- Email: [email protected]
Periglomerular cell action on mitral cells in olfactory bulb shown by current source density analysis
Brain Research, 308 (1984) 223-233 Elsevier
223
BRE 10266
Periglomerular Cell Action on Mitral Cells in Olfactory Bulb Shown by Current Source Density Analysis DIANE P. MARTINEZ and WALTER J. FREEMAN
Department (~lPhysiology-Anatomy, University of California, Berkeley, CA 94720 ( U.S.A.) (Accepted January 31st, 1984)
Key words: current source-sink density (CSD) - - electrophysiology - - CNS - - olfactory bulb - - periglomerular ceils - - rabbit
The sign of action of periglomerular (PG) cells on the apical dendrites of mitral cells in olfactory bulb glomeruli was investigated by constructing current source density (CSD) profiles from potentials evoked by primary olfactory nerve (PON) and lateral olfactory tract (LOT) stimulation. Evoked potentials were recorded and averaged from anesthetized rabbit simultaneously with a 1 x 16 array of electrodes positioned perpendicular to the bulbar surface. A one-dimensional CSD analysis with depth was made in the center of PON- and LOT-evoked potential activity. CSD was plotted vs depth for specific times during the average evoked potential (AEP): at the surface peaks of the first surface-negative wave (NI), the first surface-positive wave (PI), and the second surface-negative wave (N2). N l. Pl and N2 corresponded to excitation, dis-excitation (equivalent to inhibition), and dis-inhibition (re-excitation) of the granule cell population through mitral cell basal dendrites. The granule cells generated both PON and LOT oscillatory AEPs. When N1 and P1 profiles or PI and N2 profiles were combined, the source and sink due to granule cell activity were minimized and another source sink pair was revealed on PON but not LOT stimulation. PON-evoked N1 + P1 or PI + N2 profiles showed a secondary soucesink pair not present with LOT stimulation. The sink was located in the glomerular layer (GL) and outer plcxiform laycr (EPL) and the source in the inner EPL. It was concluded that long-lasting excitation of the mitral cells was taking place at the GL and GL/EPL border. This excitation was ascribed to concomitant PG cell activity, possibly in combination with prolonged monosynaptic PON excitation of the apical dendrites. The results support the occurrence of direct excitatory action of PG cells onto mitral cells.
INTRODUCTION
from the P O N and e x c i t a t o r y d e n d r o d e n d r i t i c input from mitral cells, and that they in turn m a k e d e n d r o -
The olfactory bulb is the site of the first synapse in
dendritic
and
axodendritic
contacts
with
mitral
the sensory system for o d o r . T h e large b u l b a r projec-
cells 22. It is a g r e e d that p e r i g l o m e r u l a r activity sup-
tion n e u r o n s , the mitral cells ( d e f i n e d h e r e to include
presses mitral cell r e s p o n s e to a test P O N shock that
tufted cells), receive input directly f r o m r e c e p t o r s
follows a c o n d i t i o n i n g P O N shock 4,5,L~,21. H o w e v e r ,
through the p r i m a r y o l f a c t o r y n e r v e ( P O N ) and send
the nature and m e c h a n i s m of this effect are at issue.
their axons in the lateral o l f a c t o r y tract. T h e r e are
O n e view is that p e r i g l o m e r u l a r synapses cause di-
two main classes of i n t e r n e u r o n s , granule cells and
rect synaptic inhibition in the mitral apical den-
p e r i g l o m e r u l a r cells. G r a n u l e cells are k n o w n to in-
drites'). A n o t h e r view24 is that the suppression of the
hibit mitral cells at their basal d e n d r i t e s ; they g e n e r -
mitral responses manifests ' p r e s y n a p t i c inhibition'.
ate the electrical current fields m a n i f e s t e d in the electroencephalographic
(EEG)
Freeman~ s h o w e d that the suppression occurs not
oscillatory p a t t e r n s of
only for the mitral cell r e s p o n s e s to P O N input but
bulbar r e s p o n s e to o d o r s and to electrical stimuli.
also p e r i g l o m e r u l a r r e s p o n s e s to P O N input. H e pro-
P e r i g l o m e r u l a r cells act o n t o mitral cell apical dendrites: the nature of their action is in dispute.
vided e v i d e n c e that p e r i g l o m e r u l a r cells excite each o t h e r 4,7 and a t t e n u a t e (i.e. partially block or finhib-
T h e r e is a g r e e m e n t in tile literature that periglo-
it') transmission in the g l o m e r u l i by releasing a sub-
m e r u l a r cells receive e x c i t a t o r y a x o d e n d r i t i c input
stance, possibly K + (ref. 11), into the e x t r a c e l l u l a r
Corre~pondence: W. J. Freeman, Department of Physiology-Anatomy, University of California, Berkeley, CA 947211. U.S.A. 110(16-89~3/84503.ql0 © 1984 Elsevier Science Publishers B.V.
224 compartment of the glomeruli. By network analysis of periglomerular, mitral and granule cell function. Freeman concluded that periglomerular cells cause direct synaptic excitation in the mitral apical dendrites~,m. In the present work we studied the action of periglomerular cells on mitral cells by measurement of buibar evoked potentials and calculation of current sources and sinks in the vicinity of the periglomerular/mitral synapses and the adjacent apical shafts. We foresaw three possible outcomes. (a) If the periglomerular cells were directly excitatory to mitral apical tufts, a current sink would be observed in the gtomerular layer (GL) and a source in the subjacent external plexiform layer (EPL), and they would persist monotonically for the duration of the evoked periglomerular activity (several tens of millisecondsS). (b) If the periglomerular cells were directly inhibitory, a current source would develop in the G L with a sink in the subjacent EPL. (c) If periglomerular action were to 'clamp' mitral cells (e.g. possibly by an increase in C1 conductance in the apical tufts), then no localized source-sink pattern would be seen. We uncovered the source-sink pair predicted as in (a) by averaging sequential alternating positive and negative peaks of oscillatory average evoked potentials ( A E P ) in order to minimize the masking effect of the powerful granule cell dipole field. We then sought to exclude forms of neural action in the bulb other than periglomerular excitation of mitral cells, that might have accounted for the newly observed source-sink pair. By its location we showed that it could be generated only by the mitral apical dendrites. By comparing responses to orthodromic and antidromic volleys we localized the active synapses to the apical tufts within the glomerular sink. By the same means and by measurement of its onset latency we excluded the role of recurrent activation of centrifugal pathways following PON stimulation, and minimized the role of possible slow EPSP component involving excitation of mitral cells by PON axon terminals. By measuring its time course we showed that it was correlated with the rate of periglomerular cell activity of single-shock PON stimulation and not with the prolonged changes in glomerular transmission that attenuate subsequent throughput 4. However, the methods we used did not distinguish which of the subtypes of periglomerular cells delivered the synap-
tic input to the mitral apical dcndritt,,s. MATERIAl. AND METHODS
Surgical techniques Adult New Zeahmd white rabbits (,2.5--3.5 kgt were anesthetized with sodium pentobarbilal (35-45 mg/kg). The dorsolateral surfacc of the otfactor~ bulb and L O T was exposed b 3 c~rbital exentcration and removal of the overlying bone and dura. "[he brain surface was covered with warm "lyrode's soh,tion car mineral oil. Evoked potentials used for CSD analysis with depth were recorded with a i x 16 array (perpendicular to the bulbar surface) of (t.0(H 5 in. diameter (3~ urn) enamelled stainless-steel wires (average interelectrode distance; 69 _+ 9/~m), aligned side-by-side in a ribbon. The array was cut at an acute angle (40 ~) and the array beveled to a knife-like edge to alloys smooth penetration of the bulb. Each interelectrode distance was measured under a m~croscope. A I × < array (placed on the surface l o t 300-urn stainlesssteel wire was used to record ew~ked potentials from the surface of the bulb in order to locate the focus of evoked potential activity. All potentials were measured with respect to a reference stainless-steel electrode located at the edge of the surmcat field (20-30 cm). This lay outside the evoked field of activity but sufficiently within the fields of bulbar E E G activntv to give a significant degree of c o m m o n mode rejection. The PON stimulating electrode was a 1 x 5 array of polished 300 um diameter stainless-steel wires. placed on the PON of the anterior bulb surface m a line perpendicular to the PON axons. The stimulus pulse was passed between the pmr of electrodes within the 1 x 5 array that evoked the lowest threshold response. The L O T was stimulated with a bipolar stainless-steel electrode placed tan the L O T surface. For PON evoked potentials the recording site was located by activating the P O N using low intensity ~5 V, 0.01 ms duration) stimuli at 2 Hz. The maximum amplitude response (focus) was found by searching with the 1 × 5 surface recording array. The I,()'F stimulating electrode was then !x)sitioned so that the focus of its bulbar response field overlapped that or the PON focus. The L O T stimulus rate was 7 ltz. With stimulating electrodes in place and the focus ot the PON and L O T responses located, the 1 > 16 re-
225 cording array was advanced perpendicularly through the epicenter of the focus.
