Phase Relations of Rhythmic Neuronal Firing in the ... - CiteSeerX

6 downloads 0 Views 209KB Size Report
neurons might play different roles in generating and/or transmitting theta rhythmic activity of the ... MS/vDB stimulation drives theta (James et al., 1977;. Wetzel et al., 1977) ...... Green JD, Arduini A (1954) Hippocampal electrical activity in arousal. J .... Stewart M, Barry M, Quirk GJ, Fox SE (1991) Firing relations of entorhinal ...
HIPPOCAMPUS 7:204–214 (1997)

Phase Relations of Rhythmic Neuronal Firing in the Supramammillary Nucleus and Mammillary Body to the Hippocampal Theta Activity in Urethane Anesthetized Rats Bernat Kocsis1,2* and Robert P. Vertes2 1Department

of Physiology, National Institute of Neurosurgery, Budapest, Hungary 2Center for Complex Systems, Florida Atlantic University, Boca Raton, Florida ABSTRACT: Structures in the caudal diencephalon including the posterior hypothalamic nucleus, the supramammillary nucleus (SUM) and the nuclei of the mammillary body (MB) occupy a strategic position in the crossroads of ascending and descending traffic between the brainstem and the limbic forebrain (septum/hippocampus). In this study we analyzed the phase relations of rhythmically discharging SUM/MB cells to hippocampal theta rhythm in urethane anesthetized rats with a dual aim of separating different functional types of SUM and MB neurons and characterizing their coupling to septohippocampal theta oscillators. We found that rhythmically firing neurons in the SUM/MB represent a functionally heterogeneous population of cells that are coupled with forebrain theta oscillators at different preferred phases. Based on their phase relations to hippocampal theta four groups of rhythmic SUM/MB cells were identified. Neurons of the first and second groups fired out-of-phase relative to each other and synchronously with the positive (8° 6 7) or negative peaks (2177° 6 7) of theta field activity in the hippocampus, recorded above the CA1 pyramidal layer. Cells of the other two groups, also forming out-of-phase counterparts, fired on the rising (97° 6 9) or falling segments (297° 6 6) of CA1 theta waves. The peaks in the phase distribution histogram were well separated, and the empty zones between them were wider (40–70°) than those comprising the phase data for different groups. The variations of phase values for individual neurons, when tested during several theta epochs, did not exceed the range of a single group. Theta field potentials were also recorded in the SUM/MB and were advanced by one quarter of the cycle (79° 6 9, range 56–99°) relative to CA1 theta oscillations. The present results indicate that, similar to other theta-generating structures, rhythmically firing neurons can be classified on the basis of their phase relations in the SUM/MB as well. Different classes of SUM/MB neurons might play different roles in generating and/or transmitting theta rhythmic activity of the limbic system. Hippocampus 7:204–214, 1997.

r 1997 Wiley-Liss, Inc. KEY WORDS: neuronal oscillation; coherence; rhythmic slow activity; septohippocampal system

INTRODUCTION Rhythmic slow activity in the hippocampus (theta rhythm) (Jung and Kornmuller, 1938; Green and Arduini, 1954) is present selectively during *Correspondence to: Bernat Kocsis, Department of Physiology, National Institute of Neurosurgery, Amerikai ut 57, H-1145, Budapest, Hungary. E-mail: [email protected] Accepted for publication 20 December 1996

r 1997 WILEY-LISS, INC.

exploratory behavior and REM sleep (Vanderwolf, 1969; Winson, 1972), and is believed to serve a critical role in mnemonic processes of the hippocampus (O’Keefe and Nadel, 1978; Larson and Lynch, 1986; Staubli and Lynch, 1987; Pavlides et al., 1988; Buzsaki, 1989; Huerta and Lisman, 1993). It is well established that cells of the medial septum and vertical limb of the diagonal band (MS/vDB) fire synchronously with theta (Petsche et al., 1962; Stumpf et al., 1962) to pace the theta rhythm. Low frequency MS/vDB stimulation drives theta (James et al., 1977; Wetzel et al., 1977) and reversible or irreversible MS/ vDB lesions eliminate theta in the hippocampus (Rawlins et al., 1979; Mitchell et al., 1982). The MS/vDB is thought to convert steady, non-rhythmic influences from the brainstem reticular formation (RF) into a rhythmical pattern which when conveyed to hippocampus induces theta; the MS/vDB is viewed as the pacemaker for the hippocampal theta rhythm (for review, Bland and Colom, 1993; Vinogradova, 1995). Brainstem RF stimulation, most effectively within nucleus pontis oralis (RPO) (Vertes, 1981; Vertes et al., 1993), elicits theta, and cells of RPO fire at enhanced rates during naturally occurring theta in behaving rats (Vertes, 1979) and during theta elicited with carbachol in urethane anesthetized rats (Nunez et al., 1991). Recent evidence indicates that RPO influences on the MS/vDB are mediated by cell groups of the caudal diencephalon, namely, the supramammillary nucleus (SUM) and the posterior nucleus of the hypothalamus (PH). It has been shown that 1) electrically or chemically induced (with carbachol) activation of the SUM/PH region generates theta (Vertes, 1981; Oddie et al., 1994), 2) RPO projects to SUM/PH and the SUM is in turn the source of strong projections to the septum as well as to the dentate gyrus and CA2/CA3a region of the Ammon’s horn (Amaral and Cowan, 1980; Haglund et al., 1984; Vertes and Martin, 1988; Vertes, 1992), 3) cells of the SUM discharge rhythmically in phase with theta, while those of PH discharge tonically (i.e.,

