Spatial organization of neuronal population responses in layer 2/3 of rat barrel cortex

Abstract

Individual pyramidal neurons of neocortex show sparse and variable responses to sensory stimuli in vivo. It has remained unclear how this variability extends to population responses on a trial-to-trial basis. Here, we characterized single-neuron and population responses to whisker stimulation in layer 2/3 (L2/3) of identified columns in rat barrel cortex using in vivo two-photon calcium imaging. Optical detection of single action potentials from evoked calcium transients revealed low spontaneous firing rates (0.25 Hz), variable response probabilities (range, 0-0.5; mean, 0.2 inside barrel column), and weak angular tuning of L2/3 neurons. On average, both the single-neuron response probability and the percentage of the local population activated were higher in the barrel column than above septa or in neighboring columns. Within the barrel column, mean response probability was highest in the center (0.4) and declined toward the barrel border. Neuronal pairs showed correlations in both spontaneous and sensory-evoked activity that depended on the location of the neurons. Correlation decreased with increasing distance between neurons and, for neuronal pairs the same distance apart, with distance of the pair from the barrel column center. Although neurons are therefore not activated independently from each other, we did not observe precisely repeating spatial activation patterns. Instead, population responses showed large trial-to-trial variability. Nevertheless, the accuracy of decoding stimulus onset times from local population activity increased with population size and depended on anatomical location. We conclude that, despite their sparseness and variability, L2/3 population responses show a clear spatial organization on the columnar scale.

Figure 1.

Sensory-evoked calcium transients in identified E1 barrel column. A, Left, Craniotomy over somatosensory cortex. Middle left, Intrinsic optical signal after stimulation of E1 whisker showing the resulting “spot.” The outline of the darkest pixels is overlaid in both images indicating the position of E1 barrel column. Middle right, Close-up view of the E1 area from a low-resolution two-photon image. Barrel field outlines obtained from cytochrome c staining are overlaid. Note the pipette tip on the right side. Right, L2/3 neurons (green) and astrocytes (yellow) stained with calcium indicator and sulforhodamine 101, respectively, in the E1 area and the surrounding tissue. A neuron within the E1 barrel border (red) was targeted with a patch electrode for cell-attached recording. B, Calcium measurement from a population of neurons from a different experiment than in A (scale bar, 30 μm). Fluorescence traces from five simultaneously recorded PW-related L2/3 neurons showing spontaneous (blue asterisks) and sensory-evoked (red arrowheads) calcium transients. Calcium transients were classified as sensory evoked if the initial fluorescence rise and peak occurred within 200 ms after either the onset (away from resting position) or the offset (return to resting position) of the whisker deflection. Note that not all whisker deflections evoke calcium transients. Yellow bars indicate stimulation periods. An example calcium transient associated with offset stimulation is shown on expanded timescale. C, Simultaneous cell-attached (CA) recording and fluorescence measurement from the L2/3 neuron from the E1 area shown in B. A spontaneous (blue asterisks) and a stimulus-evoked (red arrowhead) spike and their associated calcium transients are expanded in the bottom. The spontaneous spike fell outside the 200 ms time windows used to classify spikes as onset (window a) or offset response (window b). D, Electrically (top, ePSTH) and optically (bottom, oPSTH) determined PSTH for the same neuron and the same trials. Only five false positives resulted for 16 min continuous imaging and 231 stimuli in this experiment. Arrows depict bins that contain responses to whisker stimulation.

Figure 2.

Sensory-evoked responses are pronounced in PW-related regions. A, Probability of spiking response to whisker deflection onset as a function of distance from the PW-related BCC in PW-related (black), septum-related (red), and SW-related (green) neurons. B, As in A, showing responses to deflection offset. C, Spatial map of response probability for whisker deflection onset (left) and offset (right) pooled for all populations and overlaid after alignment to barrel centers. Minimum (green dashed line), maximum (red dashed line), and average (black lines) barrel column borders are shown. Arcs run vertically, rows run horizontally, and black crosses indicate BCC. Scale bar, 200 μm. D, Distribution of onset responses for PW-related, septum (Sept)-related, and SW-related neurons. Box representations show median (broken line), 10th, 25th, 75th, and 90th percentiles and range. E, Mean probability of spiking response in PW-related neurons to whisker deflection onset (black) and offset (red) as a function of distance from BCC. Error bars show SEM. F, Spontaneous rate of calcium transients as a function of distance from BCC; colors as in A. G, Distribution of spontaneous activity for PW-related, septum-related, and SW-related neurons.

Figure 3.

