from the shapes of PSP-based tuning functions and their relations to spike threshold; ‘thresholding’ processes could generate sharp tuning to PRR, even when the tuning of the underlying depolarizations (PSP tuning) is comparatively broad. For example, spikes might be elicited over a narrow range of PRRs if the neuron’s threshold for spike initiation was slightly less than the peak stimulus-elicited depolarizations. Alternatively, the PSP tuning for PRR could be nearly as sharp as that for spike rate measures. We, therefore, investigated the extent to which interval tuning based on spike rate paralleled that for membrane potential.
For all other neurons, PRR tuning functions derived from PSP amplitude measures qualitatively mirrored those based on spike rate, but were broader; that is, the classification of each cell as low-, band- or high-pass was the same for spike rate and PSP amplitude measures. Neurons of the type shown in Fig. 6 (inhibition with weak excitation at slow PRRs) generally showed the greatest PSP amplitude-based selectivity for PRR. In most cases, the low PRR side of these tuning functions was as steep as the corresponding aspect of the spike rate curves (panels with solid black tuning curves, Fig. 7g, h, j). This property accounts for the remarkable feature that neurons of this type generally do not spike over a considerable range of slow PRRs.
For low-pass interval-counting neurons (Fig. 3a), we calculated the octave difference between the PRR of maximum response (spike rate or membrane depolarization) and the faster PRR at which the response decreased to half-maximal (stars in Fig. 10). As for the band-pass and high-pass neurons, spike rate-based selectivity was generally greater than that of membrane depolarization.
Tuning for AM rate or PRR has been found for midbrain auditory neurons across a wide taxonomic range (Langner 1992), but the underlying mechanisms remain unclear. Our results provide the first comparison of how PRR, and therefore interval duration, is represented by spike-rate versus membrane potential responses across a population of interval-counting auditory neurons. We have shown that interval selectivity is mechanistically diverse. For example, low-pass and band-pass selectivity to slow PRRs was generally associated with temporally summating depolarizations and a relative lack of inhibition. PSP tuning to PRR was much broader than spike-rate based tuning for these neurons. In contrast, neurons that were high-pass or band-pass to intermediate PRRs generally showed prominent inhibition with only slightly broader PSP-based tuning. In a few cases, however, tuning based on spike rate was much sharper than the underlying PSP tuning, e.g., Fig. 7e, f.
The distributions of best PRRs did not differ between these species. This result supports the notion that interval-counting neurons are not functionally specialized to detect a particular call type; the advertisement calls of R. pipiens and H. regilla have PRRs of approximately 15 and 90 pps, respectively, whereas their aggressive calls have the opposite PRR relation (fast for R. pipiens, slow for H. regilla).
Comparisons with other systems
These and previous findings for auditory neurons in the IC of anurans (Edwards et al. 2007) parallel recent whole-cell recordings from electrosensory cells in the posterior exterolateral nucleus of ‘pulse-type’ mormyrid electric fish (Carlson 2009) and auditory neurons in the IC of mice (Geis and Borst 2009) and bats (Gittelman et al. 2009). Electrosensory neurons also show high-pass or band-pass interval selectivity, i.e., respond selectively for short or intermediate intervals, and the mechanisms for achieving this selectivity appear to be similar for both systems (Carlson 2009; Pluta and Kawasaki 2010). For example, enhancement of excitation, also seen in midbrain neurons of ‘wave-type’ electric fish (Fortune and Rose 2000), appears to overcome inhibition at short interpulse intervals, particularly in auditory and electrosensory cells that show strong high-pass selectivity. Further, like band-pass and high-pass interval-counting auditory neurons, many of the electrosensory counterparts showed PSP-based interval selectivity that was nearly as sharp as that derived from spike-rate measurements. Similarly, Gittelman et al. (2009) have shown that the spike rate-based FM selectivity of some IC neurons in bats is much greater than that of the underlying depolarization. Thus, in all of these systems, thresholding properties enhanced spike-rate-based selectivity over the underlying PSP tuning. In cases where threshold for spike initiation was slightly less than the peak depolarizations reached at the optimal PRR, spike-rate tuning was markedly enhanced over the PSP-based tuning. An analogous process has also been reported for enhancing the orientation selectivity (Carandini and Ferster 2000) and direction selectivity (Priebe and Ferster 2005) of visual cortical neurons, suggesting that threshold-based enhancement is a general property of information processing in sensory systems.
Recent whole-cell recordings from the IC of mice (Geis and Borst 2009) suggest that interplay between inhibition and excitation underlies band-suppression selectivity for sinusoidal AM, much like that observed in the anuran IC. Band-suppression (also called band-reject) neurons are a subclass of interval-counting neurons (Edwards and Rose 2003; Leary et al. 2008) that also show long-pass duration selectivity. It remains to be seen whether AM band-suppression neurons in mammals, as in anurans, show interval-counting properties.
Functional implications of mechanistic diversity: temporal summation versus interplay between inhibition and rate-dependent excitation
Previous work in anurans has demonstrated that IC auditory neurons show level-tolerant interval selectivity (reviewed in Rose and Gooler 2007). This general invariance of interval tuning over different sound pressure levels is surprising given our finding that PSP responses were often more broadly interval tuned than spike rate responses. In such cases, e.g., Fig. 7c, thresholding presumably generates a sharply tuned output. Hence, if PSP amplitude increased with sound pressure level, spikes would be elicited at PRRs that were ineffective at lower sound levels and broadening of the spike-rate-based interval tuning would occur. However, as presented earlier, this was not the case; increasing sound level minimally altered peak depolarization. Further work is needed to test the generality of these results. For many of the neurons recorded in this study that showed inhibition, PSP responses were almost as sharply interval-tuned as the spike rate responses. Further, peak depolarizations were, in many cases, well above threshold for spike initiation. As a result, these neurons responded selectively and strongly for a particular range of PRR.
