J Neurophysiol 97: 1413-1427, 2007.
First published December 6, 2006; doi:10.1152/jn.00384.2006
0022-3077/07 $8.00
Antidromic Activation Reveals Tonotopically Organized Projections From Primary Auditory Cortex to the Central Nucleus of the Inferior Colliculus in Guinea Pig
Hubert H. Lim and
David J. Anderson
Kresge Hearing Research Institute, Department of Biomedical Engineering, and Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, Michigan
Submitted 11 April 2006;
accepted in final form 5 December 2006
 |
ABSTRACT
|
|---|
The inferior colliculus (IC) is highly modulated by descending projections from higher auditory and nonauditory centers. Traditionally, corticofugal fibers were believed to project mainly to the extralemniscal IC regions. However, there is some anatomical evidence suggesting that a substantial number of fibers from the primary auditory cortex (A1) project into the IC central nucleus (ICC) and appear to be tonotopically organized. In this study, we used antidromic stimulation combined with other electrophysiological techniques to further investigate the spatial organization of descending fibers from A1 to the ICC in ketamine-anesthetized guinea pigs. Based on our findings, corticofugal fibers originate predominantly from layer V of A1, are amply scattered throughout the ICC and only project to ICC neurons with a similar best frequency (BF). This strict tonotopic pattern suggests that these corticofugal projections are involved with modulating spectral features of sound. Along the isofrequency dimension of the ICC, there appears to be some differences in projection patterns that depend on BF region and possibly isofrequency location within A1 and may be indicative of different descending coding strategies. Furthermore, the success of the antidromic stimulation method in our study demonstrates that it can be used to investigate some of the functional properties associated with corticofugal projections to the ICC as well as to other regions (e.g., medial geniculate body, cochlear nucleus). Such a method can address some of the limitations with current anatomical techniques for studying the auditory corticofugal system.
|
INTRODUCTION
|
|---|
The inferior colliculus (IC) is an obligatory center of convergence for most ascending auditory information from lower brain stem nuclei (Aitkin and Phillips 1984
; Casseday et al. 2002; Ehret 1997b; Malmierca et al. 2002). The IC also receives a large number of descending projections from higher auditory and nonauditory centers (Coomes et al. 2005; Winer 2005b; Winer et al. 1998, 2002), indicating that the IC is highly modulated by feedback mechanisms that may provide selective hearing of important sound features. Traditionally, it was thought that these descending projections terminated mainly in the extralemniscal regions of the IC (Faye-Lund 1985; Herbert et al. 1991; Huffman and Henson 1990; Winer et al. 1998). However, there is some anatomical evidence suggesting that a substantial number of direct projections from the primary auditory cortex (A1) to the ICC exist (Saldana et al. 1996) and appear to be glutamatergic (Feliciano and Potashner 1995).
An interesting question is how these A1 to ICC projections are spatially organized. Saldana et al. (1996) showed in rats that anterograde tracers injected into certain regions within the primary auditory cortex (Te1) stained a large number of terminal fibers in certain regions within the ipsilateral ICC in a topographic manner. They proposed that this topographic organization was tonotopic based on the frequency organization of Te1 and ICC in rats that appeared to align with their injection sites and corresponding stained regions, respectively. Andersen et al. (1980b) injected anterograde tracers into cat A1 and observed stained terminal fibers within the ICC in a topographic pattern. They inferred that this topographic projection pattern was tonotopic based on electrophysiological mappings of the injected regions. Bajo and Moore (2005) demonstrated in gerbil that retrograde tracers injected into the ICC resulted in labeled cells within the ipsilateral A1 that appeared to be organized in a tonotopic pattern. They confirmed that the injected cells were contained within the ICC based on retrograde labeling of cells within brain stem nuclei that do not project to the other IC regions. More recently, Bajo et al. (2006) injected anterograde and retrograde tracers into the auditory cortex and IC, respectively, of the ferret and observed projections from the primary auditory fields to the ICC. The projections appeared to be tonotopically organized but were restricted to the dorsomedial region of the central nucleus, suggesting possible differences among species.
The preceding findings provide some evidence for a general tonotopic organization of corticofugal projections from A1 to the ICC. However, from these studies, it is not yet clear as to how individual A1 neurons spatially project to the ICC, both along the frequency and isofrequency dimensions. This is due to limitations in current methods used for studying auditory corticofugal projections. For anatomical studies, anterograde tracers can be injected into a small region within A1, and the corresponding stained terminal fibers within the ICC can be assessed (and vice versa with retrograde tracers). However, the extent of staining is not spatially localized enough to isolate the effects of individual or small groups of neurons. For smaller injections, there is always the possibility that cells are not stained, something that can lead to the erroneous inference that certain projections do not exist. Furthermore, it is difficult to directly determine the functional properties associated with the stained cells. For electrophysiological studies, A1 neurons can be stimulated and the corresponding neural activity can be recorded from individual ICC neurons. However, it is difficult to confirm if those responses are caused purely by monosynaptic rather than polysynaptic projections from A1. Thus a more refined method for studying individual or small groups of neural projections from A1 to ICC is needed to better understand how cortical neurons spatially activate and modulate the ICC.
In a previous study investigating the potential features of an auditory midbrain implant (AMI) (Lim and Anderson 2006), we electrically stimulated the ICC and recorded the corresponding neural activity in A1 of guinea pigs. Considering that we could record antidromically activated spikes in A1, it suggests that direct projections from A1 to the ICC exist in the guinea pig, which have been observed in tracer studies by Coomes et al. (2005). It also suggests that it may be possible to study the spatial organization of monosynaptic corticofugal projections to the ICC using antidromic stimulation (Asanuma et al. 1968; Bereshpolova et al. 2006; Palmer and Rosenquist 1974; Schofield et al. 1987; Swadlow 1988). Therefore the goal of this study was to test the hypothesis that A1 projections are amply distributed and tonotopically organized throughout the ICC, which thus far has been inferred from some anatomical results but not yet confirmed with electrophysiological methods. To achieve our goal, we electrically stimulated different regions along the frequency and isofrequency dimensions of the ICC and recorded the antidromically activated spikes across different layers and regions within A1 using multi-site probes in ketamine-anesthetized guinea pigs. The ability to acoustically stimulate and record neural activity across multiple sites within the ICC and A1 enabled us to identify the different cortical layers (using current source density analysis) and best frequency (BF) regions. Histological techniques were used to locate the placement of our sites along the isofrequency dimension of the ICC. We then characterized the spatial projection pattern of A1 fibers throughout the ICC.
|
METHODS
|
|---|
Basic surgical procedures and methods for stimulation and recording were similar to those presented in previous studies (Lim and Anderson 2003, 2006). Experiments were performed on young pigmented guinea pigs (Elm Hill Breeding Labs, Chelmsford, MA) in accordance with policies of the University of Michigan Committee on the Care and Use of Laboratory Animals. These experiments were performed in two parts to investigate how A1 fibers are organized along the frequency and isofrequency dimensions of the ICC: experiment 1 (8 animals, 402–448 g) in which we electrically stimulated different regions within an ICC lamina and recorded from antidromically activated cells in different layers of the ipsilateral A1 (similar BF as the ICC lamina); and in experiment 2 (5 animals, 320–588 g), in which we electrically stimulated different BF regions within the ICC and recorded from antidromically activated cells in different BF regions of the ipsilateral A1.
