Megevand et al NeuroImage 2008.pdf

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy NeuroImage 42 (2008) 591–602

Contents lists available at ScienceDirect

NeuroImage j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / y n i m g

A mouse model for studying large-scale neuronal networks using EEG mapping techniques Pierre Mégevand a,b,⁎, Charles Quairiaux a, Agustina M. Lascano a,b, Jozsef Z. Kiss a, Christoph M. Michel a,b,⁎ a

Fundamental Neuroscience Department, Geneva University Medical School, Rue Michel-Servet 1, 1211 Geneva 14, Switzerland Functional Brain Mapping Laboratory, Neurology Clinics, Clinical Neuroscience Department, Geneva University Hospital and Medical School, Rue Micheli-du-Crest 24, 1211 Geneva 14, Switzerland

b

a r t i c l e

i n f o

Article history: Received 7 February 2008 Revised 17 April 2008 Accepted 7 May 2008 Available online 21 May 2008

a b s t r a c t Human functional imaging studies are increasingly focusing on the identification of large-scale neuronal networks, their temporal properties, their development, and their plasticity and recovery after brain lesions. A method targeting large-scale networks in rodents would open the possibility to investigate their neuronal and molecular basis in detail. We here present a method to study such networks in mice with minimal invasiveness, based on the simultaneous recording of epicranial EEG from 32 electrodes regularly distributed over the head surface. Spatiotemporal analysis of the electrical potential maps similar to human EEG imaging studies allows quantifying the dynamics of the global neuronal activation with sub-millisecond resolution. We tested the feasibility, stability and reproducibility of the method by recording the electrical activity evoked by mechanical stimulation of the mystacial vibrissae. We found a series of potential maps with different spatial configurations that suggested the activation of a large-scale network with generators in several somatosensory and motor areas of both hemispheres. The spatiotemporal activation pattern was stable both across mice and in the same mouse across time. We also performed 16-channel intracortical recordings of the local field potential across cortical layers in different brain areas and found tight spatiotemporal concordance with the generators estimated from the epicranial maps. Epicranial EEG mapping thus allows assessing sensory processing by large-scale neuronal networks in living mice with minimal invasiveness, complementing existing approaches to study the neurophysiological mechanisms of interaction within the network in detail and to characterize their developmental, experience-dependent and lesioninduced plasticity in normal and transgenic animals. © 2008 Elsevier Inc. All rights reserved.

Introduction The complex sensory, motor and cognitive functions of the cerebral cortex are mediated by large-scale networks linking groups of neurons in separate cortical areas into functional entities (Bressler, 1995; Fuster, 2006; Mesulam, 1998). Plasticity of these networks is thought to be crucial during development, learning, and functional recovery following brain lesions (Callan et al., 2003; Price and Friston, 2002; Sigman et al., 2005; Tombari et al., 2004). Structural and functional neuroimaging in humans has greatly added to the current understanding of large-scale network anatomy and physiology. However, the precise neuronal and molecular structure of large-scale brain networks, the mechanisms of communication between the network modules, and the exact temporal

⁎ Corresponding authors. Fax: +41 22 379 5452. E-mail address: [email protected] (C.M. Michel). 1053-8119/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2008.05.016

structure of information flow in these networks are still poorly understood (Bressler and Tognoli, 2006; Fingelkurts and Fingelkurts, 2006; McIntosh, 2000; Mesulam, 1990; Schnitzler and Gross, 2005). One way to study information processing in large-scale brain networks in humans are the evoked or event-related potentials (Shah et al., 2004), which are characterized by a series of components that reflect different stages or steps of information processing performed by the network (see e.g. Linden, 2007 for a recent review). By recording these evoked responses simultaneously from multiple sensors distributed over the whole scalp and applying source localization algorithms to these multichannel data, the putative brain areas involved in the processing of the stimuli can be identified and the temporal dynamics of the network can be studied (Michel et al., 1999, 2001, 2004). However, the spatial resolution of these recordings is limited and systematic studies on the neuronal and molecular basis of network functioning, on early post-natal network development, lesion-induced plasticity or pharmacological effects are not possible.

Author's personal copy 592

P. Mégevand et al. / NeuroImage 42 (2008) 591–602

Recording setup and procedure

A mouse model of large-scale network function would open the possibility to investigate these questions. Existing electrophysiological (Benison et al., 2007) and optical (Ferezou et al., 2006) approaches assess the function of local cortical areas in the living animal with high spatial and temporal resolution. However, their invasive nature makes them impractical for studying network function repeatedly in the same animal, a prerequisite for development and plasticity studies. Implanted chronic electrodes also have high local spatial and temporal resolution (Buzsaki, 2004; Hollenberg et al., 2006), but research using these methods often focuses on the detailed description of a limited cortical area, overlooking the large scale of global brain networks. On the other hand, the temporal resolution of functional magnetic resonance imaging (fMRI; de Zwart et al., 2005) and other methods that measure vascular or metabolic correlates of neuronal function is insufficient to describe the fast temporal dynamics – on the order of milliseconds – that characterize network activity. It would therefore be of interest to develop a minimally invasive method for repeatedly assessing largescale network function in mice with millisecond temporal resolution. We previously showed that epicranial recording of the electroencephalogram (EEG) is a minimally invasive approach for repeatedly assessing somatosensory evoked potentials (SEP) in anesthetized mice (Troncoso et al., 2000, 2004). Here, in order to describe more completely the large-scale somatosensory cortical network, we simultaneously recorded the response of the brain to mechanical stimulation of the mystacial vibrissae using 32 electrodes regularly distributed over the skull bones. We considered the data as topographic representations of the surface-recorded potential field and analyzed the spatiotemporal dynamics of these potential maps using quantitative, reference-free methods originally developed in the human EEG, MEG and event-related potential studies mentioned above (Michel et al., 1999, 2001, 2004). In this paper we present the results of our studies on the feasibility, stability, and reproducibility of this mapping approach by comparing the data between animals and within the same animal measured twice. Because of the electromagnetic inverse problem, the location of the sources that generated the surface maps cannot be determined unambiguously (Fender, 1987). Even if the lissencephalic cortex of the mouse produces simpler extracranial field potentials than those observed in humans, it is impossible to resolve whether the positive and negative potentials of a given map reflect the activity of two separate brain regions or volume-conducted activity from one region. Therefore, we complemented the epicranial SEP mapping with 16channel intracortical recordings of the local field potential (LFP) in several cortical areas, applying current source-density (CSD) analysis to describe the local profile of whisker-evoked activity across cortical layers (Mitzdorf, 1985).

An array of 32 stainless steel electrodes (500-μm diameter) held by a stereotaxic manipulator was lowered into contact with the skull surface. The electrodes were kept in position in the horizontal plane by a perforated Plexiglas grid and were freely movable in the vertical axis. The tip of the electrodes was immersed in EEG paste (EC2, Grass Technologies, West Warwick, RI) before making contact with the skull surface. Electrode impedance was around 50 kΩ. Electrodes were not aligned in a rectangular grid; rows were offset so that the electrodes formed equilateral triangles in the horizontal plane (distance between electrode rows 1.33 mm), giving a constant 1.54-mm distance between an electrode and all its immediate neighbors (see Fig. 1A). Electrode coordinates were (in mm, anteroposterior/lateral with respect to bregma):+ 2.67/ 1.54, +2.67/0, + 1.33/2.31, + 1.33/0.77, 0/3.08, 0/1.54, 0/0, −1.33/ 3.85, − 1.33/2.31, −1.33/0.77, − 2.67/4.62, −2.67/3.08, − 2.67/ 1.54, − 4/3.85, − 4/2.31, − 4/0.77, − 2.67/0 (reference electrode), + 4/0 (ground electrode). Signals were amplified (1000× gain), filtered (0.5-Hz high-pass, 1500-Hz low-pass), digitized (16-bit resolution, 5-kHz sampling rate), displayed online and stored on hard drive using a 64-channel conventional human EEG system (hardware: EAAS-111.64, M&I, Prague, Czech Republic; software: EASYS2, Neuroscience Technology Research, Prague, Czech Republic). At the end of the first recording session, the skull surface was cleaned, the skin was sutured and mice were returned to their cage. SEP were recorded again two weeks later in the same mice under identical conditions.

Methods

Intracortical LFP recordings

Fifteen male C57BL/6J mice aged 3–6 months and housed in individual cages were used for the epicranial SEP recordings and twelve for the intracortical LFP recordings. All procedures were in accordance with Swiss laws and were approved by the Ethics Committee on Animal Experimentation of Geneva University Medical School and by the Veterinary Office of Geneva.