Heterogeneous population response to angular whisker deflection. A, Average activity of one neuron in response to 223 whisker deflections in eight directions (0–315°). Arrows depict onset (black) and offset (red) deflection (neuron 1 in B and C). B, Image of neurons (green) and astrocytes (orange) located within barrel column borders. Scale bar, 30 μm. C, Polar plots of mean percentage responses of the same neurons depicted in B to randomly deflecting the principal whisker in eight different directions (onset deflections in black; offset deflections in red; 223 deflections; circles in units of 10% response probability). D, Scatter plots showing max/mean response versus the number of stimuli leading to AP responses for each PW-related (n = 130; left) and septum-related (n = 54; right) neuron (onset in black; offset in red). The max/mean response is defined as the ratio of maximum spike probability from all eight stimulation angles to mean response probability for all angles. The bias, defined as the max/mean response expected for a neuron with no angular preference and a given number of observed APs, is shown in green. E, Probability distribution of bias-corrected max/mean response for both PW-related (left) and septum-related (right) neurons (0.02 bin width; onset in black; offset in red; see Materials and Methods). Distributions interpolated using cubic spline. Neurons consistently responding only to one direction display values >0, up to a theoretical maximum of 7. Neurons responding evenly to all angles display 0. Shaded region corresponds to values >0.30, the value corresponding to a tuning p value of 0.1 in a linear regression (supplemental Fig. 3D, available at www.jneurosci.org as supplemental material). F, Spatial map of angular tuning. The maximal-response direction of each neuron was color coded (color scale, bottom right) and plotted on an averaged barrel-centered map for deflection onset (left) and offset (right). Minimum (green dashed line), maximum (red dashed line), and average (black lines) barrel borders are shown. Arcs run vertically, rows run horizontally, and black crosses indicate BCC. Scale bar, 200 μm. G, Angle of directional tuning vector as a function of angular anatomic coordinate for all PW-related neurons.

Figure 4.

Variability of population response. A, Calcium traces from three neurons showing spontaneous activity (black dashes) and responses to onset (red dashes) and offset (green dashes) whisker deflection for four single trials (orange boxes, trials a–d). B, Raster plot for seven consecutive deflections in 10 neurons (including same neurons as in A; a–d same trials as shown in B). C, Percentage of neurons in a PW-related population that were activated during 100 subsequent whisker onset deflections (left, red squares) or during spontaneous control trials (right, black circles). Arrows indicate occurrences of 90% (a) and 50% (b) activated. D, Spatial location of neurons in example B. Colored neurons depict active neurons in response to onset (red, top) or offset (green, below) whisker deflection. Neurons that were not active are show in black. Scale bar, 50 μm. Note that onset c repeats in onset response f.

Figure 5.

Distribution of percentage of neurons activated is exponential for small populations. A, Average fraction of activated neurons in response to onset whisker stimulation (red bars) and during spontaneous activity (black bars) in PW-, septum (Sept)-, and SW-related populations. B, Probability histogram of percentage of population activated from single whisker deflections (left) and during spontaneous (Spont) activity (right) for populations of different sizes (ranging from 5 to 12 neurons) compared with predicted probability based on independence and measured response probabilities from populations containing 10 neurons. Note log₁₀ scale. C, Average probability distributions of the number of repeated activations of exact-same subsets of neurons over multiple trials. All trials with a given fraction of neurons active (Fig. 4 C, a and b) were searched for repeating spatial patterns in which the exact same neurons were active and inactive (Fig. 4 D, c and f) in response to whisker stimulation (n = 4 animals, 800 total stimulations), and distributions of the number of exact repetitions were calculated. Numbers to the right of each distribution indicate how many neurons were active in the pattern of a possible 10 neurons. Note that diagrams depict how many neurons were active (red) and not active (black) in each distribution of a total of 10 possible neurons. Patterns that occur only once appear in the repeats, 1 bin.

Figure 6.

Pairwise correlations depend on location. Whisker deflection increases the correlation between pairs of neurons and is dependent on both distance between neurons and distance from BCC. A, Calcium transients from four simultaneously imaged neurons during four trials of on and off whisker deflection (orange box). Red asterisks represent calcium transients detected as being action potential generated (see Materials and Methods). B, Pseudocolored representation of a population with each neuron colored as a function of correlation when neuron a responds to whisker deflection (top). Correlation between all neurons within the field of view (bottom). C, Comparison of pairwise correlations from evoked and spontaneous (Spont) activity for neurons pooled from all PW related populations. Represented is median (broken line), 10th, 25th, 75th, and 90th percentiles and range. D, Pairwise correlations for both evoked (black) and spontaneous (red) as a function of neuronal pair distance from BCC (left) and distance between neurons (right) for all PW-related neuronal populations. E, Plot of the dependence of pairwise correlations on distance between neurons and on distance from BCC.

Figure 7.

Stimulus representation in local populations depends on location. A, Mean detection rate (white contours) as a function of false-positive rate for populations containing between one neuron (single arrow) and 14 neurons (double arrow). For each combination of detection rate and false-positive rates, the percentage of information captured by the decoder is color coded. B, Average decoded information (α = 0.08) for five-neuron populations as a function of distance from BCC. Each point corresponds to an average of all possible randomly selected five-neuron combinations from within a single population of simultaneously recorded neurons (n = 29 populations). Distance from BCC was calculated as a mean over all neurons in the population to the center of the intrinsic imaging spot. A significant association was observed (p = 0.02). C, Probability distribution (p) of angle decoding error (in degrees) for increasing population size (n). The peak at zero error for larger population sizes indicates improving accuracy (n = 26 populations). D, Average decoded information for deflection angle from five-neuron populations as a function of distance from BCC. Each point corresponds to an average of all possible randomly selected five-neuron combinations from within a single population of simultaneously recorded neurons (n = 22 populations). Distance to BCC was calculated as a mean over all neurons in the population. No significant association was observed (p = 0.21).