Anesthesia and surgery
The animals were initially anesthetized with an intramuscular injection of ketamine hydrochloride (40 mg/kg) and xylazine (5 mg/kg) with additional supplements to maintain an areflexive state. After placing each animal into a stereotaxic device (David Kopf Instruments, Tujunga, CA), we stereotaxically inserted silicon-substrate multi-site Michigan probes (Center for Neural Communication Technology, University of Michigan, Ann Arbor, MI) (Anderson et al. 1989; Drake et al. 1988; Najafi et al. 1985) into the right ICC and A1 (Fig. 1). The ICC probe (Fig. 1A) had two shanks separated by 500 µm (all distances are center to center) where each shank had eight iridium sites linearly spaced at intervals of 200 µm. Each shank was 15 µm thick, 8 mm long, and tapered in width from 400 µm to a few micrometers at the tip. Each site had an area of 
400 µm2 and was activated to an impedance of <100 k
(at 1 kHz). Activating the sites from iridium to iridium oxide (via cyclic voltammetry) enabled us to use them for both recording and stimulation (up to a conservative limit of 60 µA) (Anderson et al. 1989; Beebe and Rose 1988; Weiland and Anderson 2000). The ICC probe was inserted 45° off the sagittal plane through the occipital cortex (not shown) and into the IC where each shank was aligned along the tonotopic axis of the ICC (Malmierca et al. 1995; Merzenich and Reid 1974; Schreiner and Langner 1997; Snyder et al. 2004). For A1, we used two different probes depending on the experimental group (experiment 1 or 2). In Fig. 1B (experiment 1), the A1 probe consisted of a single shank with 16 sites linearly spaced at intervals of 150 µm. The shank was 15 µm thick, 5 mm long, and tapered in width from 200 µm to a few micrometers at the tip. The probe was inserted approximately perpendicular to the cortical surface along a specific BF column of A1 (Abeles and Goldstein 1970; Merzenich et al. 1975; Redies et al. 1989; Wallace et al. 2000). This enabled us to simultaneously record from different cortical layers. To record along the tonotopic gradient of A1, we used the probe shown in Fig. 1C (experiment 2). This probe had eight shanks each separated by 200 µm and each shank had two electrode sites separated by 50 µm. Each shank was 15 µm thick, 1.6 mm long, and tapered in width from 50 µm to a few micrometers at the tip. The probe was inserted approximately perpendicular to the cortical surface where each shank was positioned into a different frequency region in A1 (Redies et al. 1989; Wallace et al. 2000). The sites were inserted to a depth approximately corresponding to the output layer (V) of A1 (see Placement of probes). For experiment 2, the ICC and A1 probes usually recorded from neurons sensitive to frequencies across a similar three- to four-octave range between 2 and 30 kHz. We were unable to access the entire frequency range in guinea pigs (0.05–45 kHz, roughly spanning 2.5 mm in ICC and 3.5 mm in A1) due to the limited distance spanned by our probe sites (1.4 mm). Thus we spanned a frequency range that allowed us to simultaneously sample reasonably low and high frequencies. All sites for both A1 probes had an area of 400 µm2 and unactivated impedances of 1–2 M. After placement of the probes, the brain was covered with agarose to reduce swelling, pulsations, and drying during the recording sessions. For experiment 1, two to five ICC probe placements and only one A1 probe placement were made. For experiment 2, only one ICC probe placement and one A1 probe placement were made.
View larger version (27K):
[in this window]
[in a new window]
|
FIG. 1. Drawings of the multi-site probes positioned along the tonotopic axis of inferior colliculus central nucleus (ICC, A), a best frequency (BF) column of primary auditory cortex (A1, B), and the tonotopic gradient of Al (C). Anatomy in A was derived from an image presented in Malmierca et al. (1995) while that in B and C was from Wallace et al. (2000) for the guinea pig (simplified and not drawn to scale). Electrode sites are represented by black dots on each shank. *, blood vessels. DC, dorsocaudal cortex; F, frequency; ICD, IC dorsal cortex; ICX, IC external cortex; SC, superior colliculus.
|
|
Stimulation and recording setup
Experiments were conducted in a sound-attenuating chamber and controlled by a computer interfaced with TDT hardware (Tucker-Davis Technology, Alachua, FL) using custom software written in MATLAB script (Mathworks, Natick, MA). To aid in the positioning of the probes, we presented varying levels of pure tones and broadband noise (50 ms in duration) to elicit acoustic driven activity in the contralateral (right) ICC and A1. The acoustic stimuli were presented via a speaker coupled to the left ear through a hollow ear bar. This speaker-ear bar system was calibrated using a 0.25-in Brüel and Kjaer condenser microphone (Naerum, Denmark) where the tip of the ear bar was inserted into a short plastic tube with the microphone inserted into the other end. The tube represented the ear canal. All neural signals were passed through analog DC-blocking and anti-aliasing filters from 1.6 Hz to 7.5 kHz. For spike analysis, these signals were then digitally filtered between 0.3 and 3 kHz. After placement of the probes, we connected the ICC probe to a custom-made optically isolated current stimulator for electrical stimulation. Each ICC site was stimulated as an active electrode where the return was through a wire positioned in a neck muscle (monopolar configuration). Electrical stimuli consisted of single biphasic, charge-balanced pulses (200 µs/phase, cathodic-leading) from 1 to 56.2 µA in dB steps. Forty trials of each stimulus at 2/s were presented in a randomized sequence across all sites and levels to reduce adaptive effects. Neural data were recorded on all A1 probe sites in response to stimulation of each ICC site. The recording ground wire was positioned under the skin 2 cm rostral to Bregma.
