Articles in PresS. J Neurophysiol (May 12, 2010). doi:10.1152/jn.00297.

2010

1 Minor contribution of principal excitatory pathways to hippocampal LFPs in the

2 anesthetized rat: a combined independent component and current source density study
3
4 Korovaichuk A1*, Makarova J1*, Makarov VA2*, Benito N1 and Herreras O1
5
6
1
7 Dept of Systems Neuroscience, Cajal Institute – CSIC, Av. Doctor Arce 37, Madrid
8 28002, Spain.
2
9 Dept of Applied Mathematics, Faculty of Mathematics, Universidad Complutense of
10 Madrid, Av. Complutense s/n, Madrid 28040, Spain.
11
12 *These authors contributed equally to this work
13
14 Running title: Excitatory generators of hippocampal LFPs
15
16
17 Correspondence should be addressed to:
18 Oscar Herreras,
19 Cajal Institute - C.S.I.C.,
20 Av. Dr. Arce 37,
21 Madrid 28002, Spain.
22 Ph # 34 915854725
23 FAX # 34 915854754
24 Email: herreras@cajal.csic.es
25

1
Copyright © 2010 by the American Physiological Society.
26
27 ABSTRACT
28
29 Analysis of local field potentials (LFPs) helps understand the function of the converging
30 neuronal populations that produce the mixed synaptic activity in principal cells.
31 Recently, using independent component analysis (ICA) we resolved ongoing
32 hippocampal activity into several major contributions of stratified LFP-generators.
33 Here, using pathway specific LFP reconstruction, we have isolated LFP-generators that
34 describe the activity of Schaffer-CA1 and Perforant-Dentate excitatory inputs in the
35 anesthetized rat. First, we applied ICA and current source density analysis to LFPs
36 evoked by electrical subthreshold stimulation of the pathways. The results showed that
37 pathway specific activity is selectively captured by individual components or LFP-
38 generators. Each generator matches the known distribution of axonal terminal fields in
39 the hippocampus and recovers the specific time-course of their activation. Second, we
40 use sparse weak electrical stimulation to prime ongoing LFPs with activity of a known
41 origin. Decomposition of ongoing LFPs yields a few significant LFP-generators with
42 distinct spatiotemporal characteristics for the Schaffer and Perforant inputs. Both
43 pathways convey an irregular temporal pattern in bouts of population activity of varying
44 amplitude. Importantly, the contribution of Schaffer and Perforant inputs to the power
45 of raw LFPs in the hippocampus is minor (7% and 5 %, respectively). The results
46 support the hypothesis on a sparse population code used by excitatory populations in the
47 entorhino-hippocampal system and they validate the separation of LFP-generators as a
48 powerful tool to explore the computational function of neuronal circuits in real time.
49

2
50
51 INTRODUCTION
52 Multiple neuronal populations form a complex system of local circuits and global
53 networks whose activity constitutes the basis for information processing in the brain.
54 The electrical transmembrane currents produced by working neurons are mixed together
55 and generate local field potentials (LFPs), a macroscopic variable that reflects the
56 information processing at mesoscopic scales. Thus, the decoding of LFPs may help us
57 understand the principles of information handling. To date, resolving the electrical
58 activity of converging neuronal populations into the original informative sources
59 remains problematic (Mitzdorf 1985; Nunez and Srinivasan 2006).
60 Approaches are emerging to analyse LFPs, such as spectral coherence, principal
61 components, or laminar population analysis, and these have proven useful to identify
62 important features in mixed deep electrical sources (for example, see Kocsis et al. 1999;
63 Einevoll et al. 2007). Recently we implemented a blind source separation strategy
64 (Makarov et al. 2010) that takes advantage of the spatial immobility of electrical brain
65 sources in the LFP's time scale. This structural feature arises from the fixed location of
66 the synaptic afferents to target cells, producing transmembrane currents whose
67 extracellular part is circumscribed to the postsynaptic cell. The approach uses
68 independent component analysis (ICA) to separate the informative sources (Bell and
69 Sejnowski 1995; Makeig et al. 2004; Stone 2004) and it allowed us to disentangle
70 intracerebral LFPs recorded in the hippocampal CA1 region into the contributions of a
71 few major LFP-generators (Makarov et al. 2010). In this study, we present a special
72 technique that permits the unequivocal association of the LFP-generators isolated with
73 the presynaptic neuronal populations and its application to quantify the independent
74 contribution of Schaffer and Perforant inputs to the total hippocampal LFP. Thus, we
75 have obtained a tool for the parallel readout of the discrete activities of the working
76 neuronal populations from standard LFPs.
77 ICA is a widely used approach to extract spatially distinct independent sources of
78 activity from mixed signals (for review see Choi et al. 2005). Simultaneous multisite
79 recordings are required to estimate the different impact of several spatially localized and
80 distant generators into each sensor. This method has been successfully applied to
81 surface EEG recordings, whereby deep strongly synchronized neural sources can be
82 discriminated (e.g., Jung et al. 2005), but their spatial localization is only poorly

3
 83 assessed and it lacks cellular correlates. These problems are significantly diminished
84 when using intracerebral local recordings, since the signals are generated in the volume
85 surrounding the array of recording sensors. The parallel arrangement of principal cells
86 and the pronounced stratification of inputs in the hippocampus also favor the
87 application of ICA and interpretation of the results obtained.
88 Our recent application of ICA to LFPs in the rat CA1 disclosed several dominant
89 generators of LFPs, producing a subcellular definition of their spatial localizations that
90 was constant across animals (Makarov et al. 2010). A causal relation between the firing
91 of local units and the activity of specific LFP-generators was also shown. This result
92 suggests that each isolated LFP-generator may have a neuronal origin with different
93 functionality. Indeed, one may expect that the activity of an afferent input to a restricted
94 dendritic field in the principal cells may be associated with a single LFP-generator
95 disentangled by ICA from the complete LFPs. To test this hypothesis, here we have
96 used selective electrical stimulation of known hippocampal pathways such as the
97 Schaffer input to CA1 and the Perforant Path (PP) input to the Dentate Gyrus (DG).
98 Selective stimulation elicits evoked LFPs whose well known spatial maps are used to
99 crosscheck the ICA-derived LFP-generators in the first part of the study.
100 Hippocampal evoked potentials are regularly employed to explore the functional
101 properties of neuronal populations. Experimental and theoretical studies show they are
102 reliable population indices of the corresponding membrane events in single neurons
103 (Andersen et al. 1971; Varona et al. 2000; López-Aguado et al. 2002). Evoked synaptic
104 inputs produce inward and outward currents covering the entire morphology of principal
105 cells with pathway specific spatial distribution that can be precisely determined by
106 current-source density (CSD) analysis (Lorente de Nó 1947; Freeman and Nicholson
107 1975). As a first step, we have assessed the spatial and temporal performance of the
108 ICA, as well as its capacity to discriminate different sources using evoked LFPs of
109 known origin.
110 Subsequently, we studied spontaneous LFPs primed by random sparse subthreshold
111 evoked activity to identify the disentangled components. The presence of evoked and
112 well-known spontaneous events specific for each pathway in the raw LFPs, and in a
113 single LFP-generator, guarantees its correct unequivocal identification. We further
114 quantified the specific contribution of LFP-generators to the global signal during
115 irregular LFP patterns. Interestingly, the two major excitatory pathways, Schaffer and

4
116 Perforant, contribute little to ongoing LFPs, producing scarce and irregular bouts of
117 activity.
118
119 METHODS
120 Experimental methods
121 Female Sprague-Dawley rats (200-220 g) were anesthetized with urethane (1.2 g/kg
122 i.p.) and fastened to a stereotaxic device, maintaining their body temperature at 37° C
123 with a heated blanket. The surgical and stereotaxic procedures performed were
124 described elsewhere (Canals et al. 2005; Makarova et al. 2008). Concentric bipolar
125 stimulating electrodes were placed in at least two of the following locations: the alveus
126 for antidromic activation (from bregma, midline, and cortical surface AP: -5.5, L: 2.6,
127 V: -1.8 mm), the ipsilateral CA3 (septal pole) for orthodromic activation of the CA1
128 region (AP: -3.2, L: 2.6, V: -3.3 mm), and the angular bundle (AP: -8, L: 4-5, V: 3.5-4
129 mm) for orthodromic activation of the granule cell population in the DG. These points
130 were approached at a 30° angle in the sagittal plane. Stimuli (0.07-0.1 ms square pulses,
131 0.1-0.8 mA) were applied to elicit the characteristic evoked potentials, which were used
132 to guide the placement of recording probes. Linear multisite probes (Neuronexus
133 Technologies, Ann Arbor, MI, USA; models: A1x16-5mm50-177 and A1x32-6mm50-
134 413) were lowered into the hippocampus (AP: 4.5.-5.5, L: 2-3 mm) and connected to a
135 multiple high-impedance headstage. The 16 site probes were used to record along the
136 main axis of CA1 pyramidal cells, whereas the 32 site probes spanned both CA1 and
137 DG/CA3 regions. A silver chloride wire in the neck skin served as a reference for
138 recording, and the signals were amplified and acquired using low noise MultiChannel
139 System (Reutlingen, Germany) recording hardware and software (50 kHz sampling
140 rate). All the experiments were performed in accordance with European Union
141 guidelines (86/609/EU) and Spanish regulations (BOE 67/8509-12, 1988) regarding the
142 use of laboratory animals.
143
144 Histology
145 Silicon probes were labeled with DiI (Molecular Probes, Invitrogen, Carisbad, CA,
146 USA) by soaking in a 1 % solution of N-N-dimethyl-formamide (Sigma-Aldrich, St.
147 Louis, MO, USA) . At the end of the experiment animals were perfused transcardially
148 with saline/heparine and paraformaldehyde (4%). Coronal cryostat sections (15 µm)

5
149 were obtained and labeled with bis-benzimide (Sigma), and they were examined under a
150 fluorescence microscope (560-610 nm) to identify the probe position (see example in
151 the Supplementary Figure SM1).
152
153 Independent Component and Current Source Density Analyses of LFPs
154 The mathematical procedure and detailed signal treatment of the ICA were
155 described elsewhere (Makarov at al. 2010). Briefly, 16 or 32 LFP signals
u (t ) = {uk (t )}k =1 recorded simultaneously can be represented as the weighted sum of the
K
156

157 activities of N neuronal sources or LFP-generators1:

N
158 u (t ) =  Vn sn (t ) − σΔVn = I n ∀n ∈ [1, N ] (1)
n =1

159 where V = [V1 , V2 , ,VN ]T is the mixing matrix composed of the so-called voltage

loadings or spatial weights of all LFP-generators; {sn (t )}n=1 are the time courses or
N
160
161 activations of the LFP-generators; σ is the conductivity of the extracellular space; and
162 {I n }nN=1 are the CSD loadings. Thus the raw LFP observed at the k–th electrode tip is a
163 linear mixture of the electrical activity of several independent LFP-generators
164 describing transmembrane currents in principle cells.
165 Figure 1A shows an example of (evoked) LFPs recorded in the CA1 by a 16-site
166 probe along the pyramidal neuron axis. The raw LFPs can be used to evaluate CSD
167 maps by approximating the Laplace operator by 1D finite differences (Fig. 1B):
1 u k −1 (t ) − 2u k (t ) + u k +1 (t )
168 CSD = − Δu = − (2)
σ σh 2
169 where h = 50 μm is the intersite distance. The CSD shows the characteristic Schaffer-
170 evoked pattern, where several neuronal events can be appreciated overlapping in space
171 and time, such as generation of a propagating population spike.