One or two craniotomies were performed in the parietal (6 mice) or the parietal and frontal bones (6 mice). The dura mater was left intact and covered with warm NaCl 0.9% in water. A linear 16-electrode probe with 50-μm inter-electrode spacing (NeuroNexus Technologies, Ann Arbor, MI) was inserted through the dura mater into the cortex perpendicular to its surface. The 14 uppermost electrodes were used to record the

Mice were anesthetized with isoflurane in 20% oxygen/80% air and placed in a stereotaxic frame. Light anesthesia during recordings was maintained with 0.5–0.8% isoflurane, so as to not completely suppress the hindlimb withdrawal and corneal reflexes. In 13 mice, deep anesthesia (1.5% isoflurane), suppressing hindlimb withdrawal and corneal reflexes completely, was also used to test the sensitivity of our method to experimental manipulation. Body temperature was maintained at 37 °C by a heating blanket connected to a rectal probe. The skin overlying the skull was anesthetized with bupivacaine, incised and retracted. The skull surface was cleaned and dried. Whisker stimulation A custom-made computer-controlled electromechanical device was used for stimulating whiskers (Troncoso et al., 2000). Stimuli consisted of 500-μm back-and-forth deflections (initial direction downwards and backwards) with 1-ms rise time, applied to all whiskers on each side of the snout, 1 cm away from the face. Two hundred stimuli were administered with a 2011-ms inter-stimulus interval. Epicranial SEP recordings

Author's personal copy P. Mégevand et al. / NeuroImage 42 (2008) 591–602

593

or stored on hard drive for post-hoc analysis using custommade programs designed in VEE PRO 6 (Agilent Technologies, Santa Clara, CA). At the end of the recording session, a lesion was made at the sites of electrode penetration by inserting a 500μm-diameter solid needle 1 mm deep into the cortex, mice were killed by an overdose of pentobarbital and the brains were processed for histology as described (Troncoso et al., 2004). Data display and analysis Data analysis was performed using the Cartool software (D. Brunet, Geneva University Hospital and Medical School, Geneva, Switzerland; http://brainmapping.unige.ch/Cartool. php) and custom-made programs designed in MATLAB (The MathWorks, Natick, MA). Epicranial SEP signals were digitally filtered between 1 and 500 Hz and computed against the average reference before subsequent analyses. Responses to 200 stimuli were averaged to obtain the SEP/LFP in individual mice. Responses to individual stimuli were visually inspected offline and responses contaminated by electromagnetic noise artifacts were excluded from the average. Baseline correction was applied using the 50-ms pre-stimulus period as baseline. The grand average SEP/LFP was averaged from the SEP/LFP of individual mice. The period of analysis lasted from 5 to 60 ms post-stimulus. Epicranial SEP maps

Fig. 1. Epicranial somatosensory evoked potential mapping in response to whisker stimulation. A. The grand average waveforms of the SEP recorded from multiple epicranial electrodes in response to left-sided whisker stimulation during the first recording session are superimposed on a template MRI brain surface. Interval: 0 to 60 ms post-stimulus. Average of 15 mice. Average number of responses to individual stimuli removed due to artifact contamination: 4.3 out of 200 (range 0–28). B. Superimposed grand average waveforms. C. The temporal extent of the 6 component maps identified by the cluster analysis as optimally summarizing the grand average map series appears as colored segments on the global field power trace. The global field power is the spatial standard deviation of all voltage values at each time point and represents the strength of the electrical field. D. The topography of the 6 component maps is color-coded (red, positive voltage; blue, negative voltage) over a MRI brain surface. E. Latencies of best correlation of each component map with the map series of individual mice (mean and SD). Latency differences between pairs of successive maps were all highly significant (all p values b 10− 4).

LFP, referenced to an electrode attached to the scalp. Signals were amplified (5000× gain) and filtered (1-Hz high pass, 500Hz low-pass) using a custom-built amplifier (Troncoso et al., 2000), digitized (16-bit resolution, 2-kHz sampling rate; DT3004, Data Translation, Marlboro, MA) and displayed online

For visualizing the spatial distribution of the surface potentials, two-dimensional color-coded voltage maps were constructed by interpolating values between electrodes using Delaunay triangulation. These maps were superimposed on an anatomical C57BL/6J mouse brain MRI image (MacKenzieGraham et al., 2004). In order to localize the electrode array with respect to the MRI space, MRI slices were examined for neuroanatomical landmarks and compared to an atlas of the C57BL/6J mouse brain (Franklin and Paxinos, 1997). Topographical mapping of event-related potentials has advantages over conventional waveform analysis when assessing large-scale network function because surface topography reflects the global configuration of the underlying neuronal activity, and different surface topographies are necessarily generated by different neuronal populations (Srebro, 1996; Vaughan, 1982). In addition, topographical mapping is a reference-independent measure that is not influenced by the choice of the reference electrode (Geselowitz, 1998; Lehmann, 1987) and does not require the arbitrary selection of waveform features for analysis. Identification of component maps To characterize the spatiotemporal dynamics of the whiskerevoked epicranial SEP, we used a modified k-means cluster analysis to identify the most dominant maps in the grand average SEP in terms of the spatial distribution of the surface potential, i.e. in terms of map topography. This method and the subsequent fitting procedure described below have proven to be powerful tools for identifying the dominant component maps in event-related potentials (Arzy et al., 2006, 2007; Murray et al., 2006; Ortigue et al., 2004; Pascual-Marqui et al., 1995; Thierry et al., 2007; for detailed descriptions and reviews, see Michel et al., 2001, 2004; Murray et al., 2008). Since different topographies of the surface potential necessarily reflect the activity of different underlying neuronal sources, the cluster



analysis provides a means of defining the different steps in the pattern of cerebral activity evoked by the stimulus, i.e. the different SEP components. The cluster analysis is exclusively based on the spatial correlation between strength-normalized potential maps. The number of clusters that optimally described the grand average SEP was determined using a modified Krzanowski–Lai criterion (Krzanowski and Lai, 1988). These cluster maps were then fitted back to the original grand average SEP by means of the spatial correlation. Each momentary map was labeled with the cluster map it best correlated with, thus identifying successive time points or time periods represented by the different cluster maps. Periods shorter than 2 ms were excluded and allocated to the preceding or following segment depending on which they correlated better with. Once the different segments were determined, the average map during each segment was calculated, representing the different component maps of the grand average SEP.

vsuperf and vdeep are the voltages at depth h obtained with the probe inserted superficially and deeper, respectively. Intracortical CSD analysis CSD analysis enhances the spatial resolution of intracortical LFP recordings by revealing the location of current sinks and sources – the generators of the field potentials – in cortical laminae, while suppressing far-field, volume-conducted potentials originating from distant neuronal structures (Mitzdorf, 1985; Nicholson and Freeman, 1975). One-dimensional CSD is calculated as the product of the second spatial derivative of the electric potential in this dimension with the conductivity tensor. Since the relatively small conductivity variations of different cortical depths only slightly affect CSD estimates, the conductivity tensor is often assumed to be constant (Mitzdorf and Singer, 1980). CSD was therefore estimated by calculating the finite-difference second spatial derivative:

Evaluation of the stability of component maps In order to statistically evaluate the spatiotemporal stability of the component maps identified by the cluster analysis, each of these maps was compared to the SEP map series of individual mice at each time point by computing the spatial correlation, a measure of the topographical similarity between two maps: n

∑ ðui d vi Þ i¼1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SC ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ∑ni¼1 u2i d ∑ni¼1 v2i where ui and vi are the voltages (vs. the average reference) at each electrode for the two maps (Brandeis et al., 1992; Khateb et al., 2003). Each time point of the response of individual mice was attributed to the component map with which its spatial correlation was highest and the time point of maximal correlation with each component map was defined as the latency of this component map in each individual evoked potential. In addition, the global explained variance for each component map and the number of time frames where each component map was present in each individual evoked potential were determined. These parameters were compared between groups or in pairs of successive component maps by two-tailed paired t-tests. In order to maintain the experiment-wise significance level at 0.05, the significance level for each test repetition was adapted using Bonferroni correction. Intracortical LFP analysis Since the length of the multi-electrode probe did not span the whole cortical thickness, separate recordings were made with the probe inserted superficially so that the uppermost recording electrode was at the level of the cortical surface and with the probe inserted deeper so that the lowermost recording electrode was 1 mm below the cortical surface. Individual averaged superficial and deep LFP were then combined; for depths between 350 and 650 μm, where 2 recordings were obtained, a linearly weighted average of both recordings was computed: v¼

ðn þ 1−hÞd vsuperf þ hd vdeep nþ1

where n is the number of depths where two recordings were obtained, h represents the depth and varies from 1 to n, and

