| 1. Laminar structure of auditory signals processing in primary auditory cortexPrimary auditory cortex (Al) contains six distinct layers, each with clearly differentiated cellular compositions and with a unique set of input and output projections. In Al, cells are vertically arranged and accompanied by highly specific interlaminar connections. This vertical microcircuitry has been considered a key element of cortical processing. However, the specific functions of each layer in information processing are still elusive.Previous works useing linear silicon probes inserted vertically into the cortex so that recording sites were located across all layers in a single column. In auditory and somato sensory cortex, both spontaneous and evoked activities were very sparse in superficial layers (L2/3), whereas L5often showed denser population activity patterns. A recent study used cell-attached recording techniques in auditory cortex of unanesthetized rats, reported only modest differences in spontaneous and sound-evoked responses across layers. However, the preliminary works on individual A1neurons of our lab revealed diverse results. Most of the neurons in L2/3showed no responses to pure tone and white noise, while the spike responses were more frequently evoked in deep layers. Given the characteristic cellular structures and interlaminar connections, the results displaying differences in neural activities across the layers seemed closer to the real patterns.Here, we investigate the laminar structure of spontaneous and sound-evoked spiking activities in primary auditory cortex of ketamine/xylazine-anesthetized rats. Pure tones of varying frequency and intensity,50ms broadband-noise and100μs click of varying intensity were used to estimate the dynamic processing properties of individual neurons across the vertical A1microcircuit.Among the total621single cells obtained by juxtacellular recording,426neurons displayed spiking activities. Depths of recordings were measured perpendicular to the cortical surface, as given by micromanipulator readings. Most (78%) of the neurons in putative superficial layers (L2/3) showed no spontaneous discharge and kept silent even under the strong driven of noise with90dB SPL intensity. The spiking neurons, including spontaneous and auditory-response neurons, were more familiar in deep layers such as L4,5and6. Spontaneous activities of single neurons were then calculated during periods without sound presentation. Undoubtedly, L2/3neurons showed the sparsest activity and L6cells presented the densest activity. The high spontaneous rates (5spikes/second) occupied more than26%of neurons in L5/6and vanished in L2/3. Our results were consistent with the extracellularly recorded population patterns and presented the "silent" or "active" characteristics of neurons in different layers.To examine the layer-specific processing of acoustic information, we afforded a stimuli sequence from noise to tone burst and click for all of the neurons.128cases had finished the three tasks. The response proportions and the dynamic processing properties such as minimum threshold (MT) and first spike latency (FSL) of different stimuli were subsequently computed and compared. Great diversities were also emerged in neurons across the laminae. The dynamic properties of noise and tone processing were similar in some respects. The FSLs of the two kinds of stimuli were shortest in L4and longest in L2/3, and MTs were lowest in L5and highest in L2/3. So, the cortical signalling streams of noise and tone stimuli begin with excitatory discharge in L4, spread through the column to L5and L6, at last to L2/3. The lower excitabilities presented by the higher thresholds maybe devote to the "silent" of L2/3neurons. There was no obvious difference in the FSL between the noise-evoked response and tone-evoked response for one neuron, nor the MT.The click-evoked responses showed enormous differences with noise and tone responses. The threshold for click stimuli was significantly higher than that of the same neuron in response to noise or tone burst and the FSL was much shorter. The stream of information was initiated from the deep layer (L6), spreading rapidly across columns to the superficial layer (L2/3). These distinct processing properties of auditory signal have not been reported.The response proportions and the multi-response to different stimuli were observed across the layers. More than65percent of neurons failed to respond to any stimuli we presented. The data support the "sparse coding", which provide efficient and energy-saving representations for natural scenes by the activity of a small fraction of neurons. Most of the auditory neurons were multi-response to2or3kinds of auditory signals with different properties. Because of the sparse representation in auditory cortex, finding the optimal stimulus for any given neuron was such a challenge. The preferred stimulus was decided by the lowest threshold for a single neuron. Laminar distributions of preferred stimuli indicated subtil functional laminar diversity or specialization. The pure tone, signal of certain frequency feature, tended to response in L4, the principal recipient layer for thalamocortical afferents with tonotopical structures. The broadband noise preferred L5, which had more complex interlaminar connections.Previous studies by extracellular population recording had demonstrated that the characteristic frequency would remain constant across A1layers. Our data in couple neurons recorded in a single penetration confirmed the CF-based column structure. Neurons in a given column tended to response for the same frequency, while other response properties varied across the cortical layers.Here, we report four major conclusions. First, neuron activities are layer-specific, with "sparse" discharge in superficial layer and "dense" in deep layer. Second, most of auditory-neurons in Al are multi-responsed. Because of the sparse representation in auditory cortex, it is hard to find the optimal stimulus for any given neuron. Third, specific connectivity patterns in the auditory cortex shape the distinct processing of different auditory signals and the flow of information. The pure tone tends to response in L4, while the broadband noise prefers L5. Fourth, there is frequency-based column structure in A1.Cells are vertically arranged according the frequency selectivity and might be organized in a more subtle fashion by the functional microcircuitry.2. Linear shifting of auditory representation by background noise in primary auditory cortexNatural acoustic signals are often accompanied with various types of background noise. Extracting sounds with significance from the competing environment presents a complex challenges for listeners. The cocktail party effect, first proposed by Colin Cherry, presents us the auditory selective attention in the signals detection.Noise with high level is usually detrimental to auditory perception. Increase of detective threshold and reduction of firing strength are generally