1
Clarification of terminology. The term source is used with different implications in
ICA and CSD analysis. In the former, source is used in a comprehensive manner to
denote active brain regions (i.e., large zones of neural activity). Here a source in the
ICA-sense refers to each of the individual coherent components of the LFPs. Neither of
these should be confounded with the formal notation of outward transmembrane
currents or current sources employed in CSD analysis. For the sake of simplicity we
maintain the latter and adopt the term LFP-generator for the ICA-derived sources.

6
 The ICA of u (t ) returns the generator's activation {sn (t )}n=1 (Fig. 1C) and the
N
172

173 mixing matrix V (Fig. 1D) for up to 16 LFP-generators. Usually only a few of them
174 (G1-G6 in the figure) have significant amplitude and different spatial distributions
175 (voltage loadings). We noticed that due to the ambiguity of the ICA (fortunately
176 insignificant for our study) the voltage loadings and activations are given in arbitrary
177 units (i.e., we can arbitrarily scale the loadings but we must simultaneously apply
178 inverse scaling to the activations). Once LFP-generators have been extracted from the
179 raw LFPs, we can analyze them as if they were active alone. For example, we can
180 construct virtual LFPs produced by a single generator, say k-th, from its voltage loading
181 and activation:
182 uGk (t) = Vk sk (t) (3)

183 Figure 1F shows LFPs contributed by G6 only. Note that the ICA ambiguity does
184 not affect the virtual LFPs, i.e.: the reconstructed LFPs are dimensional. The virtual
185 LFPs can be used to evaluate the CSD created by a single generator (Fig. 1G shows the
186 CSD due to G6), and thus, the raw and generator-based CSDs can be compared (Figs.
187 1B and 1G). We also noticed that using Eqs. (1)-(3) the CSD related to the k-th
188 generator can be evaluated by CSDGk = Ik sk (t).
189 For the ICA we made use of the infomax algorithm proposed by Bell and Sejnowski
190 (1995) and further modified by Amari et al. (1996) and Lee et al. (1999). The algorithm
191 is implemented in the EEGLAB Matlab toolbox (Delorme and Makeig 2004).
192
193 Comparison of LFP-generators: Distance measure
194 Each application of the ICA on LFPs, to either different recordings or for different
195 time windows, provides several LFP-generators, and thus, we need a method to
196 compare and classify them. Each generator is characterized by its activation sk (t ) and

197 spatial loading Vk (Figs. 1C and 1D). Activations are time specific and therefore, they

198 cannot be used for comparison. By contrast, the loadings depend on the spatial
199 distribution of an axon's terminals and hence, they must be similar for LFP-generators
200 corresponding to the same population of presynaptic neurons. Thus, to define a measure
201 of similarity between a pair (k , m) of LFP-generators, we used the distance measure
202 between their voltage loadings (Makarov et al. 2010):

7
 Vk ,Vm
203 d (Vk ,Vm ) = 1 − (4)
Vk Vm

Vk ,Vm =  VkVm + κ∇Vk ∇Vm + κ 2 ΔVk ΔVm dx,

2
204 Vk = Vk ,Vk
Ω

205 where κ = 0.05 mm2 is the dimensional constant. The distance measure is bounded
206 0 ≤ d ≤ 1 and d = 1 for two orthogonal (completely different) loadings, whereas d = 0
207 for two equivalent (identical) loadings.
208 To partition a set of voltage loadings into disjointed clusters, we use hierarchical
209 clustering based on the introduced distance measure with the average-linkage algorithm.
210
211 Identification of stable LFP-generators
212 An ICA may theoretically isolate as many LFP-generators as the number of LFP
213 signals (16 or 32 in the present experiments) yet obviously, only a few of them will be
214 statistically significant and stable. For example in Fig. 1C, the first six generators have
215 high amplitude and they are clearly related to the stimulus, while the other ten may
216 capture noise and therefore, they may vary strongly from one recording to another. In
217 our former study (Makarov et al. 2010), we considered those LFP-generators
218 contributing more than 5% to the total LFP variance (power) as statistically significant.
219 The number of reliable components that can be extracted in practice depends on many
220 factors such as the particular region under study, the signal to noise ratio, and the
221 relative strength and spatial extension of active components. Some neuronal populations
222 (e.g., hippocampal pyramidal cells) fire very rarely (Ranck 1973; Thompson and Best
223 1989) and hence, the postsynaptic activity of their targets may contribute little to the
224 total LFPs. In this study, we consider an isolated LFP-generator with the variance
225 contribution below 5% to be significant, as long as it is unequivocally related to a
226 known pathway and provides a clean CSD spatial map (low noise and a flat baseline).
227 To identify the most stable LFP-generators in spontaneous LFPs we used the two-step
228 algorithm (Makarov et al. 2010). First, we identified "global" generators through an
229 ICA of a sufficiently long recording (tens of seconds). Second, we divided the same
230 recording into short-term contiguous epochs (each 1 s long) and we applied an ICA to
231 each of these segments. We expect that stable strong LFP-generators should be
232 presented in all epochs, whereas noisy, unstable, temporal or weak generators will
233 fluctuate and hence, their spatial loading will differ from epoch to epoch. Then, we

8
234 identified those that are most stable over time by calculating the similarity measure (4)
235 between the "global" and "local" generators.
236
237 Power of LFP-generators
238 Using Eq. (3) we reconstructed virtual LFPs corresponding to the generator Gk , and
239 we then defined the mean power of Gk (measured in mV2) as:
1 T 
240 PGk = max 
m∈[1,K ] T

u 2
m (t) dt 

(5)
0

241 where T is the time interval for averaging (about a few minutes in our experiments). To
242 evaluate the evolution of the power of the generator’s activations over time, we use the
243 convolution with the appropriately scaled square kernel H :
244 PGk (t) =  H(t − τ )s (τ ) dτ
2
k (6)

245
246 Analysis of evoked LFPs
247 Evoked LFPs were recorded for a series of stimuli of varying intensity (short, 0.1
248 ms, square pulses at a rate of 0.1-1 Hz), from a low subthreshold to supramaximal (i.e.,
249 producing maximal population spike, PS). The stimulus intensity was measured in units
250 of the threshold intensity to evoke a PS and stimulus artifacts were removed, except
251 where indicated. Each series of evoked LFPs was analyzed as one long signal
252 containing either all stimuli, or selected groups of evoked responses. Only the period of
253 interest (40-50 ms) immediately after the stimulus was analyzed for each profile and
254 thus, intermediate periods were removed offline and the time windows containing
255 responses were then concatenated into a continuous unique signal. This procedure does
256 not affect the applicability of ICA but rather, it improves its efficiency by increasing the
257 relative variance of the signal of interest.
258
259 Priming of LFPs with evoked activity
260 In order to identify the presynaptic populations corresponding to LFP-generators,
261 we stimulated known afferent pathways with random sparse subthreshold pulses. The
262 rationale was that the minimal electric activation of a pathway emulates its spontaneous
263 excitation. Therefore, the evoked and spontaneous activities of a given pathway must
264 have the same spatial distribution as defined by the axon's terminals. Thus, two types of
265 LFPs with the same spatial characteristics must be captured by one and the same LFP-
9
266 generator and hence, the presence of evoked responses in the activation of the LFP-
267 generator sk (t ) assists in its identification. In preliminary experiments, we found that

268 subthreshold (<0.5 x threshold) stimuli delivered at an average rate of 1 Hz do not

269 significantly change the variance of the total signal. We also found that strong (near
270 threshold and suprathreshold) evoked responses produce a complex mixture of
271 components when analyzed by ICA/CSD (see Fig. 1). Apparently such phenomena are
272 not present in the low amplitude events of ongoing LFPs and thus, we chose low
273 stimulus intensities to prevent the generation of PS components (i.e., we achieve
274 vanishing recruitment of intrinsic currents into fEPSPs during synchronous activation,
275 Herreras 1990). For Schaffer activation, we chose stimuli that produced fEPSPs of ≤ 0.5
276 mV in the stratum radiatum and for the PP input, the intensity was adjusted to raise a
277 positive field in the Hilus ≤ 1 mV (below 0.5 mV in the DG molecular layer). To
278 analyze the LFPs we used a two epoch recording: 1) spontaneous and 2) spontaneous
279 plus evoked activities. Thus, we can examine whether evoked activity modifies the level
280 of spontaneous activity on the same or another LFP-generator. In accordance with the
281 stability analysis of the LFP-generators (Makarov et al. 2010), we found that mixing
282 sparse subthreshold evoked activity into ongoing LFPs neither modifies their
283 characteristics nor affects the spatial profiles of the LFP-generators, as long as the probe
284 array remained in place.
285
286 Reconstruction of LFPs for specific pathways
287 Reconstructed LFPs recover the correct polarity and provide a good quantitative
288 estimate of the pathway-specific variance and power. The power was calculated in the
289 electrode that showed maximum amplitude during irregular LFP patterns (i.e. excluding
290 theta periods) over a total of 3 min of 20 s randomly chosen segments within 1 hour. A
291 major advantage of pathway-specific LFPs is that they provide simplified CSD maps
292 describing a unique spatial structure of inward and outward currents. These can easily
293 be matched to those obtained from customary hippocampal evoked fields, whose
294 uncomplicated nature and well-defined distribution along the anatomy of principal cells
295 assist path identification (Lømo 1971; Herreras et al. 1987; Herreras 1990; Leung et al.
296 1995).
297
298