_____________________________________

HIPPOCAMPAL THETA AND SUM/MB NEURONAL FIRING

non-rhythmically) with the theta (Kirk and McNaughton, 1991; Kocsis and Vertes, 1994a, 1994b; Bland et al., 1995), and 4) reversible (with procaine) inactivation of the SUM/PH blocks spontaneous or RPO-elicited theta as well as the burst discharge of the septal pacemaking cells (Kirk and McNaughton, 1993a; Oddie et al., 1994). It is well documented that the mammillary body (MB), and to a lesser extent the SUM, receives descending projections from the septum as well as from the hippocampus/parahippocampus (i.e., mainly, from the subiculum and entorhinal cortex) (Swanson and Cowan, 1975, 1979; Shibata, 1988; Allen and Hopkins, 1989; Gonzalo-Ruiz et al., 1992; Hayakawa and Zyo, 1993, 1996). The MB is a major component in Papez’s circuit (Papez, 1937) and recent evidence suggests that the MB may receive ‘‘thetarhythmical’’ signals from the hippocampus. A subset of neurons of the medial and lateral MB has been shown to oscillate rhythmically in the theta frequency range in vitro slices (Alonso and Llinas, 1992; Llinas and Alonso, 1992) and MB cells have been reported to fire rhythmically with theta (Kocsis and Vertes, 1994a; Bland et al., 1995) in urethane anesthetized rats. A very powerful method for investigating the relationship between the rhythmical activity of separate structures is the analysis of the relative phase between neuronal discharge and field potentials of these structures. For example, phase analysis led to the identification of cells of the lateral entorhinal cortex (EC) responsible for generating the large rhythmical sink at the hippocampal fissure during theta (Stewart et al., 1991), and to the detection of cells of the medial EC likely driving dentate granule cells on the positive peak of the local theta (Stewart et al., 1992). In the hippocampus, functionally different types of neurons discharge at different phases of the theta rhythm; that is, pyramidal cells and interneurons of CA1 region fire out-of-phase relative to each other and close to the negative and positive peaks of the local theta field, respectively (Ranck, 1973; Buzsaki et al., 1983; Fox et al., 1986). Phase analysis of rhythmically bursting MS/vDB cells has not revealed a single preferred phase of firing to theta, for the whole population (Stewart and Fox, 1989). The MS/vDB, however, contains several types of cells (e.g. cholinergic and GABAergic) which also might fire at different phases of theta rhythm (Stewart and Fox, 1989). Although recent reports have described the discharge properties of SUM and MB neurons, less attention has been given to a possible functional differentiation between subsets of these cells. SUM neurons contain several neurochemical markers (Haglund et al., 1984; Leranth and Nitsch, 1994; Magloczky et al., 1994; Hayakawa and Zyo, 1994; Leranth and Kiss, 1996) and appear to differ with respect to whether they project to the septum, to the hippocampus or by way of collaterals to both structures. These anatomical differences suggest a functional differentiation. In this study, we analyzed the phase relationships of the rhythmically discharging SUM/MB cells to the theta rhythm as well as phase relationships of field potentials in SUM/MB to the theta with the dual aim of determining phase relationships between SUM/MB neurons and theta and possibly describing different functional types of SUM and MB neurons.

205

MATERIALS AND METHODS The experiments were performed on male Sprague-Dawley rats weighing 250–325 g. Rats were initially anesthetized with methoxyflurane and fitted with a polyethylene catheter that was inserted in the jugular vein for the administration of urethane. Rats were maintained on urethane anesthesia throughout the duration of the experiment. Urethane was dissolved in deionized water to achieve 75–80% solution, and administered in amounts sufficient for complete anesthesia as indicated by the lack of the withdrawal reflex but where theta rhythm could be elicited by gentle tail pinch. When necessary, additional doses of urethane were given iv. A pair of Teflon-coated stainless steel wires twisted together (diameter 125 mm, tips separated dorso-ventrally by 0.75–1.0 mm) was inserted in the right dorsal hippocampus (AP 23.7, Lat 2.2 mm) to record hippocampal EEG activity. The electrodes were positioned to sample both superficial and deep theta oscillations (the tip of the deep electrode was 3.5–4.0 mm below the surface of the skull). SUM and MB unit firing was recorded with glass microelectrodes (4–6 MV) filled with 0.5 M sodium acetate and 2% pontamine sky blue. All signals were referred to a common point. A stainless steel screw driven into the bone above the frontal cortex, anterior to the bregma, served as indifferent electrode. The signals were amplified (Grass Instruments, model 7P511) and filtered with filter settings of 1–100 Hz (field potentials) or 0.3–10 kHz (unit recordings). With 26 dB slope of the filter’s frequency response curve and filtering 50% of the amplitude at 1 Hz, all components of the EEG signals above 3 Hz are passed undistorted and therefore the non-systematic phase shift in the theta range between 3 and 6 Hz (i.e. 90% and 95% amplitude) is negligible. The signals were stored on magnetic tape and spectrum analysis was performed off line. For phase analysis, the spike trains were preprocessed using a voltage/time window discriminator (Frederick Haer) and together with EEG recordings were sampled at 250 Hz with a 12 bit A/D converter (RC Electronics). For separation of spikes in multiunit recordings or in noisy recordings (signal-tonoise ratio less than 3:1), a K-means clustering algorithm was used, as described in detail previously (Kocsis and Vertes, 1994a). It should be noted that in eight of 27 cases even the clustering failed to give reliable separation and these recordings were considered to reflect multiunit activity. In all spike trains, including those reflecting multiunit activity and burst firing, the time of appearance of each spike was taken as the moment of the event. Autospectra and coherence functions were calculated using a customized program, as described in detail in previous reports (Kocsis and Vertes, 1992, 1994a). In brief, 4-s long segments (‘windows’) of the original EEG signals and sequences of standardized pulses representing the unit spike trains, convolved with a sinc (sinx/x) function, after augmenting the segments by zeros to obtain 1,024 point arrays, were subjected to fast Fourier transform. The final autospectra of each signal and cross-spectra for each signal-pair were obtained by averaging the smoothed raw spectra calculated for seven to 50 contiguous windows (Christakos