CSD∝

vh−Δh −2 vh þ vhþΔh Δh2

where vh is the voltage at depth h and Δh is the distance between electrodes (Freeman and Nicholson, 1975; Quairiaux et al., 2007). The intracortical LFP was spatially smoothed before estimating the CSD (Freeman and Nicholson, 1975). To compute the smoothing and CSD at the extremities of the electrode, virtual voltage values were extrapolated by assuming no voltage decay above the uppermost and below the lowermost electrodes (Vaknin et al., 1988). To better visualize the CSD profiles, color-coded plots were computed using linear interpolation. Absolute onset latencies were determined for each penetration as the first time point where any post-stimulus CSD trace was greater than 4 times the standard deviation of its pre-stimulus 50-ms baseline for at least 2 ms consecutively, starting 5 ms post-stimulus. Onset latencies were compared between areas using one-way ANOVA followed by post-hoc Games–Howell tests, performed with the SPSS 14 software (SPSS Inc., Chicago, IL). Since the variances of latencies were not equal across cortical areas, results of the ANOVA were only considered as indicative of statistical significance. However, the Games–Howell test accommodates unequal variance between groups (Chen et al., 2007). The significance level for the post-hoc tests was set at 0.05. Results Epicranial SEP mapping Whisker stimulation evoked a complex pattern of brain activity (Figs. 1A, B) that was summarized into six epicranial SEP component maps by the cluster analysis (Figs. 1C, D). Importantly, fitting these maps back to the data revealed that each map is present during a certain consecutive time period, i.e. that the evoked potentials are characterized by a progression of quasi-stable processing states (see Fig. 1E). The SEP began with a focal voltage-positive response over the parietal cortex contralateral to stimulation. This map configuration (map 1) lasted from 5 to 8 ms post-stimulus. This positivity then spread towards electrodes overlying the frontal cortex, while a focal and strong negative potential appeared over central parietal sites (map 2, 8–13.5 ms). The next component


(map 3, 13.5–18 ms) was characterized by a frontal positivity on the hemisphere contralateral to stimulation and the appearance of a second focal positivity over the parietal cortex ipsilateral to stimulation, as well as a central parietal and contralateral occipital diffuse negativity of low intensity. The positivity became less intense and more diffuse during the next component (map 4, 18–27 ms), involving contralateral frontal and parietal as well as central regions, and was then again more focused over the contralateral parietal cortex during component maps 5 (27–50 ms) and 6 (50–60 ms), while the negativity was located over the most posterior electrodes. The SEP maps in response to stimulation of rightsided whiskers were essentially mirror images of that to leftsided stimulation (see Fig. 5). These data suggest that whisker stimulation evoked a stereotypical spatiotemporal pattern of brain activity, reflecting the activation of a distributed neuronal network including parietal and frontal areas and involving both hemispheres. Laminar pattern of cortical responses to whisker stimulation Intracortical LFP recordings and CSD analysis were performed in those cortical areas that were suggested as possible sources of the recorded maps, i.e. in areas located underneath

595

focal maxima or minima of the epicranial SEP component maps: S1 (Woolsey and Van der Loos, 1970), the frontal vibrissa motor cortex (corresponding to the rostral part of cytoarchitectonic field AGm in rats; Brecht et al., 2004), and a cortical area situated medial to the recordings made in S1 and probably corresponding to the caudal vibrissa motor cortex (caudal part of AGm in rats; Brecht et al., 2004; Franklin and Paxinos, 1997). In addition, recordings were performed in S2, which lies mostly beyond our epicranial electrode array, as it is known to respond to whisker stimulation (Benison et al., 2007; Carvell and Simons, 1986). As a control, intracortical LFP recordings and CSD analysis were performed in the primary visual cortex. Local whisker-evoked activity in S1 began on average 5.3 (standard deviation, SD 0.3) ms post-stimulus with two sinks, one at the layer III–IV border that shifted to layer II–III after approximately 3 ms and one in deep layer V and superficial layer VI (Figs. 2, 3). These sinks were flanked by 3 sources in layer I, deep layer IV–superficial layer V, and deep layer VI. This configuration inverted between 25 and 30 ms post-stimulus to supragranular and infragranular sources flanked by sinks. In S2, the response began 6.1 (SD 0.6) ms post-stimulus with 2 sinks at the layer III–IV and layer V–VI borders, flanked by a superficial and a deep source. The 2 sinks merged together

Fig. 2. Current source-density analysis of intracortical responses to whisker stimulation. Color-coded plots of the averaged second spatial derivatives of LFP recordings are shown. Sinks are coded in yellow and red, sources in blue. The same color scale applies for all CSD plots. Each CSD is averaged from 6 mice. The approximate extent of cortical layers (labeled in Roman numerals) is indicated on the left-most part of the CSD plots. The times of onset of the 6 component maps of the epicranial SEP are indicated as dotted lines. AGm: vibrissa motor cortex.



Fig. 3. Absolute onset latencies of intracortical activity in response to whisker stimulation. The mean and SD of absolute onset latency of intracortical activity are plotted for each cortical area. Latencies are averaged from 6 mice except for S1 in the ipsilateral hemisphere, where one outlier was removed. ANOVA showed a significant effect of the cortical area on the latency (F6,34 = 15.92; p b 0.001). The asterisks indicate significant post-hoc Games–Howell tests (⁎, p b 0.05; ⁎⁎, p b 0.01).

after approximately 4 ms to form a large sink occupying layers II to V. This pattern was gradually replaced by the inverse configuration between 20 and 30 ms post-stimulus. The onset of local activity in S1 and S2, as determined by CSD analysis, coincided with the onset of the epicranial response to whisker stimulation, characterized by a focal parietal positivity contralateral to stimulation (Fig. 1D, map 1, Fig. 2). Local activity in S2 might have influenced the most lateral epicranial electrodes through volume conduction. CSD analysis did not show local activity in other cortical areas during map 1 (Fig. 2). Similarly, during maps 5 and 6, which were topographically similar to map 1 (Fig. 1), local activity was recorded mostly in S1 and S2, other cortical areas appearing largely inactive. Whisker-evoked activity in the vibrissa motor cortex (rostral AGm) started 7.5 (SD 0.7) ms post-stimulus with a sink at the layer III–IV border, which shifted to layer II–III after approximately 4 ms, a superficial source, and a much fainter infragranular sink and deep source. Activity in the rostral AGm started significantly later than in S1 and S2 (Fig. 3). Between 25 and 30 ms post-stimulus, the supragranular source–sink complex inverted to a fainter sink–source configuration. In the caudal part of AGm, whisker-evoked activity started 6.4 (SD 1.4) ms post-stimulus and consisted of a complex pattern of faint, brief sinks and sources involving all cortical layers, with a slightly stronger superficial sink and infragranular source appearing approximately 9–10 ms post-stimulus. Between 15 and 20 ms post-stimulus, the configuration changed to a faint infragranular sink and superficial source. The onset of intracortical activity in the rostral AGm cortex co-occurred with the extension of the surface positivity towards the frontal cortex contralateral to stimulation (Fig. 1, map 2, Fig. 2). Map 2 was also characterized by a focal central parietal negativity. The temporal extent of the superficial sink-infragranular source in the caudal AGm contralateral to stimulation was related to that of the surface-recorded negativity.

Activity in S1 ipsilateral to whisker stimulation appeared 12.9 (SD 2.3) ms post-stimulus with faint supragranular and infragranular sinks flanked by 3 sources. This configuration lasted until approximately 30 ms post-stimulus. In the ipsilateral S2, activity appeared 8.4 (SD 2.4) ms poststimulus, became larger after about 4 ms, and consisted of a layer IV–V sink, a fainter supragranular sink, and superficial and deep sources, lasting until approximately 30 ms poststimulus. The onset of activity in ipsilateral S2 was slightly earlier than in S1, but this difference was not found to be statistically significant. Otherwise, activity in ipsilateral S1 was significantly later than in all other cortical areas investigated (Fig. 3). The onset of local activity in the ipsilateral S1 coincided with the appearance of a small focal surface positivity over the parietal cortex ipsilateral to whisker stimulation (Fig. 1, map 3). Local activity in ipsilateral S2 began earlier, but grew more intense at approximately the time of onset of activity in ipsilateral S1 (Fig. 2). The very lateral position of the surface positivity might reflect a contribution of S2 through volume conduction. In the caudal AGm ipsilateral to stimulation, whiskerevoked activity appeared 7.1 (SD 1.2) ms post-stimulus with several sinks and sources involving all cortical layers. The response lasted until approximately 15 ms post-stimulus. Response to whisker stimulation in the ipsilateral rostral AGm cortex was extremely faint and was not observed in all mice. The additional intracortical LFP recordings and CSD

Fig. 4. The temporal extent and topography of the component maps of the grand average (n = 15) epicranial SEP are shown for the first (A) and second (B) recordings, performed two weeks apart. There were only minor, non-significant differences in the latency of best correlation of each component map between both recording sessions. Average number of responses to individual stimuli removed due to artifact contamination: first recording, 4.3/200 (range 0–28); second recording, 2/200 (range 0–6); no significant difference between recordings.


analyses in the primary visual cortex (V1) revealed only small contralateral responses evoked by whisker stimulation in only 3 out of 6 mice. These responses were very variable between individual animals but generally involved the infragranular layers (data not shown). Virtually no response to ipsilateral whisker stimulation was observed in V1. Interindividual stability of epicranial SEP maps The stability of the whisker-evoked brain response across individual mice was assessed by comparing the grand average component maps to the complete map series of each animal by means of the spatial correlation. The average latencies of highest spatial correlation between each component map and the SEP maps of individual mice were very similar across mice, especially for the shorter-latency components (Fig. 1E). The differences in latency of best correlation between pairs of successive component maps were all significant (all p values b 10− 4 for each comparison of successive map latencies, 5 two-tailed paired t-tests, significance level for each test = 0.01), indicating that the spatiotemporal pattern of neuronal network activity evoked by whisker stimulation was similar in each mouse.