Placement of probes
Poststimulus time histograms (PSTHs) and frequency response maps (FRMs) were plotted to confirm that our sites were positioned within the desired BF regions within the ICC and A1. Methods for probe placement, and obtaining FRMs and BFs are presented in detail in our previous study (Lim and Anderson 2006). For experiment 1 (using Fig. 1B probe), we needed to identify the different cortical layers, especially the main output layer (V) of A1. This was possible by using current source density (CSD) analysis (Mitzdorf 1985; Muller-Preuss and Mitzdorf 1984). Evoked potentials were first recorded in response to 60 dB SPL broadband noise (40 trials). CSD profiles (Fig. 2B) were then calculated by taking the discrete approximation of the second spatial derivative of the averaged evoked potentials (Fig. 2A) recorded on our A1 sites using the finite difference formula
where 
is the evoked potential, z is the depth location of each site along the A1 probe, 
z is the differentiation step size, and 
z is the component of conductivity in the z direction. z was equal to our A1 site spacing of 150 µm and z was set to 1 because we were not concerned with absolute CSD values. CSD profiles for the two edge sites could not be calculated using the preceding equation because it requires evoked potential profiles from two neighboring sites. This one-dimensional CSD approximation has shown to provide a consistent representation for the current sinks and sources associated with columnar synaptic activity in the auditory cortex (Muller-Preuss and Mitzdorf 1984). The main input layer of A1, which approximately corresponds to layer III/IV (Huang and Winer 2000; Smith and Populin 2001), was taken as the site with the shortest latency current sink (Fig. 2B) and PSTH response (Fig. 2C). We selected this site as the reference depth because it could be consistently identified across animals. All other sites were labeled in reference to this site where a more positive depth corresponded to a deeper site. Layer V could then be identified with respect to the main input layer site. Based on depth measurements made in guinea pig A1, the main input layers of III/IV tend to span 300–400 µm and layer V tends to span 500–600 µm (M. N. Wallace, personal communication). Although the thicknesses of the layers may slightly vary across BF regions and animals, we still observed almost all the lowest antidromic threshold spikes within a localized range of depths (see Experiment 1: antidromic properties) consistent with anatomical findings (Coomes et al. 2005; Winer and Prieto 2001) and demonstrating the reliability of using CSD analysis for identifying different cortical layers. For experiment 2, we used the eight-shank probe shown in Fig. 1C to record from different frequency regions of A1. We did not have a high-density probe available that could simultaneously record across multiple layers and BF regions. Our probe only had two sites per shank, so we could not perform CSD analysis to identify layer V. Instead, we inserted the probe into A1 until we recorded antidromically activated spikes in response to ICC stimulation. We positioned the probe to maximize the number of sites that recorded the lowest threshold antidromic spikes, which corresponded to layer V based on our results from experiment 1. Some sites that were not optimally positioned in layer V exhibited higher than normal antidromic spike thresholds that could artificially reduce the true spread of antidromic activation across the tonotopic gradient of A1 to what is recorded by our A1 probe. However, such effects do not alter our overall findings and are further discussed in our results sections for experiment 2.
View larger version (38K):
[in this window]
[in a new window]
|
FIG. 2. Averaged evoked potential profiles (A), current source density (CSD) profiles (B), and poststimulus time histograms (PSTHs; C) recorded on sites along the A1 probe (aligned along a cortical BF column as in Fig. 1B) in response to broadband noise (50-ms duration, 60 dB SPL, 40 trials). CSD profiles were calculated from the evoked potential profiles (for more details, see Placement of probes). Scale bar: 2 mV for evoked potentials and 40 mV/mm2 for CSDs. Grayscale for the PSTHs corresponds to total spikes where <2 indicates that all values less than or equal to 2 were set to white and >25 indicates that all values greater than or equal to 25 were set to black to improve visualization of the data; 0 µm corresponds to the main input layer site of A1 in which more positive values indicate deeper locations.
|
|
ICC probe histology
For experiment 1, we needed to identify the location of our ICC shank placements. During the experiment, we dipped the ICC probe (with 2 shanks) into a red stain [10% 1,1-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (di-I); Molecular Probes, Eugene, OR] before each insertion into the brain. An effective protocol consisted of repeatedly (10 times) dipping the probe into the stain for 10 s and letting it dry in air for 10 s (DiCarlo et al. 1996). At the end of each experiment, the animal was decapitated, and the head was immersed into 10% formaldehyde for 5 days. The brain was removed and placed in 20% sucrose solution until the tissue sank. Frozen sections 40 µm thick were cut in the sagittal plane. Figure 3 shows an example of a sagittal section with six red dots (labeled with asterisks). Because the ICC probe was inserted 45° off the sagittal plane, each shank tract appeared as a dot in the sagittal section. Once the dots were located, we were able to measure their distances from the caudal (XCR) and dorsal (XDV) edge of the IC. Because the ICC laminae are roughly aligned 45° from the horizontal plane as shown in Fig. 1A (Malmierca et al. 1995; Merzenich and Reid 1974; Schreiner and Langner 1997; Snyder et al. 2004), we multiplied XDV by cos(45°) to obtain the approximate dorsomedial-to-ventrolateral distance (XDmVl) along the laminae. In reality, each ICC lamina deviates from this 45° orientation to varying extents along the caudorostral and dorsoventral dimensions and is positioned in different locations relative to the dorsal and caudal edges of the IC. Thus we cannot claim that a certain location in one BF lamina is in the same relative location in another BF lamina even though they may have the same (XDmVl, XCR) coordinates. Also the dimensions of a lamina can vary for different BF regions and sized animals. However, this method allowed us to systematically identify the locations of our shank placements relative to the caudal and dorsal edges of the IC. Because we also acoustically identified the stimulation site within the target BF lamina for each shank, the measured coordinates indirectly correspond to the locations of our stimulation sites along that lamina. We measured the total distance, TCR, from the caudal edge to the IC-SC division so that we could gauge the relative location of our sites along the caudorostral dimension across animals (for examples, see Fig. 7). We were unable to identify any consistent landmarks to gauge our relative location along the dorsoventral (i.e., dorsomedial-to-ventrolateral) dimension. Because XCR, XDV, and TCR change along the mediolateral dimension (for different sagittal sections), we only measured these parameters along one sagittal section corresponding to a normalized distance of 0.31 from the lateral edge of the IC to the midline of the brain to be consistent across animals. Using acoustic stimulation, we identified the sites that were located within the ICC. For each animal, only sites located within one target BF lamina of the ICC were used for stimulation and data analysis.
