Articles in PresS. J Neurophysiol (November 24, 2004). doi:10.1152/jn.00556.2004
1
HETEROGENEOUS INTEGRATION OF BILATERAL WHISKER SIGNALS BY NEURONS IN PRIMARY SOMATOSENSORY CORTEX OF AWAKE RATS Michael C. Wiest1,4*, Nick Bentley5, Miguel A.L. Nicolelis1,2,3,4
1
Department of Neurobiology, Box 3209, Duke Medical Center, Durham, NC 27710, 2Department of Biomedical Engineering, Box 90281, 136 Hudson Hall, Duke University, Durham, NC 27708, 3 Department of Psychological and Brain Sciences, Box 90086, 229 Soc-Psych Building, Durham, NC 27708, 4Duke University Center for Neuroengineering, Box 3209, Duke Medical Center, Durham, NC 27710, 5Department of Neurobiology and Anatomy, Wake Forest University School of Medicine, Winston-Salem, NC 27157. 7,139 words, 34 pages Revision 2 Submitted November 15, 2004
* Correspondence should be addressed to: Department of Neurobiology, Box 3209, Room 339 Bryan Research Building Duke University Medical Center 101 Research Drive Durham, North Carolina 27710 Phone : (919) 668-7851 Fax: (919) 668-2248 Email:
[email protected]
Copyright © 2004 by the American Physiological Society.
2 Bilateral single-unit recordings in primary somatosensory cortex (S1) of anesthetized rats have revealed substantial cross-talk between cortical hemispheres (Shuler, 2001), suggesting the possibility that behaviorally relevant bilateral integration could occur in S1. To determine the extent of bilateral neural responses in awake animals, we recorded S1 multi- and single-unit activity in head-immobilized rats while stimulating groups of 4 whiskers from the same column on both sides of the head. Results from these experiments confirm the widespread presence of single units responding to tactile stimuli on either side of the face in S1 of awake animals. Quantification of bilateral integration by multiunits revealed both facilitative and suppressive integration of bilateral inputs. Varying the interval between left and right whisker stimuli between 0 and 120 ms showed the temporal integration of bilateral stimuli to be dominated on average by suppression at intervals around 30 ms, in agreement with comparable recordings in anesthetized animals. Contrary to the anesthetized data, in the awake animals we observed a high level of heterogeneity of bilateral responses and a strong interaction between synchronous bilateral stimuli. The results challenge the traditional conception of highly segregated hemispheric processing channels in the rat S1 cortex, and support the hypothesis that callosal crossprojections between the two hemispheres mediate rats’ known ability to integrate bilateral whisker signals. Key words: tactile; multi-electrode; vibrissa; barrel; callosal.
3 INTRODUCTION One approach to understand the coding of whisker stimuli in the rat S1 focuses on a one-to-one relationship between anatomically discrete cortical “barrels” in layer IV and a single principal whisker on the contralateral whisker pad (Armstrong-James, 1995; Brumberg, 1996; Petersen, 2002). Because of the relative scarcity of connections between layer IV excitatory cells in different barrels, and because these cells tend to respond to principal whisker stimulation with the shortest latency, the layer IV barrel field can be interpreted as an array of dedicated detectors for deflections of each contralateral whisker. The fact that whiskers other than the principal whisker do modulate responses in a barrel can be explained in a number of ways. Brumberg et al proposed that inhibitory circuitry within barrels functions to “increase the ‘principal whiskerness’ of the cortical column” by suppressing weaker responses so that “only…the initial perturbation of the principal whiskers will overcome the…cortical inhibition” (Brumberg, 1996). For Armstrong-James (Armstrong-James, 1995), a “transient insularity” between barrels includes the whole cortical column centered over a barrel. That is, for some time after each stimulus the barrel columns can encode the activity of single whiskers. Similarly, Petersen et al (Petersen, 2002) have recently noted that the earliest evoked spike after an isolated punctate singlewhisker stimulus contains most of the information about which single whisker was stimulated. From this they concluded that later activity, reflecting interactions between different barrels, was largely redundant for neural coding of stimulus identity.