Collection and storage of A EP data The ew)ked potentials were amplified with 16 fixed-gain (10,It00) preamplifiers (0.1 H z - 3 kHz band width), multiplexed and digitized sequentially at 12-us intervals across the array. A digitizing interval of 1-ms was found to suffice for the observed rates of change in evoked potential, with correction for the sequential sampling procedure (_+6-_+90 ,us) by linear interpolation. The responses to 200-400 stimuli were summed and stored in an Interdata 716 computer. AEPs selected for later CSD analysis displayed oscillatory waveforms that were symmetrical waveshapes across the zero isopotential (the layer of potential reversal). In the case of PON stimuli they revealed a baseline shift of the oscillation which also reversed polarity at the same depth as the oscillatory component ~4. The baseline shift was a slow monotonic component (a surface negative, deep positive dipole) of the PON evoked potential which was absent with L O T stimulation and with high PON stimulus intensity or bulbar damage. When this shift was present the A E P oscillated about a baseline shifted above or below the prestimulus baseline. Presence of a baseline shift in PON AEPs was inferred for reasons given previously 4.5 to indicate activity of the periglomerular cell population. At the completion of each A E P data gathering session, with the 1 x 16 array still in place, the array and amplifiers were calibrated as a unit by applying to the tissue an 80 Hz sine wave current (100/~A) with electrodes near the reference and ground electrodes. These artificial signals were recorded, averaged and stored on disk for later use as amplitude correction factors for the AEPs obtained from this experiment. A similar procedure was used to evaluate the resistivity of bulbar tissue as a function of depth through the layers~L The depth of the MCL and the widths of the EPL, GL and PON along the recording sites were determined from Nissl-stained frozen sections. The 16 electrode positions with respect to the bulbar laminae were determined by noting at what point along the array the polarity of the A E P reversed (zero isopotential) during the first peak of the oscillation. The
A E P had a characteristic pattern at each bulbar lamina 14,1s2° with the zero isopotential for PON stimulation occurring 100#m above the MCL during NI and at the MCL thereafter along the axis of symmetry s. Since interelectrode distances were fixed, and the thicknesses of the bulbar layers were relatively constant over animals in the lateral area of the bulb used for this study, the positions in particular lamina of all the electrodes could be determined for each experiment by this method. When results from several experiments were combined, CSD values from corresponding laminae were aligned by matching the zero isopotential points from each experiment.
CSD calculations The analysis was performed in one dimension (depth) for bulbar potentials after considering the following conditions. (1) The location of the recording site was on a relatively flat bulbar area, and the activity was restricted to that area, allowing the effects of tissue curvature to be neglected. (2) Recording along the axis of symmetry of the focus of a synchronously activated population of bulbar neurons which extended through the depth of the focus~ ensured restriction of extracellular current normal to the bulbar surface along that axis. (3) Electrode penetration and placement were normal to the bulbar surface at the epicenter. (4) Change in potential with time across recording sites due to preparation instability was minimized by using near-simultaneous recording with the 1 x 16 array (12us time interval between sequential electrode samples). In order to extract the signal from the background E E G in the lightly anesthetized rabbit, the digitized waveform was averaged for n = 200-40(I consecutive stimuli: this reduced background noise by a factor of n °-5. AEPs were calculated from the summed amplitudes, corrected for differences in channel gains and plotted in time series (V vs t) and depth profiles (V vs z) at selected latencies. The following formula was used to obtain CSD values: {V[z+n.A(z,t)]-2V(z,t)+ V[z-n.A(z,t)]} CSD(z,t) = 0 (n'Az)2 For depth z at time t, CSD (z,t) is an approximation of the second spatial derivative, which is proportional (by the inverse of the factor (?, tissue resistivity in the z direction) to the rate of change in current density. The resistivity of bulbar tissue was measured
226
A B
/
///
......
C
surlace
PON
SOURCE.
I
/ ,t GI
X
•
5
/ PG
/
/ ./
. . . . . . .
/ //
EEPL
MCL IPL
GRL
1looju
\ \
.........
25
msec
Fig. 1. A: sixteen averaged evoked potentials (n = 200) recorded from the rabbit olfactory bulb in response to stimulation of the primary olfactory nerve (PON), using a 1 x 16 electrode array. The recording site for each electrode within the bulbar laminae is indicated by lines drawn to the diagram at left. illustrating the bulbar lamina including the primary olfactory nerve (PON) at the surface the olfactory glomerulus (dashed circle) with a nearby perigtomerular cell (PG), the mitral cell (M), apical dendrite, cell body, basal dendrites and axon and a granule cell (G) interneuron. Each AEP is digitized at l-ms intervals with a prestimulus baseline of 30 ms followed by the shock artifact and the oscillatory response. PON AEPs also oscillate about a baseline shift (negative above the turn over and positive below). Negatively down. Abbreviations: GL. glomerular layer; EPL, external plexiform layer; MCL. mitral celt layer; IPL, internal plexiform layer; G R L granule cell layer. B: current source density (CSD) values vs time calculated for nearest neighbor electrodes using the digitized PON AEPs recorded with the l x 16 electrode array placed across the bulbar lamina. Amplitudes are in arbitrary units, with resistivity across the lamina assumed constant (see text). C: CSD calculations using the second nearest neighbor electrodes. ( t h e a v e r a g e w a s 343 _ 65 o h m ' c m 2 / c m )
across the
Since interelectrode distances were not equal, n'dz
layers and f o u n d to h a v e insignificant e f f e c t on C S D
w a s r e p l a c e d by t h e a p p r o p r i a t e v a l u e . B e t t e r signal-
profiles 13, so w a s a s s u m e d to be c o n s t a n t in t h e s e cal-
t o - n o i s e ratio w a s o b t a i n e d b y u s i n g n = 2 ( s e c o n d
culations. V ( z , t ) w a s the p o t e n t i a l a m p l i t u d e at p o i n t
n e a r e s t n e i g h b o r ) , but at t h e e x p e n s e o f spatial r e s o -
z and t i m e t w i t h r e s p e c t to a distant r e f e r e n c e . T h e
lution 2J5. B o t h nearest n e i g h b o r a n d s e c o n d n e a r e s t
p o t e n t i a l s at the a d j a c e n t e l e c t r o d e s A z ~ m superfi-
neighbor CSD maps were calculated and compared
cial
to e x a m i n e the c o m p r o m i s e b e t w e e n r e s o l u t i o n and
and
deep
V[z+n'A(z,t)]
to
z
were
V[z-n-A(z,t)]
respectively, where n -
and
I or n = 2.
noise.
227 The CSD value was calculated for each sampling
o
time of the A E P , starting with the prestimulus interval of approximately 10 ms (used to define the baseline level of zero potential and zero CSD), for each of the 14 (n = l) or 12 (n = 2) electrodes. The CSD waveforms (CSD vs t) and depth profiles (CSD vs z) were plotted. Negative CSD values indicated net inward flow of current or a sink: positive CSD values indicated a source. As a check on the validity of the results, a sum of the sources was c o m p a r e d to a sum of the sinks. If the ratio of sources plus sinks to the sum of their absolute values for a depth profile was greater than 0.25, the
-10
N1
0
I0 i b
GRL.