10
299 RESULTS
300 We first tested the spatial and temporal performance of the ICA on evoked LFPs
301 obtained by activation of the Schaffer-CA1 and PP-Dentate inputs. Subsequently, we
302 employed the pathway specificity of the ICA-derived LFP-generators to explore the
303 spontaneous population activity of these pathways in ongoing LFPs.
304
305 Spatial and temporal precision of pathway-specific LFP-generators
306
307 SEPARATION OF ARTIFACTS AND NEURAL ACTIVITY. The capacity of the ICA to suppress
308 EEG artifacts is well established (see Vigario et al. 2000; Castellanos and Makarov
309 2006, and references therein) and thus, we assess here its applicability to isolate
310 stimulus artifacts (due to electric shock) in deep brain recordings. We antidromically
311 stimulated the CA1 pyramidal population from the alveus (Fig. 2A, cartoon inset) and
312 the electric shock produced a large negative spike-like potential (>40 mV, truncated at
313 the black dot in Fig. 2A,a) that may reverse polarity outside the hippocampal tissue.
314 This artifact was followed by evoked LFPs with the characteristic profile of the
315 antidromic PS (Varona et al. 2000). The corresponding CSD map (Fig. 2A,b) shows the
316 somatodendritic sink (blue, electrodes 7-11) surrounded by passive sources
317 (yellow/red). The polarity of currents reverses in the ensuing slower tail, as expected for
318 somatic recurrent inhibition (active somatic currents flanked by passive basal and apical
319 dendritic currents).
320 The ICA successfully isolated the artificial potential as a single generator, G1, that
321 was devoid of neuronal activity (Figs. 2A,c and 2A,d, blue curves). We noted that the
322 spatial voltage loading of the artificial generator is essentially flat (Fig. 2A,c blue line).
323 Such a generator, which enters all electrodes at the same strength or with a linear decay,
324 has a null CSD [Eq. (2)], which means that its origin lies outside the recording zone and
325 that the generator is volume propagated, as expected. The second strongest LFP-
326 generator, G2, contains the main part of the neuronal activity, including the PS and
327 inhibitory currents. The LFP variance explained by the remaining G3-G16 is below 5%
328 (Fig. 2A.e). Thus, when evoked LFPs contain a large artifact the resolution of the mixed
329 neuronal activity significantly deteriorates. Hence, in subsequent analysis we routinely
330 cut off the stimulus artifacts before applying the ICA.

11
331 We found that the results shown in Fig. 2A were stable in all animals and the
332 clustering analysis of voltage loadings of all the LFP-generators in the 7 animals
333 provided a well-defined partition into two groups (Fig. 2B,a: intracluster distance <
334 0.15, intercluster distance > 0.5). This is evident in Figure 2B,b which shows the mean
335 (± s.d.) spatial loadings for the artificial G1 and neuronal G2 generators averaged over
336 loadings belonging to the corresponding cluster ( n = 7 ).
337
338 SUBTHRESHOLD ACTIVATION OF AN AFFERENT PATHWAY IS CAPTURED BY A SINGLE LFP-
339 GENERATOR WITH PRECISE SPATIOTEMPORAL DEFINITION. Let us now examine the ICA’s
340 performance on LFPs evoked by activation of Schaffer-CA1 and PP-DG inputs (a
341 detailed description of evoked LFPs and their CSD analysis can be found in the
342 Supplementary Material, Fig. SM1). We first studied the excitatory Schaffer input to the
343 CA1 that was activated by subthreshold (0.5 x threshold) stimuli to minimize the
344 secondary intrinsic currents (Herreras 1990). Figure 3A shows raw profiles of evoked
345 LFPs and the corresponding CSD map. It can be appreciated a single active sink
346 centered in the st. radiatum surrounded by passive current sources in the soma region
347 and apical distal dendrites.
348 Application of the ICA identified 16 LFP-generators and only one of them, G1, had
349 statistically significant amplitude (Fig. 3B). The activation of G1 matches the time
350 course of the raw LFP at the site of maximum amplitude (electrode # 9 in Fig. 3A,a). G1
351 has a bell-shaped spatial distribution (Fig. 3B,c, blue curve) that is maximal in the apical
352 dendrites and its polarity reverse in the soma layer, in close correspondence to the
353 spatial profile of LFPs calculated for the time instant marked by the arrow in Fig. 3A,a
354 (Fig. 3B,c, black line). The spatiotemporal dynamics of the CSD obtained from the
355 LFPs reconstructed from G1 matched the CSD map calculated from the raw LFP profile
356 (Figs. 3B,e vs. 3A,b). We noticed that the three-layered source-sink-source distribution
357 caused by excitatory synaptic and return currents was not decomposed by the ICA into
358 three different components but rather, it was explained by the single G1 LFP-generator.
359 The transmembrane source/sink loop of current raised by a discrete patch of active
360 membranes is contained within the shape of the spatial weights of the ICA-isolated
361 LFP-generator [loading vector V1 in Eq. (1)].
362 Similar results were repeatedly obtained in five animals and in a clustering analysis
363 of the loading curves obtained in different experiments, the inter-loading distance was

12
364 below 0.1 for all generators (Fig. 3C,a). In one experiment we obtained two significant
365 generators: one of them (#2a in Fig. 3C,a) had a similar voltage loading as that
366 described above (Fig. 3B,c), while the other (#2b in Fig. 3C,a) was significantly
367 different. Hence, in this experiment it would appear that some spontaneous LFP event
368 of a different origin overlapped with the evoked activity, which was then isolated into a
369 separate LFP-generator. The mean (± s.d.) voltage loading for the Schaffer synaptic
370 input averaged over n = 5 experiments is shown in Figure 3C,b.
371 In a representative experiment that examines PP-DG evoked responses (Fig. 4),
372 subthreshold (0.5 x threshold) activation of the PP produced the expected negative LFP
373 in the molecular layer of both blades and a strong positive potential across the Hilus
374 (Fig. 4A). The CSD map revealed strong sinks in both molecular layers (sk1, sk2) and a
375 weaker sink (sk3) produced by CA1 pyramidal cells in the st. lacunosum, in close
376 correspondence to the known anatomical terminations of these fibers. Passive sources
377 (return currents) appeared in the GCL (cartoon in Fig. 4A, right). The ICA of LFPs
378 again rendered only one significant LFP-generator, G1 (activation in Fig. 4B,a and
379 loading in Fig. 4B,d), while the other 15 were negligible (Fig. 4B,b). The CSD map of
380 the virtual LFPs reconstructed for this unique generator [Eq. (3)] matches the CSD of
381 the raw signal (Figs. 4B,c vs. 4A). Likewise the voltage and CSD loading curves (Fig.
382 4B,d) match the spatial profiles for the raw signals. Thus, one ICA-isolated LFP-
383 generator may contain synaptic activity from several separate populations, as long as
384 they are simultaneously activated by a common input.
385 Qualitatively, the same results were observed when we repeated this analysis in five
386 animals. However, since the separation between the GC layers varies due to the curved
387 arrangement of this population, the anatomical normalization of the recording tracks
388 across animals is not ideal. This precludes direct comparison of spatial loadings
389 obtained in different experiments. Figure SM2 in the Supplementary Material shows the
390 results from the other four experiments.
391
392 SEPARATION OF THE ACTIVITY EVOKED BY MULTIPLE PATHWAYS. Due to the volume
393 propagation of extracellular currents, a LFP may contain activity from several domains
394 within the same neurons or even from different distant populations and at a distance
395 from the recording electrode. Thus, we tested whether the ICA can decompose LFPs

13
396 with a contribution from two pathways into the original Schaffer and PP inputs to the
397 CA1 and DG, respectively.
398 Conveniently, the pyramidal and granule cell populations are spatially segregated
399 and they can be covered by different groups of electrodes in the 32-site linear probe. In
400 the two representative experiments shown in figure 5, a series of subthreshold paired-
401 pulses of stimuli were delivered at a rate of 0.2 Hz rate to the Schaffer and PP pathways.
402 These pathways were either stimulated with a 100 ms delay or simultaneously (left and
403 right columns in Fig. 5, respectively). The stimulus intensity was maintained constant
404 for the Schaffer input, while for the PP input it increased over time for non-coincident
405 pairs of stimuli (Fig. 5A) while it decreased for the coincident pairs (Fig. 5B), yet
406 always remaining below a 0.5 x threshold. The LFPs recorded for the two stimulus
407 configurations by the 32-site linear probe extended from the CA1 st. oriens to the lower
408 GCL can be seen in figure 5A (see Fig. SM1 in the Supplementary Material for greater
409 detail).
410 Non coincident stimuli. An electric shock in the ipsilateral CA3 produces the
411 Schaffer fEPSP in CA1, as well as a local multiphasic field in the CA3/DG. The CSD
412 map shows the characteristic source-sink-source distribution of currents along the CA1
413 pyramidal cell axis and a local source-sink dipole (Fig. 5B, Sch). This latter dipole
414 contains a mixture of the antidromic sink in the area of the granule cell and of the local
415 recurrent orthodromic fEPSPs in CA3 (this local response may contain PP activation in
416 other experiments, depending on the position of the stimulating electrode). Activation of
417 the PP pathway produced an excitatory sink in the molecular layer (100 µm above the
418 CA3 sink), as well as the corresponding somatic sources in both GCLs (Fig. 5B, PP). A
419 weaker sink was also observed in the st. lacunosum of the CA1 region.
420 The ICA of the LFPs yielded three significant LFP-generators, G1-G3, with little if
421 any cross-contamination (Fig. 5C). Noticeably, the time-course of G1 activation shows
422 increasing pulses in agreement with the increasing amplitude of the PP stimuli. In
423 addition, the beginning of the G1 pulses coincides with the onset of the stimuli in this
424 pathway. The other two LFP-generators, G2 and G3, exhibit pulses of constant
425 amplitude (as for the Schaffer stimulation), whose onset coincide with the beginning of
426 the stimulation to the Schaffer input. Thus, the LFP-generator G1 exclusively represents
427 the activity in the molecular layer of the DG (compare also to Fig. 4), whereas the CA3-
428 evoked LFPs were decomposed into two generators: G2 responsible for the Schaffer-