206

KOCSIS AND VERTES

et al., 1991). The 95% confidence intervals of Benignus (1969, 1970) were used as a test of zero coherence for each individual cell. Theta frequency was defined as the frequency at the peak of the coherence and autospectra functions. For recordings showing significant coherence between different signals, phase spectra were calculated using the co-spectrum (in-phase spectral density) and the quad-spectrum (out-of-phase spectral density) (Bendat and Piersol, 1966), as described previously (Kocsis et al., 1990). The phase spectra are given in a form in which phase values are expressed in degrees between 21807 and 11807. Relative phase of 07 indicates in-phase firing with theta EEG; that is, spikes predominantly appearing at the positive peak of the field potentials. When the spikes followed the positive peak by less than a half cycle, the phase values fell between 07 and 21807 and when the spikes precede the positive EEG peak, the phase is greater than 07. The degree of significance of the phase estimate depends on the coherence between the two signals at that frequency (Jenkins and Watts, 1968). Accordingly, the significant coherent frequency band is marked in the figures (see regions between the vertical markers in Figures 1 and 5). Group data are expressed as mean and the standard error of the mean (SEM). To decide whether phase values for different neurons showed significant grouping around preferred phase(s), the phase distribution for the population of all SUM/MB cells was compared with the uniform distribution using the chi-square statistics. The methods used here for phase estimates are different from those used in some other studies. In most studies the phase of unit firing is calculated from cross-correlograms obtained either by averaging the slow waves triggered by the unit spikes (Bland et al., 1995) or by calculating the mean of the delay between the spikes and some event in the theta cycle (most frequently the positive peak) (Kirk and McNaughton, 1993b). The problems with these phase estimates were addressed by Fox et al. (1986) and resolved using their ‘phase of firing histograms’ and statistical analyses which offered an appropriate solution to most of the previous problems. They estimated the unit’s firing phase as a weighted mean of the probability of spikes occurring at different phases of preselected (i.e., of standard length) theta cycles. The phase, as calculated from the cross-correlograms (including phase of firing histograms), however, only represents the relationship between the spike train and the EEG at the frequency of the dominant rhythm. Spikes appearing during shorter or longer theta cycles will not be correctly accounted for and, unless treated as by Fox et al. (1986), will make the phase estimate at the peak frequency less accurate. By contrast, the phase spectrum provides information regarding phase difference between two signals as it varies as a function of frequency. In frequency domain analysis, all signals are represented by a sum of sine-waves of different frequencies composing these signals and the contribution of each of these components to the original signal is characterized by their power. The relationship between different signals is characterized by the coherence and phase spectra representing the coherence/phase relations between their sine-wave components at each frequency. The relative phase at the theta frequency, therefore, will be ‘‘clean’’ of the effect of spikes related to other components running faster or slower than the dominant rhythm. In addition, the variations of phase values

at frequencies around the peak frequency could give valuable information about the coupling between the two signals (see e.g. Kocsis et al., 1990).

RESULTS Phase relationships between the firing of 27 supramammillary and mammillary body (MB) neurons and the hippocampal theta rhythm was analyzed in 17 rats. Eighteen cells (12 in SUM; six in MB) were identified by deposits of pontamine sky blue through the recording microelectrode; nine others were identified by referencing depths of recording to histological reconstructions of the electrode tract. The discharge of all SUM/MB neurons showed statistically significant coherence with the hippocampal theta rhythm. As described earlier, there was no significant correlation between firing of these units with hippocampal EEG during non-theta states. One third of the cells fired in a burst-like manner similar to the complex spikes in the hippocampus, whereas others fired only single spikes (see Kocsis and Vertes, 1994a). Non-theta cells recorded in these experiments (n 5 141, all located outside of the SUM and MB, see Kocsis and Vertes, 1994a) were not included in the present analysis.

Phase Relations Between SUM/MB Units and Hippocampal Theta Stable relative phase between signals recorded in the hippocampus and SUM/MB was maintained during each theta epoch characterized by rhythmic slow activity in both structures at a regular frequency. Such events, either occurring spontaneously or evoked by sensory stimulation, were separated from the record and analyzed for coherence and relative phase. SUM/MB neurons usually fired around one of the following four phases: in-phase with ‘CA1’ theta, in-phase with ‘dentate’ theta, or with nearly equal phase delays (i.e. about 1907 and 2907) from these two states. Figure 1 shows representative examples of each of these groups. As shown in the raw records, the action potentials for two cells occurred mainly at the positive and negative peaks of ‘CA1’ theta, respectively, (Fig. 1C and 1D), whereas those for two other cells were neither in nor out of phase with CA1 theta. Instead, most of these spikes appeared on the rising (Fig. 1A) or falling phases (Fig. 1B) of CA1 field potentials. When a neuron fired at the positive peak of hippocampal theta (i.e., in-phase with CA1 field potentials), its relative phase was defined as 07. Similarly, a relative phase of 11807 or 21807 represents out-of-phase firing. Figures 1A–D show three phase spectra along with the associated coherence functions for each of the four categories of cells. The first phase spectrum (on the left) demonstrates out-of-phase rhythmicity between field potentials at the two hippocampal recording sites (CA-DG) and the two others relate the unit firing to deep (DG-SUM) and superficial (CASUM) theta rhythms. The relative phase between hippocampus and unit discharge was constant within the coherent frequency

_____________________________________

HIPPOCAMPAL THETA AND SUM/MB NEURONAL FIRING

207

FIGURE 1. Phase characteristics of four different types of SUM/MB neurons. Each cell, recorded in the SUM, is represented by a 1-s raw record of spontaneous firing during theta along with hippocampal (CA1) field potentials (left column) and by coherence and phase spectra (right column) calculated for each pair of three signals: SUM spike train and EEGs recorded in the CA1 region and near the hippocampal fissure (DG). In each case (A–D), all three signals were highly coherent in a narrow range around the theta frequency. The accuracy of the phase estimate is the highest at the peak of the coherence function. The range of reliable phase estimates

are shown by vertical markers. At frequencies outside these markers the phase is meaningless (as shown for the second neuron, B; see Materials and Methods). The relative phase between theta oscillations in the two hippocampal recordings was 1807, i.e. out of phase (CA-DG). A: SUM cell firing on the positive-going segment of CA1 theta, i.e. SUM oscillations leading CA1 theta (CA-SUM) and lagging DG theta (DG-SUM) by one quarter of the theta cycle. B: SUM cell firing on the negative-going segment of CA1. C: SUM cell firing at the positive peak of CA1 theta, i.e. in-phase with CA1. D: SUM cell firing out-of-phase with CA1 theta EEG.

band (i.e. in the area between vertical markers in Figure 1; see Materials and Methods). Phase values in Figure 1A, for instance, demonstrate that the SUM unit led ‘CA1’ theta by about 807 and lagged ‘dentate’ theta by 1007, at each frequency between 3 and 6