I~,I~i L
L
EPL
I GL
I PONI
-1 0
P1
data were discarded. RESUI_TS
Averaged evoked potentials and CSD profiles jor PON stimuli A representative set of digitized PON averaged evoked potentials (n = 200) are shown in Fig. 1A. The waveforms were initially surface negative/deep positive (peak N1), then surface positive/deep negative (peak PI), and so forth for peaks N2, P2, etc. The PON A E P s showed a baseline shift of the oscillatory potential in a negative direction on the surface and in a positive direction deep, with the zero isopotential for both the monotonic shift and the oscillatory c o m p o n e n t located just above the mitral cell layer. The depth profiles for both P O N and L O T A E P s showed a near symmetry about the reversal point (i.e. surface N1 = - d e e p N l), validating the elimination of the effects of bulbar curvature on the potential amplitudes. Fig. 2 shows the potential vs depth (normalized to _+1, where 1 equals maximum amplitude of the oscillation at any point) of the PON A E P s for peaks N1 and P 1 of the same experiment. Profiles at N 1 and P2 were similar to each other but opposite to that at P1. The potential reversal occurred midway between electrodes 8 and 9 at N 1 and N2 and slight d e e p e r at P2. These profiles were used to determine the array position with respect to the anatomical turnover point (located a p p r o x i m a t e l y 100~tm above the M C L at the peak of N1 (see ref. 8)) for each set of A E P s and thereby locate the 16 recording sites in specific laminae. The sets of nearest neighbor and second nearest
1.0 ~oo~
SINK
+
SOURCE
Fig. 2. Profiles of potential amplitude (solid triangles) vs bulbar depth for PON AEPs, showing the potential polarity and reversal site for the first two peaks of the oscillation: N1 (a) and PI (b). Amplitudes were scaled by dividing each of the 16 values for a peak by the spatial maximum. Location of bulbar lamina is indicated along the depth (x-axis) (superficial layers at right). Electrode 1 (left) is deepest. Current source density depth profile (solid lines) calculated for PON averaged evoked potentials using the second nearest neighbor method, showing the location of sources (downward deflection) and sinks (upward deflection). Dashed lines are the results of including resistivity variations across the lamina in the CSD calculations, a: profile at the peak of N1, showing superficial sinks and deep sources; b: map at the peak of P1 (when granule cell dipole field is of opposite polarity from its orientation at N1 ), with superficial sources and deep sinks. CSD amplitude scaled to extremumat +1 o r - l . neighbor PON CSDs vs time calculated for the A E P s in Fig. 1A are shown in Fig. 1B, C. Although the data from the second nearest neighbor calculation did not span the bulbar thickness that the nearest neighbor calculation did and was less localized, the variance was reduced. Fig. 2 shows the CSD vs depth profiles for N1 and P1 using the second nearest neighbor data from Fig. IC. W h e n comparing A E P and CSD profiles, the maximum negative and positive amplitudes of the field potential at the times N 1 and P2 were seen not to correspond in depth with the locations of maximal sources and sinks. In Fig. 3 the results of 9 PON experiments are averaged to form one CSD profile for each peak using nearest neighbor calculations. The distribution of
228 became prominent and an accompanying source ap-
-10
peared (Fig. 4). This secondary dipole consisted of a sink in the G L and superficial third of the EPL and a source in the deep half of the EPI~ and in the MCLI
NI
The magnitude of the secondary dipole dccrcased from the time of N1 + P1 to the time of Pt .÷ N2, but the location of its source and sink remained the same. This indicated that this secondary s o u r c e - s i n k pair
i
o.L
l l tL
IoL I.o.I
was slowly varying, which was a requirement for using NI + P1 or P1 + N2 combinations to reveal it. In Fig. 5, CSD profiles constructed at the time
-I,0
J
I
PI
7
NI + PI
o
I
1.0
SINK
+
1.o
-loj c
-I0
N2
SOURCE
oL I
0 1o SINK
1.0
I00#
+
-'O c
SOURCE
Fig. 3. PON-evoked CSD profile for peaks N1 (a), P1 (b) and N2 (c) using the combined results from 9 animals. The nearest neighbor CSD calculation (mean + S.E.) is plotted vs bulbar depth for the 3 different peak times. CSD in arbitrary units, sources and sinks in the bulb (mean + 1 S.E.) for PON stimulation is shown at N1, P1 and N2. During N1 (Fig. 3a), a sink occurred in the deep half of the
X~V
d
,o - 10~
lO0~,l
SINK
SOURCE
EPL with a source in the superficial G R L and IPL. The peak amplitudes of the source and sink were located slightly deeper at P1 and N2 and were reversed during P1 (Fig. 3b, c). The turnover occurred approximately 5 0 - 6 0 # m above the M C L during N1 and at the IPL/MC1 border during P1 and N2. In both the nearest neighbor and second nearest neighbor calculations there appeared to be an additional sink located more superficially than the major sink at both N1 and P1 peaks and perhaps N2. W h e n the CSD profiles of N1 and P1 or of P1 and N2 were combined, the primary s o u r c e - s i n k pair seen in Figs. 2 and 3 disappeared. The more superficial sink
~O Fig. 4 Effects of minimizing the source-sink pair due to the granule cell dipole field on the PON-evoked CSD profiles. By adding together PON CSD profiles for N1 and P1 from each of 9 experiments, the granule cell source-sink pair was combined with its mirror image and minimized. The same procedure was applied to the profiles for P1 and N2. Additional sources and sinks are revealed with both second nearest neighbor and nearest neighbor calculation, a: second nearest neighbor combined profile for N1 and PI: b: second nearest neighbor combined profile for P1 and N2: c: nearest neighbor combined profile for N1 and PI: d: nearest nmghbor combined profile for P1 and N2. Mean of 9 experiments ± S.E. CSD in arbitrary units.
229 a
potential amplitude vs depth for L O T A E P peaks
-10
N1, PI and N2 showed that the sign of the potential at various depths in the bulb matched that of the PON A E P graphs for N1, P1 and N2. The set of nearest neighbor and second nearest neighbor L O T CSDs vs time and depth were calculated for the same experiments and array positions as
1.0
those for the PON CSDs. The extent of s o u r c e - s i n k density along the z axis for the L O T stimulus was more localized than the activity due to PON stimulus,
b
-I0
i.e. there was less activity near the bulbar surface
t
with LOT stimulation. The averaged second nearest neighbor CSD profiles for 4 LOT experiments are shown in Fig. 6. In both nearest neighbor and second a I0
-1.0
SINK
l'O0;u
÷
N1
SOURCE
Fig. 5. PON-evoked CSD profiles constructed for two different times shortly after the stimulus, a: at time T1, 2-3 ms before the start of the first peak (N1) and during postsynaptie activity due to the PON action potential; b: at time T2, on the rising slope of the first peak (N 1), when the sources and sinks due to bulbar cell activity become prominent. Second nearest neighbor calculation: mean of 7 experiments _+S.E. CSD in arbitrary units.
0
10
b
o.L
I'fiJL
EpL
I
oL
o.L
liltlL
E.L
] °~
I Po.
t
-1.0
(T1) of the PON action potential (before the rise of N 1) and also at the time (T2) of the initial slope of N 1
P1
are shown. Seven PON experiments were averaged and the mean ± S.E. plotted vs depth (two PON experiments were eliminated because of interference from the shock artifact at short latencies). Fig. 5a shows the CSD profile at the time of the PON action potential. A s o u r c e - s i n k pair appeared at the surface with a potential reversal occurring very close to the
c
I .o.I
-1.0
EPL/GL border. Fig. 5b shows the CSD profile 2 - 3 ms later, on the slope of NI. A s o u r c e - s i n k pair appeared at the surface but with a potential reversal occurring deeper in the EPL.
Averaged evoked potentials and CSD pn?files for L 0 T.stimuli Like PON AEPs, the LOT AEPs were oscillating potentials, with peaks N1, Pl, N2, P2, etc.3. A brief LOT-mitra[ spike was seen at very short latencies: the slower PON spike was absent. The L O T NI wave was short in duration compared to the PON N 1 wave and there was no baseline shift. The profiles of field
N2o
1.0¸
loo~
SBNK
+
SOURCE
Fig. 6. LOT-evoked CSD profiles calculated using the second nearest neighbor method for peaks N1 (a), PI (b) and N2 (c). The mean of 4 experiments _+ S.E. is plotted vs bulbar depth, Comparison of profiles for NI and P1 shows that they tire reversed for almost every point, i.e. sources for N I become sinks for P1, while sinks for NI become sources for PI. CSD in arbitrary units.
230 pair (consisting of a sink in the deep EPI.. M(:L, I Pl and superficial G R L , and a source in the middle and superficial EPL) persisted during !~ter oscillations.
-1.0
o
NI + Pl
DISCUSSION SINK
10 -LO
Pl + N2
+
IOOju
SOURCE
O
1.0 L
-I,O
N1
+
PI
0
~+~_~....p~k, ,._..,4...~
.....
•
SINK
,.o loo3
-I.O
Pl + N2
k,, ]_ , ~ 0 , x~l ~.