View larger version (152K):
[in this window]
[in a new window]
|
FIG. 3. Sagittal section of the IC corresponding to a normalized distance of 0.31 from the lateral edge of the IC to the midline of the brain. Red dots labeled with asterisks correspond to the location of the shanks within the ICC for 3 probe placements (ICC probe had 2 shanks). One site from each shank placement was positioned within the target best frequency lamina and used for antidromic stimulation. Thus those red dots indirectly correspond to the relative location of the sites across an ICC lamina (for more details, see ICC probe histology). Scale bar equals 0.5 mm.
|
|
View larger version (37K):
[in this window]
[in a new window]
|
FIG. 4. PSTHs (0.2-ms bins) recorded on 14 sites of the A1 probe (in Fig. 1B) aligned along a cortical column (BF = 12 kHz). Each column corresponds to a different stimulus level, whereas each row corresponds to a different cortical depth relative to the main input layer site in which more positive values indicate deeper locations. Both orthodromic and antidromic activity are displayed;  , antidromic PSTH peaks.
|
|
View larger version (14K):
[in this window]
[in a new window]
|
FIG. 5. A: example PSTH (0.5-ms bins) recorded in A1 that consisted of a short-latency, precisely stimulus-locked antidromic spike peak followed by more temporally diffuse orthodromic activity in response to single pulse ICC stimulation. B: same data on an expanded time scale. C: PSTH for same recording site in A but in response to a 200-ms pulse train (120 pulse/s) demonstrating the ability of the antidromic spike peak to follow high pulse rates, whereas the orthodromic activity rapidly diminishes by the 3rd pulse.
|
|
View larger version (14K):
[in this window]
[in a new window]
|
FIG. 6. Experiment 1: distributions for the depths (A), thresholds (B), and latencies (C) corresponding to the lowest threshold antidromic spike recorded in A1 for each stimulated ICC site (n = 43, experiment 1).
|
|
View larger version (18K):
[in this window]
[in a new window]
|
FIG. 7. Experiment 1: the locations of the sites along a target ICC lamina (BF labeled in bottom left corner) for each animal are displayed as shaded circles where the site is located at the center. The circle radius corresponds to the maximum distance a corticofugal fiber can be located from the site and still be activated with a current level equal to the antidromic threshold for that site. The maximum distances were calculated from the fitted equation presented in Fig. 14 (for more details on this equation, see Experiment 2: ICC spread of activation). XDmVl, dorsomedial-to-ventrolateral location; XCR, caudorostral location; TCR, total distance between caudal edge and the IC-SC division along XCR axis (for more details on how these coordinates were measured, see ICC probe histology).
|
|
Data analysis
SPIKE ACTIVITY.
Multi-unit activity was recorded in A1 in response to electrical stimulation of the ICC. This activity consisted of both antidromic and orthodromic spikes. Any value exceeding 3.5 times the SD of the background (without spikes) neural signal was detected as a spike and an algorithm was used to determine the time of occurrence of the largest peak (positive or negative) of that spike. We then displayed the spike activity as PSTHs across 40 trials. For electrical stimulation in the ICC, we observed stimulus artifacts that usually lasted less than a millisecond. Only at high levels close to 56.2 µA did we observe artifacts that lasted between 1 to 2 ms. Because we did not observe stimulus-driven spike activity with latencies <2 ms, we just zeroed out all the PSTHs from 0 to 2 ms from stimulus onset for analysis.
ANTIDROMIC IDENTIFICATION.
In Fig. 4, PSTHs recorded across 14 sites along the single shank A1 probe (Fig. 1B, experiment I) for varying stimulus levels are presented. Each column corresponds to a different stimulus level, whereas each row corresponds to a different depth along a BF column of A1. As shown in Fig. 2, we were able to identify the main input layer site in A1 using CSD analysis. In Fig. 4, each A1 site was labeled with a depth relative to the main input layer site. A depth of –300–0 µm approximately corresponds to layers III/IV and a depth of 150–600 µm approximately corresponds to layer V. As the stimulus level was increased, orthodromic activity increased on all sites especially for depths between –300 and 0 µm. In deeper regions (300–600 µm), antidromic activity (arrows) was also present. The short latency and precisely stimulus-locked nature of these antidromic PSTH peaks is characteristic of antidromic activity. To confirm that these responses were antidromic, we performed a pulse rate test. Figure 5A shows an example of an orthodromic A1 response preceded by an antidromic PSTH peak. Figure 5B shows the same data plotted on an expanded time scale. Studies have shown that orthodromic activity from the ICC to A1 (via the thalamus) cannot follow pulse rates >100 pulses/s, whereas antidromic activity can follow pulse rates significantly higher than 100 pulses/s (Mitani and Shimokouchi 1985; Rose and Metherate 2001). Figure 5C presents an example where the orthodromic onset response was rapidly diminished by the third pulse, whereas the antidromic spike activity robustly followed a pulse train of 120 pulses/s. Although the collision test is generally more reliable in identifying antidromically activated spikes (Fuller and Schlag 1976; Swadlow 1998), we could more efficiently identify antidromic activity for multi-unit recordings by stimulating with pulse rates >100 pulses/s (Rose and Metherate 2001) and isolating the remaining units. Furthermore, the precise latencies and sudden appearance of antidromic spikes with stimulation level (for example, see Fig. 4) made it readily possible to differentiate antidromic from orthodromic activity.
ANTIDROMIC THRESHOLDS.
Antidromic threshold corresponded to the stimulus level that caused a sudden increase in antidromic activity from no activity. For example, in Fig. 4, the threshold for antidromic activity was 8.9 µA, and it appeared at a depth of 450 µm. As the stimulus level was increased, antidromic activity began to appear on other sites. Usually we observed these antidromic spikes in layer V of A1 (see Experiment 1: antidromic properties). For experiment 2, we were unable to perform CSD analysis because there were only two sites along each shank of the A1 probe (Fig. 1C). Instead, we inserted the probe into A1 until we observed antidromic activity on our recording sites generally with the lowest thresholds. For data analysis, we only used the lower antidromic threshold site along each shank, resulting in a total of eight sites for each A1 probe placement.
ANTIDROMIC ACTIVATION SPREAD.