597

in mice. We used the well-studied whisker-evoked potentials to probe our method (for recent reviews on the whisker somatosensory cortex, see Brecht, 2007; Petersen, 2007). The result of the spatiotemporal analysis of the evoked potential maps after whisker stimulation is in agreement with previous reports using local recordings and shows that somatosensory stimulation activates a neuronal network involving somatosensory and motor cortical areas of both hemispheres, both serially and in parallel. Our method allows detecting these different activation areas in one single recording and unravelling the temporal dynamics of activation of these areas. Intracranial recordings under the areas of potential maxima and minima confirm the interpretation of the maps with respect to the location of the generators in the brain. Our results are similar to recently reported work using voltagesensitive dye imaging to map the response of the sensorimotor cortical network to whisker stimulation (Ferezou et al., 2007). They also confirm earlier findings regarding whisker-evoked activity in the primary (e.g. Petersen et al., 2003; Rojas et al., 2006) and secondary somatosensory cortex (e.g. Benison et al., 2007) and the vibrissa motor cortex (Ahrens and Kleinfeld, 2004; Farkas et al., 1999) contralateral to stimulation as well as the primary (Pidoux and Verley, 1979; Shuler et al., 2001) and

Intraindividual stability of epicranial SEP maps The stability of whisker-evoked cerebral activity in individual mice was studied by recording the SEP twice in the same animals with a 2-week interval. The two recordings yielded a highly similar spatiotemporal pattern of SEP maps (Fig. 4). The cluster analysis showed nearly identical component maps for both grand averages, with almost no difference in the latency of best correlation of each map (data not shown). There was no significant difference in global explained variance when comparing the grand average component maps of the first recording with the individual map series of the first and the second recordings; the same was true for the grand average component maps of the second recording. This shows that the intraindividual variability of the SEP maps over time was also limited. Effect of anesthetic depth on epicranial SEP maps The effect of varying the depth of anesthesia on epicranial SEP mapping was studied in 13 mice during the first recording session with deep (1.5% isoflurane) anesthesia. The amplitude of the whisker-evoked brain responses was lower under deep anesthesia, as illustrated by the amplitude of the global field power (Fig. 5). The clustering analysis summarized the response into 3 maps only. Activity appeared to be restricted to the parietal cortex contralateral to stimulation; no focused activity was seen in either the frontal cortex or the hemisphere ipsilateral to stimulation (Fig. 5B). The average time spent in map 1 was longer under deep than light anesthesia (18 vs. 4 ms, p b 0.001, two-tailed paired t-test). These results suggest that deep anesthesia altered the propagation of whisker-evoked potentials beyond the somatosensory cortices contralateral to stimulation. Discussion We developed a multichannel epicranial EEG recording and analysis technique to describe large-scale neuronal networks

Fig. 5. The effect of anesthetic depth on epicranial SEP mapping. The temporal extent and topography of the component maps of the grand average (n = 13) epicranial SEP in response to stimulation of right-sided whiskers are shown under light (0.8% isoflurane, A) and deep (1.5% isoflurane, B) anesthesia. Responses under deep anesthesia were of lesser amplitude and complexity; the 3 component maps that summarized the data showed that focused activity was restricted to the parietal cortex contralateral to stimulation. The average time spent in component map 1 was significantly longer under deep than light anesthesia (18 vs. 4 ms, p b 0.001, two-tailed paired t-test). Onset latency differences for maps 5 and 6 between groups were non-significant (two-tailed paired t-tests). Average number of responses to individual stimuli removed due to artifact contamination: light anesthesia, 1.9/200 (range 0–6); second recording, 1.8/200 (range 0–18); no significant difference between recordings.



secondary somatosensory cortex (Carvell and Simons, 1986) ipsilateral to stimulation. In addition, our CSD profiles give (to our knowledge) hitherto unreported insight into the laminar pattern of response to whisker stimulation in the mouse cortex in S2 and the rostral vibrissa motor cortex contralateral to stimulation, as well as in the hemisphere ipsilateral to stimulation. We propose that epicranial SEP mapping usefully complements existing approaches to investigate with minimal invasiveness and repeatedly the functionality, integrity, vulnerability, and plasticity of large-scale cortical networks in an animal model. Interpretation of epicranial SEP maps The potential field generated by synchronous synaptic activity in neurons of a given cortical area can be modeled by an equivalent dipole whose direction is perpendicular to the cortical surface (Lopes da Silva and Van Rotterdam, 2005). In the mouse cerebral cortex, the absolute amplitude (regardless of polarity) of the potential field generated by activity in a cortical area is expected to be maximal at the region of skull surface overlying this area, and focal maximal or minimal potentials with steep gradients recorded by epicranial SEP mapping should reflect activity in cortical areas lying close to the electrode that recorded the voltage peak. Not every deflection in voltage should be taken to mean local cortical activity, especially if it is spatially diffuse rather than focal. For example, the focal positivity in the parietal cortex contralateral to whisker stimulation in map 1 (Fig. 1D) likely reflects an underlying dipole caused by intracortical activity in S1, whereas the diffuse, low-intensity minimum recorded over the opposite hemisphere in the same map probably corresponds to the volume-conducted negative pole of the same dipole. Our intracortical recordings concurred with the proposed interpretation of the surface maps. Sites where epicranial focal maxima or minima were recorded lay over cortical areas that showed local intracortical activity as determined by the CSD analysis. In particular, there was good temporal correspondence between the onset of appearance of a surface-recorded focus and the onset of intracortical activity (Fig. 2). Conversely, areas such as the frontal or occipital cortex ipsilateral to stimulation, where no clear surface focus appeared for the whole analysis period, showed no evidence of local intracortical activity. The large-scale cortical network activated by whisker stimulation Our data suggest that cortical activity evoked by whisker stimulation initially involves the primary somatosensory cortex contralateral to stimulation. A single potential maximum with very steep gradient dominates the map during this early period (map 1 in Fig. 1D). The intracranial recordings support this interpretation (Fig. 2). The electrode in the contralateral S1 underneath the map maximum is the first to show significant activation during this early period. The CSD configuration shows two sinks at the layer III–IV and V–VI borders flanked by 3 sources, the largest of which spans the deepest cortical layers. This profile is similar to previously reported findings (see e.g. Agmon and Connors, 1991; Jellema et al., 2004; Lecas, 2004). These findings are consistent with previous observations that thalamic input from the ventroposteromedial nucleus to S1 terminates mainly in layer IV and lower layer III, with a smaller projection to the layer V–VI border (Bernardo and Woolsey, 1987). Neurons in these layers are the