SOURCE
.,-k •
,
~,~
Fig. 7. Results of combining the LOT-evoked CSD profiles of N1 with P1 and P1 with N2, for both calculations methods. From the second nearest neighbor data NI and PI (a) and PI and N2 are combined (b); and from the nearest neighbor data, N1 and P1 (c) and P1 and N2 (d) are combined. Minimizing the granule cell dipole field by combining the CSD profiles at the times the dipole is of reversed polarity shows that for LOT stimulation, almost flat source-sink profiles result, especially during the early part of the AEP (a and c). Mean of 4 experiments ___S.E. CSD in arbitrary units. nearest neighbor results there was a large sink in the d e e p half of the E P L and a matching source in the G R L and IPL during N1 and N2. This s o u r c e - s i n k pair was reversed during P1. The t u r n o v e r point between the source and sink occurred at the M C L for N1, P1 and N2. The L O T C S D profiles for N1 and P1 were combined to minimize the p r i m a r y s o u r c e - s i n k pair (Fig. 7), leaving a flat s o u r c e - s i n k profile with N1 + P1 combinations (Fig. 7a, c). H o w e v e r , when P1 and N2 were a d d e d , a secondary s o u r c e - s i n k pair appears at this later time (Fig. 7b, d). This secondary
PON-evoked source-sink pair Following low-level P O N or l . O T stimulation the olfactory bulb generates a dipole field of current with one pole in the deep half of the EPL and the other pole in the G R L . This dipole field regularly reverses s o u r c e - s i n k polarity as the A EP oscillates. N 1 corresponds to phase I l I of Rail and Shepherd2(L which they showed was generated by the granule cell population upon excitation by mitrai cells through the LOT. O u r results show that with L O T stimulation this granule cell activity forms the onlv detectable dipole field in the bulb during the early part of the A E P . However. with P O N stimulation a secondary dipole field is uncovered when the granule cell activity is minimized by summing over positive and negative peaks. This secondary dipole has the following properties. (1) The s o u r c e - s i n k pair generating this dipole field is more superficial than the granule cell s o u r c e - s i n k pair (the secondary sink occurs in the superficial E P E and in the GL. while the source is in the deep half of the E P L and M C L I: (2) the secondary dipole is more prominent during N1 and P1 than during P1 and N2: (3] it lasts b e y o n d the distribution ot arrival times of the action potentials in the PON terminals3: and (4) this secondary dipole does not appear during N1 -,- PI C S D profiles constructed from L O T - e v o k e d field pore ntiats. The only cell processes spanning the secondary s o u r c e - s i n k pair location that are oriented in such a way as to generate this dipole field are the mitral cell apical dendrites O t h e r cell types and processes in this region (from G L to M C L ) are not omented in the corresponding direction, would generate closed (not dipole) fields, are not numerous, or do not extend across the s o u r c e - s i n k location. The mitrat apical dendrite is capable of carrying sufficient intracellular current 20 to generate a m e a s u r a b l e dipole field resulting from a s o u r c e - s i n k current loop. If the rail ral apical dendrite is generating the secondary s o u r c e sink. then activation of the mitral apical dendrite must be either by excitation near the surface or b~ m-
231 hibition along the shaft near the cell body in order to produce the CSD profiles in Fig. 4. Since no synapses have been seen along the mitral shaft other than those just outside the glomerulus, it is unlikely that inhibition takes place along the shaft. Evidence exists of inhibitory input at the cell body v~,> due to granule cell dendrites. However, this input would be activated by L O T stimulation as well as PON stimulation. Therefore, deep inhibition along the mitral shaft does not appear to be the generator of the secondary PON source-sink pair. Conversely, superficial excitation of the mitral apical dendrite may occur in several ways. (1) The PON terminals may excite the apical dendrite within the glomerulus, and depending on the time course of the resulting EPSP, may account for the long-lasting (30-50 ms) secondary source-sink. Slow EPSPs have been described in cells of mammalian sympathetic ganglia le, but their properties include long synaptic delays (200-300 ms) and durations of many seconds. (2) Superficial excitation occurring only on PON stimulation could also be due to activation of an excitatory periglomerular cell population. Periglomerular cells make synaptic contact with mitral cells within the glomerulus (dendrodendritic) and just outside the glomerulus (axodendritic) on the apical shaft. The PSTH of periglomerular cells in response to a PON stimulus shows a monotonic increase in unit activity lasting up to 100 ms, long enough to last through several oscillations of the A E P 5. (3) A third possibility for producing a superficial excitation is the accumulation of a substance in the glomerulus 5. This process has been suggested to result in partially depolarized axon terminals, reduction of the output of transmitter from PON axons, causing attenuation of transmission through glomeruli on paired shock stimulation *-7. The attenuation appears to be a secondary effect of synaptic transmission in the glomerulus, since its decay rate is an order of magnitude slower than the decay rate of induced firing probability of glomerular units as seen in PSTHs. It appears then that the most likely generator of the PON secondary source-sink is either a slow EPSP due to the afferent volley, or excitatory PG synaptic activity onto the mitral cell. Recent work r with intracellular mitral cell recording in the isolated turtle olfactory bulb has shown the existence of a slow, excitatory potential produced only by PON
I
I
G.L
M
I[Icl
EP,
I G,
I"oNI
-1.0 .,"°x
•-..~ ,,.
0
i.'xl
}///%" iI [ \,
"V'\..J\// Y/
1.C
"
SINK
loo,u
+
SOURCE
Fig. 8. PON-evoked depth profiles using the results from one experiment (see Fig. 5 for the mean of 4 experiments), constructed at two times following the stimulus artifact: at time T1 (solid line), 2 ms before the start of the N1 peak and at time T2 (dashed line) at the start of the N1 peak. Thc source-sink pair at T1 is much more superficial than the pair at T2, which is similar in distribution to the combined N1 + P1 profiles with PON stimulation. CSD is in arbitrary units. stimulation and generated by synaptic activity at the glomerular level. This potential differed from an earlier and more superficially generated brief EPSP, presumably due to PON terminals. These findings and the locations of synaptic interactions on the apical dendrite suggest that the PON-generated EPSP should result in a dipole field located more superficially than a dipole due to periglomerular cell activity {activity which extends deep to the glomerulus on the apical shaft). In an attempt to verify this, the CSD profiles of Figs. 5 and 8 were examined. Fig. 5a, at time T1, should show early EPSP activity due to afferent input while Fig. 5b, at time T2, should show the activity of periglomerular cells becoming prominent, before the granule cell dipole becomes dominant. The CSD configuration at time T2 is very similar to that obtained when N1 + P1 or PI + N2 profiles are combined, i.e. when the granule cell dipole is minimized. However, Fig. 5b shows a definite current sink in the superficial third of the EPL, where PON afferents do not extend. The CSD profiles at these two times for one experiment are shown in Fig. 8. In this data there was little detectable PON action potential, unlike the other PON A E P sets. The reversal level of the secondary source-sink for this experiment at T1 was clearly more superficial than the reversal level 2 ms later at time 1"2. Although the results of CSD profiles made at very short latencies are not conclusive, they do suggest
232 that the P O N EPSP dipole and P G EPSP dipole are distinguishable with this m e t h o d , as well as by intracellular recording tT. It is also possible that a longlasting P O N EPSP could mask a dipole field generated by mitral apical dendrites in response to periglomerular cell activity, at least near the surface of the bulb. The extension of the dipole sink well into thc upper E P L during the initial slope of N1 and with N I + P1 or P1 + N2 CSD profiles indicates that some activity may be taking place in the u p p e r E P L that is not due to granule cell activity. In o r d e r for this secondary dipole to be due to the p e r i g l o m e r u l a r cell population, the p e r i g l o m e r u l a r cells must be excitatory to the mitral cells. There does not a p p e a r to be direct inhibition (excluding non-synaptic attenuation) taking place at the glomerular layer, although if a subpopulation of p e r i g l o m e r u l a r cells (e.g. the d o p a m i n e r gic cells) were inhibitory, their action may be overshadowed by excitation of mitral cells due to the PON or another larger subpopulation of periglomerular cells (e.g. the G A B A e r g i c cells). This latter hypothesis is compatible with G A D - c o n t a i n i n g periglomerular cells, since G A B A is thought to be responsible for presynaptic inhibition in the spinal cord by depolarizing primary afferent terminals l,~'. L O T - e v o k e d s o u r c e - s i n k pair
The secondary L O T - e v o k e d s o u r c e - s i n k pair revealed in Fig. 7b, d after minimization of the granule cell dipole is not as p r o m i n e n t as the P O N - e v o k e d s o u r c e - s i n k pair discussed above. The pair consists of a source in the middle E P L and a sink in the 1 P L - M C L - d e e p EPL. This pair appears to occur later during the A E P oscillation and therefore makes
REFERENCES 1 Curtis, D. R., Pre- and non-synaptic activities of GABA and related amino acids in the mammalian nervous system. In F. Fonnum (Ed.), Amino Acids as Chemical Transmitters, Plenum Press, New York, 1978, pp. 55-86. 2 Freeman, J. A. and Nicho!son, C., Experimental optimization of current source-density technique for anuran cerebellum, J. Neurophysiot., 38 (1975) 369-382. 3 Freeman, W. J., Spatial divergence and temporal dispersion in primary olfactory nerve of cat, J. Neurophysiok, 35 (1972) 733-744. 4 Freeman, W. J., Attenuation of transmission through glomeruli of olfactory bulb on paired shock stimulation, Brain Research, 65 (1974) 77-90. 5 Freeman, W. J., Relation of glomerular neuronal activity
more likely the involvement o[ siow-conductmg (~ slow-acting synaptic pathways. Atso, the tocatiot~ o! the source and sink m a k e it possible for the granuk. cell to generate this dipole. MitJal cells are excluded from this configuration by the ,:mal~,sis of Rail md Shepherd2~L The granule cell population is known to recetvc thc majority of centrifugal input to the bulb. These centrifugal fibers terminate on several parts of the granule cell. One mechanism by which granule cells might give the observed late response is, for e x a m p l e , exci~ tation of the anterior olfactory nucleus by L O T antidromic stimulation of en passant synapses_ resulting in o r t h o d r o m i c excitation to the basal dendrites of granule cells by ipsilateral collaterals of commissural axons. It is unlikely that some intrinsic inputs to the granule cell (e.g. mitral collaterals, short-axon cells) were responsible for this L O T - e ~ o k e d secondary dipole, because this pair was not observed on PON stimulation of mitral cells. Its failure to a p p e a r (if it is based on a neural loop through lhe forebrain) can bc ascribed to the greater temporal dispersion ol I~(Yt activity after P O N stimulation as c o m p a r e d with the L O T stimulation. Since the superficial source---sink pair did not a p p e a r on I,OT stimulation, there is no reason to suppose that centrifugal axons to the (;1~ were responsible for its a p p e a r a n c e following PON stimulation. ACKNOWI,EDGEMENT This study was s u p p o r t e d by G r a n t MH06686 from the National Institute of M e n t a l Health.