14
429 mediated CA1 excitation (compare to Fig. 3) and G3 corresponding to the local
430 CA3/DG activation. The CSD maps of the virtual LFPs reconstructed from each of the
431 isolated generators (Fig. 5D) also confirm the pathway and population specific nature of
432 the LFP-generators. Therefore, ICA provides an accurate decomposition of multiple
433 inputs that are not temporally coincident.
434 Coincident stimuli. In principle, the tight temporal coincidence of multiple sources
435 decreases the efficiency of ICA to separate them. Thus we checked its performance on
436 two precisely synchronized different inputs. When the stimuli were delivered
437 simultaneously to the Schaffer and Perforant pathways (Fig. 5, right column) there
438 appeared to be a complex mixture of the sources and sinks in the raw CSD map.
439 Notably, the proximity of the CA3 and GC layers led to fusion of their respective sinks
440 into a unique slightly drifting sink (Fig. 5B), which made the sinks indistinguishable and
441 complicated its direct interpretation.
442 Even in this case, the ICA still rendered three significant LFP-generators. Each of
443 the generators had voltage loadings similar to those found in the non-coincident case
444 (Fig. 5C, left vs. right), although their separation was less effective. As such, the
445 strongest LFP-generator, G1, responsible for the PP input, introduced some
446 contamination into the other two. This cross-contamination could be appreciated by
447 comparing the loading curves in the left and right columns. We noticed that in the lower
448 part of the loading curve (DG/Hilus) for the Schaffer G2 (arrow pointing to red curve),
449 there was a spatial weighting analogous (disregard the polarity) to that of the PP-
450 specific G1 generator (arrow pointing to blue curve). The CSD maps for the
451 reconstructed LFPs (Fig. 5D) show that the Schaffer-specific G2 generator contains
452 some current belonging to the CA3 and FD regions (left vs. right columns in Fig. 5D).
453 Nevertheless, even in this extreme case of the perfect time coincidence of different
454 inputs, the specific activations of the LFP-generators confirm an adequate separation.
455 Indeed, only G1 exhibits decaying amplitude of activation in agreement with the
456 decaying intensity of the stimulus applied to the PP input, while G2 has stationary
457 amplitude in agreement with constant stimulation of the Schaffer pathway. All these
458 results were reproduced in four additional animals.
459
460 Schaffer and PP population activity in ongoing LFPs
461

15
462 We have shown that ICA can separate Schaffer and PP evoked activities mixed in
463 LFPs and we have described their spatiotemporal characteristics. Neither of the spatial
464 weight curves (loadings) for these pathways matched those found in our former study
465 that described the most powerful generators in spontaneous LFPs (Makarov et al. 2010).
466 This observation suggests that the LFP-generators describing the activity induced by the
467 excitatory pathway may contribute weakly to the spontaneous LFPs. Thus, we sought
468 after the LFP-generators for the Schaffer and PP pathways in ongoing LFPs by mixing
469 in subthreshold stimuli that facilitate the identification of the generators.
470
471 IDENTIFICATION OF SCHAFFER AND PP POPULATION ACTIVITIES IN ONGOING LFPS. A
472 fragment of ongoing LFPs was analyzed during which both pathways were activated in
473 a random-like manner at an average rate of 1 Hz each (Fig. 6A). The vertical dashed red
474 and blue lines mark the onset of the Schaffer and PP stimuli, respectively. Due to the
475 weak intensity of the electrical stimulation, the evoked responses are hardly discernible
476 in the raw LFPs. By applying an ICA, we found five main LFP-generators (Figs. 6B and
477 6C), two of which exclusively contain the evoked responses to the Schaffer (G2) and PP
478 (G4) stimuli (Fig. 6C). We noticed that the activations of the other LFP-generators (G1,
479 G3, and G5) did not correlate with the onset of the stimulus, indicating that there was
480 little if any cross-contamination between these generators. The remarkable specificity of
481 the activities contained within G2 and G4 (Fig. 6B) was proof that both are truly
482 independent and well-isolated LFP-generators responsible for the Schaffer and PP
483 inputs, respectively. The shape of the voltage loadings (spatial weight curves) for G2
484 and G4 are essentially identical to those found in experiments where only evoked
485 activity was assessed (compare Fig. 6C, red and blue curves to Fig. 3B,c and Fig. 4B,d,
486 respectively, and also to Fig. 5C red and blue curves). The portion of the variance
487 explained by the two primed pathways (bars in red and blue) is small (below 5%, Fig.
488 6D). This reduced contribution of the Schaffer and Perforant pathways to LFPs makes
489 their identification in ongoing spontaneous LFPs extremely difficult. As such, LFP-
490 generators containing their activities fall below the level of significance and hence they
491 can be overlooked.
492
493 RECONSTRUCTION OF PATHWAY-SPECIFIC LFPS AND THEIR SUBCELLULAR SPECIFICATION.
494 Disentangling the raw LFPs into LFP-generators enables the reconstruction of virtual

16
495 LFPs produced by a specific pathway during ongoing activity and their separate
496 analysis. An example is illustrated in Figure 7 for an LFP fragment containing three
497 evoked (subthreshold) pulses (applied to Sch or PP pathways; the onset of each is
498 marked by a dashed line in Fig. 7A). The LFP was recorded from the pyramidal/oriens
499 border in CA1 to the lower blade of the DG. The corresponding CSD map shows a
500 confusion of sources and sinks produced by the ongoing spatiotemporal activity of
501 different origins (Fig. 7B).
502 The ICA of the LFP fragment rendered five LFP-generators (Fig. 7C) and their
503 spatial loadings and time course of activation can now be used to identify pathway-
504 specific evoked activity. By searching for the evoked responses amongst all generators,
505 and comparing their activations and voltage loadings to those shown in Figs. 6B and 6C,
506 respectively, we identified G2 and G4 as responsible for the Schaffer and Perforant
507 inputs, respectively. Due to the ambiguity of the ICA (see Methods), the time course of
508 the activations are given in arbitrary units and are not directly comparable. However, we
509 can reconstruct the virtual LFPs created by each single generator. Figure 7D shows
510 LFPs reconstructed for the Schaffer generator (G2). The spontaneous population
511 activity and two of the evoked events can be tracked down in the raw signal. The
512 reconstructed evoked potential profile shows no difference to its raw counterpart, as
513 expected (see the ellipses in Figs. 7A and 7D, amplified in the right insets). By visual
514 inspection of the raw and Schaffer reconstructed LFPs, it is patent that ongoing raw
515 LFPs are most prominent in the hilus of the DG (lower electrodes), while Schaffer
516 mediated activity is more conspicuous in the CA1, with some activity in the CA3/DG
517 region that is probably due to recurrent excitation of presynaptic pyramidal cells in the
518 CA3 region.
519 As further confirmation of the specificity of the Schaffer generator (G2) identified,
520 we compared its activation (Fig. 7B, G2) with the high-frequency activity (about 100-
521 200 Hz) associated to sharp-waves, an electrographic event known to be caused by
522 synchronous activation of the CA3 region (i.e., Schaffer-mediated: Buzsáki et al. 1983).
523 Indeed, the specific bouts of activity contained within the Schaffer G2 generator were
524 coherent with ripple-like fast activity in the CA1 perisomatic LFPs (black upper trace
525 in Fig. 7B; bandpass: 100-500 Hz).
526 To obtain the precise localization and magnitude of the underlying transmembrane
527 currents, we evaluated the CSD of the reconstructed LFPs (Fig. 7E). As expected, the

17
528 Schaffer population activity presented a main sink in the st. radiatum surrounded by
529 passive sources (return currents) in the st. pyramidale and lacunosum, matching the
530 locations of the well-known CSD map for evoked Schaffer activity (Fig. 3). The CSD
531 maps of isolated pathway-specific LFPs contrasted with those obtained for the mixed
532 signal (compare Fig. 7B to 7E). It is worth mentioning that the temporal activation of
533 isolated components corresponds to that of the population synaptic currents (e.g.,
534 compare the evolution of the CSD map at electrodes 8-9 in Fig. 7E to the corresponding
535 activation in Fig. 7C, red trace). Thus, LFP-generators isolated by the ICA enable the
536 population-specific transmembrane current density to be examined, rejecting
537 contamination of the other inputs in ongoing LFPs.
538
539 MINOR CONTRIBUTION OF EXCITATORY PATHWAYS TO HIPPOCAMPAL LFPS. The Schaffer
540 and Perforant generators (G2 and G4) identified that describe the activation of the
541 corresponding pathways enabled the spontaneous activity of these pathways to be
542 studied and quantified (i.e., the presynaptic CA3 and entorhinal afferent populations,
543 respectively). In general, we found that the spontaneous activity (during irregular LFPs)
544 of both the Schaffer and PP generators represents an irregular pattern of short bouts over
545 a rather flat baseline. In the typical traces of the activities of the Schaffer and PP
546 generators and their wavelet spectra, bouts (hundred milliseconds time scale) of variable
547 amplitude appear in average at 1 Hz rate (Fig. 8A). Smaller and shorter events can also
548 be appreciated, which may be physiologically relevant. The main frequency content of
549 the activity concentrates in the 0.5-4 Hz range for both generators. To track the
550 evolution of the generator's power over time, we calculated time envelopes of this
551 activity [see Methods, Eq. (6)] where large events and baseline activity can be better
552 distinguished (Fig. 8B).
553 The dimensional power (measured in mV2) that contributed to the ongoing LFPs for
554 each synaptic pathway can be estimated from the virtual LFPs reconstructed from the
555 corresponding LFP-generators [see Methods, Eqs. (3) and (5)]. The Schaffer and PP
556 inputs produce 0.045 and 0.031 mV2 on average (7% and 5 % of the total power),
557 respectively (Fig. 8C), while the contributions of the other generators ranged between
558 0.13-0.24 mV2 in four animals. This result corroborates the relatively small contribution
559 of the excitatory pathways into the total LFP variance observed above (Fig. 6D), but
560 now excluding the evoked activity. We also noticed that the powers of the Schaffer and