4 In the supra- and infra-granular layers inter-columnar connections are more common, and receptive fields in those layers are correspondingly larger (Armstrong-James, 1992; Ghazanfar, 2000; Schubert, 2001). Moreover, these multi-whisker receptive fields are dynamic over post-stimulus time (Ghazanfar, 1999, 2001a). It is more difficult to interpret the responses of cells in these layers primarily in terms of the stimulation of individual whiskers. A complementary view emphasizes the integrative function of S1. Thus, although Simons (Simons, 1985) considered that “each barrel is the morphological correlate in layer IV of a functional cortical column that extends throughout the thickness of cortex,” he also concluded “that an important function of the SI cortex is to integrate information from different, individual vibrissae on the contralateral face.” He suggested inter-columnar integration functioned to make SI neurons sensitive to spatiotemporal patterns of whisker activity characterizing the movement of particular objects across the contralateral whisker pad (Simons, 1985), as have others (Ghazanfar, 1997; Shimegi, 2000). Even for coding single-whisker stimuli, Ghazanfar et al showed (Ghazanfar, 2000) that the relative contribution of temporal interactions between neurons increases with the number of whiskers to be discriminated. Inhibitory inter-columnar interactions have also been suggested to maintain a dynamic range and prevent response saturation (Mirabella, 2001). It is important to note that the different approaches to inter-columnar interactions sketched above need not be exclusive, but could reflect different processing strategies appropriate to different behavioral contexts. This idea has
5 arisen in the whisker sensory-motor literature, in terms of different S1 neuronal response properties in the quiet waking state as compared to the whisking state (Fanselow, 1999; Moore, 1999; Nicolelis, 2002; Castro-Alamancos, 2004; Moore, 2004). These authors proposed that greater response magnitudes during the quiet state could perform a gross “detection” function, while relatively less spike adaptation during the whisking state would allow for finer “discrimination” of spatiotemporal patterns of stimulation. Based on the finding that the horizontal extent of cortex activated by stimulation of a vibrissa is relatively smaller during whisking than during an isolated passive contact, some authors (Moore, 1999; Castro-Alamancos, 2004; Moore, 2004) have further speculated that a more spatially “focused” sensory representation during whisking could improve tactile discrimination. All the coding studies mentioned above have in common that the whisker cortex is interpreted as a topographic map of whisker deflections on the contralateral whisker pad, in line with the cortical homunculus as a map of somatosensory activity in general. A topographic organization does not prohibit lateral interactions, but implies that they should be relatively local. In particular, SI cortex is understood to encode stimulations of the contralateral whisker pad. Such a view is apparently supported by behavioral experiments demonstrating that certain unilateral tactile detection tasks do not require ipsilateral S1 (Hutson, 1986). This picture is overly restrictive. It has long been known that callosal cross-projections integrate the two S1 hemispheres (White, 1977; Olavarria,
6 1984; Koralek, 1990; Cauller, 1998). Interhemispheric connections are concentrated in the homologous portions of the barrel field, innervating primarily but not exclusively the supra- and infra-granular layers (Hayama, 1997). Those callosal connections that do innervate layer 4 are concentrated in the septal regions outside the barrel centers (Olavarria, 1984; Hayama, 1997), as are unilateral inter-barrel connections in layer IV (Kim, 1999). In the present study we focus on the bilateral integration of whisker sensory signals in S1 through these interhemispheric connections, which offers an advantageous window on the larger issue of the dynamic integration of distributed sensory representations, because the peripheral signals to be integrated are clearly segregated in space and may be independently manipulated up to the level of the S1 cortex. Pidoux and Verley (Pidoux, 1979) recorded S1 LFP responses to ipsilateral whisker stimulation in anesthetized rats, the first electrophysiological evidence of cross-talk between the S1 hemispheres. Shuler et al (Shuler, 2001) demonstrated that whisker stimulation on either side of the face of an anesthetized rat could evoke responses in the same layer V S1 neuron. This result has been recently confirmed by intracellular recordings (Manns, 2004). Other studies have shown that the responses of individual neurons in S1 are suppressed in the absence of activity in the contralateral hemisphere (Shin, 1997). Rema and Ebner (Rema, 2003) found similar response suppression as well as changes in whisker-pairing plasticity in all layers, due to contralateral lesions.
7 One potentially relevant function of the commissural inter-hemispheric connections is to transfer learned associations from one hemisphere to the other (Ebner, 1962; Calford, 1990). Such transfer of plasticity on relatively long time scales might be imagined to preserve the unambiguous representation of the contralateral whisker pad in each S1 hemisphere implied by the traditional models mentioned above. But if bilateral receptive fields (Shuler, 2001) (Manns, 2004) observed in anesthetized preparations remain functional in the behaving animal, the conception of the S1 barrel cortex as an encoder for exclusively contralateral whisker activity will need to be refined or replaced. Previous behavioral experiments in our lab have demonstrated that rats can integrate whisker signals from both sides of the face in a tactile discrimination task (Krupa, 2001b; Shuler, 2002). In fact rats integrated bilateral whisker signals even in an aperture width discrimination task that does not force them to use a bilateral strategy (Oliveira, 2003; Krupa, 2004a), suggesting that bilateral integration is a natural function of the rat whisker system. S1 is the first place where bilateral whisker sensory afferents converge (Waite, 1969; Smith, 1973; Erzurumlu, 1980). Accordingly, Shuler et al (Shuler, 2002) found that intact S1 cortex in both hemispheres was required for successful performance of a bilateral tactile discrimination task. If S1 is an actual site of integration, rather than merely a conduit for sensory information that is integrated at higher processing stages, then the influence of ipsilateral as well as contralateral stimuli should be measurable in S1. It remains unknown whether the bilateral connections between the S1
8 hemispheres are fully functional on millisecond time scales in awake, behaving animals. Our experiments addressed this question using bilateral multi-electrode recordings in S1 of head-immobilized awake rats during controlled bilateral stimulation of columns of whiskers. Following Shuler et al (Shuler, 2001) we looked for evidence of bilateral integration in layer V, which contains the main output neurons of S1.