to glomerular transmission attenuauon, Brain Research~ ~5 (1974) 91-107. 6 Freeman. W. J.. A model for mutual excitanon in a neuron population in olfactory bulb. IEEE Trans. biomed. Eng., 21 (1974) 350-358. 7 Freeman. W I . Stability characteristics of positive feedback in a neural population, IEEE Trans. biomed. En~.. 21 (1974) 358-364. 8 Freeman. W. J.. Mass Action in ttze Nervous System, Academic Press. New York, 1975. 9 Getchell. T. V and Shepherd. G. M., Short-axon cells in the otfactor~ bulb: dendrodendritic synaptie interaction. J. Physiol. (Lond.) 251 (1975) 523-548. 10 Gonzalez-Estrada. M. T. and Freeman- W. J.. Effects ot carnosine on olfactory bulb EEG. evoked potentials and DC potentials, Brain Research, 202 f1980~ 373- 386.
233 ll Jahr, C. E. and Nicoll, R. A., Primary afferent depolarization in the in vitro frog olfactory bulb, J. Physiol. (Lond.), 318 (1981) 375-384. 12 Libet, S. B. and Tosaka, T., Slow inhibitory and excitatory postsynaptic responses in single cells of mammalian sympathetic ganglia, J. Neurophysiol., 32 (1969) 43-50. 13 Martinez, D. P., Current source density analysis of olfactory bulb evoked potentials, Ph. D. Thesis, University of California, Berkeley, 1982. 14 Mori. K. and Takagi, S. F., An intracellular study of dendrodendritic inhibitory synapses on mitral cells in the rabbit olfactory bulb, J. Physiol. (Lond.), 279 (1978) 569-588. 15 Nicholson, C., Generation and analysis of extracellular field potentials. In Electrophysiological Techniques, 1979 Short Course. Society for Neuroscience, Bethesda, Maryland, 1979. 16 Nishi, S., Minota, S. and Karczman, A. G., Primary afferent neurons. The ionic mechanism of GABA-mediated depolarization, Neuropharm., 13 (1974) 215-219. 17 Nowycky, M. C., Mori, K. and Shepherd, G. M., Blockade of synaptic inhibition reveals long-lasting synaptic excitation in isolated turtle olfactory bulb, J. Neurophysiol., 46 (1981) 649-658.
18 Phillips, C. G., Powell, T. P. S. and Shepherd, G. M., Responses of mitral cells to the stimulation of the lateral olfactory tract in the rabbit, J. Physiol. (Lond.), 168 (1963) 65-88. 19 Price, J. L. and Powell, T. P. S., The mitral and short axon cells of the olfactory bulb, J. Cell Sci., 7 (197(I) 631-651. 20 Rall, W. and Shepherd, G. M., Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in the olfactory bulb, J. Neurophysiol., 31 (1968) 884-915. 21 Shepherd, G. M., Synaptic organization of the mammalian olfactory bulb, Physiol. Rev., 52 (1972) 864-917. 22 Shepherd, G. M., The Synaptic Organization of the Brain, Oxford University Press, New York, 1974. 23 Sotelo, C., General features of the synaptic organization in the central nervous system. In R. Paoletti and A. N. Davison (Eds.), Advances in Experimental Medicine and Biology, Vol. 13, Chemistry and Brain Development, Plenum Press, New York, 1971, pp. 239-280. 24 Voronkov, G. S. and Gusel'nikova, K. G., Presynaptic inhibition in the frog olfactory bulb, Neurosci. Transl., 7 (1969) 775-777.
223
BRE 10266
Periglomerular Cell Action on Mitral Cells in Olfactory Bulb Shown by Current Source Density Analysis DIANE P. MARTINEZ and WALTER J. FREEMAN
Department (~lPhysiology-Anatomy, University of California, Berkeley, CA 94720 ( U.S.A.) (Accepted January 31st, 1984)
Key words: current source-sink density (CSD) - - electrophysiology - - CNS - - olfactory bulb - - periglomerular ceils - - rabbit
The sign of action of periglomerular (PG) cells on the apical dendrites of mitral cells in olfactory bulb glomeruli was investigated by constructing current source density (CSD) profiles from potentials evoked by primary olfactory nerve (PON) and lateral olfactory tract (LOT) stimulation. Evoked potentials were recorded and averaged from anesthetized rabbit simultaneously with a 1 x 16 array of electrodes positioned perpendicular to the bulbar surface. A one-dimensional CSD analysis with depth was made in the center of PON- and LOT-evoked potential activity. CSD was plotted vs depth for specific times during the average evoked potential (AEP): at the surface peaks of the first surface-negative wave (NI), the first surface-positive wave (PI), and the second surface-negative wave (N2). N l. Pl and N2 corresponded to excitation, dis-excitation (equivalent to inhibition), and dis-inhibition (re-excitation) of the granule cell population through mitral cell basal dendrites. The granule cells generated both PON and LOT oscillatory AEPs. When N1 and P1 profiles or PI and N2 profiles were combined, the source and sink due to granule cell activity were minimized and another source sink pair was revealed on PON but not LOT stimulation. PON-evoked N1 + P1 or PI + N2 profiles showed a secondary soucesink pair not present with LOT stimulation. The sink was located in the glomerular layer (GL) and outer plcxiform laycr (EPL) and the source in the inner EPL. It was concluded that long-lasting excitation of the mitral cells was taking place at the GL and GL/EPL border. This excitation was ascribed to concomitant PG cell activity, possibly in combination with prolonged monosynaptic PON excitation of the apical dendrites. The results support the occurrence of direct excitatory action of PG cells onto mitral cells.