first ones to be activated following somatosensory stimulation (Armstrong-James et al., 1992). Physiological and modeling studies in monkey primary sensory cortices have established that the earliest sensory-evoked CSD configurations are generated by depolarization of both thalamocortical axon terminals and their main targets in layer IV, spiny stellate cells. This combination of pre- and postsynaptic events in turn contributes to the generation of the initial surface sensoryevoked potentials (Peterson et al., 1995; Schroeder et al., 1991; Steinschneider et al., 1992; Tenke et al., 1993). Similarly, we suggest that the initial current sinks evoked by whisker stimulation in the mouse S1 correspond to the depolarization of thalamocortical axon terminals and of cortical neurons at the layer III–IV and (to a lesser extent) V–VI borders, and that this depolarization is reflected in the focal positivity of epicranial map 1. The second map identified in our epicranial recordings (map 2 in Figs. 1D, 2, 3) shows a propagation of the potential maximum to more anterior electrodes. In addition, a strong focal minimum on parietal midline electrodes is noted. The intracranial CSD profiles show, besides the appearance of large supra- and infragranular current sinks in S1 that likely reflect activation of neurons in these layers (Armstrong-James et al., 1992), a strong activation in S2, consisting of two sinks at the layer III–IV and V–VI borders. This S2 activation is probably due to direct projections from the ventroposteromedial nucleus of the thalamus that terminate in the same layers as those directed to S1 (Pierret et al., 2000; Wise and Jones, 1978). Unfortunately, S2 lies mostly beyond our epicranial electrode array and probably contributes relatively little to the epicranial SEP maps through volume conduction. On the other hand, the intracranial CSD profiles also show activation in the frontal vibrissa motor cortex (rostral AGm) that involves mostly supragranular layers. This CSD profile is similar to previously reported findings in the rat (Ahrens and Kleinfeld, 2004) and is likely to be due to connections between the somatosensory and the motor cortices (Farkas et al., 1999; Izraeli and Porter, 1995; Welker et al., 1988). These connections might play an important role in informing sensory areas about motor planning and in modulating exploratory motor behavior as a function of sensory input (Kleinfeld et al., 2006). Map 2 is also characterized by a strong focal negativity over central parietal sites. It seems to at least partly reflect activity of the caudal part of area AGm, as indicated by the onset of activation in the intracranial electrode positioned in this area. Sensory-evoked potentials were previously reported using epidural electrodes placed close to the parietal midline in rats (Miyazato et al., 1995; Stienen et al., 2003). The physiological implication of whisker-related activity in the medial parietal cortex is unclear. Lesions of the caudal AGm and neighboring areas in rats were reported to cause a behavioral syndrome comparable in some aspects to multimodal sensory neglect (King and Corwin, 1990), suggesting that this cortical area might be implicated in building multimodal spatial representations and in orienting behavior accordingly. It must be noted that the low intensity of intracortical activity compared to the strong surface negativity suggests that caudal AGm might not be the sole generator. Further research is needed to clarify the role of medial parietal cortical areas in sensory processing. Map 3 shows, in addition to a contralateral frontal positive potential, a focal small surface positivity over the parietal cortex of the hemisphere ipsilateral to stimulation. The CSD profile of the intracranial recordings indeed indicates onset of


activity in the ipsilateral hemisphere at this latency. However, it involves S2 more strongly than S1. As already mentioned above, our epicranial electrode array does not extend to S2. This is probably the reason why this activity that is clearly seen in the intracranial electrode is only seen relatively feebly by the most lateral electrodes. Nevertheless, the intracranial CSD profiles show that the ipsilateral S1 cortex is also (though weakly) activated at this time period. In both areas, supragranular and infragranular layers are involved; in S2, the lower part of layer IV also appears to be involved. Map 4 is characterized by a diffuse positivity over the central and contralateral frontal and parietal areas, without any steep and focused voltage gradient. Since several cortical areas show a change in their CSD configuration during this time period, it might be that no single area is sufficiently consistently activated to generate an equivalent dipole strong enough to show on the surface recordings. The ongoing CSD activity in ipsilateral S2 during map 4 is largely unseen by the epicranial electrodes, probably due to the lateral location of S2. Maps 5 and 6 show a focused positivity over the parietal cortex contralateral to stimulation. The CSD profiles confirm that activity is confined to contralateral S1 and S2 during this period. The positive wave over the somatosensory cortex, consistently reported in previous SEP studies (Di and Barth, 1991; Rojas et al., 2006), is of the same polarity as that during map 1, whereas the CSD configuration in S1 and S2 during maps 5 and 6 reverses as compared to their initial configuration. This illustrates the complexity underlying the generation of the EEG signal, which represents only the equivalent dipolar component of the multipolar sources inside the cortex (Tenke et al., 1993). The fact that most of the anatomical connections between the somatosensory cortices and the other cortical areas involved in the somatosensory network are reciprocal suggests that the activity in S1 and S2 during maps 5 and 6 might be influenced by feedback from these other areas. Dynamics of large-scale neuronal networks The cluster analysis and the subsequent fitting of the cluster maps in the individual data revealed that the evoked responses were characterized by a series of distinct map configurations, each one remaining stable for a given period of time and then quickly changing into a new configuration in which it remained stable again. This characteristic has been repeatedly described for human event-related potentials and human spontaneous EEG and it has been postulated that human large-scale neuronal networks evolve through a sequence of quasi-stable states, the so-called microstates (Koenig et al., 2002; Lehmann and Skrandies, 1980; Lehmann et al., 1998; Wackermann et al., 1993). It has been proposed that these microstates represent the basic building blocks of cognition, the different steps in the stream of information processing. Each of these microstates represents a stable pattern of the large-scale network activity (for a discussion of this concept see Bressler and Tognoli, 2006; Changeux and Michel, 2006; Fingelkurts and Fingelkurts, 2006; Lehmann, 1987; Michel et al., 1999). The fact that these metastable states are also observed in the SEP of an anesthetized mouse supports the hypothesis that information processing occurs through a stream of discrete units or epochs rather than in a continuous flow of neuronal activity (for a comprehensive discussion of this dichotomy, see Fingelkurts and Fingelkurts, 2006). Our data suggest that these discrete blocks of microstates do not

599

only appear in cognitive processing in humans, but may represent a fundamental property of large-scale neuronal network functioning in the mammalian cerebral cortex. This is further supported by our intracranial recordings that confirm these periods of stable sink-source patterns across the different cortical layers, corresponding to the different periods of stable surface maps. Interindividual and intraindividual stability of the whisker-evoked cortical response In order to evaluate the capacity of epicranial mapping to assess large-scale network function reliably and repeatedly, we looked at the stability of SEP maps across and within individual mice. The temporal sequence of epicranial maps evoked by whisker stimulation was very stable and similar across mice (Fig. 1E). Furthermore, repeating epicranial SEP mapping in the same mice after two weeks yielded almost identical maps (Fig. 4). Thus, both the interindividual and the intraindividual variability of SEP are limited. This extends our previous findings about the temporal stability of epicranial SEP waveforms (Troncoso et al., 2000) and suggests that the large-scale cortical network activated by whisker stimulation is a fundamental component underlying sensorimotor processing in mice (Kleinfeld et al., 2006). This low intra- and interindividual variance is an important prerequisite for using the method as an animal model for studying the development and plasticity of large-scale neuronal networks. Effects of anesthesia on the activity of the large-scale somatosensory network As a first approach to test the sensitivity of the method to detect changes of the large-scale somatosensory network, we evaluated the effect of varying the level of anesthesia on the whisker-evoked brain response (Fig. 5). Our epicranial waveforms over S1 under deep isoflurane anesthesia are similar to those recorded with epidural electrodes in similarly anesthetized rats (Rojas et al., 2006). SEP mapping showed that whisker-evoked activity was now mostly restricted to the parietal cortex contralateral to stimulation, suggesting an alteration in the propagation of activity from somatosensory areas to the other regions involved in the network. It was recently found that whisker-evoked neuronal firing in S1 of isoflurane-anesthetized mice was more strongly inhibited than subthreshold activity as isoflurane concentration was increased (Berger et al., 2007). It is tempting to suggest that this disproportional reduction in firing is reflected in our data by the relative preservation of the response in the somatosensory cortex contralateral to stimulation (Fig. 5, map 1) and by the absence of propagation of whisker-evoked activity to the other cortical areas of the large-scale somatosensory network. An alternative interpretation of our findings stems from recent evidence that evoked potentials may be generated by phase resetting of ongoing brain oscillations in addition to stimulus-evoked neuronal activity (Fell et al., 2004; Makeig et al., 2002). In particular, EEG phase resetting might be a relatively greater contributor to event-related responses in higher-order cortical areas compared to primary sensory cortices (Shah et al., 2004). Since deep isoflurane anesthesia markedly reduces and alters the spontaneous EEG (our own observations; Rojas et al., 2006), its potential impact on the phase-resetting component of evoked potential generation



might interfere relatively more with the SEP in cortical areas downstream of the somatosensory cortex, as observed here. Although the mechanisms underlying the effects of varying anesthetic depth on SEP maps are yet incompletely understood, epicranial SEP mapping is able to resolve differences in the spatiotemporal pattern of sensory-evoked responses across experimental conditions. The effects of anesthetic depth on SEP maps shown here raise the issue of comparing somatosensory processing in the waking versus lightly anesthetized state. Multichannel epicranial SEP recordings in the waking mouse would clearly be of great interest to address arousal- and behavioral-state-dependent cortical function. However, the propagation of whisker-evoked activity to several cortical areas in both hemispheres that we observed under light isoflurane anesthesia is similar to that reported following passive whisker stimulation in awake, headfixed mice using voltage-sensitive dye imaging (Ferezou et al., 2007). This suggests that light isoflurane anesthesia does not cause major disturbances of whisker-evoked activity in cortical networks. Mouse epicranial SEP mapping and intracortical CSD analysis as a model approach to cortical network function The stability of whisker-evoked responses suggests that epicranial SEP mapping is adequate for repeated, minimally