These observations provide insights into the functional consequences and constraints potentially associated with the different types of interval tuning mechanisms reported in the current study. Interval tuning that results from temporal summation of EPSPs, for example, can be quite sharp if depolarizations at the best PRR slightly exceed spike-initiation threshold and do not increase appreciably above that value with increasing sound pressure level. However, neurons of this type may have a restricted dynamic range over which they can code increasing sound level with increasing spike rate, while maintaining interval selectivity.
Female anurans experience calls of increasing amplitude as they approach a calling male, and are capable of discriminating among calls that differ in temporal properties even when call amplitudes are quite high (Gerhardt 2001; Gerhardt and Huber 2002; Wells and Schwartz 2007). Similarly, males are able to detect changes in the amplitude of a neighbor’s calls and identify those changes with the particular call type (i.e., aggressive vs. mating calls) that differ primarily in temporal structure (Brenowitz and Rose 1994; Rose and Brenowitz 1997). Because of dynamic range constraints, many temporal summation-type neurons, with differing thresholds, would be required for these behaviors. In this coding scheme, additional cells would be recruited into the active population as stimulus amplitude increased. In contrast, neurons in which depolarization responses were nearly as sharply interval-tuned as the spike-rate responses could function over a larger dynamic range and code increasing call amplitude in their increasing spike rate. In this latter case, concurrent excitatory and inhibitory conductances might increase in parallel as stimulus amplitude is raised. Because the differential between membrane potential and synaptic reversal potential is greater for excitation versus inhibition, greater depolarization and, therefore, increased spike rate should occur as sound amplitude is increased. The sharp PRR tuning of the membrane potential responses would ensure, however, that cells of this type generate spikes to only a narrow range of PRRs.
Finally, our results raise the question of the functional relations of temporal summation-type neurons to other interval-selective cells. It is unclear whether temporal summation-type neurons, which are rarely encountered in H. regilla and R. pipiens, represent recent mechanistic solutions for processing slow PRRs or vestiges of an ancestral interval counting mechanism. Cells of this type are not unique for processing low temporal frequencies; ‘long-interval’ neurons (Edwards et al. 2008) respond selectively for slow PRRs and are commonly found in both species. In contrast, interval-counting neurons that show interplay of rate-dependent excitation and inhibition (Edwards et al. 2007) are highly selective for intermediate or fast PRRs. Cells of this latter ‘resettable’ type appear to be required for processing the intermediate or fast PRRs that characterize the calls of many derived anuran species (Rose and Brenowitz 2002; Schwartz et al. 2010), and may have evolved to overcome the limitations associated with temporal summation-type neurons. Intracellular studies investigating the mechanisms underlying interval selectivity are currently limited to two species (H. regilla and R. pipiens; Edwards et al. 2007, 2008) representing two highly derived lineages within the anuran clade (Duellman and Trueb 1986). Further comparative studies of temporal processing are needed, therefore, to address the question of whether temporal summation-type neurons represent a pleisiomorphic condition.
Relations to previous work
These new data extend our previous findings indicating that the best PRR for interval-counting neurons is predicted by the PRR at which rate-dependent enhancement of excitation is first seen and EPSP duration (Edwards et al. 2007). Inhibition strength was positively correlated with interval-number threshold, but not best PRR. We now show, however, that the rarely encountered interval-counting neurons that are selective for slow PRRs appear to have little or no inhibition, whereas inhibition is prominent in interval-counting neurons that have best PRRs in the mid to fast range. These results suggest that inhibition functions, in part, to attenuate responses at slow PRRs.
Overall, our findings are generally consistent with a recent model (Buonomano 2000) in which interval selectivity results from interplay between excitation, inhibition and short-term synaptic plasticity. It remains to be determined whether rate-dependent depression of inhibition, a component of Buonomano’s model, plays a role in the interval tuning of IC neurons. Many interval-counting neurons show band-pass or low-pass interval selectivity, raising the question of what underlies decreases in depolarizations and spike rate at PRRs above the best rate. Our earlier work (Edwards et al. 2007) provided some evidence that intervals shorter than the best rate are less effective because afferents to the IC neurons fail to sustain responses at fast PRRs; that is, the fast PRR side of the tuning function appears to be determined by an input property versus a result of processing within the IC. For low-pass cells, synaptic depression of excitation may limit responses to fast PRRs. Alternatively, inhibition, which appears to play little role at slow PRRs, may facilitate and limit depolarization at fast PRRs. Further work is needed to address these issues.
Finally, it is becoming increasingly clear that excitation and inhibition can be integrated in diverse ways (Grothe 1994; Buonomano 2000; Large and Crawford 2002) to generate selectivity for a wide range of temporal parameters of sounds, including interval length (Edwards et al. 2007, 2008), duration (Casseday et al. 1994; Leary et al. 2008; Aubie et al. 2009), sinusoidal frequency modulation (Casseday et al. 1997) and direction of frequency modulation (Zhang et al. 2003). These and future studies should provide considerable insights into how temporal patterns of activity, representing temporal information in acoustic signals, are decoded in the brain.
All procedures were approved by the University of Utah Institutional Animal Care and Use Committee.