MATERIALS AND METHODS Head-immobilized awake electrophysiology experiments. A sample of 5 female Long-Evans rats were used in the present experiments. Details of surgery and recording procedures may be found elsewhere (Nicolelis, 1994; Nicolelis, 1997). Briefly, surgery was performed under pentobarbital anesthesia to implant rats with pairs of 16-electrode arrays (2×8) bilaterally in S1. The arrays were positioned stereotactically and by means of tactile stimulation during surgery to record in cortical layer V (-3 mm caudal from bregma, 5.5 mm mediolateral, and ~1.3 mm depth from brain surface). During the same surgery a brass head-post (1/4” width, Small Parts, Inc.) was fixed to the dental-cement head-cap for head-immobilized awake recordings. Rats were given at least 7 days after surgery to recover. Control of whisker stimulus delivery was achieved by combining headimmobilization with a computer-controlled multiple individual-whisker stimulator (Krupa, 2001a) adapted for use on an awake animal. The head-restraint setup is illustrated in Figure 1A. Specifically, each of two stimulator arms was fitted with a
9 thread noose, which could be constricted around an individual whisker or group of whiskers. The tip of one stimulator arm is indicated by an arrow in Figure 1A (left panel). In this photo the stimulator noose is attached to a single whisker; however, for all experiments in this study the stimulator noose was attached to the bottom four large whiskers in a single arc (B-E) (Figure 1A, right panel). Animals were partially acclimated before surgery to restraint in a Plexiglas restraint-tube with mild food deprivation and calorie-dense liquid reward during sessions in the restraint-tube, and wearing light drawstring ‘jackets’ (Fig 1A) to minimize traction against the Plexiglas tube and prevent the subject from using her forepaws to remove the stimulator arms. The whisker stimulator was used to apply controlled 5 ms rostral deflections of the bottom 4 whiskers in a single column on each side of the subject’s face. To lasso 4 whiskers, the noose was first opened wide, then fine forceps were used to pull through the whiskers one at a time before tightening the noose. Two basic protocols were run. In both experiments, stimuli were presented in random order at random inter-trial intervals of between 2-4 seconds. In Experiment 1 the stimulus set consisted of a left column deflection, a right column deflection, and simultaneous bilateral deflections. In Experiment 2 the stimulus set consisted of a left then right column deflection at Inter-Stimulus Intervals (ISIs) of 0, 30, 60, 90, and 120 ms. TEMPO software (Reflective Computing, St. Louis, MO) controlled the stimulus administration and sent time-
10 stamped stimulus events with millisecond accuracy to the file of neural
recordings described below. The Duke University Institutional Animal Use Committee approved all surgical and behavioral methods.
11 Single-unit (SU), multi-unit (MU), and local field potential (LFP) activity were recorded through a Many-neuron Acquisition Processor (MAP; Plexon, Inc., Dallas, TX) using electrode arrays built in-house by Gary Lehew. Units were sorted first online during each recording session, to set each channel’s threshold for recording a spike (digitized at 40 kHz) and graphically defining unit waveforms on each channel with spikes significantly (~2 standard deviations or more) above background noise (Nicolelis, 2003). These online-sorted units, which may include spikes from multiple neurons, are termed MUs. Some units were further sorted offline according to their clustering in the principle component space representing the spike waveforms. Those sorted units that showed a distinct cluster in principle component space from the cluster of noise waveforms, and displayed fewer than 0.1% of inter-spike intervals within a refractory period of 1 ms, are termed SUs. Histology of coronal slices of the experimental animals’ brains confirmed the localization of the electrode arrays in each animal to the infragranular layers of S1. A Nissl stained slice from one animal, containing the track of an electrode, is shown in Figure 1B. Data analysis. Significant firing rate responses of SUs were identified from Post-Stimulus Time Histograms (PSTH) for each session using the standard statistical function built-in to the Nex data analysis software package (Nex Technologies; (Abeles, 1982)) which assumes Poisson spike count statistics to derive confidence intervals on the spike counts in each PSTH bin. To identify
12 significant SU responses we used a bin width of 10 ms and set the confidence level to 99%. To achieve greater precision in response latencies and magnitudes we constructed the MU PSTHs using 1 ms bins. However, despite the greater numbers of spikes generally in the MU PSTHs, bin counts could fluctuate above and below the 99% confidence limits, adding uncertainty to a latency calculation. To circumvent this problem but avoid smoothing the histograms, which would introduce complexities into the determination of confidence intervals, we applied a method based on the statistical distribution of cumulative-summed spike counts (e.g. (Ghazanfar, 2001b; Ushiba, 2002)). This approach takes into account the statistics of all the spiking activity up to a given moment, as opposed to considering each bin independently. Additionally, the method requires no assumption of a particular parametric form of the spiking statistics. To find the onsets of significant deviations from pre-stimulus activity, we calculated the deviation Dn of the post-stimulus cumulative summed spike count from the expected cumulative sum, based on the average prestimulus firing rate , at each post-stimulus bin n: n
Dn = i =1
Here
i
n
Posti
i =1
i
is the expected number of spikes per bin calculated from the base-line,
and Posti is the observed spike count in post-stimulus bin i. To assess the significance of deviations from the expected cumulative sum, an empirical distribution of the cumulative-summed spike count at each time bin before the stimulus was constructed from 1000 bootstrapped samples
13 (with replacement) of the prestimulus spike histogram.
The prestimulus
cumulative sums were counted starting from 200 ms before stimulus onset. This empirical distribution of prestimulus firing was then used to find the post-stimulus bin (if any) at which the cumulative-summed post-stimulus spike count exceeded or was less than 99% of the cumulative spike counts from the baseline distribution (Martinez, 2002). This bin was recorded as the onset of an excitatory or inhibitory response. The response offset was identified as the first zero crossing of the derivative of the cumulative deviation from the baseline spike count, meaning that the cumulative sum was no longer deviating from its expected growth with time. This procedure identified the time at which the PSTH returned to a baseline firing rate. The magnitude of the response was quantified as the number of excess spikes between onset and offset, compared to the baseline expected number, divided by the number of trials. This procedure gives the average number of spikes per trial, over and above the baseline number expected, fired by an MU in response to a particular whisker column stimulation.