invasive functional assessment of the cortical somatosensory network. Most importantly, CSD analysis in cortical areas selected from the surface recordings brings further detail about the local processing of somatosensory input. Fig. 6 shows surface SEP waveforms and maps in mice (A) and healthy human subjects (B) as well as intracranial SEP recordings from subdural electrodes in an epileptic patient (C). Human data are consistent with previously published surface and subdural SEP recordings (Allison et al., 1989a,b; Urbano et al., 1997; Valeriani et al., 1998; van de Wassenberg et al., 2008). In both species, surface waveforms and topographic mapping show two successive positivities overlying the frontoparietal cortex. Furthermore, human subdural SEP recordings display the same polarity reversal across the central sulcus at both latencies, suggesting that activity in the cortex surrounding the sulcus is similar at both these moments. However, these data do not allow concluding unambiguously whether or not two components with similar polarity and topography but separated in time are generated by the same neuronal events. Indeed, our CSD analysis in mouse S1 (Fig. 6D) indicates that this is not the case, at least in mice. Thus, mouse epicranial EEG mapping coupled to intracortical CSD analysis reveals crucial information about the genesis of the surface SEP that would have been impossible to uncover from the results of human scalp recordings or invasive subdural recordings. Of course, the point here is not to establish direct analogies

Fig. 6. A. Grand average epicranial waveform and instant maps of left whisker-evoked SEP in mice (same dataset as in Fig. 1). B. Grand average scalp waveform and instant maps of SEP to electrical median nerve stimulation in 44 healthy human subjects. 256-channel EEG was continuously acquired while square current pulses (2000 repetitions, 200-μs duration, 267-ms inter-stimulus interval, amplitude just sufficient to elicit slight thumb adduction) were administered to the left median nerve. Black dots in A and B illustrate the location of the electrode whose waveform is shown. C. Average waveforms and maps of SEP to electrical stimulation of the left median nerve (stimulation parameters identical to those in B) in an epileptic patient implanted with subdural electrode strips spanning the central sulcus (indicated by an arrow). D. Grand average whisker-evoked CSD profile in mouse S1 (same dataset as in Fig. 2). Waveforms: positive voltages upwards; vertical bar in A: 50 μV in A, 2 μV in B, 30 μV in C; horizontal bar in A: 5 ms; red color: positive voltages in A–C, current sinks in D; blue color: negative voltages in A–C, current sources in D. Note that despite similar polarities and topographies of SEP maps at the selected time points in both mice and humans, the CSD profile in the mouse suggests that these two temporally separated evoked potential components are not generated by the same neuronal structures.


between mouse and human SEP components, but rather to illustrate how similar surface potentials may be generated by different neuronal events. In any case, care is needed when generalizing from results obtained in a given species and sensory modality. For instance, although our CSD configurations in S1 are in good agreement with recently published profiles in rats (Jellema et al., 2004; Lecas, 2004), they differ somewhat from those observed in monkey S1 (Lipton et al., 2006; Schroeder et al., 1995). Some degree of difference is also apparent with respect to other sensory modalities in rodents (Barth and Di, 1990; Heynen and Bear, 2001) and monkeys (Schroeder et al., 1991; Steinschneider et al., 1992). These differences likely reflect the adaptation of cortical sensory processing to species- and modality-specific demands (Hirsch and Martinez, 2006). Some technical limitations must also be kept in mind. The spatial extent of the electrode array is restricted by the insertion on the skull of temporal and neck muscles, so that the array covers mostly the frontal, parietal and occipital cortices. Thus, some somatosensory areas and most of the auditory cortex lie beyond the array. The spatial resolution is limited by the number of electrodes that can be placed over the skull and by the blurring of electrical potentials generated by the brain as they traverse the cerebrospinal fluid, meninges and skull (Nunez and Srinivasan, 2006). This technique is therefore most useful as a first step in approaching cortical function at the global network level, particularly with respect to temporal characteristics of network activities. If more local details are of interest, the method needs to be complemented by other, more spatially precise techniques, such as the multichannel intracranial recordings in areas of interest as demonstrated here. The major advantage of epicranial SEP mapping (besides its spatial extension to the global network level) is that it can be repeated several times in the same animal and thus allows studying how network function changes over time. This approach will therefore be suitable for studying large-scale network plasticity during the early postnatal development of the somatosensory system, as well as after changes in sensory experience and localized ischemic lesions to the cerebral cortex. Combining this approach with transgenic mouse strains will give insight into the role played by specific proteins in network plasticity. Acknowledgments We thank Cynthia Saadi for technical assistance with the histological preparations. The Cartool software (http:// brainmapping.unige.ch/Cartool.php) is developed by Denis Brunet, from the Functional Brain Mapping Laboratory, Geneva, supported by the Center for Biomedical Imaging (CIBM), Geneva and Lausanne, Switzerland. This work was supported by the Swiss Academy of Medical Sciences grant 323600-111505 (MD– PhD Program of the Swiss Universities) to P.M., the Swiss National Science Foundation grant 31-64030.00, the Eagle Foundation and the European Community Grant Promemoria No. 512012-2005 to J.Z.K., and the Swiss National Science Foundation grant 320000-111783 to C.M.M. References Agmon, A., Connors, B.W., 1991. Thalamocortical responses of mouse somatosensory (barrel) cortex in vitro. Neuroscience 41, 365–379. Ahrens, K.F., Kleinfeld, D., 2004. Current flow in vibrissa motor cortex can phase-lock with exploratory rhythmic whisking in rat. J. Neurophysiol. 92, 1700–1707.