INTRODUCTION
from the P O N and e x c i t a t o r y d e n d r o d e n d r i t i c input from mitral cells, and that they in turn m a k e d e n d r o -
The olfactory bulb is the site of the first synapse in
dendritic
and
axodendritic
contacts
with
mitral
the sensory system for o d o r . T h e large b u l b a r projec-
cells 22. It is a g r e e d that p e r i g l o m e r u l a r activity sup-
tion n e u r o n s , the mitral cells ( d e f i n e d h e r e to include
presses mitral cell r e s p o n s e to a test P O N shock that
tufted cells), receive input directly f r o m r e c e p t o r s
follows a c o n d i t i o n i n g P O N shock 4,5,L~,21. H o w e v e r ,
through the p r i m a r y o l f a c t o r y n e r v e ( P O N ) and send
the nature and m e c h a n i s m of this effect are at issue.
their axons in the lateral o l f a c t o r y tract. T h e r e are
O n e view is that p e r i g l o m e r u l a r synapses cause di-
two main classes of i n t e r n e u r o n s , granule cells and
rect synaptic inhibition in the mitral apical den-
p e r i g l o m e r u l a r cells. G r a n u l e cells are k n o w n to in-
drites'). A n o t h e r view24 is that the suppression of the
hibit mitral cells at their basal d e n d r i t e s ; they g e n e r -
mitral responses manifests ' p r e s y n a p t i c inhibition'.
ate the electrical current fields m a n i f e s t e d in the electroencephalographic
(EEG)
Freeman~ s h o w e d that the suppression occurs not
oscillatory p a t t e r n s of
only for the mitral cell r e s p o n s e s to P O N input but
bulbar r e s p o n s e to o d o r s and to electrical stimuli.
also p e r i g l o m e r u l a r r e s p o n s e s to P O N input. H e pro-
P e r i g l o m e r u l a r cells act o n t o mitral cell apical dendrites: the nature of their action is in dispute.
vided e v i d e n c e that p e r i g l o m e r u l a r cells excite each o t h e r 4,7 and a t t e n u a t e (i.e. partially block or finhib-
T h e r e is a g r e e m e n t in tile literature that periglo-
it') transmission in the g l o m e r u l i by releasing a sub-
m e r u l a r cells receive e x c i t a t o r y a x o d e n d r i t i c input
stance, possibly K + (ref. 11), into the e x t r a c e l l u l a r
Corre~pondence: W. J. Freeman, Department of Physiology-Anatomy, University of California, Berkeley, CA 947211. U.S.A. 110(16-89~3/84503.ql0 © 1984 Elsevier Science Publishers B.V.
224 compartment of the glomeruli. By network analysis of periglomerular, mitral and granule cell function. Freeman concluded that periglomerular cells cause direct synaptic excitation in the mitral apical dendrites~,m. In the present work we studied the action of periglomerular cells on mitral cells by measurement of buibar evoked potentials and calculation of current sources and sinks in the vicinity of the periglomerular/mitral synapses and the adjacent apical shafts. We foresaw three possible outcomes. (a) If the periglomerular cells were directly excitatory to mitral apical tufts, a current sink would be observed in the gtomerular layer (GL) and a source in the subjacent external plexiform layer (EPL), and they would persist monotonically for the duration of the evoked periglomerular activity (several tens of millisecondsS). (b) If the periglomerular cells were directly inhibitory, a current source would develop in the G L with a sink in the subjacent EPL. (c) If periglomerular action were to 'clamp' mitral cells (e.g. possibly by an increase in C1 conductance in the apical tufts), then no localized source-sink pattern would be seen. We uncovered the source-sink pair predicted as in (a) by averaging sequential alternating positive and negative peaks of oscillatory average evoked potentials ( A E P ) in order to minimize the masking effect of the powerful granule cell dipole field. We then sought to exclude forms of neural action in the bulb other than periglomerular excitation of mitral cells, that might have accounted for the newly observed source-sink pair. By its location we showed that it could be generated only by the mitral apical dendrites. By comparing responses to orthodromic and antidromic volleys we localized the active synapses to the apical tufts within the glomerular sink. By the same means and by measurement of its onset latency we excluded the role of recurrent activation of centrifugal pathways following PON stimulation, and minimized the role of possible slow EPSP component involving excitation of mitral cells by PON axon terminals. By measuring its time course we showed that it was correlated with the rate of periglomerular cell activity of single-shock PON stimulation and not with the prolonged changes in glomerular transmission that attenuate subsequent throughput 4. However, the methods we used did not distinguish which of the subtypes of periglomerular cells delivered the synap-
tic input to the mitral apical dcndritt,,s. MATERIAl. AND METHODS
Surgical techniques Adult New Zeahmd white rabbits (,2.5--3.5 kgt were anesthetized with sodium pentobarbilal (35-45 mg/kg). The dorsolateral surfacc of the otfactor~ bulb and L O T was exposed b 3 c~rbital exentcration and removal of the overlying bone and dura. "[he brain surface was covered with warm "lyrode's soh,tion car mineral oil. Evoked potentials used for CSD analysis with depth were recorded with a i x 16 array (perpendicular to the bulbar surface) of (t.0(H 5 in. diameter (3~ urn) enamelled stainless-steel wires (average interelectrode distance; 69 _+ 9/~m), aligned side-by-side in a ribbon. The array was cut at an acute angle (40 ~) and the array beveled to a knife-like edge to alloys smooth penetration of the bulb. Each interelectrode distance was measured under a m~croscope. A I × < array (placed on the surface l o t 300-urn stainlesssteel wire was used to record ew~ked potentials from the surface of the bulb in order to locate the focus of evoked potential activity. All potentials were measured with respect to a reference stainless-steel electrode located at the edge of the surmcat field (20-30 cm). This lay outside the evoked field of activity but sufficiently within the fields of bulbar E E G activntv to give a significant degree of c o m m o n mode rejection. The PON stimulating electrode was a 1 x 5 array of polished 300 um diameter stainless-steel wires. placed on the PON of the anterior bulb surface m a line perpendicular to the PON axons. The stimulus pulse was passed between the pmr of electrodes within the 1 x 5 array that evoked the lowest threshold response. The L O T was stimulated with a bipolar stainless-steel electrode placed tan the L O T surface. For PON evoked potentials the recording site was located by activating the P O N using low intensity ~5 V, 0.01 ms duration) stimuli at 2 Hz. The maximum amplitude response (focus) was found by searching with the 1 × 5 surface recording array. The I,()'F stimulating electrode was then !x)sitioned so that the focus of its bulbar response field overlapped that or the PON focus. The L O T stimulus rate was 7 ltz. With stimulating electrodes in place and the focus ot the PON and L O T responses located, the 1 > 16 re-
225 cording array was advanced perpendicularly through the epicenter of the focus.
Collection and storage of A EP data The ew)ked potentials were amplified with 16 fixed-gain (10,It00) preamplifiers (0.1 H z - 3 kHz band width), multiplexed and digitized sequentially at 12-us intervals across the array. A digitizing interval of 1-ms was found to suffice for the observed rates of change in evoked potential, with correction for the sequential sampling procedure (_+6-_+90 ,us) by linear interpolation. The responses to 200-400 stimuli were summed and stored in an Interdata 716 computer. AEPs selected for later CSD analysis displayed oscillatory waveforms that were symmetrical waveshapes across the zero isopotential (the layer of potential reversal). In the case of PON stimuli they revealed a baseline shift of the oscillation which also reversed polarity at the same depth as the oscillatory component ~4. The baseline shift was a slow monotonic component (a surface negative, deep positive dipole) of the PON evoked potential which was absent with L O T stimulation and with high PON stimulus intensity or bulbar damage. When this shift was present the A E P oscillated about a baseline shifted above or below the prestimulus baseline. Presence of a baseline shift in PON AEPs was inferred for reasons given previously 4.5 to indicate activity of the periglomerular cell population. At the completion of each A E P data gathering session, with the 1 x 16 array still in place, the array and amplifiers were calibrated as a unit by applying to the tissue an 80 Hz sine wave current (100/~A) with electrodes near the reference and ground electrodes. These artificial signals were recorded, averaged and stored on disk for later use as amplitude correction factors for the AEPs obtained from this experiment. A similar procedure was used to evaluate the resistivity of bulbar tissue as a function of depth through the layers~L The depth of the MCL and the widths of the EPL, GL and PON along the recording sites were determined from Nissl-stained frozen sections. The 16 electrode positions with respect to the bulbar laminae were determined by noting at what point along the array the polarity of the A E P reversed (zero isopotential) during the first peak of the oscillation. The
A E P had a characteristic pattern at each bulbar lamina 14,1s2° with the zero isopotential for PON stimulation occurring 100#m above the MCL during NI and at the MCL thereafter along the axis of symmetry s. Since interelectrode distances were fixed, and the thicknesses of the bulbar layers were relatively constant over animals in the lateral area of the bulb used for this study, the positions in particular lamina of all the electrodes could be determined for each experiment by this method. When results from several experiments were combined, CSD values from corresponding laminae were aligned by matching the zero isopotential points from each experiment.
