Asymmetry in corticofugal modulation of frequency-tuning in mustached bat auditory system

Results

Centripetal and Centrifugal BF Shifts. Electric stimulation of the DSCF areas of the right (n = 24) or left (n = 24) evoked the BF shifts of contralateral DSCF and DPD neurons. There was no sign of hemispheric difference in the effect of the electric stimulation.

Fig. 2A shows the arrays of poststimulus-time histograms representing the frequency-response curves of a left DSCFd neuron tuned to 61.5 kHz (Fig. 2Aa1). When right DSCFd neurons tuned to 61.2 kHz were electrically stimulated, the BF of this recorded neuron shifted from 61.5 kHz to 61.4 kHz (Fig. 2Aa2), exhibiting a centripetal BF shift. The shifted BF returned (i.e., recovered) to the control BF ≈180 min after the electric stimulation, but the response to 61.4 kHz did not (Fig. 2Aa3), taking ≈210 min for the complete recovery (Fig. 2Aa4). After the recovery of the frequency-response curve of this neuron (Fig. 2Ab1), BMI was applied to the stimulation site. Then, the BF of the neuron shifted from 61.5 kHz to 61.3 kHz (Fig. 2Ab2) and recovered ≈35 min after the BMI application (Fig. 2Ab4). This centripetal BF shift was larger but recovered faster than the BF shift evoked by electric stimulation.

The electrode used to record the above neuron was moved ≈40 μm, because its action potentials became small. Then, another neuron, also tuned to 61.5 kHz, was recorded (Fig. 2Ba1). When right DSCFv neurons tuned to 61.3 kHz (instead of right DSCFd neurons) were electrically stimulated, the BF of the neuron centrifugally shifted from 61.5 kHz to 61.6 kHz (Fig. 2Ba2), recovering to the control BF ≈270 min after the electric stimulation (Fig. 2Ba4). After the recovery (Fig. 2Bb1), BMI was applied to the stimulation site. Then, the BF of the neuron shifted from 61.5 kHz to 61.3 kHz (Fig. 2Bb2). That is, BMI evoked a centripetal BF shift, which recovered ≈35 min after the BMI application (Fig. 2Bb4).

All of the DPDv neurons studied showed basically the same changes in BF, as did all of the DSCFd neurons (Table 1). In Fig. 2C, a left DPDv neuron was tuned to 60.9 kHz (Fig. 2Ca1). When 61.2-kHz-tuned right DSCFd neurons were electrically stimulated, the BF of the neuron centripetally shifted from 60.9 kHz to 61.0 kHz (Fig. 2Ca2) and then returned to the control BF ≈270 min after the stimulation (Fig. 2Ca4). After the recovery of the BF (Fig. 2Cb1), BMI was applied to the stimulation site, evoking a centripetal BF shift larger than that evoked by the electric stimulation (Fig. 2Cb2). This BF shift recovered ≈35 min after the BMI application (Fig. 2Cb4).

The BF shift of the same neuron as that shown in Fig. 2C was further studied by stimulating right DSCFv neurons (Fig. 2Da) and, later, by applying BMI (Fig. 2Db). The neuron showed a centrifugal BF shift for the electric stimulation (Fig. 2Da2) and a centripetal BF shift for BMI (Fig. 2Db2). The BF shifts of DPD neurons evoked by BMI were also larger and shorter-lasting than those evoked by electric stimulation, as found in the DSCF neurons.

In four single cortical and two single collicular neurons, we found that a single neuron could show both centripetal and centrifugal BF shifts, depending on whether electric stimulation was delivered to the opposite DSCFd or -v neurons. In 13 DSCFd and 14 DSCFv electric stimulation sites, we found that electric stimulation at a single cortical location evoked both centripetal and centrifugal BF shifts of contralateral DSCF and DPD neurons, depending on the locations of the recorded neurons, as described below.

BF-Shift-Difference Curves. A “BF-shift-difference” curve indicates a change in the frequency axis around the BF of stimulated neurons (13, 15). The amount of the BF shifts of contralateral DSCF and DPD neurons was related to the difference in BF between paired recorded and stimulated neurons, so that all of the centripetal and centrifugal BF shifts were separately pooled and plotted as the function of BF differences. The BF-shift-difference curves in Fig. 3 (open circles) are flat near the point of origin, because the BFs of recorded neurons matched to those of stimulated ones within 0.2 kHz did not shift, as reported for the BF shifts of ipsilateral cortical DSCF, thalamic DSCF, and collicular DPD neurons evoked by cortical electric stimulation (16, 17).

Both the maximum centripetal and centrifugal BF shifts of contralateral DSCF neurons evoked by electric stimulation were 0.20 kHz (n = 18), with no difference in amount between centripetal and centrifugal BF shifts (P > 0.1). BF shifts increased with BF differences within the range of ± 0.7 kHz, decreased for BF differences >0.7 kHz, and were nonexistent at BF differences >1.3 kHz (Fig. 3 A and B, open circles). BMI applied to DSCFd or DSCFv neurons always evoked centripetal BF shifts (0.30 kHz at the maximum, n = 8) that were larger than those evoked by electric stimulation (P < 0.0001; filled circles in Fig. 3 A and B). BMI-evoked centripetal BF shifts, even when the BF difference between paired recorded and stimulated neurons was 0.10 kHz.