601

Allison, T., McCarthy, G., Wood, C.C., Darcey, T.M., Spencer, D.D., Williamson, P.D., 1989a. Human cortical potentials evoked by stimulation of the median nerve. I. Cytoarchitectonic areas generating short-latency activity. J. Neurophysiol. 62, 694–710. Allison, T., McCarthy, G., Wood, C.C., Williamson, P.D., Spencer, D.D., 1989b. Human cortical potentials evoked by stimulation of the median nerve. II. Cytoarchitectonic areas generating long-latency activity. J. Neurophysiol. 62, 711–722. Armstrong-James, M., Fox, K., Das-Gupta, A., 1992. Flow of excitation within rat barrel cortex on striking a single vibrissa. J. Neurophysiol. 68, 1345–1358. Arzy, S., Thut, G., Mohr, C., Michel, C.M., Blanke, O., 2006. Neural basis of embodiment: distinct contributions of temporoparietal junction and extrastriate body area. J. Neurosci. 26, 8074–8081. Arzy, S., Mohr, C., Michel, C.M., Blanke, O., 2007. Duration and not strength of activation in temporo-parietal cortex positively correlates with schizotypy. NeuroImage 35, 326–333. Barth, D.S., Di, S., 1990. Three-dimensional analysis of auditory-evoked potentials in rat neocortex. J. Neurophysiol. 64, 1527–1536. Benison, A.M., Rector, D.M., Barth, D.S., 2007. Hemispheric mapping of secondary somatosensory cortex in the rat. J. Neurophysiol. 97, 200–207. Berger, T., Borgdorff, A., Crochet, S., Neubauer, F.B., Lefort, S., Fauvet, B., Ferezou, I., Carleton, A., Luscher, H.R., Petersen, C.C., 2007. Combined voltage and calcium epifluorescence imaging in vitro and in vivo reveals subthreshold and suprathreshold dynamics of mouse barrel cortex. J. Neurophysiol. 97, 3751–3762. Bernardo, K.L., Woolsey, T.A., 1987. Axonal trajectories between mouse somatosensory thalamus and cortex. J. Comp. Neurol. 258, 542–564. Brandeis, D., Naylor, H., Halliday, R., Callaway, E., Yano, L., 1992. Scopolamine effects on visual information processing, attention, and event-related potential map latencies. Psychophysiology 29, 315–336. Brecht, M., 2007. Barrel cortex and whisker-mediated behaviors. Curr. Opin. Neurobiol. 17, 408–416. Brecht, M., Krauss, A., Muhammad, S., Sinai-Esfahani, L., Bellanca, S., Margrie, T.W., 2004. Organization of rat vibrissa motor cortex and adjacent areas according to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells. J. Comp. Neurol. 479, 360–373. Bressler, S.L., 1995. Large-scale cortical networks and cognition. Brain Res. Brain Res. Rev. 20, 288–304. Bressler, S.L., Tognoli, E., 2006. Operational principles of neurocognitive networks. Int. J. Psychophysiol. 60, 139–148. Buzsaki, G., 2004. Large-scale recording of neuronal ensembles. Nat. Neurosci. 7, 446–451. Callan, D.E., Tajima, K., Callan, A.M., Kubo, R., Masaki, S., Akahane-Yamada, R., 2003. Learning-induced neural plasticity associated with improved identification performance after training of a difficult second-language phonetic contrast. NeuroImage 19, 113–124. Carvell, G.E., Simons, D.J., 1986. Somatotopic organization of the second somatosensory area (SII) in the cerebral cortex of the mouse. Somatosens. Res. 3, 213–237. Changeux, J.-P., Michel, C.M., 2006. Mechanisms of neural integration at the brain scale level: the neuronal workspace and microstate models. In: Grillner, S., Graybiel, A.M. (Eds.), Microcircuits: The Interface between Neurons and Global Brain Function. Dahlem Workshop Report. MIT Press, Cambridge, MA, pp. 347–370. Chen, C.M., Lakatos, P., Shah, A.S., Mehta, A.D., Givre, S.J., Javitt, D.C., Schroeder, C.E., 2007. Functional anatomy and interaction of fast and slow visual pathways in macaque monkeys. Cereb. Cortex 17, 1561–1569. de Zwart, J.A., Silva, A.C., van Gelderen, P., Kellman, P., Fukunaga, M., Chu, R., Koretsky, A.P., Frank, J.A., Duyn, J.H., 2005. Temporal dynamics of the BOLD fMRI impulse response. NeuroImage 24, 667–677. Di, S., Barth, D.S., 1991. Topographic analysis of field potentials in rat vibrissa/barrel cortex. Brain Res. 546, 106–112. Farkas, T., Kis, Z., Toldi, J., Wolff, J.R., 1999. Activation of the primary motor cortex by somatosensory stimulation in adult rats is mediated mainly by associational connections from the somatosensory cortex. Neuroscience 90, 353–361. Fell, J., Dietl, T., Grunwald, T., Kurthen, M., Klaver, P., Trautner, P., Schaller, C., Elger, C.E., Fernandez, G., 2004. Neural bases of cognitive ERPs: more than phase reset. J. Cogn. Neurosci. 16, 1595–1604. Fender, D.H., 1987. Source localization of brain electrical activity. In: Gevins, A.S., Rémond, A. (Eds.), Handbook of electroencephalography and clinical neurophysiology, vol. 1. Methods of analysis of brain electrical and magnetic signals. Elsevier, Amsterdam, pp. 355–403. Ferezou, I., Bolea, S., Petersen, C.C., 2006. Visualizing the cortical representation of whisker touch: voltage-sensitive dye imaging in freely moving mice. Neuron 50, 617–629. Ferezou, I., Haiss, F., Gentet, L.J., Aronoff, R., Weber, B., Petersen, C.C., 2007. Spatiotemporal dynamics of cortical sensorimotor integration in behaving mice. Neuron 56, 907–923. Fingelkurts, A.A., Fingelkurts, A.A., 2006. Timing in cognition and EEG brain dynamics: discreteness versus continuity. Cogn. Process 7, 135–162. Franklin, K.B., Paxinos, G., 1997. The Mouse Brain in Stereotaxic Coordinates. Academic Press, San Diego. Freeman, J.A., Nicholson, C., 1975. Experimental optimization of current source-density technique for anuran cerebellum. J. Neurophysiol. 38, 369–382. Fuster, J.M., 2006. The cognit: a network model of cortical representation. Int. J. Psychophysiol. 60, 125–132. Geselowitz, D.B., 1998. The zero of potential. IEEE Eng. Med. Biol. Mag. 17, 128–132. Heynen, A.J., Bear, M.F., 2001. Long-term potentiation of thalamocortical transmission in the adult visual cortex in vivo. J. Neurosci. 21, 9801–9813. Hirsch, J.A., Martinez, L.M., 2006. Laminar processing in the visual cortical column. Curr. Opin. Neurobiol. 16, 377–384.



Hollenberg, B.A., Richards, C.D., Richards, R., Bahr, D.F., Rector, D.M., 2006. A MEMS fabricated flexible electrode array for recording surface field potentials. J. Neurosci. Methods 153, 147–153. Izraeli, R., Porter, L.L., 1995. Vibrissal motor cortex in the rat: connections with the barrel field. Exp. Brain Res. 104, 41–54. Jellema, T., Brunia, C.H., Wadman, W.J., 2004. Sequential activation of microcircuits underlying somatosensory-evoked potentials in rat neocortex. Neuroscience 129, 283–295. Khateb, A., Michel, C.M., Pegna, A.J., O'Dochartaigh, S.D., Landis, T., Annoni, J.M., 2003. Processing of semantic categorical and associative relations: an ERP mapping study. Int. J. Psychophysiol. 49, 41–55. King, V., Corwin, J.V., 1990. Neglect following unilateral ablation of the caudal but not the rostral portion of medial agranular cortex of the rat and the therapeutic effect of apomorphine. Behav. Brain Res. 37, 169–184. Kleinfeld, D., Ahissar, E., Diamond, M.E., 2006. Active sensation: insights from the rodent vibrissa sensorimotor system. Curr. Opin. Neurobiol. 16, 435–444. Koenig, T., Prichep, L., Lehmann, D., Sosa, P.V., Braeker, E., Kleinlogel, H., Isenhart, R., John, E.R., 2002. Millisecond by millisecond, year by year: normative EEG microstates and developmental stages. NeuroImage 16, 41–48. Krzanowski, W.J., Lai, Y.T., 1988. A criterion for determining the number of groups in a data set using sum-of-squares clustering. Biometrics 44, 23–34. Lecas, J.C., 2004. Locus coeruleus activation shortens synaptic drive while decreasing spike latency and jitter in sensorimotor cortex. Implications for neuronal integration. Eur. J. Neurosci. 19, 2519–2530. Lehmann, D., 1987. Principles of spatial analysis. In: Gevins, A.S., Rémond, A. (Eds.), Handbook of electroencephalography and clinical neurophysiology, vol.1. Methods of analysis of brain electrical and magnetic signals. Elsevier, Amsterdam, pp. 309–354. Lehmann, D., Skrandies, W., 1980. Reference-free identification of components of checkerboard-evoked multichannel potential fields. Electroencephalogr. Clin. Neurophysiol. 48, 609–621. Lehmann, D., Strik, W.K., Henggeler, B., Koenig, T., Koukkou, M., 1998. Brain electric microstates and momentary conscious mind states as building blocks of spontaneous thinking: I. Visual imagery and abstract thoughts. Int. J. Psychophysiol. 29, 1–11. Linden, D.E., 2007. What, when, where in the brain? Exploring mental chronometry with brain imaging and electrophysiology. Rev. Neurosci. 18, 159–171. Lipton, M.L., Fu, K.M., Branch, C.A., Schroeder, C.E., 2006. Ipsilateral hand input to area 3b revealed by converging hemodynamic and electrophysiological analyses in macaque monkeys. J. Neurosci. 26, 180–185. Lopes da Silva, F., Van Rotterdam, A., 2005. Biophysical aspects of EEG and magnetoencephalogram generation. In: Niedermeyer, E., Lopes da Silva, F. (Eds.), Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Lippincott Williams and Wilkins, Philadelphia, pp. 107–125. MacKenzie-Graham, A., Lee, E.F., Dinov, I.D., Bota, M., Shattuck, D.W., Ruffins, S., Yuan, H., Konstantinidis, F., Pitiot, A., Ding, Y., Hu, G., Jacobs, R.E., Toga, A.W., 2004. A multimodal, multidimensional atlas of the C57BL/6J mouse brain. J. Anat. 204, 93–102. Makeig, S., Westerfield, M., Jung, T.P., Enghoff, S., Townsend, J., Courchesne, E., Sejnowski, T.J., 2002. Dynamic brain sources of visual evoked responses. Science 295, 690–694. McIntosh, A.R., 2000. Towards a network theory of cognition. Neural Netw. 13, 861–870. Mesulam, M.M., 1990. Large-scale neurocognitive networks and distributed processing for attention, language, and memory. Ann. Neurol. 28, 597–613. Mesulam, M.M., 1998. From sensation to cognition. Brain 121 (Pt 6), 1013–1052. Michel, C.M., Murray, M.M., Lantz, G., Gonzalez, S., Spinelli, L., Grave de Peralta, R., 2004. EEG source imaging. Clin. Neurophysiol. 115, 2195–2222. Michel, C.M., Seeck, M., Landis, T., 1999. Spatiotemporal dynamics of human cognition. News Physiol. Sci. 14, 206–214. Michel, C.M., Thut, G., Morand, S., Khateb, A., Pegna, A.J., Grave de Peralta, R., Gonzalez, S., Seeck, M., Landis, T., 2001. Electric source imaging of human brain functions. Brain Res. Brain Res. Rev. 36, 108–118. Mitzdorf, U., 1985. Current source-density method and application in cat cerebral cortex: investigation of evoked potentials and EEG phenomena. Physiol. Rev. 65, 37–100. Mitzdorf, U., Singer, W., 1980. Monocular activation of visual cortex in normal and monocularly deprived cats: an analysis of evoked potentials. J. Physiol. 304, 203–220. Miyazato, H., Skinner, R.D., Reese, N.B., Boop, F.A., Garcia-Rill, E., 1995. A middle-latency auditory-evoked potential in the rat. Brain Res. Bull. 37, 247–255. Murray, M.M., Imber, M.L., Javitt, D.C., Foxe, J.J., 2006. Boundary completion is automatic and dissociable from shape discrimination. J. Neurosci. 26, 12043–12054. Murray, M.M., Brunet, D., Michel, C.M., 2008. Topographic ERP Analyses: a step-by-step tutorial review. Brain Topogr. 20, 249–264. Nicholson, C., Freeman, J.A., 1975. Theory of current source-density analysis and determination of conductivity tensor for anuran cerebellum. J. Neurophysiol. 38, 356–368. Nunez, P.L., Srinivasan, R., 2006. Current sources in inhomogeneous and isotropic media. Electric fields of the brain: the neurophysics of EEG. Oxford University Press, Oxford, pp. 244–274. Ortigue, S., Michel, C.M., Murray, M.M., Mohr, C., Carbonnel, S., Landis, T., 2004. Electrical neuroimaging reveals early generator modulation to emotional words. NeuroImage 21, 1242–1251. Pascual-Marqui, R.D., Michel, C.M., Lehmann, D., 1995. Segmentation of brain electrical activity into microstates: model estimation and validation. IEEE Trans. Biomed. Eng. 42, 658–665. Petersen, C.C., 2007. The functional organization of the barrel cortex. Neuron 56, 339–355.