In experiment 2, we investigated the extent of activation spread across the tonotopic gradient of A1 to determine if corticofugal fibers from different BF regions project down to a specific BF region in the ICC or if a strict BF-aligned projection pattern exists. We also investigated the extent of activation spread within the ICC. In particular, we wanted to determine the amount of current required to activate corticofugal fibers located a certain distance away from the stimulation site. This was important not only to gauge how close our activated fibers were from our stimulated sites, but also to provide insight into some of the biophysical and anatomical properties of these fibers. Measuring the activation spread in A1 was directly possible from our recording setup since we simultaneously recorded from eight sites positioned across the tonotopic gradient of A1 in response to stimulation of a certain BF ICC region. Measuring the activation spread within the ICC was obtained indirectly from the A1 spread data. As we will demonstrate in our experiment 2 results, corticofugal fibers from a certain BF region only project to a similar BF region in the ICC. This one-to-one tonotopic alignment allowed us to directly scale our A1 spread data to obtain the activation spread within the ICC. The following equation describes this scaling process
 | (1) |
The BF gradients with respect to distance along the tonotopic dimension have shown to be approximately linear on an octave scale for both the ICC (Snyder et al. 2004) and A1 (Arenberg et al. 2000; Bakin et al. 1996) in guinea pigs. Thus it is possible to scale the spread of antidromic activation along the tonotopic gradient of A1 (in µm) by the A1 BF gradient to get the total BF range of activated corticofugal fibers (in octaves). By multiplying this range by the inverse of the ICC BF gradient (and assuming a one-to-one tonotopic projection from A1 to ICC), we can obtain the activation spread of corticofugal fibers along the tonotopic gradient of the ICC. The 0.5 scale factor is applied to obtain the distance of activated fibers from the stimulation site assuming the site was located at the midpoint of the total activated region. This ICC activated distance provides a measure for how far fibers along the tonotopic gradient of the ICC can be activated from a stimulated site for a certain current level. Assuming an isotropic, homogeneous volume conductor model with passive current spread through the tissue medium, we can fit the ICC activated distance (D) versus current level (I) data to the theoretical equation
 | (2) |
where Io is the current threshold for activation of a cell/fiber (at D = 0) and k is a constant that accounts for the resistivity of the medium (Asanuma and Sakata 1967; Bagshaw and Evans 1976). The ICC does have a well-defined layered structure that is anatomically similar along the tonotopic axis (Oliver 2005). Thus estimating the ICC as infinite dimensional homogeneous layers is a reasonable assumption.
|
RESULTS
|
|---|
All analyses are based on data from 13 animals. The results are separated into two groups corresponding to experiments 1 and 2. For experiment 1 (8 animals), we electrically stimulated three to eight different locations within an ICC lamina and recorded the antidromically activated spikes across different layers along a BF column of A1 (similar BF as ICC lamina). Only one BF region was investigated for each animal. The BFs corresponding to all eight animals were 9, 10, 12, 14, 20, 20, 22, and 23 kHz. There were a total of 44 stimulated ICC sites. Stimulation of one site did not elicit any antidromic activity in A1 even at our maximum level of 56.2 µA. For experiment 2 (5 animals), we electrically stimulated different frequency regions within the ICC and recorded the antidromically activated spikes across different frequency regions within A1 (approximately in layer V). For each animal, there was only one ICC probe placement (with 
2 valid shanks) and one A1 probe placement. Across animals we had a total of eight different ICC shank placements aligned along the tonotopic axis of the ICC in which 51 sites (of 64 because 8 sites aligned along a shank) could be used for stimulation. For each stimulated ICC site, we analyzed the spike activity recorded on eight sites aligned along the tonotopic gradient of A1. Stimulation of all 51 ICC sites elicited antidromic activity in A1.
Experiment 1: antidromic properties
Electrical stimulation of different regions within an ICC lamina antidromically activated fibers projecting from A1. The antidromic spikes were recorded from different cortical layers, usually in layer V but also within layers III/IV and VI. In attempt to identify the main output region of A1, we determined the depth corresponding to the lowest antidromic threshold for each stimulated ICC site. Figure 6A presents a histogram of these different depth values. Almost all depths were between 450 and 600 µm, which corresponds to the middle and deeper parts of layer V approximated from the histological depth measurements made in guinea pig A1 by M. N. Wallace (personal communication), and is consistent with anatomical results by Coomes et al. 2005. This is also consistent with findings in cat where Winer and Prieto (2001) showed that corticocollicular projections (A1 to the IC) originated from the bottom two (of 3) subdivisions of layer V. However, some of the lowest thresholds did occur in the upper layers (III/IV). It is possible that we may have recorded from dendritic processes extending from output neurons located in layer V that were closer to our recording sites than the actual layer V neurons. This is consistent with findings presented by Schofield et al. (1987), who observed, for projections from the visual cortex to the superior colliculus, that antidromically activated dendritic processes extending from corticotectal cells located in layer V could elicit spike activity in layers III/IV of the cortex. The one depth value at 1,350 µm (layer VI) may be a result of an antidromically activated fiber originating from layer V but that passes in close proximity to a deeper recording site. There is some evidence in guinea pigs that a small percentage (<10%) of the corticofugal neurons that project to the IC from the ipsilateral auditory cortex originate from layer VI (B. R. Schofield, personal communication), which could also explain why we occasionally observed antidromic spikes in deeper layers.
The antidromic thresholds corresponding to the depths in Fig. 6A are plotted in B. Although there were a few high thresholds and even one case where no antidromic activity was elicited, the majority (70%) of thresholds were <10 µA (almost all were <30 µA). Based on current spread data for mammalian CNS stimulation (Ranck 1975), 10 µA (200-µs monopolar cathodal pulse) would not be able to activate fibers more than 100 µm away. A current level of 30 µA would not be able to activate fibers more than 200 µm away. Considering that corticofugal fibers projecting from A1 to the ICC have shown to be much thinner (<1 µm in diameter) (Winer 2005b) than the fibers used to obtain those estimates suggests that ICC stimulation should exhibit even less spread of antidromic activation. Later, we present data supporting this inference (see Experiment 2: ICC spread of activation). Therefore to have antidromically activated corticofugal fibers with low stimulus levels for almost all our ICC locations throughout different BF regions (see next sections), fibers directly projecting from A1 must be amply scattered throughout the ICC. Otherwise, higher thresholds would have been observed.
Most of the antidromic latencies we observed (Fig. 6C) were 2–3 ms with some as large as 10 ms. These values are consistent with results presented by Mitani et al. (Mitani and Shimokouchi 1985) where they found latencies between 1.1 and 6.7 ms in cat. Schofield et al. (1987) also found latencies between 2 and 10 ms for antidromically activated visual cortical fibers that projected down to the superior colliculus in rats. Considering that these antidromically activated fibers are only traveling between two regions rather than several synapses raises the question as to why these latencies are generally long. There is anatomical evidence that these corticofugal fibers are very thin (<1 µm in diameter) (Winer 2005b). Thus the antidromically activated spikes from the ICC to A1 would exhibit slow conduction velocities thus longer latencies due to the small diameters of these corticofugal fibers.