RESULTS Bilateral integration by single units in S1 We isolated 361 layer V SUs recorded over 38 sessions in 5 animals. 198 (55%) of the 361 SUs showed significant responses to stimulation on at least one side of the face. 96 (27%) of the 361 SUs responded to both ipsilateral and contralateral stimulation. That is, 49% of those responding to contralateral
14 stimulation also responded to ipsilateral stimulation. Several examples of SU responses to ipsilateral and contralateral stimulation are shown in Figure 2. These results strongly suggest that the firing rates of individual neurons in S1 of awake rats are driven by both ipsilateral and contralateral stimuli. Most SU responses were initially excitatory, occasionally followed by a period of significantly reduced firing (e.g. the SU in row 2 of Fig. 2). However, ‘direct’ inhibitory responses were observed occasionally, such as the SU on the bottom row of Figure 2. The firing of 11 SUs was suppressed by stimulation on either side of the face; while 19 other SUs reduced their firing in response to one unilateral stimulus but not the other. Such inhibitory responses thus accounted for 30 (15%) of the significant SU responses. Sachdev et al (Sachdev, 2000) have reported direct inhibitory responses in S1 of awake rats in response to stimulation of single whiskers. In that study most neurons exhibiting direct inhibition were found to be located over the septal regions outside the central barrel region of the cortex, in layers II,III, IV, and V. They also noted that repetitive stimulation faster than 6 Hz tended to turn the inhibitory responses into excitation. Swadlow (Swadlow, 2003) has described the electrophysiology of layer IV inhibitory interneurons in rabbits, believed to relay feedforward inhibition to excitatory cortical cells. Interestingly, in freely moving rats engaged in an active whisker-dependent discrimination task, 54% of infragranular S1 neurons
15
responded to multi-whisker deflections with purely inhibitory or inhibitory followed by excitatory responses, indicating a greater role for inhibitory coding in the discriminating animals as compared to the passively stimulated animals (Krupa, 2004b).
16 Multiunit responses in S1 to ipsi- and contra-lateral stimuli In the same experiments we recorded 1103 multi-unit channels (MUs) that showed significant modulation in response to bilateral stimulation of 4 whiskers in a single column on each side of the rat’s face. Overall, 984 MUs significantly
17 modulated their firing rates in response to contralateral stimulation, and 585 responded to ipsilateral stimulation. Of those that responded to contralateral stimulation, 50% (491 MUs) responded to ipsilateral stimulation as well, in agreement with the results observed for SUs (49%). Figure 3 shows several examples of MU responses to ipsilateral and contralateral whisker stimulation. Figure 4a shows the distribution of response latencies corresponding to contralateral, ipsilateral, and synchronous bilateral stimulation. The average
response latencies for contralateral, ipsilateral, and bilateral stimuli were 14.67 ± 0.02 ms, 23.78 ± 0.04 ms, and 15.07 ± 0.02 ms (standard errors; standard deviations 19, 24, and 17 ms, respectively). For comparison, response latencies
18 in S1 SUs in pentobarbital anesthetized rats, for contralateral and ipsilateral whisker stimuli similar to those used in the present study, were 11 ± 3.4 ms and 23 ± 4.7 ms, respectively (Shuler, 2001), consistent with the present study. One potential confound in interpreting the bilaterally evoked responses we observed is the possibility that a startle reaction due to an ipsilateral stimulus might lead to a movement of contralateral whiskers against the (stationary) stimulator noose, leading to a response due to contralateral self-stimulation. Such a response could be misinterpreted as an evoked ipsilateral response. This scenario can be ruled out as follows: First, sessions with the stimulator attached on only one side of the face continue to show ipsilateral responses. Four examples are shown in Figure 5. This rules out the possibility that false ipsilateral responses resulted from contralateral self-stimulation against the stimulator noose. That leaves the possibility that a contralateral movement “in air” could generate a false ipsilateral response in S1.
19
This is possible in principle because Fee et al (Fee, 1997) found that whisking movements modulate spiking activity in S1. However, they found that the whiskers’ movement signal proceeded to S1 by “direct sensory activation,” implying that a signal due a motor reaction would take at least the time of the motor reaction plus the time of the sensory relay to reach S1. The timing of ipsilateral responses is inconsistent with what we know about the latency of a startle movement, the fastest possible motor response the rats could make. Using the same type of head restraint in a separate series of experiments (Wiest MC, 2003), we were able to record occasional startle responses to whisker stimulation. These manifested themselves as a distinct early peak (~30 ms) in
20 the latency distribution of licking responses to stimulation. Since a fast motor reaction can only occur at least 30 ms after stimulation, a false ipsilateral response due to a motor reaction would be observed only at longer latencies still. The average latency of ipsilateral responses was 24 ms, and most of the significant ipsilateral responses were observed at latencies less than 30 ms. Therefore those ipsilateral responses we observed could not have been due to a motor reaction.
Finally, the ipsilateral responses we observed are similar to
those recorded in anesthetized animals, where whisker movements did not occur. Figure 4b shows the distribution of response magnitudes for the three stimulation conditions, quantified in terms of excess spikes per trial. The average response magnitudes were 0.82 ± 0.02, 0.64 ± 0.04, and 0.83 ± 0.03 spikes above background per trial, in response to contralateral, ipsilateral, and bilateral stimulation, respectively (standard deviations 0.76, 0.84, and 0.84 spikes per trial). Direct inhibitory responses were rare, accounting for only 32 of 984 (3%) responses to contralateral stimulation, 30 of 585 (5%) responses to ipsilateral stimulation, and 35 of 1103 (3%) responses to simultaneous bilateral stimulation. The lower incidence of inhibitory MU responses (3-5%) as compared to SU responses (15%) suggests that inhibitory modulation within an MU is sometimes masked by excitatory responding from other neurons in the same MU.