Peterson, N.N., Schroeder, C.E., Arezzo, J.C., 1995. Neural generators of early cortical somatosensory evoked potentials in the awake monkey. Electroencephalogr. Clin. Neurophysiol. 96, 248–260. Petersen, C.C., Grinvald, A., Sakmann, B., 2003. Spatiotemporal dynamics of sensory responses in layer 2/3 of rat barrel cortex measured in vivo by voltage-sensitive dye imaging combined with whole-cell voltage recordings and neuron reconstructions. J. Neurosci. 23, 1298–1309. 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. Pierret, T., Lavallee, P., Deschenes, M., 2000. Parallel streams for the relay of vibrissal information through thalamic barreloids. J. Neurosci. 20, 7455–7462. Price, C.J., Friston, K.J., 2002. Functional imaging studies of neuropsychological patients: applications and limitations. Neurocase 8, 345–354. Quairiaux, C., Armstrong-James, M., Welker, E., 2007. Modified sensory processing in the barrel cortex of the adult mouse after chronic whisker stimulation. J. Neurophysiol. 97, 2130–2147. Rojas, M.J., Navas, J.A., Rector, D.M., 2006. Evoked response potential markers for anesthetic and behavioral states. Am. J. Physiol., Regul. Integr. Comp. Physiol. 291, R189–196. Schnitzler, A., Gross, J., 2005. Normal and pathological oscillatory communication in the brain. Nat. Rev., Neurosci. 6, 285–296. Schroeder, C.E., Tenke, C.E., Givre, S.J., Arezzo, J.C., Vaughan Jr., H.G., 1991. Striate cortical contribution to the surface-recorded pattern-reversal VEP in the alert monkey. Vis. Res. 31, 1143–1157. Schroeder, C.E., Seto, S., Arezzo, J.C., Garraghty, P.E., 1995. Electrophysiological evidence for overlapping dominant and latent inputs to somatosensory cortex in squirrel monkeys. J. Neurophysiol. 74, 722–732. Shah, A.S., Bressler, S.L., Knuth, K.H., Ding, M., Mehta, A.D., Ulbert, I., Schroeder, C.E., 2004. Neural dynamics and the fundamental mechanisms of event-related brain potentials. Cereb. Cortex 14, 476–483. Shuler, M.G., Krupa, D.J., Nicolelis, M.A., 2001. Bilateral integration of whisker information in the primary somatosensory cortex of rats. J. Neurosci. 21, 5251–5261. Sigman, M., Pan, H., Yang, Y., Stern, E., Silbersweig, D., Gilbert, C.D., 2005. Top–down reorganization of activity in the visual pathway after learning a shape identification task. Neuron 46, 823–835. Srebro, R., 1996. A bootstrap method to compare the shapes of two scalp fields. Electroencephalogr. Clin. Neurophysiol. 100, 25–32. Steinschneider, M., Tenke, C.E., Schroeder, C.E., Javitt, D.C., Simpson, G.V., Arezzo, J.C., Vaughan Jr., H.G., 1992. Cellular generators of the cortical auditory evoked potential initial component. Electroencephalogr. Clin. Neurophysiol. 84, 196–200. Stienen, P.J., Haberham, Z.L., van den Brom, W.E., de Groot, H.N., Venker-Van Haagen, A.J., Hellebrekers, L.J., 2003. Evaluation of methods for eliciting somatosensory-evoked potentials in the awake, freely moving rat. J. Neurosci. Methods 126, 79–90. Tenke, C.E., Schroeder, C.E., Arezzo, J.C., Vaughan Jr., H.G., 1993. Interpretation of highresolution current source density profiles: a simulation of sublaminar contributions to the visual evoked potential. Exp. Brain Res. 94, 183–192. Thierry, G., Martin, C.D., Downing, P., Pegna, A.J., 2007. Controlling for interstimulus perceptual variance abolishes N170 face selectivity. Nat. Neurosci. 10, 505–511. Tombari, D., Loubinoux, I., Pariente, J., Gerdelat, A., Albucher, J.F., Tardy, J., Cassol, E., Chollet, F., 2004. A longitudinal fMRI study: in recovering and then in clinically stable sub-cortical stroke patients. NeuroImage 23, 827–839. Troncoso, E., Muller, D., Czellar, S., Zoltan Kiss, J., 2000. Epicranial sensory evoked potential recordings for repeated assessment of cortical functions in mice. J. Neurosci. Methods 97, 51–58. Troncoso, E., Muller, D., Korodi, K., Steimer, T., Welker, E., Kiss, J.Z., 2004. Recovery of evoked potentials, metabolic activity and behavior in a mouse model of somatosensory cortex lesion: role of the neural cell adhesion molecule (NCAM). Cereb. Cortex 14, 332–341. Urbano, A., Babiloni, F., Babiloni, C., Ambrosini, A., Onorati, P., Rossini, P.M., 1997. Human short latency cortical responses to somatosensory stimulation. A high resolution EEG study. NeuroReport 8, 3239–3243. Vaknin, G., DiScenna, P.G., Teyler, T.J., 1988. A method for calculating current source density (CSD) analysis without resorting to recording sites outside the sampling volume. J. Neurosci. Methods 24, 131–135. Valeriani, M., Restuccia, D., Di Lazzaro, V., Le Pera, D., Barba, C., Tonali, P., Mauguiere, F., 1998. Dipolar sources of the early scalp somatosensory evoked potentials to upper limb stimulation. Effect of increasing stimulus rates. Exp. Brain Res. 120, 306–315. van de Wassenberg, W., van der Hoeven, J., Leenders, K., Maurits, N., 2008. Multichannel recording of median nerve somatosensory evoked potentials. Neurophysiol. Clin. 38, 9–21. Vaughan Jr., H.G., 1982. The neural origins of human event-related potentials. Ann. N. Y. Acad. Sci. 388, 125–138. Wackermann, J., Lehmann, D., Michel, C.M., Strik, W.K., 1993. Adaptive segmentation of spontaneous EEG map series into spatially defined microstates. Int. J. Psychophysiol. 14, 269–283. Welker, E., Hoogland, P.V., Van der Loos, H., 1988. Organization of feedback and feedforward projections of the barrel cortex: a PHA-L study in the mouse. Exp. Brain Res. 73, 411–435. Wise, S.P., Jones, E.G., 1978. Developmental studies of thalamocortical and commissural connections in the rat somatic sensory cortex. J. Comp. Neurol. 178, 187–208. Woolsey, T.A., Van der Loos, H., 1970. The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex. The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res. 17, 205–242.