Hz. The SUM-hippocampus phase spectra in Figure 1B, for a different neuron, show the reverse relationship. The firing of this unit lagged rhythmic activity in CA1 by 757 and, accordingly, led ‘dentate’ field potentials by 1057. Phase delays of the two other

208

KOCSIS AND VERTES

cells (Fig. 1C and 1D) were 07 and 1807 relative to ‘CA1’ and 1807 and 07 relative to ‘dentate’ theta. The results shown in Figure 1 are representative for the larger group of experiments (10 rats, 18 units) in which the two hippocampal recordings were used for references. The average phase between ‘CA1’ and ‘dentate’ theta rhythms in these experiments was 1817 6 127 (range: 1507–2407). Distribution of phase data for these 18 units (Fig. 2A) shows noticeable grouping in four distinct regions around 07, 1907, 2907, and 1807. The regions were limited to relatively narrow ranges (approximately 407 wide), and were well separated from each other. In fact, no cells were found in the intermediate zones (i.e. 6247–717 and 61207–1627). The average phase of the four groups, relative to ‘CA1’ theta, were 87 6 7 (n 5 6), 977 6 9 (n 5 4), 2977 6 6 (n 5 5), and 21777 6 7 (n 5 3). To decide whether the multiple peaks in the phase distribution were significant or could have been caused by chance we tested the goodness-of-fit of the uniform distribution to the data. The fine resolution (18 bins, representing 207 each) shown in Figure 2A could not be used directly because for a reliable chi-square test the data should be arranged so that expectations in each category was close to 5, or higher (see e.g. Bailey, 1995). Since we observed four peaks separated by four ranges with few or no units, the total range of 07–3607 had to be divided into at least eight categories. For n 5 18 observations, however, this would still yield a low expected number of cells (2.25) in each category. In order to further amalgamate comparable categories with minimal loss of physiological meaning we transformed all phase data to one quarter of the theta cycle into the 07–907 range. This also allowed to include more cells recorded with only one theta signal from the hippocampus, as described below. In seven rats (nine SUM/MB units) the phase between the two hippocampal recordings was less than 207 and it was concluded that both electrodes recorded the same field, the location of which in these experiments was not precisely identified. Since the phase switch between the two out-of-phase theta oscillations in the hippocampus of urethane anesthetized rats occurs within a narrow layer above the fissure (Buzsaki et al., 1983; Bland, 1986), we assumed that the EEG signal reflected the activity of the superficial or the deep theta oscillators and not the field between them. Accordingly, these nine cells could be combined with the others, having first transformed all phase data to one quarter of the theta cycle: 07/1807 to 1907/2907 range. Figure 2B shows that when this is done for all cells, the 27 SUM/MB neurons could be classified into two groups. The first group contained cells firing in-phase (13 6 37, n 5 11) with one of the hippocampal theta oscillations (and out of phase with the other), while the second group consisted of neurons leading one of the hippocampal oscillators (average phase of 797 6 27; n 5 16) or lagging the other (by about 907). No cells were found to fire with phase delays between 247 and 607 from either hippocampal theta recordings. The phase distribution was significantly different from the uniform distribution (chi-square test, P 5 0.01). Phase data of the 18 cells with two out-of-phase hippocampal theta recordings converted into the 07–907 range were also tested separately and proved significantly different from uniform distri-

bution. Furthermore, since all four groups contained a nearly equal number of cells this allows the conclusion that there are four preferred phases of SUM/MB firing. SUM and MB neurons (Figs. 2C,D) were similar in that neurons of both groups fired at one or the other of the two preferred phases. SUM cells were about equally distributed among the two phase groups (i.e., in phase or 907 out of phase), while four of six MB cells fired 907 before the positive peak in the CA1 theta wave. The number of cells and their anatomical distribution (see Figure 1 in Kocsis and Vertes, 1994a) did not allow a more detailed analysis of possible differences in phase values for cells located in different parts within the SUM and MB. No differences were seen between phase relations of neurons firing single or complex spikes (compare filled and empty circles in Figure 2). There was, however, a significant relationship between phase distributions and unit-EEG coherence values (indicated by unequal chi-square/n values for two groups of cells with coherence .0.5 or ,0.5, respectively; coherence is shown as distance of circles from the center in Figure 2): the stronger the relationship of rhythmic unit discharge to theta, the closer its relative phase was to 07 or 907. The classification of a cell was stable across different theta epochs. For 16 SUM/MB cells, the phase of rhythmic firing was calculated for several (up to eight) discontinuous theta epochs occurring spontaneously or elicited by tail pinch. All phase values obtained during different theta epochs for a particular neuron fell within the range of the above four groups. Although individual phase variations for some neurons (mostly those in the 1907 groups) were as high as 50–607 (see Fig. 3A), these variations were always scattered around the group average and therefore never resulted in any uncertainty in the classification of the neuron. The phase distribution histogram of the individual phase data (Fig. 3B, right) also showed characteristic peaks close to 07 and 907 as seen with the distribution of phase averages.

Field Potentials in the SUM/MB In seven rats field potentials were recorded with a macroelectrode in the posterior hypothalamus at different depths ventrally through the levels of the SUM and MB (Fig. 4). In each of these animals, a theta rhythmicity highly coherent (0.86 6 0.13) with hippocampal EEG was recorded in the SUM/MB area. The average phase was 797 6 97 (range 567–997) relative to ‘CA1’ and 2847 6 97 (range between 2567 and 21107) relative to ‘dentate’ theta. Thus, at theta frequency (i.e., the peak frequency in the autospectra and coherence function), SUM/MB field always led ‘CA1’ and lagged ‘dentate’ rhythm by about one quarter of the theta cycle. The relative phase between the two field oscillations, however, was not constant for different frequencies below and above theta. Rather, the phase between the two signals showed a linear increase (in absolute values) within the coherent frequency band; that is, phase differences were larger for faster than for slower oscillations. For example, in the experiment shown in Figure 4, the phase changed from 2407 at 2.4 Hz to 21207 at 5.2 Hz. Such phase relationships indicate a constant time lag between