18
561 PP generators have smaller standard errors relative to the powers of G1, G3, and G5,
562 which suggests that the Schaffer and PP inputs described are more stationary in time
563 and between different animals than the other three LFP-generators.
564
565 DISCUSSION
566 Studying the mechanisms of neural information processing requires certain
567 knowledge of the circuitry, and of the causal and functional associations between
568 activities of different neuronal groups. Here, we have shown that an ICA can effectively
569 separate standard LFPs in the hippocampus into pathway-specific isolated activities,
570 called here LFP-generators, describing the inputs of multiple presynaptic populations.
571 The main requisite for the separation is the different spatial distribution of
572 transmembrane currents along the somatodendritic axis of principal cells, whereas the
573 temporal pattern of the pathway's activation is less important (e.g., they can be rhythmic
574 or irregular).
575 Using selective subthreshold electrical activation of major excitatory synaptic
576 pathways, such as the Schaffer and PP inputs, we have described spatiotemporal
577 features of LFP-generators specific to each. Subsequently, the Schaffer and PP-
578 generators found were used to quantify the pathway-specific contribution to ongoing
579 LFPs. Hence, we have consistently shown in all studied animals that the contribution of
580 the major excitatory pathways to the power of ongoing LFP signals is minor (about 7%
581 and 5%, respectively). We have also shown that both Schaffer and Perforant pathways
582 have irregular temporal patterns with bouts of population activity of varying amplitude
583 and with the frequency concentrated in the 0.5 - 4 Hz range.
584
585 Methodological considerations
586 A critical consideration regarding the ICA is that the components found may have
587 little functional relevance unless accurate information on their anatomical basis is
588 provided. All the same, ICA can efficiently separate signals coming from distant point
589 sources (Choi et al. 2005). There are few studies on ICA of deep brain recordings (see
590 Tanskanen et al. 2005; Makarov et al. 2010). In deep recordings, the signals generated
591 in the volume surrounding the electrode array (i.e., the sources or LFP-generators) are
592 essentially extended spatially and their spatial scales may be larger than the
593 interelectrode distance. Still, the same mathematical principles are applicable to LFP

19
594 recordings (Makarov et al. 2010). Disentangling such LFP-generators is based: i) on the
595 coincident activation of individual homologous fibers over the millisecond time scale;
596 and ii) on their different spatial distributions. The first condition provides the necessary
597 synchronization of homologous axons to raise significant macroscopic synaptic field
598 potentials, the main contributors to normal LFPs (Mitdorf 1985; Nunez and Srinavasan
599 2006). The second condition is ensured by the distinct distributions of afferent axons
600 from different afferent populations. The accuracy of the spatial resolution is therefore
601 optimized to the subcellular level of the active neurons. We have shown that the spatial
602 CSD-loading curves of the Schaffer and PP LFP-generators match the configuration of
603 the inward and outward currents for known membrane events (Kandel et al. 1961;
604 Andersen et al. 1971; Herreras 1990). Admittedly, some factors can modify the
605 expected spatial pattern of field potentials, such as the heterogeneous electrical
606 properties of the conducting medium, spatial cancellation, and frequency scaling of
607 currents in the volume conductor (e.g., López-Aguado et al. 2001; Bédard et al. 2004;
608 Makarova et al. 2008). In earlier experimental and modeling work we reported some
609 particular cases where some of these may be relevant (Varona et al. 2000; López-
610 Aguado et al. 2001, 2002; Makarova et al. 2008). Such studies cannot easily be
611 extrapolated on the distribution of LFP-generators. Since the present analysis is
612 performed on locally recorded LFPs we speculate that the factor potentially most
613 relevant is the mutual spatial cancellation between individual synaptic inputs when
614 these are scattered through extended dendritic domains (e.g., Makarova et al. 2010). The
615 excitatory Schaffer and PP inputs are strongly stratified, hence we would not expect a
616 strong bias introduced by this factor either.
617 The current approach is not dependent on the presence of rhythmic activity or
618 stereotyped LFP events and hence, it can be applied to a wide variety of behaviors.
619 Indeed, all the examples shown in this report were chosen amongst epochs of irregular
620 activity, whose behavioral correlates are very rich (see Jarosiewicz and Skaggs 2004).
621 In neuroscience, ICA is frequently used to suppress artifacts in EEG recordings (see
622 Castellanos and Makarov 2006). We have shown that artifacts due to electric shocks
623 within brain tissue can also be isolated and suppressed by the ICA. However, strong
624 artifacts (up to 100 times higher than the neuronal activity) decrease the efficiency of
625 the ICA for the simultaneous separation of other mixed components of neuronal origin.
626 Hence, we suggest that artifacts should be removed before performing an ICA.

20
627 The weak contribution of the Schaffer and Perforant pathways to ongoing LFPs
628 makes their identification extremely difficult (i.e., LFP-generators can be overlooked).
629 Therefore, we used selective low magnitude electrical stimulation of the known
630 excitatory pathways to ensure the reliable identification of LFP-generators describing
631 the activity of these pathways. Thus, the identification was based on: i) a comparison of
632 spatial voltage loadings of the LFP-generators found for evoked responses with those
633 found for ongoing LFPs; and ii) the recognition of evoked pulses in the time courses of
634 LFP-generators obtained by ICA of composite epochs (i.e., a recording consisting of
635 two epochs: spontaneous LFPs and spontaneous plus a few evoked LFPs).
636 The polarity of the CSD-loading is essential to define whether an LFP-generator
637 describes excitatory or inhibitory input. However, the ICA cannot unequivocally
638 determine this feature (see Hyvärinen and Oja 2000). Although this problem is hard to
639 solve for spontaneous LFPs and it requires special biophysical assumptions (Makarov et
640 al. 2010), we found that mixing sparse subthreshold evoked potentials in LFPs can be
641 used to resolve the problem. The presence of evoked pulses in the time course of an
642 LFP-generator is related to the stimulated neuronal population and it helps identify the
643 correct polarity. The subsequent application of CSD analysis on virtual LFPs
644 reconstructed from a single generator enables the resultant partial CSD map to be
645 directly compared with the raw CSD and with CSD maps of well-known evoked
646 responses of specific pathways. It is worth noting, that CSD map of isolated generators
647 show no alternations in terms of the current’s sinks and sources, as expected for a
648 specific synaptic input. By contrast, raw CSD maps may contain such alternations due
649 to the spatiotemporal mixing of the activities of several generators and DC filtering (see
650 Brankačk et al. 1993).
651
652 When and what can an ICA discriminate in LFPs?
653 To correctly interpret LFP-generators isolated by an ICA it is relevant to ask
654 whether each of them corresponds to postsynaptic activity from a single neuron group
655 or it may contain activity from multiple cell types. This problem is simplified greatly by
656 the use of linear probes in the hippocampus where each subfield has only one type of
657 principal cell, namely pyramidal and granule cells. Each synaptic input produces loops
658 of inward/outward currents with a different spatial distribution within the same core
659 conductor. Typically the maximum potential is attained at the center of the synaptic

21
660 contact made by the isolated pathway (Herreras 1990). However, distant recording sites
661 also contain activity, albeit of reduced magnitude. Therefore, steadily correlated
662 recording sites correspond to anatomical subdomains of one type of principal cell, while
663 the subcellular precision can be achieved through a CSD analysis of the distribution of
664 spatial weights in isolated generators. This procedure enables the separation of inputs
665 into a single population according to known stratification along the dendritic arbor.
666 However, when linear recording probes extend through several subfields (e.g., CA1
667 and DG/CA3), the interpretation may not be that trivial due to volume propagation of
668 currents and/or the simultaneous activation of populations in both subfields. The
669 electrical fields created by two physically separated cells extend beyond their respective
670 physical limits, mixing in local recordings. This is no handicap for the present approach
671 as the temporal activation for each generator has no spatial dimension (i.e., it contains
672 the coherent activity of a group of electrodes regardless of their location). Accordingly,
673 the CSD examination reveals whether the activity can be circumscribed to one or
674 several physically separated cell generators. In the present study we found that the two
675 excitatory generators appear to have true activity in more than one region. Thus the
676 activity of the entorhino-hippocampal pathway (PP) was contained in only one LFP-
677 generator, but it was divided between both blades of the DG. The parsimonious
678 explanation is that the presynaptic population activates two or more physically separate
679 target populations through bifurcating axons, in agreement with anatomical and
680 electrophysiological findings (Lømo 1971; Hjorth-Simonsen and Jeune 1972). Hence,
681 activation is identical and simultaneous in both blades and thus, it can be described by a
682 single LFP-generator isolated by the ICA. With regards the Schaffer generator, although
683 CA3 activation rendered a single component in CA1 whose spatial loading curve
684 matches the termination of Schaffer collaterals (Herreras 1990), prominent activity was
685 also obtained near the stimulating electrode in the CA3/DG. The origin of this local
686 activity varied somewhat in different experiments and was most commonly found in the
687 molecular layers of the DG for evoked responses, probably due to backfiring of the PP
688 fibers (Wu et al. 1998). During spontaneous LFPs, the CA3/DG activity associated to
689 the Schaffer generator is less prominent and it most likely reflects the synchronous
690 activity of CA3 pyramidal cells due to recurrent excitation in this area (Miles and Wong
691 1987). Thus, the CA3 pyramidal population is presynaptic for both CA1 and itself, and