Experiment 1: spatial distribution within the ICC
To assess how A1 neurons project to different laminae within the ICC, we plotted the location of all our stimulation sites within the target ICC lamina for each animal (Fig. 7). For each plot, we used the total distance (TCR) from the caudal edge of the IC to the IC-SC division as the maximum value along the caudorostral (XCR) axis. Along the dorsomedial-to-ventrolateral (XDmVl) axis, we did not have any consistent landmarks. So we used a maximum distance of 2.5 mm for all the plots. Each site location is defined by a shaded circle (except for 1 location labeled with an asterisk where no antidromic activity was elicited using our maximum level of 56.2 µA) in which its border corresponds to the maximum distance from the site (center) that a corticofugal fiber can be activated with a current level equal to the antidromic threshold for that site. The diameters of the circles were calculated using our ICC spread Eq. 2 obtained from experiment 2, which is presented in Fig. 14 (see Experiment 2: ICC spread of activation). That equation is consistent with the current spread data presented in Fig. 1 of Ranck (1975) and represents the upper spread limit for activating A1 fibers within the ICC. From our data, we can only claim that each shaded circle corresponds to either a single neuron or a small number of neurons projecting from a BF column of A1 to an ICC lamina. We used principal component analysis and spike waveform features (amplitude and duration) to extract single-unit data from our multi-unit A1 recordings. However, many antidromic spikes exhibited similar waveform shapes making it difficult to determine if our single-unit data corresponded to one neuron or multiple neurons. Furthermore, slight movements of our sites along a BF column as we stimulated different ICC locations could result in different waveform properties for the same A1 neuron.
View larger version (21K):
[in this window]
[in a new window]
|
FIG. 8. Experiment 1: histogram of the antidromic thresholds for all the sites located within laminae with BFs between 9 and 14 kHz (A) and 20 and 23 kHz (B) and corresponding to the circles presented in Fig. 7.
|
|
View larger version (25K):
[in this window]
[in a new window]
|
FIG. 10. Experiment 2: BF mapping plot demonstrating the tonotopic nature of A1 to ICC projections. The BF of the A1 site with the lowest antidromic threshold for a stimulated ICC site is plotted against the BF of that ICC site (n = 51). Diagonal line depicts perfect mapping which is not always possible due to the set geometry of the electrode sites thus inherent BF misalignment. Thus dots are used to indicate the closest alignment possible.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
FIG. 11. Experiment 2: plots showing how antidromic activity appears on different A1 sites as the ICC stimulation level is increased above threshold for a typical example (A), the greatest spread case across all our stimulated sites (B), and a discontinuous type (C). Distance between sites is 200 µm.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
FIG. 12. Experiment 2: spread of antidromic activation (mean ± SD) along the tonotopic gradient of A1 as a function of ICC stimulation level above threshold (A) and absolute ICC current levels (B). These values exclude data from the discontinuous types (see Fig. 10C). The number of sites used to calculate the mean and SD for each level in A is 38, 38, 38, 38, 35, 33, 33, 26, 22, 18, 14, 7, and 4, whereas for B, it is 4, 7, 14, 18, 22, 26, 33, 33, 35, 38, 38, 38, and 38.
|
|
View larger version (34K):
[in this window]
[in a new window]
|
FIG. 13. Experiment 2: curves showing how BF changes as a function of recording site location along the tonotopic axis of A1 (A and B) and the ICC (C and D) in the same animals used to obtain the A1 antidromic spread data. Only BFs  5 kHz were included in B and D to calculate constant slope values from the fitted lines. n = 5 for A and B and n = 8 for C and D.
|
|
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 14. Experiment 2: spread of antidromic activation (mean ± SD) along the tonotopic axis of the ICC as a function of absolute ICC current levels (thin line). These values were calculated from the data in Fig. 12B using Eq. 1 and only include data for ICC sites with BFs 5 kHz. The number of sites used to calculate the mean and SD for each level is 1, 4, 11, 15, 19, 23, 30, 30, 32, 35, 35, 35, and 35. The dark line is a fitted curve of Eq. 2 to the maximum spread value for each current level (R2 = 0.9, P<<0.001) and describes how far from the stimulated site (D) fibers can be activated for a given current level (I). For more details on these calculations, see Data analysis: antidromic activation spread.
|
|
In general, Fig. 7 demonstrates that fibers projecting from A1 are amply distributed throughout the ICC laminae. Based on stained sections of the guinea pig ICC (particularly in Figs. 1E, 2, and 6 presented by Malmierca et al. 1995), the isofrequency layers corresponding to the BF regions presented in our Fig. 7 roughly span a caudorostral distance of 2 mm and a dorsomedial-to-ventrolateral distance of 2–2.5 mm. Across animals, our sites spanned a caudorostral distance (XCR) of 1.9 mm and a dorsomedial-to-ventrolateral distance (XDmVl) of 2.1 mm in which stimulation of all but one of these sites (asterisk) elicited antidromic activity in A1. Some of the circles did span larger areas with diameters up to several hundred micrometers. However, if we assume that only a few or no A1 neurons project to the ICC, then the circles should span almost the entire lamina and/or exhibit greater overlap. Furthermore, stimulation of most (70%) of our sites activated fibers within about a 100 µm distance making it unlikely that there were only a few fibers we coincidentally positioned our sites next to. It is important to note that we could not differentiate between activation of passing fibers versus terminal fibers. However, tracer studies have shown widespread termination of corticollicular fibers throughout the ICC (Saldana et al. 1996), and it is unlikely that fibers would traverse across all the locations of our sites without making any synaptic terminations.
Based on our data, there does appear to be some differences based on BF region. Figure 7, A–D, corresponds to mid-BF regions (9–14 kHz) and generally consists of smaller circles than those in Fig. 7, E–H, which correspond to higher BF regions (20–23 kHz). A greater number of larger circles suggest that fewer corticofugal fibers span the ICC lamina and/or higher current levels are required to activate the fibers. Figure 8 better depicts this difference by displaying the antidromic threshold distributions for the mid BF (Fig. 8A) and higher BF (Fig. 8B) regions. All but one location for the mid BF regions exhibited antidromic thresholds that were <10 µA (50% were <5 µA), whereas thresholds were scattered across a larger range for the higher BF regions (12% were <5 µA). Thus lower BF A1 neurons appear to project more densely to an ICC lamina than higher BF neurons. Another possibility is that higher BF ICC regions may consist of corticofugal fibers with thinner diameters (different biophysical properties), which would result in higher current levels for activation (BeMent and Ranck 1969b; Ritchie 1982; Rubinstein 1991; Waxman and Swadlow 1977).
One interesting observation we observed was the difference in projection patterns between the laminae in Fig. 7, G and H. Considering that both correspond to similar BF regions (and dimensions) yet Fig. 7H exhibits a greater number of larger circles suggests that different ICC projection patterns may exist for neurons located within different locations along the isofrequency dimension of A1 for higher BF regions. Such an organization might reflect a topographic map of corticofugal projections aligned along the isofrequency dimension that are associated with sound features other than frequency. However, it is also possible that for the 22- to 23-kHz BF ICC lamina there are just fewer corticofugal fibers in more caudal and ventrolateral locations where most of the placements were made in Fig. 7H. At least this would suggest that unlike the mid BF A1 neurons that appear to project amply throughout an ICC lamina, some higher BF A1 neurons may exhibit more spatially specific projection patterns within the ICC. In future studies, we will need to systematically map the antidromic responses along the isofrequency dimension of A1 for stimulation of different locations along an ICC lamina to confirm these inferences.