21 Integration of synchronous bilateral stimuli in S1 The observation that single and multi-units respond to ipsilateral whisker stimulation clearly indicates that bilateral stimuli are integrated in rat S1. However, the average neuronal response to synchronous bilateral stimulation (0.83 ± 0.03 spikes per trial) is not significantly different from the response elicited by contralateral stimulation alone (0.82 ± 0.02 spikes per trial). To go beyond such average measures, Figure 6a shows a histogram of the ratios of bilateral response magnitude to contralateral response magnitude for each MU for all rats. The distribution of ratios is peaked near one, but many are less than or greater than one, indicating that ipsilateral stimulation can result in either inhibitory or excitatory modulation of the contralateral response. To assess whether the facilitative or suppressive effects of ipsilateral stimulation on the contralateral response were statistically significant for the different MUs, we estimated the statistical uncertainty in the response magnitudes as the squareroot of the total excess spike count in the response window (Poisson statistics). Then, at the 95% confidence level, the bilateral response was significantly lower than the contralateral response (i.e. suppressive modulation) for 265 MUs (out of 345 MUs whose ratio was < 1), while the bilateral response was larger compared to the contralateral response (i.e. facilitative modulation) for 392 MUs (out of 455 whose ratio was > 1). An additional 194 MUs (not depicted in the histogram of
22
Fig. 6a) responded significantly to bilateral but not to contralateral stimulation alone. For these units also, the ipsilateral stimulation may also be considered to have increased, or facilitated, the contralateral response.
23 Given that ipsilateral stimulation can significantly modify the response to contralateral stimulation, we also asked how the two responses added. Figure 6b plots the distribution of the ratios of bilateral response magnitudes to the sum of response magnitudes for each side, to assess whether bilateral whisker signals are summed sub- or supra-linearly. The trend is toward sub-linear summation, but many MUs showed supra-linear integration. Specifically, 552 MUs were significantly sub-linear (out of 663 with ratios < 1, 2 sigma confidence level), and 254 MUs were significantly supra-linear (out of 313 with ratios > 1). To these supra-linear units may be added 127 more that responded significantly to bilateral but not to contralateral or ipsilateral stimulation alone. Thus, in awake S1, supra-linear summation of synchronous bilateral inputs is almost as common as sub-linear summation.
24
Temporal integration of asynchronous bilateral stimuli in S1 The above results demonstrate that simultaneous ipsilateral and contralateral whisker signals interact in each S1 hemisphere. To characterize the modulation of these bilateral interactions by the interstimulus interval (ISI) between the left and right stimuli, we recorded S1 responses at varying ISIs. The net excitatory neuronal responses at each ISI are shown in Figure 7. The significant dip at 30 ms shows that response suppression dominates at ISIs around 30 ms. This is another confirmation that ipsilateral stimulation can modulate the spiking activity in S1 on behaviorally relevant time scales.
25 DISCUSSION The main result of the present study is that neurons in S1 of awake rats integrate bilateral inputs over more than a hundred milliseconds. This result suggests the potential importance of bilateral integration in S1 for whisker-guided behavior. Moreover, this finding further challenges the traditional conception of the organization of rat somatosensory cortex into independent representational “modules” in each hemisphere. In awake, immobilized rats, neural responses in S1 to punctate whisker stimuli were similar in terms of response latencies to those recorded in anesthetized animals (Shuler, 2001). In terms of bilateral integration, 49% of SUs responding to contralateral whisker column stimulation also responded to ipsilateral stimulation, compared to 62% in the anesthetized study (Marshall Shuler, personal communication—the published report (Shuler, 2001) contains the percentage that responded to either of two whisker columns (73%)). In these terms, the degree of bilateral integration is similar in the anesthetized and waking S1 (but see below).
Principles of spatiotemporal integration of multiple unilateral stimuli Previous studies of spatiotemporal integration of multi-whisker inputs in the rat whisker sensory system have found evidence of both inhibitory and excitatory interactions. Extra- and intra-cellular studies of S1 responses to combinations of unilateral stimuli found inhibitory interactions dominating between pairs of neighboring whiskers (Simons, 1985; Brumberg, 1996) (Higley,
26 2003) at ISIs around 10-20 ms. (Moore (Moore, 2004) has recently suggested that these relatively slow temporal interactions are supplemented by resonances at higher frequencies that could account for rats’ texture discrimination ability.) Pairs of synchronous stimuli were found to sum linearly (Simons, 1989; Mirabella, 2001), whereas (Mirabella, 2001) reported that synchronous stimulation of 4 whiskers usually resulted in sub-linear summation. These studies focused on responses of neurons in middle and supragranular layers of S1 in immobilized or anesthetized rats. (Brumberg, 1996) suggested that surround inhibition could function for contrast enhancement in layer IV and to “effectively increase the ‘principal whiskerness’ of the cortical column,” in line with a model of barrel columns as “single-whisker processing units.” (Mirabella, 2001) suggested that inhibitory interactions could serve to avoid response saturation and maintain a dynamic response range under naturalistic multiwhisker inputs.
Recent recordings in freely moving rats engaged in an
active whisker-dependent discrimination task have suggested (Krupa, 2004b) that inhibitory response modes in all layers are significantly more common in the discriminating animals as compared to passively stimulated animals. In undrugged rats, similar temporal integration of multiple unilateral inputs in S1 layer V was reported in (Fanselow, 1999). SU responses to the second of a pair of unilateral electrical stimulations of the facial afferent infraorbital nerve in awake rats were suppressed most strongly at ISIs around 25-50 ms. This suppression was most pronounced during rats’ awake quiet behavioral state as compared to the whisking and active exploring states.