22
692 therefore the ongoing activity in the Schaffer generator should appear simultaneously in
693 both regions, as happens to be the case.
694
695 Excitatory currents contribute little to hippocampal LFPs
696 Some LFP events and rhythmic oscillations have been studied intensively and the
697 neuronal circuits producing them have therefore been clarified. For example, sharp
698 hippocampal waves are known to be mediated by the excitatory input from CA3 to CA1
699 populations (Buzsáki et al. 1983). Here, we also observed a correlation between sharp
700 waves and the ICA isolated Schaffer generator describing the synaptic activity produced
701 by CA3 pyramidal cells in CA1. Some LFP events, such as the theta rhythm, have a
702 mainly inhibitory origin (see Soltesz and Deschennes 1993), whereas others are still
703 under study, such as gamma oscillations (Mann et al. 2005; Csicsvari et al., 2003), fast
704 ripples (Ylinen et al. 1995) and dentate spikes (Bragin et al. 1995). In the light of the
705 proportion of excitatory and inhibitory neurons in the hippocampus (Aika et al. 1994)
706 the small contribution of the major excitatory pathways to hippocampal LFPs might
707 appear unexpected (Schaffer and PP-generators explain 7% and 5% of the power). The
708 other three LFP-generators show a significant phase-spike relationship with about 60%
709 of the putative interneurons in CA1 (Makarov et al. 2010), suggestive of their inhibitory
710 nature (i.e., the corresponding synaptic currents are produced by local inhibitory
711 circuits). Circumstantial data in the literature argues in favor of the dominance of
712 inhibitory over excitatory currents in ongoing population activity (see Herreras et al.
713 1988; Hajos et al. 2000; Oren et al. 2010), although direct evidence is scarce. The
714 dominance of inhibitory currents was found in synaptic noise recorded from membranes
715 of cortical pyramidal cells (Rudolph et al. 2005; see also Borg-Graham et al. 1998;
716 Anderson et al. 2000). Besides the firing rate of several types of interneurons in the
717 hippocampus is much higher than that of pyramidal cells (Somogyi and Klausberger
718 2005). Along with the functional coordination of some interneuron networks
719 (Whittington et al. 1995), the hypothesis that inhibition prevails over excitation in
720 hippocampal LFPs becomes more plausible. Nonetheless, the precise percentage of
721 excitatory contribution to LFPs found here may vary in the freely moving animal, since
722 anesthesia modifies the overall pattern of LFPs (Buzsáki et al. 1986; Brankačk et al.
723 1993).

23
724 In any case, the results presented here point to a very low output rate of CA3
725 pyramidal cells during irregular LFPs, in agreement with previous data (Ranck 1973;
726 Thompson and Best 1989). In terms of the PP generator, the overall behavior was
727 similar (occasional synchronous bouts of irregular activity), again indicating a low
728 output rate from entorhino-hippocampal projection cells during irregular patterns of
729 LFP. The presynaptic origin of this generator is suggested by the limited evoked
730 responses following PP stimulation. The identity of the entorhinal projecting cells is not
731 clearly established in the literature and difficulties may arise from the fact that the
732 entorhino-hippocampal projection is multifarious. Unitary studies have not clarified this
733 issue, since most interest has focused on the relationship of entorhinal units with theta
734 (Alonso and Llinás 1989; Chrobak and Buzsaki 1994; Frank et al. 2001). Thus stellate
735 and pyramidal-like cells have been deemed candidates for this projection. Indeed,
736 interesting functional data supporting the identity of the PP generator revealed here is
737 that the bouts of activity are not related to hippocampal sharp waves (Chrobak and
738 Buzsáki 1994), which is consistent with the poor temporal correspondence between
739 bouts of activity in the Schaffer and PP generators.
740 To conclude we propose the combined use of ICA and CSD as a high resolution
741 method for the subcellular identification of LFP components in regular structures, and
742 for the quantification of their ongoing activity.
743
744 ACKNOWLEDGEMENTS
745 We thank Mark Sefton at BiomedRed for editorial support, S. Canals for critical
746 reading and comments, and L. López-Mascaraque and N. Salvador for help with
747 histological processing.
748
749 GRANTS
750 This work was supported by the grants BFU2007-66621/BFI and FIS2007-65173
751 from the Spanish Ministry of Education and Science, and S-SEM-0255-2006 from the
752 Comunidad Autónoma de Madrid.
753
754

24
755 REFERENCES

756 Aika Y, Ren JQ, Kosaka K Kosaka T. Quantitative analysis of GABA-like-

757 immunoreactive and parvalbumin-containing neurons in the CA1 region of the rat

758 hippocampus using a stereological method, the dissector. Exp Brain Res 99: 267-

759 276, 1994.

760 Alonso A, Llinás RR. Subthreshold Na+-dependent theta-like rhythmicity in stellate

761 cells of entorhinal cortex layer II. Nature 342: 175-177, 1989.

762 Amari S, Cichocki A, Yang H. A new learning algorithm for blind source separation.

763 Adv Neural Inf Process Syst 8: 757-763, 1996.

764 Andersen P, Bliss TVP, Skrede KK. Lamellar organization of hippocampal excitatory

765 pathways. Exp Brain Res 13: 222-238, 1971.

766 Anderson JS, Carandini M, Ferster D. Orientation tuning of inputconductance,

767 excitation and inhibition in cat primary visual cortex. J Neurophysiol 84: 909–926,

768 2000.

769 Bédard C, Kröger H, Destexhe A. Modeling extracellular field potentials and the

770 frequency-filtering properties of extracellular space. Biophys J, 86: 1829–1842,

771 2004.

772 Bell A, Sejnowski, T. An information-maximization approach to blind separation and

773 blind deconvolution. Neural Computation 7: 1129-1159, 1995.

774 Borg-Graham LJ, Monier C, Frégnac Y. Visual input evokes transient and strong

775 shunting inhibition in visual cortical neurons. Nature 393: 369–373, 1998.

776 Bragin A, G Jandó, Nádasdy Z, van Landeghem M, Buzsáki G. Dentate EEG spikes

777 and associated interneuronal population bursts in the hippocampal hilar region of the

778 rat. J Neurophysiol 73: 1691-705, 1995.

779 Brankačk J, Stewart M, Fox SE. Current source density analysis of the hippocampal

25
780 theta rhythm: associated sustained potentials and candidate synaptic generators.

781 Brain Res 615: 310-327, 1993.

782 Buzsáki G, Leung LS, Vanderwolf CH. Cellular bases of hippocampal EEG in the

783 behaving rat. Brain res 287: 139-171, 1983.

784 Buzsáki G, Czopf J, Kondákor I, Kellényi L. Laminar distribution of hippocampal

785 rhythmic slow activity (RSA) in the behaving rat: current-source density analysis,

786 effects of urethane and atropine. Brain Res 365: 125-37, 1986.

787 Canals S, López-Aguado L, Herreras O. Synaptically-recruited apical currents are

788 required to initiate axonal and apical spikes in hippocampal pyramidal cells:

789 modulation by inhibition. J Neurophysiology 93: 909-918, 2005.

790 Castellanos NP, Makarov VA. Recovering EEG brain signals: Artifact suppression

791 with wavelet enhanced independent component analysis. Journal of Neuroscience

792 Methods 158: 300-312, 2006.

793 Choi S, Cichocki A, Park HM, Lee SY. Blind source separation and independent

794 component analysis: A review. Neural Information Processing - Letters and

795 Reviews 6: 1-57, 2005.

796 Chrobak JJ, Buzsaki G. Selective activation of deep layer (V-VI) retrohippocampal

797 cortical neurons during hippocampal sharp waves in the behaving rat. J Neurosci

798 14(10): 6160-70, 1994.

799 Csicsvari J, Jamieson B, Wise KD, Buzsáki G. Mechanisms of gamma oscillations in

800 the hippocampus of the behaving rat. Neuron 37: 311-322, 2003.

801 Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single trial

802 EEG dynamics including independent component analysis. J Neurosci Meth 134: 9-

803 21, 2004.

804 Einevoll GT, Pettersen KH, Devor A, Ulbert I, Halgren E, Dale AM. Laminar

26
805 population analysis: estimating firing rates and evoked synaptic activity from

806 multielectrode recordings in rat barrel cortex. J Neurophysiol 97(3): 2174-90, 2007.

807 Frank LM, Brown EN, Wilson MA. A comparison of the firing properties of putative

808 excitatory and inhibitory neurons from CA1 and the entorhinal cortex. J

809 Neurophysiol 86(4): 2029-40, 2001.

810 Freeman JA, Nicholson C. Experimental optimization of current source-density

811 technique for anuran cerebellum. J Neurophysiol 38: 369-382, 1975.

812 Hajos N, Katona I, Naiem SS, MacKie K, Ledent C, Mody I, Freund TF.

813 Cannabinoids inhibit hippocampal GABAergic transmission and network

814 oscillations. Eur J Neurosci 12: 3239-3249, 2000.

815 Herreras O, Solís JM, Martín del Río R, Lerma J. Characteristics of CA1 activation

816 through the hippocampal trisynaptic pathway in the unanaesthetized rat. Brain Res

817 413: 75-86, 1987.

818 Herreras O, Solís JM, Herranz AS, Martín del Río R, Lerma J. Sensory modulation

819 of hippocampal transmission. II. Evidence for a cholinergic locus of inhibition in the

820 Schaffer-CA1 synapse. Brain Res 451: 303-313, 1988.

821 Herreras O. Propagating dendritic action potential mediates synaptic transmission in

822 CA1 pyramidal cells in situ. J Neurophysiol 64: 1429-1441, 1990.

823 Hjorth-Simonsen A, Jeune B. Origin and termination of the hippocampal perforant

824 path in the rat studied by silver impregnation. J Comp Neurol 144: 215-232, 1972.

825 Hyvärinen A, Oja E. Independent component analysis: Algorithms and applications

826 Neural Networks 13: 411-430, 2000.

827 Jarosiewicz B, Skaggs WE. Level of arousal during the small irregular activity state in

828 the rat hippocampal EEG. J Neurophysiol 91(6): 2649-57, 2004.

829 Jung KY, Kim JM, Kim DW, Chung CS. Independent component analysis of

27
830 generalized spike-and-wave discharges: primary versus secondary bilateral

831 synchrony. Clin Neurophysiol 116: 913-919, 2005.

832 Kandel ER, Spencer WA, Brinley Jr FJ. Electrophysiology of hippocampal neurons.

833 I. Sequential invasion and synaptic organization. J Neurophysiol 24: 225-242, 1961.

834 Kocsis B, Bragin A, Buzsáki G. Interdependence of multiple theta generators in the

835 hippocampus: a partial coherence analysis. J Neurosci 19(14): 6200-12, 1999.