_____________________________________

HIPPOCAMPAL THETA AND SUM/MB NEURONAL FIRING

209

FIGURE 2. A: Summary distribution of the mean phase of firing of cells in the SUM and MB simultaneously recorded with two out-of-phase hippocampal EEGs (n 5 18). Top, polar scatter plot of mean phase (angles relative to CA1 positive peak, 07) vs. coherence (distance from the origin) values measured at theta frequency; bottom, phase distribution histogram of cell firing with respect to the reference CA1 theta rhythm. B: Summary of all theta cells recorded in the SUM/MB area including those accompanied with two or one

hippocampal recordings (n 5 27). Since the phase of hippocampal theta in the latter case could only be estimated as w 6 n p 1807, cell firing was referenced to one quarter of the theta cycle. C: Cells histologically identified in the SUM (n 5 12). D: Cells histologically identified in the MB (n 5 6). Filled circles in the polar plots represent cells firing complex spikes (Kocsis and Vertes, 1994a), open circles represent cells firing single spikes. Histogram bin width: 207.

the two signals (Kocsis et al., 1990), which in this experiment was 72 ms. As shown in Figure 4A the coherence between field potentials in the SUM and the hippocampus were very high in the SUM layer, dropped to 0.3–0.6 at 1.5–2.0 mm dorsal to SUM and was

again high near the hippocampus, about 5 mm dorsal to SUM. The finding that relative phase of the dorsally recorded potentials was 07 to ‘dentate’ theta in the entire coherent frequency band indicated that they were due to volume conduction from the hippocampal fissure (Fig. 4B,C, 5 mm).

210

KOCSIS AND VERTES

In two experiments, the signal from the microelectrode was passed through two filters with different bands to record both unit activity and field potentials in SUM/MB. One of these units fired nearly in-phase (57) with ‘CA1’ theta, lagging the local field potentials by 957. The discharge of the other unit (shown in Fig. 5) was shifted by one quarter of a cycle relative to the hippocampal (dentate) theta and appeared at the negative phase of the local field potentials.

FIGURE 4. Phase relations of field potential oscillations at different depths in the posterior hypothalamus to theta activity recorded near the hippocampal fissure. A: Phase and coherence (horizontal axis) at different depths (vertical axis). Insert shows the anatomical sketch; pc, posterior commissure; 3V, third ventricle; fr, fasciculus retroflexus. B: Two-second sample records of field potentials at the hippocampal fissure (dark lines) and hypothalamus (thin lines) at dorso-ventral coordinates of 5 mm (DH) and 9 mm (SUM). C: Spectral characteristics of theta activity in DH (left) and SUM (right). Note clear theta peaks in the autospectra (upper traces) and coherence functions (middle traces). Theta oscillations in DH were in-phase with hippocampal EEG at all frequencies between 2.4 and 5.2 Hz (bottom left). Field theta in SUM (bottom right) lagged hippocampal (‘dentate’) EEG by 907 at the frequency of the autospectral peak (theta, 3.6 Hz). The Hipp-SUM phase changed from 2407 to 21207 at different frequencies within the coherent frequency band.

FIGURE 3. Variability of phase relations of SUM/MB cell firing to hippocampal theta rhythm. Each circle represents a phase value calculated for a separate segment of recording with uninterrupted theta. The number of segments varied between one and eight for different neurons. A: Phase values measured during different theta segments for different SUM/MB cells (n 5 27). Each group of circles placed along one vertical represents one neuron (i.e. all circles are plotted along 27 verticals and each of them contains one to eight circles). Units were grouped according to their mean phase. B: Left, polar scatter plot of phase vs. coherence values measured at theta frequency for different theta segments. Filled and open circles as in Figure 2. Right, phase distribution histogram of cell firing.

DISCUSSION The principal conclusion of the present study is that rhythmically firing neurons in the SUM/MB represent a functionally heterogeneous population of cells that are coupled with forebrain theta oscillators at different preferred phases and might therefore play different roles in theta rhythmic activity of the limbic system. Based on their phase relations to hippocampal theta, four groups of rhythmic SUM/MB cells were identified. Neurons of two of the groups fired out-of-phase relative to each other and synchronously

_____________________________________

HIPPOCAMPAL THETA AND SUM/MB NEURONAL FIRING

FIGURE 5. Simultaneous recording of theta EEG at the hippocampal fissure (Hipp) and rhythmic field (SUM/f ) and unit spike activities (SUM/s) in the SUM. A: Two-second sample of the three signals. B: Coherence functions for each pair of signals. C: Phase spectra calculated during two different theta segments. Note different shapes of the phase spectra for Hipp-SUM/f (i.e. changing with frequency) and Hipp-SUM/s (constant within the coherent frequency band).

with the positive or negative peaks of theta field activity in the hippocampus. Cells of the other two groups, also out-of-phase counterparts to each other, fired on the rising or falling segments of CA1 theta waves. Theta field potentials could also be recorded in the SUM/MB and were advanced by one quarter of the cycle relative to CA1 theta oscillations. The present findings extend those of earlier studies in two important respects and consequently provide a more accurate and detailed understanding of SUM/MB phase data. First, for most cells, we recorded theta activity from both the CA1 and dentate layers which is necessary when comparing neuronal firing to the different hippocampal theta generators. Secondly, we analyzed phase values for each cell, rather than pooling the data (population averages), with the intent of revealing potential functional groups of SUM/MB theta cells. This approach seemed reasonable since no preferred phase exists for an entire population of cells in any