Experiment 2: tonotopic organization
Results from experiment 1 demonstrated that corticofugal neurons project from A1, mainly from layer V, to different regions within an ICC lamina. These projections were assessed for BF-aligned regions between A1 and the ICC. For experiment 2, we were interested in characterizing the distribution of these corticofugal projections along the tonotopic dimension. Figure 9 presents multi-site PSTHs recorded across eight A1 sites located in different BF regions in response to stimulation of sites within different BF ICC regions for two stimulus levels. The colorscale was adjusted to provide better visualization of the antidromic activity. For ICC site 1, the orthodromic activity was still present and partially masked the antidromic component of the response for both levels. Generally, stimulation of a site within a low BF ICC region elicited antidromic activity in a low BF region of A1. Similarly, stimulation of a higher BF ICC site elicited antidromic activity in a higher BF A1 region. From these plots, it is evident that the projection pattern from A1 to the ICC is tonotopically organized such that A1 neurons project to ICC neurons with a similar BF. Even when the stimulus level is increased 5–10 dB above antidromic threshold, the responses maintain localization (see next section).
We quantified the tonotopic projection pattern from A1 to the ICC across all our animals by plotting the BF of the A1 site with the lowest antidromic threshold against the BF of the stimulated ICC site (Fig. 10). Perfect BF alignment (diagonal line) was not always possible due to the set geometry of our electrode sites. Thus we presented the closest possible BF-aligned sites with dots. Stimulation of 69% of our ICC sites elicited antidromic activity with the lowest threshold on the closest BF-aligned A1 sites. The fact that stimulation of almost all of our sites (48 of 51) elicited antidromic activity with the lowest threshold on an A1 site that was at most one site away from the closest BF site demonstrates that corticofugal projections from A1 to the ICC are tonotopically organized, at least for the BF regions we assessed (3–30 kHz). Furthermore, stimulation of almost all the sites (47 of 51) elicited antidromic activity with a current level <30 µA with more than half of them (27 of 51) being <10 µA. Combined with the inferences made from experiment 1, these low thresholds across different BF regions suggest that corticofugal fibers are amply scattered throughout the ICC along the frequency and isofrequency dimensions.
Experiment 2: A1 spread of activation
We have shown that A1 neurons tonotopically project to different ICC neurons. However, the question arises as to how spatially specific these tonotopic projections are. To answer this question, we assessed the spread of antidromic activation along the tonotopic gradient of A1 in response to varying levels of stimulation within the ICC. Figure 11A presents a typical example where a black dot indicates that antidromic activity was present on an A1 site for a given stimulus level. At threshold and even up to 8 dB above threshold, antidromic activity was present on only one A1 site. As the stimulus level was increased, neighboring sites gradually began to exhibit antidromic activity. Even for the case where we observed the greatest spread across all our sites and animals (Fig. 11B), antidromic activity was still spatially localized to two sites at threshold that then gradually spread to neighboring sites for higher levels. These results, along with the findings from Fig. 10, suggest that A1 neurons do not project to multiple BF regions within the ICC but rather project only to ICC neurons with a similar BF in a one-to-one pattern.
We did observe cases where stimulation of an ICC site elicited a discontinuous activation pattern in A1 (Fig. 11C). At threshold, localized antidromic activity was observed. However, as the stimulus level was increased, antidromic activity appeared on a more distant A1 site in a discontinuous pattern that did not usually spread to other sites. This pattern may be caused by activation of corticofugal fibers passing across different BF regions en route to their target ICC lamina. Stimulation of 25% (13 of 51) of our ICC sites elicited discontinuous antidromic patterns, which included discontinuities over one site to as many as five sites. The example in Fig. 11C was discontinuous over three sites.
To summarize our spread results, we plotted the mean activation spread and SD for different stimulus levels above threshold in Fig. 12A. Data for the discontinuous cases were excluded from this analysis. Activation spread was plotted in terms of distance rather than site number, where the distance between two sites was 200 µm. On average, antidromic activity did not span a distance of >200 µm across A1 (appeared on only 1 site) even at 6 dB above threshold. As the stimulus level was increased, antidromic activity gradually appeared across a larger region of A1 but was still only 400–500 µm at 24 dB above threshold. Figure 12B plots A1 spread distance as a function of absolute ICC current level to display the extent of antidromic activation across A1 independent of thresholds, and is used for further analysis in the next section.
In Fig. 12 there was some variation in activation spread across stimulation sites for each level. This may be attributed to differences in biophysical properties of A1 neurons projecting into the ICC. However, some technical limitations likely contributed to this variation. Corticofugal fibers projecting from the vicinity of our A1 recording sites may not have always been in direct contact with our stimulation sites as was evident from Fig. 7. If an ICC site was not in direct contact with its target fiber (within a given BF lamina), then higher levels of stimulation would be required to activate that fiber. Furthermore, depending on how close fibers in neighboring laminae were to this stimulation site would affect the extent of activation spread at higher stimulus levels. Also we could not guarantee that all our A1 sites were optimally positioned near neurons with the lowest antidromic thresholds due to the set geometry of our recording probe. Depending on how much additional current within the ICC was required to antidromically activate neurons close to those A1 sites could further alter the spread of antidromic activation across A1.
Experiment 2: ICC spread of activation
As explained in METHODS, assuming a one-to-one tonotopic projection pattern from A1 to the ICC and the existence of linear BF gradients (on an octave scale) enables us to directly scale our A1 spread data to obtain current-distance values for activation of corticofugal fibers across the tonotopic gradient of the ICC using Eq. 1. We showed in the previous section that a one-to-one tonotopic pattern exists. Figure 13 demonstrates that the BF gradients in A1 and the ICC are approximately linear. In Fig. 13, A and C, the inclusion of sites with BFs <5 kHz resulted in nonlinear curves. Therefore we omitted those sites and recalculated the BF gradients in Fig. 13, B and D, by fitting straight lines to all the curves. For A1, we obtained a gradient (slope inverse value) of 711.9 µm/octave. This is consistent with results presented by Arenberg et al. (2000) where they obtained values ranging from 203 to 712 µm/octave. The fact that our value equaled their upper limit may reflect differences in calculation methods for these BF shifts. Our results are closer to the value of 650 µm/octave that can be estimated from the optical imaging results presented by Bakin et al. (1996). For the ICC, we obtained a gradient of 583.3 µm/octave, which is consistent with the values (380–716 µm/octave) presented by Snyder et al. (2004).