836 Lee T, Girolomi M, Sejnowski T. Independent component analysis using an extended

837 infomax algorithm for mixed subgaussian and supergaussian sources. Neural

838 Comput 11: 417-441, 1999.

839 Leung, LS, Roth L, Canning KJ. Entorhinal inputs to hippocampal CA1 and dentate

840 gyrus in the rat: a current-source-density study. J Neurophysiol 73: 2392-2403,

841 1995.

842 Lømo T. Patterns of activation in a monosynaptic cortical pathway: the perforant path

843 input to the dentate area of the hippocampal formation. Exp. Brain Res 12: 18–45,

844 1971.

845 López-Aguado L, Ibarz JM, Herreras O. Activity-dependent changes of tissue

846 resistivity in the CA1 region in vivo are layer-specific: modulation of evoked

847 potentials. Neuroscience 108: 249-262, 2001.

848 López-Aguado L, Ibarz JM, Varona P, Herreras O. Structural inhomogeneities

849 differentially modulate action currents and population spikes initiated in the axon or

850 dendrites. J Neurophysiol 88: 2809-2820, 2002.

851 Lorente de Nó R. Analysis of the distribution of action currents of nerves in volume

852 conductors. In A study of nerve physiology. Stud Rockefeller Inst Med Res Repr

853 132: 384-477, 1947.

854 Makarov VA, Makarova J, Herreras O. Disentanglement of local field potential

28
855 sources by independent component analysis. Journal of Computational
856 Neuroscience DOI: 10.1007/s10827-009-0206-y, 2010.
857 Makarova J, Gómez-Galán M, Herreras O. Layer specific changes in tissue

858 resistivity and spatial cancellation of transmembrane currents shape the voltage

859 signal during spreading depression in the CA1 vivo. Eur J Neurosci 27: 444-456,

860 2008.

861 Makarova J, Makarov VA, Herreras O. A model of sustained field potentials based

862 on gradients of polarization in single neurons. J Neurophysiol DOI:

863 10.1152/jn.01045.2009, 2010.

864 Makeig S, Debener S, Onton J, Delorme A. Mining event-related brain dynamics.

865 Trends Cognit Sci 8: 204-210, 2004.

866 Mann EO, Radcliffe CA, Paulsen O. Hippocampal gamma-frequency oscillations:

867 from interneurones to pyramidal cells, and back. J Physiol (Lond) 562: 55-63, 2005.

868 Miles R, Wong RK. Inhibitory control of local excitatory circuits in the guinea-pig

869 hippocampus. J Physiol 388: 611-29, 1987.

870 Mitzdorf U. Current source-density method and application in cat cerebral cortex:

871 investigation of evoked potentials and EEG phenomena. Physiol Rev 65: 37-100,

872 1985.

873 Nunez PL, Srinivasan R. Electric Fields of the Brain: The Neurophysics of EEG. 2nd

874 edition. New York: Oxford University Press, 2006.

875 Oren I, Hajos N, Paulsen O. Identification of the current generator underlying

876 cholinergically induced gamma frequency field potential oscillations in the

877 hippocampal CA3 region. J Physiol 588(Pt 5): 785-97, 2010.

878 Ranck JB Jr. Studies on single neurons in dorsal hippocampal formation and septum in

879 unrestrained rats. Part 1. Behavioral correlates and firing repertoires. Exp Neurol 40:

29
880 462-531, 1973.

881 Rudolph M, Pelletier JG, Paré D, Destexhe A. Characterization of synaptic

882 conductances and integrative properties during electrically induced EEG-activated

883 states in neocortical neurons in vivo. J Neurophysiol 94(4): 2805-21, 2005.

884 Soltesz I, Deschenes M. Low- and high-frequency membrane potential oscillations

885 during theta activity in CA1 and CA3 pyramidal neurons of the rat hippocampus

886 under ketamine-xylazine anesthesia. J Neurophysiol 70(1): 97-116, 1993.

887 Somogyi P, Klausberger T. Defined types of cortical interneurone structure space and

888 spike timing in the hippocampus. J Physiol (Lond) 562: 9-26, 2005.

889 Stone JV. Independent Component Analysis: A Tutorial Introduction. MIT Press, 2004.

890 Tanskanen JM, Mikkonen JE, Penttonen M. Independent component analysis of

891 neural populations from multielectrode field potential measurements. J Neurosci

892 Methods 145(1-2): 213-32, 2005.

893 Thompson LT, Best PJ. Place cells and silent cells in the hippocampus of freely-

894 behaving rats. J Neurosci 9: 2382–2390, 1989.

895 Varona, P, Ibarz JM, López-Aguado L, Herreras O. Macroscopic and subcellular

896 factors shaping CA1 population spikes. J Neurophysiol 83:2192-2208, 2000.

897 Vigario R, Sarela J, Jousmaki V, Hamalainen M, Oja E. Independent component

898 approach to the analysis of EEG and MEG recordings. IEEE Trans Biomed Eng 47:

899 589–593, 2000.

900 Whittington MA, Traub RD, Jefferys JGR. Synchronized oscillations in interneuron

901 networks driven by metabotropic glutamate receptor activation. Nature 373:

902 612–615, 1995.

903 Wu K, Canning KJ, Leung LS. Functional interconnections between CA3 and the

904 dentate gyrus revealed by current source density analysis. Hippocampus 8: 217-30,

30
905 1998.

906 Ylinen A, Bragin A, Nádasdy Z, Jandó G, Szabó I, Sik A, Buzsáki G. Sharp wave-

907 associated high-frequency oscillation (200 Hz) in the intact hippocampus: network

908 and intracellular mechanisms. J Neurosci 15: 30-46, 1995.

909

31
910 FIGURE LEGENDS.
911 Fig. 1. General strategy of LFP decomposition and the identification of generators. The
912 diagram shows the sequential steps from an LFP spatial map recorded along the CA1
913 axis containing evoked responses to a twin pulse in the ipsilateral CA3 (Schaffer
914 response). A similar procedure is followed for spontaneous LFPs. The simultaneous
915 recording of 16 sites along the main axis of the principal cell (pyramidal neurons) (A) is
916 analyzed by CSD (B) to obtain the map of inward and outward population currents
917 (sinks are in blue and sources in yellow-red). Multiple membrane events overlap
918 spatiotemporally. The LFP profile is also analyzed by an ICA that provides a set of 16
919 generators each with a temporal activation (C) and spatial distribution (D). Of the 16
920 generators, only a few (#1 to 6 in the example) have a significant variance, while the
921 others are rejected for analysis. The second spatial derivative of the spatial weight
922 curves (or voltage loadings) provides the CSD loadings (E), a spatial index that reflects
923 the distribution of membrane currents along the conductor. Since the ICA does not
924 ensure the correct polarity or absolute values of amplitude, we reconstructed the LFP for
925 each generator isolated (F). Such partial LFPs regain the correct polarity and true
926 amplitude. A CSD can then be applied to the reconstructed LFP, which provides
927 simplified spatiotemporal maps of inward/outward currents for unique spatially
928 coherent membrane events (G). These maps can be compared with the CSD of raw
929 LFPs (curved dashed arrow) to find specific membrane events buried within the mixed
930 CSD (closed dashed white lines). It can be shown that the spatial profiles of partial
931 CSDs (E) match the spatial profile of the CSD loadings (small dashed arrow) and hence,
932 their true polarity is assessed. Arrowheads mark the time of stimulus (artifacts have
933 been removed for analysis). a.u. (arbitrary units).
934
935 Fig. 2. Segregation of artifacts and neural activity by the ICA. The figure corresponds to
936 the analysis of an LFP obtained during antidromic activation of the CA1 field. The
937 alvear electric shock produced a large voltage artifact (dot in A,a) followed by an
938 antidromic population spike. The artifact is nearly deleted in the corresponding CSD
939 (A,b) while locally generated active sink and passive sources are visible. The ICA
940 revealed only two significant generators (B,b). G1 contains the stimulus artifact (B,a)
941 that enters all electrodes with similar weight (flat voltage loading in A,b). G2 contains
942 all the neural activity, displaying a bell-shaped voltage loading curve spanning the

32
943 activated somatodendritic domains (A,c). B: The cluster analysis of all voltage loading
944 curves obtained in 7 animals establishes two groups (blue and green) corresponding to
945 the artifacts and population spikes respectively. The average spatial loading curves is
946 shown in B,b (mean ± s.d.).
947
948 Fig.3. Precise spatial and temporal performance of ICA on LFPs produced during
949 subthreshold activation of the Schaffer input to CA1. A: An electric shock in the
950 ipsilateral CA3 evoked a negative potential in the apical dendritic tree (a). The
951 schematic neuron indicates the approximate recording position. The CSD reveals
952 inward currents (sinks) centered in st. radiatum flanked by passive sources in the soma
953 layer and st. lacunosum (b). B: The ICA applied to the LFP profile reveals a unique
954 significant component (b) with a temporal course matching the voltage change in central
955 electrodes (a). C: The voltage and CSD loadings (blue plots) coincide with the spatial
956 profiles of the raw LFP and CSD (black plots; measured at the instant marked by the
957 small arrows in A). The CSD calculated for the reconstructed LFP for G1 (c) is similar
958 to the raw CSD (A,b). D: cluster analysis for voltage loadings obtained in four animals
959 show a unique group (a). The average voltage loading (mean ± s.e.m) shown in (b) was
960 built using the st. pyramidale as a spatial reference.
961
962 Fig.4. Precise spatial and temporal performance of the ICA on LFPs produced during
963 subthreshold activation of the entorhinal input to the DG. A: An electric shock in the
964 ipsilateral perforant path (PP) evoked a negative potential in both molecular layers of
965 the DG. A sample record from electrode 13 of a 32-site probe is shown in (a). The raw
966 CSD (b) shows active sinks in both molecular layers (sk1 and sk2) and a weaker sink in
967 the st. lacunosum of the CA1 (sk3). Passive sources can be appreciated in the granule
968 cell layers (GCL). The schematic neurons show the approximate site of recordings. B:
969 The ICA applied to the LFP profile reveals a unique significant component (b) with a
970 temporal course matching the voltage change in the molecular layer (a). The CSD
971 calculated for the reconstructed LFP for G1 (c) is similar to the raw CSD (A,b). The
972 voltage and CSD loadings (d) coincide with the spatial profiles of the raw LFP and
973 CSD. Note that a unique ICA component captures coherent activity in physically
974 separated cell generators as long as they are synaptically activated by a common input.
975