211

structure showing theta rhythmic activity. On the other hand, phase analysis has proved valuable for understanding the relationship of cells to the theta rhythm for structures in which preferred phases could be assigned to subgroups of cells (Ranck, 1973; Buzsaki et al., 1983; Holsheimer et al., 1983; Fox et al., 1986; Alonso and Garcia-Austt, 1987; Alonso et al., 1987). The results of the present study indicate that rhythmically firing cells of the SUM/MB, like those of other structures, can be classified on the basis of their phase relations to the theta rhythm. In this regard, three features of the phase distribution deserve special attention. First, the peaks in the phase distribution histogram were well separated; the empty zones between them were usually wider (40–707) than those occupying the phase data for each of the groups. Second, the preferred phases were even more narrowly defined for those recordings in which the coupling between the two oscillations was very strong (i.e., high coherence values). Third, each cell remained tied to a single group; that is, when phase comparisons were made over several theta epochs, even at different frequencies, the phase of firing always fell within the range of a single group. These characteristics support the existence of separate classes of theta cells in the SUM/MB. Nevertheless, it should be mentioned that the present study only included 27 neurons yielding an average of seven cells in each group. The observed multipeaked phase distribution was statistically tested against a flat distribution. It cannot be excluded that the distribution itself may take on a different shape with more data. It is also possible that the peaks were clearer if recordings were taken from freely moving animals. In our sample, the numbers of cells in the 07/1807 and the 1907/2907 group-pairs were nearly equal. Kirk and McNaughton (1993b) reported that the majority of their sample of SUM cells fired on a specific phase of theta: the negative-going phase of their single hippocampal (probably dentate) reference (i.e., 1907 according to our classification). Bland et al. (1995) described two populations of SUM/MB cells, those firing on the falling and rising phases of hippocampal theta. They also recorded only one theta signal from the hippocampus, most likely from the region of the hippocampal fissure. When combining SUM and MB cells and converting their phase data to our nomenclature, their group averages were 1477 and 21087. The spread of the distribution was not described in detail, but according to the standard errors of their group averages the 1477 group might have contained cells with phases close to 07 and to 1907 while the other group (21087) probably mainly contained neurons lagging the CA1 theta by a quarter of a cycle (2907), but also a few cells with relative phases of 1807. The co-existence of theta cells firing out-of-phase relative to each other within the same structure is not a unique feature of the SUM/MB. The two major classes of units in the Ammon’s horn, the pyramidal cells and interneurons, fire on opposite phases, usually close to the negative and positive peaks of the local theta wave, respectively (Ranck, 1973; Buzsaki et al., 1983). A similar relationship has also been reported for granule cells and interneurons of the dentate gyrus during theta in the urethane anesthetized rat (Wolfson et al., 1981). In these structures, the rhythmic

212

KOCSIS AND VERTES

inhibitory modulation of principal cells by local interneurons is thought to be involved in generating this out-of-phase firing. The MS/vDB also contains reciprocally connected excitatory and inhibitory theta neurons (Petsche et al., 1962; Alonso et al., 1987; Brazhnik and Vinogradova, 1988; Stewart and Fox, 1989; Colom and Bland, 1991); the phase of firing to theta of the two sets, however, is unknown. Preliminary observations by Stewart and Fox (1989) indicate that atropine-resistant and atropinesensitive MSvDB neurons (presumably GABAergic and cholinergic cells, respectively) fire on opposite phases of theta. Since both the cholinergic and GABAergic septohippocampal projections are involved in hippocampal theta-generation, precise characterization of their phase relationship would be essential for understanding the MS/vDB mechanisms producing theta. A similar argument could be made for SUM/MB cells. The out-of-phase firing of SUM/MB cells may result from local inhibitory/excitatory interconnections and/or specific actions of different populations of SUM/MB cells on neurons firing on opposite phases of the theta cycle in other structures. Data are presently lacking for either possibility. Interestingly, a subsets of neurons in the SUM and MB fire complex spikes resembling pyramidal cells of the hippocampus while others fire single spikes only. The pharmacological nature of these SUM/MB cells are not known and the present study did not provide evidence for their firing at opposite phases of the theta cycle, as in the hippocampus. Recent findings indicate that the SUM sends frequency coded signal to the MS/vDB pacemaker in the control of theta (Kirk and McNaughton, 1993a; Oddie et al., 1994). The recent demonstration that SUM cells project to both cholinergic and GABAergic MS/vDB neurons (Borhegyi et al., 1994) leads one to speculate that different populations of SUM cells, most likely those firing at 07/1807, could drive out-of-phase rhythmic oscillations in the septum. Functional differentiation of SUM-septal neurons is also indicated by the finding of Leranth and Kiss (1996) demonstrating that although the majority of these cells form asymmetric contacts on MS cholinergic neurons only a subset of them is aspartate/glutamatergic. SUM/MB and hippocampal field potentials are coherent within a relatively wide band over which their relative phase is linearly dependent on frequency. This indicates that SUM/MB field theta follows hippocampal theta with a constant time delay. As the frequency increases, or as the cycle length decreases, the same time delay (in ms) will comprise a larger part of the cycle resulting in larger phase values (see Kocsis et al., 1990). The origin of the rhythmic drive reaching the SUM/MB 70–80 ms after the negative peak of CA1 theta (or the positive peak of dentate theta, see Fig. 5) is not now known. It may involve the well-established subicular to MB projection (Swanson and Cowan, 1975; Allen and Hopkins, 1989) or alternatively a MS/vDB to SUM/MB projections (Swanson and Cowan, 1979; Gonzalo-Ruiz et al., 1992; Kiss et al., 1995; Hayakawa and Zyo, 1993, 1996). Descending rhythmic influences could also activate SUM/MB cells firing in-phase with the local field theta; that is, mainly those cells belonging to 1907/2907 group. The proportion of cells firing in-phase with SUM/MB field potentials was higher in MB than in SUM indicating that a larger percentage of MB than SUM

cells are possibly driven from forebrain sites. This is consistent with our partial coherence analysis of the relationship of SUM/MB cells to theta (Kocsis and Vertes, 1994a) and recent work of Kirk et al. (1996) showing that inactivation of the MS/vDB disrupts the theta rhythmical discharge of MB but not SUM cells. The present and previous analyses indicate that the SUM contains cells that likely drive or are driven by the septal pacemaker.

Acknowledgments This work was supported by OTKA grant T17778 (B.K.) and by NIH grant NS 35883 (R.P.V.).