27 However, not all unilateral interactions are inhibitory: synchronous multiple-whisker deflections have also been reported to sum supralinearly. Shimegi et al (Shimegi, 2000) reported supra-linear summation of neighboring single-whisker stimuli in 37% of S1 neurons, primarily by cells in supragranular layers. They noted that enhanced responses were often selective for specific combinations of stimulus features such as whisker position, angle of deflection, and relative timing of multiwhisker stimuli, supporting the idea that S1 cells code for spatiotemporal patterns of whisker stimulation. Ghazanfar and Nicolelis (Ghazanfar, 1997) found that 81% of SUs in S1 layer V showed supralinear responses to simultaneous stimulation of triplets of whiskers as compared to single whisker stimulations. They reported a bias toward supralinear summation of within-column triplets as compared to within-row triplets. They pointed out that their findings were consistent with the reports of sublinear summation of unilateral stimuli, since sublinear spatial summation was reported in studies of responses in granular and supragranular layers to pairs of neighboring whiskers, as opposed to the layer V responses to stimulation of triplets of relatively distant whiskers in their study. Together these findings imply dynamic excitatory as well as inhibitory nonlinear interactions among multiple whisker stimuli, which could function to encode spatiotemporal patterns of unilateral whisker stimulations.
Principles of spatiotemporal integration of bilateral stimuli (Shuler, 2001) found primarily inhibitory interactions between pairs of bilateral whisker column stimuli in S1 layer V of anesthetized rats, with greatest
28 suppression occurring at ISIs of 30-90 ms. The responses in the anesthetized animals increased monotonically on average from ISIs of 60 ms up to 180 ms. In our study, the net excitatory responses increased from ISIs of 30 ms to ISIs of 60 or 90 ms, but were relatively smaller for ISIs of 120 ms (Figure 6a). Thus bilateral integration appears somewhat more dynamic in the waking animals. Furthermore, the anesthetized study found only a marginal difference between the responses to synchronous bilateral stimuli as compared to contralateral stimulation alone. This was expected because contralateral stimuli universally suppressed the responses to ipsilateral stimuli. Since the latency for the contralateral response is shorter, the ipsilateral excitatory response is suppressed before it occurs. We also found no significant difference between the average response to contralateral and bilateral stimulation. However, in the awake rats many individual units showed significant supralinear spatial summation (Figure 5b) as well as sublinear summation of synchronous bilateral inputs. This difference suggests that the profile of excitatory and inhibitory contributions to ongoing S1 responses to bilateral stimuli is substantially altered in the anesthetized animal. The heterogeneity of responses we observed in the awake animals could function to enhance discrimination of patterns of whisker stimulation (Maass, 2002). We conclude that individual neurons in the cortical whisker representation in the awake rat are influenced by nonlinear combinations of inputs from multiple whiskers on both sides of the face, over time spans up to and greater than a hundred milliseconds. Moreover, the spatiotemporal integration of these inputs
29 can depend strongly on the animal’s behavioral state (Fanselow, 1999; Moore, 1999; Nicolelis, 2002; Castro-Alamancos, 2004; Krupa, 2004b; Moore, 2004). The fact that rats can perform a variety of unilateral and bilateral whisker tasks points to the possibility that the degree of tactile bilateral integration through the corpus callosum can be modulated according to the animal’s current behavioral context. This would amount to using different sensory codes for different behavioral contexts. If this is the case, mechanisms for adaptive control of bilateral connectivity revealed in the whisker system may provide insight into dynamical mechanisms for adjusting the functional connectivity within and among other brain areas as well.
Acknowledgements: This work was supported by NIH grants DE11451 and DE13810. References Abeles M (1982) Quantification, smoothing, and confidence limits for single-units' histograms. J Neurosci Methods 5:317-325. Armstrong-James M (1995) The nature and plasticity of sensory processing within adult rat barrel cortex. In: Cerebral Cortex Volume 11 The Barrel Cortex of Rodents (Jones E, and Diamond, IT, ed), pp 333-373. New York: Plenum Press. Armstrong-James M, Fox, K, Das-Gupta, A (1992) Flow of excitation within rat barrel cortex on striking a single vibrissa. Journal of Neurophysiology 68:1345-1358. Brumberg J, Pinto, DJ, Simons, DJ (1996) Spatial gradients and inhibitory summation in the rat whisker barrel system. Journal of Neurophysiology 76:130-140. Calford MaT, R (1990) Interhemispheric transfer of plasiticity in the cerebral cortex. Science 249:805-807. Castro-Alamancos M (2004) Absence of rapid sensory adaptation in neocortex during information processing states. Neuron 41:455-464. Cauller L, Clancy, B, Connors, B (1998) Backward cortical projections to primary somatosensory cortex in rats extend long horizontal axons in layer 1. J Comp Neurol 390:297-310.
30 Ebner F, Myers, RE (1962) Corpus callosum and the interhemispheric transmission of tactual learning. J Neurophysiol 25:380-391. Erzurumlu R, Kilackey, H (1980) Diencephalic projection of the brainstem trigeminal complex in the rat. Neuroscience 5:1891-1901. Fanselow E, and Nicolelis, MAL (1999) Behavioral modulation of tactile responses in the rat somatosensory system. Journal of Neuroscience 19:7603-7616. Fee M, Mitra, PP, Kleinfeld, D (1997) Central versus peripheral determinants of patterned spike activity in rat vibrissa cortex during whisking. J Neurophysiol 78:1144-1149. Ghazanfar A, and Nicolelis, MAL (1997) Nonlinear processing of tactile information in the thalamocortical loop. Journal of Neurophysiology 78:506-510. Ghazanfar A, and Nicolelis, MAL (1999) Spatiotemporal properties of layer V neurons of the rat primary somatosensory cortex. Cerebral Cortex 9:348361. Ghazanfar A, and Nicolelis, MAL (2001a) The structure and function of dynamic cortical and thalamic receptive fields. Cerebral Cortex 11:183-193. Ghazanfar A, Krupa, DJ, and Nicolelis, MAL (2001b) Role of cortical feedback in the receptive field structure and nonlinear response properties of somatosensory thalamic neurons. Exp Brain Res 141:88-100. Ghazanfar A, Stambaugh, CR, Nicolelis, MAL (2000) Encoding of tactile stimulus location by somatosensory thalamocortical ensembles. J Neurosci 20:3761-3775. Hayama T, and Ogawa, H (1997) Regional differences of callosal connections in the granular zones of the primary somatosensory cortex in rats. Brain Research Bulletin 43:341-347. Higley M, Contreras, D (2003) Nonlinear integration of sensory responses in the rat barrel cortex: An intracellular study in vivo. J Neurosci 23:1019010200. Hutson K, and Masterton, RB (1986) The sensory contribution of a single vibrissa's cortical barrel. J Neurophysiol 56:1196-1223. Kim U, and Ebner, FF (1999) Barrels and septa: separate circuits in rat barrel field cortex. J Comp Neurol 408:489-505. Koralek K, Olavarria, J, Killackey, H (1990) Areal and laminar organization of corticocortical projections in the rat somatosensory cortex. J Comp Neurol 299:133-150. Krupa D, Brisben, AJ, Nicolelis, MAL (2001a) A multi-channel whisker stimulator for producing spatiotemporally complex tactile stimuli. Journal of Neuroscience Methods 104:199-208. Krupa D, Matell, MS, Brisben, AJ, Oliveira, LM, Nicolelis, MAL (2001b) Behavioral properties of the trigeminal somatosensory system in rats performing whisker-dependent tactile discriminations. Journal of Neuroscience 21:5752-5763.