33
976 Fig. 5. Discrimination of multiple evoked synaptic inputs in mixed LFPs. Subthreshold
977 stimuli were applied to both the ipsilateral CA3 (Schaffer) and the PP, either as paired
978 pulses (100 ms delay, left column) or synchronously (right column). PP stimuli were
979 modulated in intensity while CA3 stimuli were constant. A: Raw LFPs recorded along a
980 track spanning the CA1 and CA3/DG up to the lower granule cell layer (GCL). B: CSD
981 maps for a sample response (raw CSD) to either the PP, the CA3 (Sch) or both.
982 Schematic neurons between panels show the approximate locus or recording. PP
983 produced a main sink in the st. moleculare (sk1) and a weaker sink in the CA1 st.
984 lacunosum (sk2). CA3 activation raised a sink in the CA1 st. radiatum (sk3) and a local
985 sink (sk4). Combined stimuli produced a complex sink in the DG (sk5) and the Schaffer
986 sink in CA1 (sk6). C: The ICA yielded three significant generators in both cases. G1
987 and G2 correspond to the PP and Schaffer generators, respectively, while G3
988 corresponds to local activity in the CA3/DG. The upper traces show the time activation
989 and the lower plots the corresponding spatial voltage and CSD loadings. Note that
990 activations vary in G1 with successive stimuli while they remain constant in G2,
991 precisely reflecting the intensity applied in each case. The spatial loadings for G1 and
992 G2 were identical as when each pathway was stimulated alone. A small contaminant
993 appeared during coincident stimulation (small arrows). D: Virtual CSD maps for sample
994 responses of each separate generator. Note the spatial coincidence of the separate virtual
995 sinks with those in the raw LFPs.
996
997 Fig. 6. Identification of the Schaffer and PP population activities in ongoing LFPs. A:
998 Segment of raw LFPs recorded from the CA1 st. pyramidale to the lower GCL in the
999 DG. LFPs were primed with subthreshold stimuli applied in a random manner to the
1000 CA3 (Sch) and PP. The vertical dashed lines mark the timing of stimulation (blue for PP
1001 and red for CA3). Evoked potentials are short and small, and hence, they do not
1002 introduce much variance in the raw signals (they are hardly discernible over this time
1003 scale). B-D: The ICA yielded five principal generators, two of which specifically
1004 contain the activation corresponding to the evoked potentials for the Schaffer (G2) and
1005 the PP (G4). B shows an amplification of the responses after each stimulus. Note the
1006 pathway specific segregation of evoked responses in the corresponding generators. The
1007 spatial loadings shown in C match those found for non-primed LFPs (G1, G3, G5) or

34
1008 for separately stimulated pathways (G2, G4). D: Relative variance of the generators
1009 isolated.
1010 Fig. 7. Characterization of pathway-specific activity in ongoing LFPs. A: Segment of
1011 raw LFPs recorded from the CA1 st. pyramidale to the lower GCL in the DG. The CA3
1012 and the PP was stimulated to prime LFPs with Schaffer and PP evoked activity (instants
1013 marked by red and blue dashed lines, respectively). The ellipse marks a Schaffer evoked
1014 response enlarged to the right. Note the subthreshold fEPSP in the st. radiatum
1015 (electrodes #3-10, asterisk). B: CSD map estimated for the raw LFPs. Note the complex
1016 mixture of currents. The Schaffer sink in the evoked response is appreciated in the
1017 amplification on the right (small arrow). C: ongoing activities of the five main
1018 generators separated by the ICA. G2 and G4 correspond to the Schaffer and PP
1019 generators, respectively. Note the characteristic spatial voltage loadings on the right.
1020 The upper (black) trace corresponds to the filtered (100-500 Hz bandpass) LFP recorded
1021 from the CA1 soma layer. Note that epochs of fast oscillations coincide with the bouts
1022 of activity in the Schaffer generator (G2), matching the features of sharp wave events.
1023 D: Reconstruction of the Schaffer-specific LFP. This virtual LFP now contains only the
1024 virtual LFP corresponding to ongoing activation of the CA3 to CA1 input. The
1025 amplification on the right shows the virtual fEPSP for a sample Schaffer response. Note
1026 that all evoked and spontaneous activity undergoes changes of identical polarity for a
1027 given electrode. E: Virtual CSD map for the ongoing Schaffer activity. Schaffer evoked
1028 sinks and sources are now like the standard evoked responses (Compare with the raw
1029 CSD in B).
1030
1031 Fig. 8. Minor contribution of the excitatory generators to the hippocampal LFPs. A:
1032 Frequency content of the ongoing Schaffer (G2) and PP (G4) generators during
1033 irregular activity in the raw LFP. The spectrogram shows dominant activity below 2-3
1034 Hz. The white trace shows the temporal activation of the generators (amplified above).
1035 The specific activity of spontaneous pathways was in both cases sparse and highly
1036 irregular. B: Time envelopes of activity (1 second time window). C: Average power of
1037 reconstructed LFPs for specific generators in 6 animals (mean ± s.e.m.) during irregular
1038 activity in the raw LFP. Note the small global contribution of excitatory generators.
1039
1040

35
 A raw LFPs B raw CSD
1
2

electrode #
CSD analysis
8 8

15
16
10 ms 5 mV 10 ms 50 m
m

ICA
C D E
generator activations voltage loadings CSD loadings
G1 1

# of electrode
nd
2
G8 8 derivative

G16 16
10 ms 8 a.u. 0 1
-1
(a.u.) -0.5 0
(a.u.) 0.5

F virtual LFPs from G6

G
1
CSD for reconstructed LFPs
2
electrode #

CSD analysis
8 8

15
16
10 ms 5 mV 10 ms 50 m
m

Fig. 1
A a raw LFP c volt. loadings
1
G2
electrode #

G1

5 mV
16
0 1 2 3
(a.u.)
b CSD for raw LFP
2 9

8 0

15 -9
1 ms

d activations e expl. variance

1.0
G1

0.5
G2
40
1
(a.u.) 0.0
G1 G2 G3-G16

B a clustering of generators b mean v. loadings

0.6 -0.25
G2
dissimilarity

depth (mm)

0.4 0

0.2 0.25
G1
0 0.5
2a 4a 3a 5a 6a 7a 1a1b 2b 5b 6b 7b 4b 3b 1c 0 3
-1 1 2
(a.u.)

Fig. 2
A a b
raw LFPs raw CSD
1
0.8
electrode #

8 0

-0.8
16 2 ms
5 mV

B a activation of G1 b
1
explained variance

0.5
2 ms 0.3 (a.u.)

0
G1 G1-G13
c d
volt. loading CSD loading e CSD for G1
1
0.8
electrode #

8 0

G1

LFP -0.8
16
0 4 8 -1 0 1 2 2 ms
(a.u.) (a.u.)
Ca clustering of generators b mean voltage loading
-0.25
0.6
depth (mm)

0
dissimilarity

0.4

0.25
0.2

0 0.5
1a 5a 2a 3a 4a 2b -3 0 3 6 9
(a.u.)

Fig. 3
 A B a b 1.0
explained variance
e28
0.5
1 mV 1 (a.u.)
G1 0.0
G1 G2-G16

raw CSD c CSD for G1 d CSD loadings volt. loadings

2
2
CA1
sk3
electrode #

10
sk1
GCL 0
DG

20
GCL
sk2
30 -2
10 ms -0.8 0 0.8 0 2 4
(a.u.) (a.u)

Fig. 4
 A non-coincident coincident
1

10
electrode #

20

5 mV
30
. . . . . . . .* * . . . . . . . . . . . . . . . . . . .
*
0.2 s 0.05 s
B
PP raw CSD Sch raw CSD PP + Sch raw CSD -4
2 x10
8

CA1
10
electrode #

sk3 sk6
sk2 0
20 sk1
sk4 sk5
GCL
DG
H -8
30
GCL
5 ms 5 ms
C activations activations

G1
G2
G3
. .. .. .. .* *
. .. .. .. .. . . . . . . * . . . .
2 (a.u.) 1 (a.u.)
volt. loadings CSD loadings volt. loadings CSD loadings
1
electrode #

10

20

30
0 2 4 -0.4 0 0.4 -4 0 4 8 -1 0 1 2
(a.u.) (a.u.) (a.u.) (a.u.)

D
2
G1 (PP) G1 (PP)
10
sk1 sk1
20 sk2 sk2
30
2
G2 (Sch) G2 (Sch)
electrode #

10 sk3 sk3
20

30 -4
2 x10
G3 (CA3) 4 G3 (CA3)
10
0
20
sk4 sk4
30 -4
5 ms 5 ms

Fig.5
A raw LFPs 1s
C
5 mV voltage loadings
1

CA1
electrode #

10

20 GCL

CA3/H
DG
30 GCL
-0.2 0 0.2 0.4
(a.u.)
B D explained variance
G1 0.5

Schaffer
0.4
G2

0.3
G3
0.2
Perforant
G4
0.1

G5 0
G1 - G5 G6 - G11
40 ms 2 (a.u)

Fig. 6
A1 raw LFPs 0.2 s 20 ms

CA1
electrode # 10 *

20
GCL

CA3/H
30
GCL 2 mV
5 mV
B2 raw CSD
0.25
electrode #

10

0
20

30 -0.25

LFP [100-500 Hz]

0.1 mV

C
activations volt. loadings
G1
Schaffer
G2

G3
Perforant
G4

G5 -0.4 0 0.4
5(a.u.) (a.u.)
D virtual LFPs reconstructed from G2
1
electrode #

10

20

30
5 mV 2 mV

E CSD for Schaffer LFP-generator (G2)

2 0.25
electrode #

10

0
20

30 -0.25

Fig. 7
 A Schaffer (G2) Perforant (G4) C
5 (a.u.)

0.4
frequency (Hz)

32

8 0.3

Power (mV2)
2
0.2

Perforant
0.5

Schaffer
5s
B 0.1
2 (a.u.) 0.2 (a.u)
0.0
G1 G2 G3 G4 G5

Fig. 8