The BF shifts of DPD neurons (Fig. 3 C and D, open circles) were smaller than those of the DSCF neurons in amount and range (P < 0.001). The centripetal BF shifts of DPD neurons evoked by BMI applied to the opposite DSCFd or DSCFv neurons (Fig. 3 C and D, filled circles) were larger than those evoked by electric stimulation (P < 0.001).

Organization for Evoking BF Shifts. The effects of electric stimulation of the DSCF area on the BFs of contralateral DSCF neurons were just opposite, depending on whether the stimulated DSCF area was DSCFd or -v and whether a recorded DSCF neuron was in the DSCFd or -v (Table 1). To confirm this observation, the distribution of DSCF neurons that evoked centrifugal or centripetal BF shifts was mapped in five animals. First, for example, the left DSCF area was electrically stimulated at different locations, and the effect of the stimulation on the BFs of contralateral DPD neurons was examined. At 6 of 14 sites, stimulated neurons evoked the centripetal BF shifts of DPDv neurons (Fig. 4A Left, open circles). At the remaining 8 sites, stimulated neurons evoked the centrifugal BF shifts of the DPDv neurons (Fig. 4A Left, filled circles). That is, DSCF neurons located in the dorsal and ventral portions of the DSCF area, respectively, evoked the centripetal and centrifugal BF shifts of DPDv neurons. Second, action potentials were recorded from neurons in the hatched portion of the left DSCFd (Fig. 4A Left), and their BFs were remeasured. Then, the right (i.e., opposite) DSCF area was electrically stimulated at different locations, and the effect of the stimulation on the BF shift of the recorded left DSCFd neurons was studied. It was found that electric stimulation at the dorsal and ventral portions of the right DSCF area, respectively, evoked centripetal and centrifugal BF shifts of the left DSCFd neurons (Fig. 4A Center). Third, BMI was applied to the neurons in the right DSCFd or DSCFv. BMI always evoked a centripetal BF shift of the left DSCFd neurons (Fig. 4A Right).

The same experiments described in the second and third steps above were repeated to examine the effects of electric stimulation or BMI applied to the right DSCF area on the left DSCFv neurons in the hatched portion of the left DSCFv (Fig. 4B Left). Electric stimulation at the seven dorsal sites evoked the centrifugal BF shifts of the left DSCFv neurons, whereas electric stimulation at the six ventral sites evoked the centripetal BF shifts of the left DSCFv neurons (Fig. 4B Center). That is, the direction of a BF shift evoked by electric stimulation of DSCFd or DSCFv neurons flip-flopped, depending on whether recorded contralateral neurons were DSCFd or DSCFv neurons. The BF shifts evoked by BMI were always centripetal, regardless of whether BMI-applied or recorded neurons were in the DSCFd or DSCFv (Fig. 4B Right).

The BF shifts of left DPDv neurons (Fig. 4C Left) were always centripetal for the right DSCFd stimulation and centrifugal for the right DSCFv stimulation (Fig. 4C Center). That is, the direction of their BF shifts flip-flopped, depending on whether DSCFd or -v neurons were stimulated. The BF shifts of left DPDv neurons were always centripetal for BMI applied to the right DSCFd or -v (Fig. 4C Right).

Distribution of Centripetal and Centrifugal BF Shifts in the DPD. Contralateral DPD neurons located within a dorsal 500-μm depth showed a centrifugal BF shift for electric stimulation of the DSCFd but a centripetal BF shift for the DSCFv stimulation. On the other hand, contralateral DPD neurons located at depths >500 μm showed a centripetal BF shift for the DSCFd stimulation but a centrifugal BF shift for the DSCFv stimulation (Fig. 5B). That is, the direction of the BF shifts of DPD neurons flip-flopped according to the depths of recorded neurons and the stimulation sites in the DSCF area, as did that of cortical DSCF neurons.

Time Courses of Cortical and Collicular BF Shifts. The contralateral cortical and collicular BF shifts showed a mean latency of 58.2 ± 16.7 min (mean ± SD; n = 181) and a mean recovery time of 211.4 ± 49.6 min (n = 181) for electric stimulation of the DSCF area (Fig. 6, open circles). On the other hand, the BF shifts showed a much shorter latency (2.7 ± 2.4 min, n = 149) and recovery time (15.9 ± 5.3 min, n = 149) for BMI applied to the DSCF area than those for the electric stimulation (Fig. 6, filled triangles or filled circles). There were no differences in latency and recovery time between the contralateral (this work) and ipsilateral (17) neurons. (The time resolution was 3.83 min in our current studies, because the sampling time of an array of poststimulus-time histograms was 3.83 min.) An intriguing phenomenon was that BMI applied to the DSCF area evoked a short-lasting centripetal BF shift, and abolished not only a long-lasting centrifugal BF shift, but also a long-lasting centripetal BF shift evoked by electric stimulation. This phenomenon was observed for BMI applied to the DSCF area at the beginning of (Fig. 6, filled triangles) or 160 min after (Fig. 6, filled circles) the electric stimulation of the DSCF area.