observed throughout the peripheral and central auditory systems. However, the characteristic frequency (CF) and the tuning sharpness of the receptive field tend to hold steady. Previous studies also demonstrate the unaffected functional properties under noisy background. Except for the predictable shift of threshold, the contineous noise do not change the preferences to envelope features of aperiodically amplitude modulation (AM) or the spatial selectivity in most Al neurons, the dynamics features remain constant for suprathreshold tone levels.The neural underpinning of the encoding manners of signals in noise is not yet completely clear. Attention triggers a cascade of events results in dynamic receptive field changes in auditory cortex to enhance figure/ground separation, by "tuning in" to salient acoustic cues and "tuning out" all others. A recent research proposes a model of contrast gain control. Neurons would adjust the gain for better sensitivities to changes in the acoustic simulations, by expanding or compressing their dynamic response ranges, according to the contrast of recent stimulation. However, the model have not mentioned the modification of the raw frequency tuning curves from the single neuron undergo variable contrast, instead of fitting methods of population units. There is no direct evidence for gain rescale of neurons.Here, we are interested in how levels of background noise influence the response representation of a target tone. We hypothesize that presence of background noise might not affect other response properties except the linear shifts of the detective thresholds. The consequent shifts of the dynamic ranges and receptive areas towards higher signal levels drive neurons to work in higher fields with unaltered functional selectivity. The current study examine the effects of background noise levels on characteristic frequency(CF), minimum threshold (MT), bandwidth (BW), shape of tonal receptive field (TRF) and firing strength of individual neurons to tone bursts, by in vivo cell-attached recording in L4of rat auditory cortex. We also concerned about the homogeneity of noise effects within different cell types, for example, monotonic and nonmonotonic neurons, excitatory regular spiking and inhibitory fast spiking neurons.The background noise intensity varied from0to48dB SPL at five steps. Cell-attached recordings were obtained from83neurons of AI, located at400-700μn below the pia, corresponding to layer4of the auditory cortex. Fixed-frequency (CF) tone bursts of different intensities were applied to explore effects of the noise levels on the intensity processing of auditory cortical neurons. The results showed that the intensity tuning ranges of monotonic excitatory neurons in response to CF-tones were reduced. The tone detection thresholds were progressively shifted towards higher tone intensity in a linear dependence over a wide range of noise levels. The shifting amplitude of the threshold varies among the neurons, and is partly influenced by the original threshold, which reflects the inherent properties of individual neuron. When tested by tones of different frequency and fixed intensity at60dB, the bandwidth (BW) of the frequency receptive field exhibited a significant compression and the responses evoked by tone stimuli depressed obviously along with the increasing noise. The shrink degree of BW was up to61%with background noise at60dB. However, the best frequency (BF) of the tuning curve under different noise conditions remained constant.We infer that the response properties under different noise could be estimated from the original TRF when corrected by the linear increment of threshold. A testing method was designed to explore the conformity of the real response parameters and the calculated response parameters. We found that the real response properties and corresponding values emerged good coincidence. The results confirmed the linear mode of changes in frequency and intensity processing derived from the increasing noise. Furthermore, the TRFs of15monotonic neurons under quiet conditions and in continuous background noise at36dB were overlapped for sharpness estimating. The exact overlap of the curves presented us a shrink tuning area with a constant shape under the two conditions. The response properties, such as BW and firing strength at the same positions of the two TRFs scaled by the vertexes under the two conditions were comparable in the values. That is, the background noise emerged an upward push effect on the tonal response area without changing of the response properties in the corresponding levels.The nonmonotonic neurons are dominated by powerful inhibitory inputs that form a closed, narrow TRF distinct from the V-shape response of monotonic neurons. For response to CF-tone bursts with different intensities in noise background, the thresholds, best intensities (peak of rate-intensity function) and firing rates displayed a linear shifting to higher tone levels along with the increase of noise levels. The increments of best intensity threshold tracked the values of thresholds shift. The dynamic range of intensity tuning was compressed in some degree, but the discharge rates at the new best intensity reset by the increasing background noise showed no obvious reduction. To our opinion, the push effect of noise background on the nonmonotonic neuron may transfer the closed TRF into V-shape like a monotonic neuron.To explore the noise effect on inhibitory neuron,3cases of fast-spiking inhibitory neurons were involved to estimate the changes of CF, MT, BW and TRF under noise background. We were not surprised in the presence of linear shift of threshold and up-shift of TRF under noise background. However, with a flater TRF than that of excitatory neurons, the shrink degree of BW was just28%with background noise at48dB. Comparison of threshold shift (determined by the linear slopes obtained in the threshold-noise level function) for the excitatory monotonic neurons and inhibitory neurons showed significant difference between the two groups. Lower mean values appeared in inhibitory neurons, which suggested the lower sensitivity of inhibitory neuron to background noise. The imbalanced and slowly change of inhibitory neuron may devote to the stronger lateral inhibition for the quickly compressed frequency receptive field of excitatory neuron under background noise.Accordingly, the study report three conclusions. First, background noise cause the linear shifting of threshold, the compress of tonal receptive range and the reduction of firing strength, and the linear up-shift of the tonal recetive area of A1neurons. Second, the frequency selectivity of neuron remains constant and the best intensity representation of nonmonotonic neuron shifts upward to high intensity level. Third, the influences of background noise to excitatory and inhibitory neurons are unbalaned. The slowly shrink in receptive field of inhibitory neuron may devote to the compressed TRF in noise of nonmonotonic neuron. |
PDF Full Text Request