CSD calculations The analysis was performed in one dimension (depth) for bulbar potentials after considering the following conditions. (1) The location of the recording site was on a relatively flat bulbar area, and the activity was restricted to that area, allowing the effects of tissue curvature to be neglected. (2) Recording along the axis of symmetry of the focus of a synchronously activated population of bulbar neurons which extended through the depth of the focus~ ensured restriction of extracellular current normal to the bulbar surface along that axis. (3) Electrode penetration and placement were normal to the bulbar surface at the epicenter. (4) Change in potential with time across recording sites due to preparation instability was minimized by using near-simultaneous recording with the 1 x 16 array (12us time interval between sequential electrode samples). In order to extract the signal from the background E E G in the lightly anesthetized rabbit, the digitized waveform was averaged for n = 200-40(I consecutive stimuli: this reduced background noise by a factor of n °-5. AEPs were calculated from the summed amplitudes, corrected for differences in channel gains and plotted in time series (V vs t) and depth profiles (V vs z) at selected latencies. The following formula was used to obtain CSD values: {V[z+n.A(z,t)]-2V(z,t)+ V[z-n.A(z,t)]} CSD(z,t) = 0 (n'Az)2 For depth z at time t, CSD (z,t) is an approximation of the second spatial derivative, which is proportional (by the inverse of the factor (?, tissue resistivity in the z direction) to the rate of change in current density. The resistivity of bulbar tissue was measured
226
A B
/
///
......
C
surlace
PON
SOURCE.
I
/ ,t GI
X
•
5
/ PG
/
/ ./
. . . . . . .
/ //
EEPL
MCL IPL
GRL
1looju
\ \
.........
25
msec
Fig. 1. A: sixteen averaged evoked potentials (n = 200) recorded from the rabbit olfactory bulb in response to stimulation of the primary olfactory nerve (PON), using a 1 x 16 electrode array. The recording site for each electrode within the bulbar laminae is indicated by lines drawn to the diagram at left. illustrating the bulbar lamina including the primary olfactory nerve (PON) at the surface the olfactory glomerulus (dashed circle) with a nearby perigtomerular cell (PG), the mitral cell (M), apical dendrite, cell body, basal dendrites and axon and a granule cell (G) interneuron. Each AEP is digitized at l-ms intervals with a prestimulus baseline of 30 ms followed by the shock artifact and the oscillatory response. PON AEPs also oscillate about a baseline shift (negative above the turn over and positive below). Negatively down. Abbreviations: GL. glomerular layer; EPL, external plexiform layer; MCL. mitral celt layer; IPL, internal plexiform layer; G R L granule cell layer. B: current source density (CSD) values vs time calculated for nearest neighbor electrodes using the digitized PON AEPs recorded with the l x 16 electrode array placed across the bulbar lamina. Amplitudes are in arbitrary units, with resistivity across the lamina assumed constant (see text). C: CSD calculations using the second nearest neighbor electrodes. ( t h e a v e r a g e w a s 343 _ 65 o h m ' c m 2 / c m )
across the
Since interelectrode distances were not equal, n'dz
layers and f o u n d to h a v e insignificant e f f e c t on C S D
w a s r e p l a c e d by t h e a p p r o p r i a t e v a l u e . B e t t e r signal-
profiles 13, so w a s a s s u m e d to be c o n s t a n t in t h e s e cal-
t o - n o i s e ratio w a s o b t a i n e d b y u s i n g n = 2 ( s e c o n d
culations. V ( z , t ) w a s the p o t e n t i a l a m p l i t u d e at p o i n t
n e a r e s t n e i g h b o r ) , but at t h e e x p e n s e o f spatial r e s o -
z and t i m e t w i t h r e s p e c t to a distant r e f e r e n c e . T h e
lution 2J5. B o t h nearest n e i g h b o r a n d s e c o n d n e a r e s t
p o t e n t i a l s at the a d j a c e n t e l e c t r o d e s A z ~ m superfi-
neighbor CSD maps were calculated and compared
cial
to e x a m i n e the c o m p r o m i s e b e t w e e n r e s o l u t i o n and
and
deep
V[z+n'A(z,t)]
to
z
were
V[z-n-A(z,t)]
respectively, where n -
and
I or n = 2.
noise.
227 The CSD value was calculated for each sampling
o
time of the A E P , starting with the prestimulus interval of approximately 10 ms (used to define the baseline level of zero potential and zero CSD), for each of the 14 (n = l) or 12 (n = 2) electrodes. The CSD waveforms (CSD vs t) and depth profiles (CSD vs z) were plotted. Negative CSD values indicated net inward flow of current or a sink: positive CSD values indicated a source. As a check on the validity of the results, a sum of the sources was c o m p a r e d to a sum of the sinks. If the ratio of sources plus sinks to the sum of their absolute values for a depth profile was greater than 0.25, the
-10
N1
0
I0 i b
GRL.
I~,I~i L
L
EPL
I GL
I PONI
-1 0
P1
data were discarded. RESUI_TS
Averaged evoked potentials and CSD profiles jor PON stimuli A representative set of digitized PON averaged evoked potentials (n = 200) are shown in Fig. 1A. The waveforms were initially surface negative/deep positive (peak N1), then surface positive/deep negative (peak PI), and so forth for peaks N2, P2, etc. The PON A E P s showed a baseline shift of the oscillatory potential in a negative direction on the surface and in a positive direction deep, with the zero isopotential for both the monotonic shift and the oscillatory c o m p o n e n t located just above the mitral cell layer. The depth profiles for both P O N and L O T A E P s showed a near symmetry about the reversal point (i.e. surface N1 = - d e e p N l), validating the elimination of the effects of bulbar curvature on the potential amplitudes. Fig. 2 shows the potential vs depth (normalized to _+1, where 1 equals maximum amplitude of the oscillation at any point) of the PON A E P s for peaks N1 and P 1 of the same experiment. Profiles at N 1 and P2 were similar to each other but opposite to that at P1. The potential reversal occurred midway between electrodes 8 and 9 at N 1 and N2 and slight d e e p e r at P2. These profiles were used to determine the array position with respect to the anatomical turnover point (located a p p r o x i m a t e l y 100~tm above the M C L at the peak of N1 (see ref. 8)) for each set of A E P s and thereby locate the 16 recording sites in specific laminae. The sets of nearest neighbor and second nearest
1.0 ~oo~
SINK
+
SOURCE
Fig. 2. Profiles of potential amplitude (solid triangles) vs bulbar depth for PON AEPs, showing the potential polarity and reversal site for the first two peaks of the oscillation: N1 (a) and PI (b). Amplitudes were scaled by dividing each of the 16 values for a peak by the spatial maximum. Location of bulbar lamina is indicated along the depth (x-axis) (superficial layers at right). Electrode 1 (left) is deepest. Current source density depth profile (solid lines) calculated for PON averaged evoked potentials using the second nearest neighbor method, showing the location of sources (downward deflection) and sinks (upward deflection). Dashed lines are the results of including resistivity variations across the lamina in the CSD calculations, a: profile at the peak of N1, showing superficial sinks and deep sources; b: map at the peak of P1 (when granule cell dipole field is of opposite polarity from its orientation at N1 ), with superficial sources and deep sinks. CSD amplitude scaled to extremumat +1 o r - l . neighbor PON CSDs vs time calculated for the A E P s in Fig. 1A are shown in Fig. 1B, C. Although the data from the second nearest neighbor calculation did not span the bulbar thickness that the nearest neighbor calculation did and was less localized, the variance was reduced. Fig. 2 shows the CSD vs depth profiles for N1 and P1 using the second nearest neighbor data from Fig. IC. W h e n comparing A E P and CSD profiles, the maximum negative and positive amplitudes of the field potential at the times N 1 and P2 were seen not to correspond in depth with the locations of maximal sources and sinks. In Fig. 3 the results of 9 PON experiments are averaged to form one CSD profile for each peak using nearest neighbor calculations. The distribution of
228 became prominent and an accompanying source ap-
-10
peared (Fig. 4). This secondary dipole consisted of a sink in the G L and superficial third of the EPL and a source in the deep half of the EPI~ and in the MCLI
NI
The magnitude of the secondary dipole dccrcased from the time of N1 + P1 to the time of Pt .÷ N2, but the location of its source and sink remained the same. This indicated that this secondary s o u r c e - s i n k pair
i
o.L
l l tL
IoL I.o.I
was slowly varying, which was a requirement for using NI + P1 or P1 + N2 combinations to reveal it. In Fig. 5, CSD profiles constructed at the time
-I,0
J
I
PI
7
NI + PI
o
I
1.0
SINK
+
1.o
-loj c
-I0
N2
SOURCE
oL I
0 1o SINK
1.0
I00#
+
-'O c
SOURCE
Fig. 3. PON-evoked CSD profile for peaks N1 (a), P1 (b) and N2 (c) using the combined results from 9 animals. The nearest neighbor CSD calculation (mean + S.E.) is plotted vs bulbar depth for the 3 different peak times. CSD in arbitrary units, sources and sinks in the bulb (mean + 1 S.E.) for PON stimulation is shown at N1, P1 and N2. During N1 (Fig. 3a), a sink occurred in the deep half of the
X~V
d
,o - 10~
lO0~,l
SINK
SOURCE
EPL with a source in the superficial G R L and IPL. The peak amplitudes of the source and sink were located slightly deeper at P1 and N2 and were reversed during P1 (Fig. 3b, c). The turnover occurred approximately 5 0 - 6 0 # m above the M C L during N1 and at the IPL/MC1 border during P1 and N2. In both the nearest neighbor and second nearest neighbor calculations there appeared to be an additional sink located more superficially than the major sink at both N1 and P1 peaks and perhaps N2. W h e n the CSD profiles of N1 and P1 or of P1 and N2 were combined, the primary s o u r c e - s i n k pair seen in Figs. 2 and 3 disappeared. The more superficial sink
~O Fig. 4 Effects of minimizing the source-sink pair due to the granule cell dipole field on the PON-evoked CSD profiles. By adding together PON CSD profiles for N1 and P1 from each of 9 experiments, the granule cell source-sink pair was combined with its mirror image and minimized. The same procedure was applied to the profiles for P1 and N2. Additional sources and sinks are revealed with both second nearest neighbor and nearest neighbor calculation, a: second nearest neighbor combined profile for N1 and PI: b: second nearest neighbor combined profile for P1 and N2: c: nearest neighbor combined profile for N1 and PI: d: nearest nmghbor combined profile for P1 and N2. Mean of 9 experiments ± S.E. CSD in arbitrary units.