The data in Fig. 14 is just a scaled version of the data presented in Fig. 12B excluding points corresponding to BFs <5 kHz. The dark line represents a least-squares fit (R2 = 0.9, P << 0.001) of Eq. 2 to the maximum spread value for each current level resulting in a k equal to 0.000504 µA/µm2 and Io equal to 1.6 µA. As discussed in the previous section, there was some variation in A1 activation spread due to misalignments between our sites and the corresponding neurons within the ICC and A1. These misalignments tend to artificially reduce the true spread of activation within the ICC. So we fit Eq. 2 to only the maximum spread value for each level. If we assume that at least some of our spread values for each level corresponded to optimally aligned sites within the ICC and A1 and to a representative sample of fibers, then the fitted equation should provide an approximate upper limit for the current-distance relationship of activation of corticofugal fibers across the tonotopic axis of the ICC. Furthermore, it appears that even across different tissue regions and anisotropic conditions, the current-distance relationships tend to be fairly similar and depend more on properties associated with conduction velocity and fiber diameter (BeMent and Ranck 1969a,b; Ranck 1975; Rubinstein 1991). Thus the fitted curve in Fig. 14 should also provide a reasonable upper estimate for current spread along the isofrequency laminae of the ICC. Our curve is consistent with data presented in Fig. 1 of Ranck (1975) in which current-distance curves across different tissue regions and preparations for 200-µs/phase cathodal pulses were compiled into one plot. Although we used biphasic cathodic-leading 200-µs/phase pulses, studies have shown that threshold increases should be minimal (<3%) at that pulse width (McIntyre and Grill 2000; Rubinstein et al. 2001). Ranck's lower bound of data points (i.e., shortest activated distance for a given current level) corresponds to fibers with slower conduction velocities as low as 9 m/s, which in turn corresponds to fibers with smaller diameters as low as 1.5 µm based on the well-known scaling factor of 6 (Hursh 1939; Ritchie 1982; Waxman and Swadlow 1977). Considering that corticofugal fibers within the ICC appear to be very thin (<1 µm in diameter) (Winer 2005b) and our fitted curve aligns with Ranck's lower bound of data points, our inference that our curve represents an upper limit for activation of corticofugal fibers throughout the ICC is reasonable.
|
DISCUSSION
|
|---|
Overall, our results demonstrate that corticofugal fibers from A1 to the ICC are amply scattered throughout the ICC and tonotopically organized in a one-to-one pattern. However, there does appear to be some differences in spatial distribution (or even biophysical properties) of the fibers across an ICC lamina that depends on BF region and possibly isofrequency location in A1. These findings also provide insight as to some of the functional effects observed in other electrophysiological studies that orthodromically stimulated pathways projecting into the ICC. Although there are limitations associated with the antidromic stimulation method, we have demonstrated the potential of combining this method with other electrophysiological techniques and electrode technologies for studying the descending auditory system that can complement anatomical studies. These findings are further discussed in the following text.
Antidromic stimulation method
As is evident from our results, antidromic stimulation combined with other electrophysiological techniques can be used as a method for studying the spatial distribution of descending monosynaptic projections from A1 to the ICC. The success of this method was attributed to the multi-site Michigan probe technology, which enables us to stimulate and record from the same electrode sites and simultaneously across multiple regions in the ICC and A1. Thus it is possible to first characterize the neurons surrounding each site within the ICC and A1 based on acoustic response properties (i.e., BF, threshold, PSTH response, binaurality, frequency bandwidth, etc.). Then different regions of the ICC can be electrically stimulated, and the antidromically activated spikes can be recorded in different layers and regions of A1. CSD analysis, histological techniques, and acoustic-driven responses can be used to identify the location of each site within the ICC and A1. As a result, the spatial projection pattern of different A1 neurons to the ICC and the electrophysiological properties associated with these projections can be identified using antidromic stimulation. This method can also be used to study monosynaptic projections from the auditory cortex to other regions, such as the medial geniculate body and cochlear nucleus.
There were a few limitations with this method we observed in this study. With respect to our probe designs, we were not able to ensure close proximity stimulation of corticofugal fibers within the ICC and recording of single output neurons within A1. Furthermore, we were not able to simultaneously record from different layers, BF regions, and isofrequency locations within A1. So we had to perform two separate experiments in which we still could not record from the lowest antidromic threshold neurons with all our A1 sites. There are now available multi-site silicon probes with as many as 64 sites (NeuroNexus Technologies, Ann Arbor, MI) and more dense probes currently being developed that can allow us to simultaneously record from different layers to identify the optimal depth site and regions to assess BF and isofrequency location effects. Using more closely spaced electrode configurations, such as tetrodes (Gray et al. 1995), we can better isolate single units and assess how individual A1 neurons project to different subcortical regions. This would also enable us to use the collision test to more accurately identify antidromic spikes and isolate slower conducting neurons (Fuller and Schlag 1976; Swadlow 1998). In terms of ensuring close proximity stimulation of corticofugal fibers, particularly the axonal terminations within the ICC, more accurate methods will need to be developed. At least based on biophysical models, stimulation of axon terminals can achieve substantially lower thresholds compared with that of passing fibers under certain stimulus conditions (Rubinstein 1993). If such models are developed for the corticofugal fibers within the ICC and we use denser probes to stimulate more sites throughout a target region, then it may be possible to identify which site is in close proximity to an axon terminal based on antidromic thresholds.
Another limitation we encountered was the inability to directly compare our ICC site locations along a lamina across different sized animals and BF regions. This is important for assessing differences in projection patterns across BF regions and identifying if any systematic spatial distribution exists along the isofrequency dimension. Similar limitations will exist for identifying recording site locations within A1. One advantage of using higher-density probes that span a larger area is that we will be able to more efficiently identify the borders along the frequency and isofrequency dimensions. In this sense, we can normalize each region with respect to these borders. It may also be possible to identify other electrophysiological properties that can further separate out different functional zones, especially for regions that do not have a systematic tonotopic organization.
Projections from A1 to ICC
Some anatomical studies have shown that a substantial number of direct projections from A1 to the ICC exist that appear to be tonotopically organized (Andersen et al. 1980b; Bajo and Moore 2005; Bajo et al. 2006; Saldana et al. 1996) and glutamatergic (Feliciano and Potashner 1995). Using antidromic stimulation, we provided further evidence confirming that fibers projecting from A1 are amply scattered throughout the ICC and are spatially organized such that A1 neurons only project to ICC neurons with a similar BF. Also consistent with anatomical studies (Bajo and Moore 2005; Bajo et al. 2006
TOP
ABSTRACT
INTRODUCTION