Discussion

It is a totally unexpected finding that the direction of the BF shifts of contralateral DSCFd and -v and DPDd and -v neurons flip-flopped, depending on whether electric stimulation was delivered to the DSCFd or -v (Table 1), although the ipsilateral DSCFd and -v and DPDd and -v neurons always show centrifugal BF shifts (16, 17). The neural mechanism and functional role of the flip-flop may be speculated on the anatomical and physiological data, as described below.

Neural Net for Contralateral Corticofugal Modulation. Suga et al. (11, 12) hypothesized that, when cortical electric stimulation evokes excitation stronger and more widespread than inhibition, centripetal BF shifts are evoked and that, when it evokes inhibition stronger and more widespread than excitation, centrifugal BF shifts are evoked. In our current studies, BMI applied to the DSCF area changed centrifugal BF shifts into centripetal BF shifts (Figs. 2 and 3) (17). This result suggests that the neural net for evoking centripetal and centrifugal BF shifts in the contralateral side is in, or originates from, the electrically stimulated DSCF area; that there are inhibitory connections between the DSCFv and DSCFd for flip-flopping the direction of BF shifts; and that the direction of a BF shift depends on the balance between excitation and inhibition inputs (11, 12). The commissural projection through the corpus callosum appears to play the major role in contralateral corticofugal modulation, because our preliminary data indicate that a lesion to the contralateral DSCF area significantly reduced the development of the collicular BF shift evoked by the cortical electric stimulation. In the ACs of the mustached bat (27) and cat (28), E-E bands are bilaterally connected, but I-E bands are not. In the cat AC, I-E bands are interconnected with cortical auditory areas in the same hemisphere (29).

On the basis of the above physiological and anatomical data, we may propose a working model for the flip-flop of BF shifts. In the model (Fig. 7), electric stimulation of the DSCFd stimulates two types of inhibitory neurons in the ipsilateral DSCFd: One evokes inhibition within the ipsilateral DSCFd, and the other evokes inhibition in the ipsilateral DSCFv. These inhibitions result in the centrifugal BF shifts in both the DSCFd and -v. Because the callosal neuron in the DSCFv is inhibited by DSCFd stimulation, the excitation in the contralateral DSCFv and inhibition in the contralateral DSCFd evoked by the callosal neurons both decrease, resulting in centrifugal and centripetal, respectively, BF shifts (Fig. 7B). When the DSCFv is stimulated, two types of inhibitory neurons and callosal neurons are excited. Therefore, inhibition in the ipsilateral DSCFd and -v increases, resulting in centrifugal BF shifts in both the ipsilateral DSCFd and -v and increases both excitation in the contralateral DSCFv and inhibition in the contralateral DSCFd, so that the contralateral DSCFd and -v, respectively, show centrifugal and centripetal BF shifts (Fig. 7C). Contralateral DPD neurons showed a flip-flop in BF shift for the electric stimulation of DSCFd or -v neurons, as did contralateral DSCF neurons.

Function of Contralateral Corticofugal Modulation. In the natural condition, acoustic stimuli always stimulate both ears, so that contralateral corticofugal modulation occurs on the ipsilateral side, in addition to ipsilateral corticofugal modulation.

DSCFd neurons are mostly I-E neurons (23) and are tuned to strong sounds (19, 20) originating from ≈30° on the contralateral side (30), whereas DSCFv neurons are mostly E-E neurons (23) and are tuned to weak sounds (19, 20) originating from ≈0° in front (23). When a weak sound at 61.00 kHz is repetitively delivered to the animal, right DSCFv neurons are mainly excited via the left ear and, presumably, show centrifugal BF shifts. Likewise, left DSCFv neurons are mainly excited via the right ear and, presumably, show centrifugal BF shifts. The amount of these BF shifts can be slightly different between corresponding right and left DSCFv neurons, because the intensity of the sound is frequently different between the two ears. However, the DSCFv neurons on the right and left sides send signals mutually for evoking centripetal BF shifts, so as to eliminate this asymmetry. The above interpretation also holds true for the interaction between the right and left DSCFd areas.

Why are BF shifts opposite between the contralateral DSCFv and DSCFd? Why does the direction of a BF shift in the contralateral DSCF area flip-flop, depending on whether the DSCFv or DSCFd is activated? When a weak sound at ≈61 kHz is repetitively delivered to the animal, the DSCFv neurons are mainly excited, so that the BF shift of the DSCFv would be larger than that of the DSCFd. To match the BF shift of the DSCFd with that of the DSCFv, the BF shift of the DSCFd is increased by the DSCFv. This interpretation also holds true for the DSCFv when a strong sound is repetitively delivered to the animal and strongly activates the DSCFd. Our working hypothesis is that the centrifugal and centripetal contralateral BF shifts mediated by commissural fibers are for equalization in the frequency representation of the DSCFd and -v areas on both sides.