REFERENCES Allen GV, Hopkins DA (1989) Mammillary body in the rat: topography and synaptology of projections from the subicular complex, prefrontal complex, and midbrain tegmentum. J Comp Neurol 286:311–336. Alonso A, Garcia-Austt E (1987) Neuronal sources of the theta rhythm in the entorhinal cortex of the rat. II Phase relations between unit discharges and theta field potentials. Exp Brain Res 67:502–509. Alonso A, Llinas RR (1992) Electrophysiology of the mammillary complex in vitro. II. Medial mammillary neurons. J Neurophysiol 68:1321–1331. Alonso A, Gaztelu JM, Buno W Jr, Garcia-Austt E (1987) Crosscorrelation analysis of septohippocampal neurons during thetarhythm. Brain Res 413:135–146. Amaral DG, Cowan WM (1980) Subcortical afferents to the hippocampal formation in the monkey. J Comp Neurol 189:573–591. Bailey NTJ (1995) Statistical methods in biology. Cambridge: Cambridge University Press. Bendat JS, Piersol AG (1966) Measurement and analysis of random data. New York: Wiley. Benignus VA (1969) Estimation of the coherence spectrum and its confidence interval using the Fast Fourier Transform. IEEE Trans Audio Electroacoustics AU-17:145–150. Benignus VA (1970) Correction to ‘‘Estimation of the coherence spectrum and its confidence interval using the Fast Fourier Transform.’’ IEEE Trans Audio Electroacoustics AU-18:320. Bland BH (1986) The physiology and pharmacology of hippocampal formation theta rhythms. Prog Neurobiol 26:1–54. Bland BH, Colom LV (1993) Extrinsic and intrinsic properties underlying oscillation and synchrony in limbic cortex. Prog Neurobiol 41:157–208. Bland BH, Konopacki J, Kirk IJ, Oddie SD, Dickson CT (1995) Discharge patterns of hippocampal theta-related cells in the caudal diencephalon. J Neurophysiol 74:322–333. Borhegyi Zs, Acsady L, Magloczky Zs, Freund TE (1994) Postsynaptic targets of supramammillary afferents in the medial septum. Eur J Neurosci Suppl 7:207. Brazhnik S, Vinogradova OS (1988) Modulation of the afferent input to the septal neurons by cholinergic drugs. Brain Res 451:1–12. Buzsaki G (1989) A two-stage model model of memory trace formation: a role of ‘‘noisy’’ brain states. Neuroscience 31:551–570. Buzsaki G, Leung LS, Vanderwolf CH (1983) Cellular basis of hippocampal EEG in the behaving rat. Brain Res Rev 6:139–171. Christakos CN, Cohen MI, Barnhardt R, Shaw CF (1991) Fast rhythms in phrenic motoneuron and nerve discharges. J Neurophysiol 66:674– 687.

_____________________________________

HIPPOCAMPAL THETA AND SUM/MB NEURONAL FIRING

Colom LV, Bland BH (1991) Medial septal cell interactions in relation to hippocampal field activity and the effects of atropine. Hippocampus 1:15–30. Fox SE, Wolfson S, Ranck JB Jr (1986) Hippocampal theta rhythm and the firing of neurons in walking and urethane anesthetized rats. Exp Brain Res 62:495–508. Gonzalo-Ruiz A, Alonso A, Sanz JM, Llinas RR (1992) Afferent projections to the mammillary complex of the rat, with special reference to those from surrounding hypothalamic regions. J Comp Neurol 321:277–299. Green JD, Arduini A (1954) Hippocampal electrical activity in arousal. J Neurophysiol 17:533–557. Haglund L, Swanson LW, Kohler C (1984) The projection of the supramammillary nucleus to the hippocampal formation: an immunohistochemical and anterograde transport study with the lectin PHA-L in the rat. J Comp Neurol 229:171–185. Hayakawa T, Zyo K (1993) Neuroanatomical study of afferent projections to the supramammillary nucleus. Anat Embryol 188:139–148. Hayakawa T, Zyo K (1994) Fine structure of the supramammillary nucleus of the rat: analysis of the ultrastructural character of dopaminergic neurons. J Comp Neurol 346:127–136. Hayakawa T, Zyo K (1996) Synaptic organization of afferent projections to the supramammillary nucleus in the rat. Anat Embryol 193:249– 257. Holsheimer J, Stok CJ, Lopes da Silva FH (1983) Theta rhythm related hippocampal cell discharges in the urethane anesthetized rat: evidence for a predominant entorhinal input. Electroencephalogr Clin Neurophysiol 55:464–467. Huerta PT, Lisman JE (1993) Heightened synaptic plasticity of hippocampal CA1 neurons during a cholinergically induced rhythmic state. Nature 364:723–725. James DTD, McNaughton N, Rawlins JNP, Feldon JA, Gray JA (1977) Septal driving of hippocampal theta rhythm as a function of frequency in the free moving male rat. Neuroscience 2:1007–1117. Jenkins GM, Watts DG (1968) Spectral analysis and its applications. San Francisco: Holden-Day. Jung R, Kornmuller A (1938) Eine Methodik der Ableitung lokalisierter Potential-Schwankungen aus subkortikalen Hirngeneiten. Arch Psychiatr Nervenkr 109:1–30. Kirk IJ, McNaughton N (1991) Supramammillary cell firing and hippocampal rhythmical slow activity. NeuroReport 2:723–725. Kirk IJ, McNaughton N (1993a) Mapping the differential effects of procaine on frequency and amplitude of reticularly elicited hippocampal rhythmical slow activity. Hippocampus 3:517–529. Kirk IJ, McNaughton N (1993b) Discharge patterns of supramammillary neurons in relations to reticularly-elicited hippocampal rhythmical slow wave activity (theta). Neurosci Abstr 19:355. Kirk IJ, Oddie SD, Konopacki J, Bland BH (1996) Evidence for differential control of posterior-hypothalamic, supramammillary, and medial mammillary theta-related cellular discharge by ascending and descending pathways. J Neurosci 16:5547–5554. Kiss J, Magloczky Zs, Borhegyi Zs, Freund TF (1995) Organization of the medial septal region in the rat brain: topographic distribution and connections of calretinin-containing cells. Eur J Neurosci Suppl 8:149. Kocsis B, Vertes RP (1992) Dorsal raphe neurons: synchronous discharge with theta rhythm of the hippocampus in the freely moving rat. J Neurophysiol 68:1463–1467. Kocsis B, Vertes RP (1994a) Characterization of neurons of the supramammillary nucleus and mammillary body that discharge rhythmically with the hippocampal theta rhythm in the rat. J Neurosci 14:7040–7052. Kocsis B, Vertes RP (1994b) Phase relationship between theta rhythmic activities in the mammillary complex (SUM/MB) and the hippocampus. Neurosci Abstr 20:341.