31 Krupa D, Nicolelis, MAL (2004a) Psychophysical analysis of unilateral and bilateral whisker stimulation in rat trigeminal somatosensory system. In: Society for Neuroscience Abstracts. San Diego, CA. Krupa D, Wiest, MC, Shuler, MG, Laubach, M, Nicolelis, MAL (2004b) Layer specific somatosensory cortical activation during active tactile discrimination. Science in press. Maass W, Natschlager, T, and Markram, H (2002) Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural Computation 14:2531-2560. Manns I, Sakmann, B, Brecht, M (2004) Sub- and suprathreshold receptive field properties of pyramidal neurones in layers 5A and 5B of rat somatosensory barrel cortex. J Physiol 556:601-622. Martinez W, and Martinez, AR (2002) Computational Statistics Handbook with Matlab. Boca Raton: Chapman & Hall/CRC. Mirabella G, Battiston, S, and Diamond, M (2001) Integration of multiple-whisker inputs in rat somatosensory cortex. Cerebral Cortex 11:164-170. Moore C (2004) Frequency-dependent processing in the vibrissa sensory system. J Neurophysiol 91:2390-2399. Moore C, Nelson, SB, Sur, M (1999) Dynamics of neuronal processing in rat somatosensory cortex. Trends Neurosci 22:513-520. Nicolelis M (1997) Reconstructing the Engram: Simultaneous, Multisite, Many Single Neuron Recordings. Neuron 18:529-537. Nicolelis M, Chapin, JK (1994) Spatiotemporal structure of somatosensory responses of many-neuron ensembles in the rat ventral posterior medial nucleus of the thalamus. Journal of Neuroscience 14:3511 - 3532. Nicolelis M, Dimitrov, D, Carmena, JM, Crist, R, Lehew, G, Kralik, JD, and Wise, SP (2003) Chronic, multisite, multielectrode recordings in macaque monkeys. PNAS 100:11041-11046. Nicolelis M, Fanselow, EE (2002) Thalamocortical optimization of tactile processing according to behavioral state. Nat Neurosci 5:517-523. Olavarria J, van Sluyters, R, Killackey, H (1984) Evidence for the complementary organization of callosal and thalamic connections withing rat somatosensory cortex. Brain Research 291:364-368. Oliveira L, Wiest, MC, Krupa, DJ, Nicolelis, MAL (2003) Integration of bilateral whisker stimuli in primary somatosensory cortex of awake rats. In: Society for Neuroscience Abstracts. New Orleans, LA. Petersen R, Panzeri, S, Diamond, ME (2002) Population coding in somatosensory cortex. Current Opinion in Neurobiology 12:441-447. Pidoux B, Verley, R (1979) Projections on the cortical somatic I barrel subfield from ipsilateral vibrissae in adult rodents. Electroencephalogr Clin Neurophysiol 46:715-726. Rema V, Ebner, FF (2003) Lesions of mature barrel field cortex interfere with sensory processing and plasticity in connected areas of the contralateral hemisphere. J Neurosci 23:10378-10387.
32 Sachdev R, Sellien, H, Ebner, FF (2000) Direct inhibition evoked by whisker stimulation in somatic sensory (SI) barrel field cortex of the awake rat. J Neurophysiol 84:1497-1504. Schubert D, Staiger, JF, Cho, N, Kotter, R, Zilles, K, Luhmann, HJ (2001) Layerspecific intracolumnar and transcolumnar functional connectivity of layer V pyramidal cells in rat barrel cortex. J Neurosci 21:3580-3592. Shimegi S, Takafumi, A, Takehiko, I, and Sato, H (2000) Physiological and anatomical organization of multiwhisker response interactions in the barrel cortex of rats. Journal of Neuroscience 20:6241-6248. Shin H-C, Won, C-K, Jung, S-C, Oh, S, Park, S, Sohn, J-H (1997) Interhemispheric modulation of sensory transmission in the primary somatosensory cortex of rats. Neuroscience Letters 230:137-139. Shuler M, Krupa, DJ, Nicolelis, MAL (2001) Bilateral integration of whisker information in the primary somatosensory cortex of rats. Journal of Neuroscience 21:5251-5261. Shuler M, Krupa, DJ, Nicolelis, MAL (2002) Integration of bilateral whisker stimuli in rats: Role of the whisker barrel cortices. Cerebral Cortex 12:86-97. Simons D (1985) Temporal and spatial integration in the rat S1 vibrissa cortex. Journal of Neurophysiology 54:615-635. Simons D, Carvell, GE (1989) Thalamocortical response transformations in the rat vibrissa/barrel system. Journal of Neurophysiology 54:615-635. Smith R (1973) The ascending fiber projections from the principal sensory trigeminal nucleus in the rat. Journal of Physiology (London) 148:423-446. Swadlow H (2003) Fast-spike interneurons and feedforward inhibition in awake sensory neocortex. Cerebral Cortex 13:25-32. Ushiba J, Tomita, Y, Masakado, Y, Komune Y (2002) A cumulative sum test for a peri-stimulus time histogram using the Monte Carlo method. J Neurosci Methods 118:207-214. Waite P (1969) Organization of whisker responses in the rat thalamus. Journal of Physiology (London) 202:51-53. White E, DeAmicis, R (1977) Afferent and efferent projections of the region in mouse SmI cortex which contains the posteromedial barrel subfield. J Comp Neurol 175:455-482. Wiest MC N, MAL (2003) Behavioral detection of tactile stimuli during 7-12 Hz cortical oscillations in awake rats. Nature Neuroscience 6:913-914.