213

Kocsis B, Gebber GL, Barman SM, Kenney MJ (1990) Relationships between activity of sympathetic nerve pairs: phase and coherence. Am J Physiol 259:R549–R560. Larson J, Lynch G (1986) Induction of synaptic potentiation in hippocampus by patterned stimulation involves two events. Science 232:985–988. Leranth C, Kiss F (1996) A population of supramammillary area calretinin neurons terminating on medial septal area cholinergic and lateral septal area calbindin-containing cells are aspartate/glutamatergic. J Neurosci 16:7699–7710. Leranth C, Nitsch R (1994) Morphological evidence that hypothalamic substance P-containing afferents are capable of filtering the signal flow in the monkey hippocampal formation. J Neurosci 14:4079–4094. Llinas RR, Alonso A (1992) Electrophysiology of the mammillary complex in vitro. I. Tuberomammillary and lateral mammillary neurons. J Neurophysiol 68:1307–1320. Magloczky Z, Acsady L, Freund TF (1994) Principal cells are the postsynaptic targets of supramammillary afferents in the hippocampus of the rat. Hippocampus 4:322–334. Mitchell SJ, Rawlins JNP, Steward O, Olton DS (1982) Medial septal area lesions disrupt u rhythm and cholinergic staining in medial entorhinal cortex and produce impaired radial arm maze behavior in rats. J Neurosci 2:292–302. Nunez A, de Andres I, Garcia-Austt E (1991) Relationship of nucleus reticularis pontis caudalis neuronal discharge with sensory and carbachol evoked hippocampal theta rhythm. Exp Brain Res 87:303– 308. Oddie SD, Bland BH, Colom LV, Vertes RP (1994) The midline posterior hypothalamic region comprises a critical part of the ascending brainstem hippocampal synchronizing pathway. Hippocampus 4:454–473. O’Keefe J, Nadel L (1978) The hippocampus as a cognitive map. Oxford: Clarendon. Papez JW (1937) A proposed mechanism of emotion. Arch Neurol Psychiatry 38:725–743. Pavlides C, Greenstein YJ, Grudman M, Winson J (1988) Long-term potentiation in the dentate gyrus is induced preferentially on the positive phase of theta-rhythm. Brain Res 439:383–387. Petsche H, Stumpf CH, Gogolak G (1962) The significance of the rabbit’s septum as a relay station between the midbrain and the hippocampus. I. The control of hippocampus arousal activity by the septum cells. Electroencephalogr Clin Neurophysiol 14:202–211. Ranck JB Jr (1973) Studies on single neurons in dorsal hippocampal formation and septum of unrestrained rats. I. Behavioral correlates and firing repertoires. Exp Neurol 41:461–555. Rawlins JNP, Feldon J, Gray JA (1979) Septo-hippocampal connections and the hippocampal theta rhythm. Exp Brain Res 37:49–63. Shibata (1988) A direct projection from the entorhinal cortex to the mammillary nuclei in the rat. Neurosci Lett 90:6–10. Staubli U, Lynch G (1987) Stable hippocampal long-term potentiation elicited by ‘theta’ pattern stimulation. Brain Res 435:227–234. Stewart M, Fox SE (1989) Firing relations of medial septal neurons to the hippocampal theta rhythm in urethane anesthetized rats. Exp Brain Res 77:507–516. Stewart M, Barry M, Quirk GJ, Fox SE (1991) Firing relations of entorhinal neurons to the hippocampal theta rhythm in walking and urethane anesthetized walking rats. Neurosci Abstr 17:1036. Stewart M, Quirk GJ, Barry M, Fox SE (1992) Firing relations of medial entorhinal neurons to the hippocampal theta rhythm in urethane anesthetized and walking rats. Exp Brain Res 90:21–28. Stumpf CH, Petsche H, Gogolak G (1962) The significance of the rabbit’s septum as a relay station between the midbrain and the hippocampus. II. The differential effects of drugs upon both the septal cell firing pattern and the hippocampus theta activity. Electroencephalogr Clin Neurophysiol 14:212–219.

214

KOCSIS AND VERTES

Swanson LW, Cowan WM (1975) Hippocampo-hypothalamic connections: origin in subicular complex, not in Ammon’s horn. Science 189:303–304. Swanson LW, Cowan WM (1979) The connections of the septal region in the rat. J Comp Neurol 186:621–665. Vanderwolf CH (1969) Hippocampal electrical activity and voluntary movement in the rat. Electroencephalogr Clin Neurophysiol 26:407– 418. Vertes RP (1979) Brain stem gigantocellular neurons: patterns of activity during behavior and sleep in the freely moving rat. J Neurophysiol 42:214–228. Vertes RP (1981) An analysis of ascending brain stem systems involved in hippocampal synchronization and desynchronization. J Neurophysiol 46:1140–1159. Vertes RP (1992) PHA-L analysis of projections from the supramammillary nucleus in the rat. J Comp Neurol 326:595–622. Vertes RP, Martin GF (1988) Autoradiographic analysis of ascending

projections from the pontine and mesencephalic reticular formation and the median raphe nucleus in the rat. J Comp Neurol 275:511– 541. Vertes RP, Colom JV, Fortin WJ, Bland BH (1993) Brainstem sites for the carbachol elicitation of the hippocampal theta rhythm in the rat. Exp Brain Res 96:419–429. Vinogradova OS (1995) Expression, control, and probable functional significance of the neuronal theta rhythm. Prog Neurobiol 45:523– 583. Wetzel W, Ott T, Matthies H (1977) Hippocampal rhythmic slow activity (theta) and behavior elicited by medial septal stimulation. Behav Biol 19:534–542. Winson J (1972) Interspecies differences in the occurrence of theta. Behav Biol 7:479–487. Wolfson S, Fox SE, Ranck JB Jr (1981) Hippocampal theta rhythm: phase relations of neuron firing and conductance in urethanized rats. Neurosci Abstr 7:451.