33 Figure Captions Figure 1 (A) Left panel: Head-immobilized whisker stimulation and multi-electrode recording setup. A rat is shown in the Plexiglas restraint tube, with her head fixed to the pin vise. The connector for an implanted 2×8 multi-electrode array is also visible at the top of her head. The arrow indicates one of the computer controlled whisker-stimulator arms, visible as a thin metal tube extending from the right of the picture to the animal’s face. In the photograph the stimulator is shown attached to a single whisker. In all experiments described in this study, the stimulator was used to apply 5 ms, ~1 mm deflections of 4 whiskers from a single arc on each side of the head. Right panel: Close-up of a whiskerstimulator noose around 4 whiskers from an arc. Visible on the other side of the subject’s head is the improved stimulator design (also shown in the left panel) with thread internal to the metal tube. (B) A Nissl stained slice from one animal showing an electrode track down to layer V (indicated by the arrow). Figure 2 Single units (SUs) respond to contralateral and ipsilateral whisker stimuli. Each row shows PSTHs of one layer V single unit in response to contralateral (left column) and ipsilateral (right column) whisker column stimuli delivered at t=0. Spike counts are based on 10 ms bins. The dashed lines show the 99% confidence limit for the spike counts, based on the pre-stimulus baseline firing rate. The bottom row is an example of an SU whose firing was suppressed by whisker stimulation, a relatively rare type of response. Figure 3 Multi units (MUs) respond to contralateral and ipsilateral whisker stimuli. Each row shows PSTHs of one layer V multi unit in response to contralateral (left column) and ipsilateral (right column) whisker column stimuli delivered at t=0, in 1 ms bins. The dashed lines show the 99% confidence limit for the spike counts, based on the pre-stimulus baseline firing rate.
Figure 4 MU response onset latencies and magnitudes. (A) Histograms of response onset latencies (in milliseconds) for bilateral (red trace), contralateral (blue trace), and ipsilateral (green trace) whisker column stimulation. Response onsets to ipsilateral stimulation tended to be longer than to contralateral or bilateral stimulation, consistent with a longer synaptic path including interhemispheric connections in the corpus callosum. The range of latencies shown in the figure includes 94%, 95%, and 88% of responses to bilateral, contralateral, and ipsilateral stimulation, respectively. This range was chosen for clarity in displaying the main peaks of the distributions; but all responses were included in the calculation of average latencies. The data are combined from 5 rats. (B) Histograms of response magnitudes (in spikes above baseline per trial) for
34 bilateral (red trace), contralateral (blue trace), and ipsilateral (green trace) whisker column stimulation. Bilateral stimulation tended to evoke a significantly smaller response compared to unilateral stimulation, although bilateral stimulation was more reliable in generating a significant response. The range of magnitudes shown includes 100%, 76%, and 75% of responses to bilateral, contralateral, and ipsilateral stimulation, respectively; so that the distribution peaks are clearly visible. Nevertheless, all responses were included in calculations of average magnitudes and integration factors (Figure 6). The data are pooled from 5 rats.
Figure 5 Sessions with the stimulator attached on only one side of the face continue to show ipsilateral responses. PSTHs are shown for four MUs recorded at different electrodes, showing significant responses to ipsilateral whisker-column stimulation. The dashed line marks the 99% confidence limit. These responses from unilateral sessions rule out the possibility that artifactual ipsilateral responses arose from a motor response that led to self-stimulation of contralateral whiskers against the stimulator arm. Figure 6 Heterogeneous integration of synchronous bilateral whisker stimuli. (A) Histogram of the ratio of response magnitudes to bilateral vs. contralateral whisker column stimulation, for all MUs recorded in 5 rats. Many MUs response ratios fall above or below 1, indicating that synchronous ipsilateral stimulation can result in either inhibitory (1) modulation of the contralateral response. (B) Histogram of the ratio of response magnitudes to bilateral stimulation vs. the sum of contralateral and ipsilateral response magnitudes, to assess whether bilateral stimuli are integrated sub- or supralinearly. A ratio greater than 1 indicates supralinear summation of the two whisker column stimuli. Most MUs showed sublinear summation of bilateral stimuli, but almost as many were supralinear. Figure 7 Temporal integration of asynchronous bilateral whisker stimuli. The net excitatory response magnitude, in spikes per trial above baseline, is plotted for varying intervals between a left and a right whisker column stimulus. Response suppression is dominant at intervals of ~30 ms. Data shown are pooled from 5 animals.