229 a
potential amplitude vs depth for L O T A E P peaks
-10
N1, PI and N2 showed that the sign of the potential at various depths in the bulb matched that of the PON A E P graphs for N1, P1 and N2. The set of nearest neighbor and second nearest neighbor L O T CSDs vs time and depth were calculated for the same experiments and array positions as
1.0
those for the PON CSDs. The extent of s o u r c e - s i n k density along the z axis for the L O T stimulus was more localized than the activity due to PON stimulus,
b
-I0
i.e. there was less activity near the bulbar surface
t
with LOT stimulation. The averaged second nearest neighbor CSD profiles for 4 LOT experiments are shown in Fig. 6. In both nearest neighbor and second a I0
-1.0
SINK
l'O0;u
÷
N1
SOURCE
Fig. 5. PON-evoked CSD profiles constructed for two different times shortly after the stimulus, a: at time T1, 2-3 ms before the start of the first peak (N1) and during postsynaptie activity due to the PON action potential; b: at time T2, on the rising slope of the first peak (N 1), when the sources and sinks due to bulbar cell activity become prominent. Second nearest neighbor calculation: mean of 7 experiments _+S.E. CSD in arbitrary units.
0
10
b
o.L
I'fiJL
EpL
I
oL
o.L
liltlL
E.L
] °~
I Po.
t
-1.0
(T1) of the PON action potential (before the rise of N 1) and also at the time (T2) of the initial slope of N 1
P1
are shown. Seven PON experiments were averaged and the mean ± S.E. plotted vs depth (two PON experiments were eliminated because of interference from the shock artifact at short latencies). Fig. 5a shows the CSD profile at the time of the PON action potential. A s o u r c e - s i n k pair appeared at the surface with a potential reversal occurring very close to the
c
I .o.I
-1.0
EPL/GL border. Fig. 5b shows the CSD profile 2 - 3 ms later, on the slope of NI. A s o u r c e - s i n k pair appeared at the surface but with a potential reversal occurring deeper in the EPL.
Averaged evoked potentials and CSD pn?files for L 0 T.stimuli Like PON AEPs, the LOT AEPs were oscillating potentials, with peaks N1, Pl, N2, P2, etc.3. A brief LOT-mitra[ spike was seen at very short latencies: the slower PON spike was absent. The L O T NI wave was short in duration compared to the PON N 1 wave and there was no baseline shift. The profiles of field
N2o
1.0¸
loo~
SBNK
+
SOURCE
Fig. 6. LOT-evoked CSD profiles calculated using the second nearest neighbor method for peaks N1 (a), PI (b) and N2 (c). The mean of 4 experiments _+ S.E. is plotted vs bulbar depth, Comparison of profiles for NI and P1 shows that they tire reversed for almost every point, i.e. sources for N I become sinks for P1, while sinks for NI become sources for PI. CSD in arbitrary units.
230 pair (consisting of a sink in the deep EPI.. M(:L, I Pl and superficial G R L , and a source in the middle and superficial EPL) persisted during !~ter oscillations.
-1.0
o
NI + Pl
DISCUSSION SINK
10 -LO
Pl + N2
+
IOOju
SOURCE
O
1.0 L
-I,O
N1
+
PI
0
~+~_~....p~k, ,._..,4...~
.....
•
SINK
,.o loo3
-I.O
Pl + N2
k,, ]_ , ~ 0 , x~l ~.
SOURCE
.,-k •
,
~,~
Fig. 7. Results of combining the LOT-evoked CSD profiles of N1 with P1 and P1 with N2, for both calculations methods. From the second nearest neighbor data NI and PI (a) and PI and N2 are combined (b); and from the nearest neighbor data, N1 and P1 (c) and P1 and N2 (d) are combined. Minimizing the granule cell dipole field by combining the CSD profiles at the times the dipole is of reversed polarity shows that for LOT stimulation, almost flat source-sink profiles result, especially during the early part of the AEP (a and c). Mean of 4 experiments ___S.E. CSD in arbitrary units. nearest neighbor results there was a large sink in the d e e p half of the E P L and a matching source in the G R L and IPL during N1 and N2. This s o u r c e - s i n k pair was reversed during P1. The t u r n o v e r point between the source and sink occurred at the M C L for N1, P1 and N2. The L O T C S D profiles for N1 and P1 were combined to minimize the p r i m a r y s o u r c e - s i n k pair (Fig. 7), leaving a flat s o u r c e - s i n k profile with N1 + P1 combinations (Fig. 7a, c). H o w e v e r , when P1 and N2 were a d d e d , a secondary s o u r c e - s i n k pair appears at this later time (Fig. 7b, d). This secondary
PON-evoked source-sink pair Following low-level P O N or l . O T stimulation the olfactory bulb generates a dipole field of current with one pole in the deep half of the EPL and the other pole in the G R L . This dipole field regularly reverses s o u r c e - s i n k polarity as the A EP oscillates. N 1 corresponds to phase I l I of Rail and Shepherd2(L which they showed was generated by the granule cell population upon excitation by mitrai cells through the LOT. O u r results show that with L O T stimulation this granule cell activity forms the onlv detectable dipole field in the bulb during the early part of the A E P . However. with P O N stimulation a secondary dipole field is uncovered when the granule cell activity is minimized by summing over positive and negative peaks. This secondary dipole has the following properties. (1) The s o u r c e - s i n k pair generating this dipole field is more superficial than the granule cell s o u r c e - s i n k pair (the secondary sink occurs in the superficial E P E and in the GL. while the source is in the deep half of the E P L and M C L I: (2) the secondary dipole is more prominent during N1 and P1 than during P1 and N2: (3] it lasts b e y o n d the distribution ot arrival times of the action potentials in the PON terminals3: and (4) this secondary dipole does not appear during N1 -,- PI C S D profiles constructed from L O T - e v o k e d field pore ntiats. The only cell processes spanning the secondary s o u r c e - s i n k pair location that are oriented in such a way as to generate this dipole field are the mitral cell apical dendrites O t h e r cell types and processes in this region (from G L to M C L ) are not omented in the corresponding direction, would generate closed (not dipole) fields, are not numerous, or do not extend across the s o u r c e - s i n k location. The mitrat apical dendrite is capable of carrying sufficient intracellular current 20 to generate a m e a s u r a b l e dipole field resulting from a s o u r c e - s i n k current loop. If the rail ral apical dendrite is generating the secondary s o u r c e sink. then activation of the mitral apical dendrite must be either by excitation near the surface or b~ m-
231 hibition along the shaft near the cell body in order to produce the CSD profiles in Fig. 4. Since no synapses have been seen along the mitral shaft other than those just outside the glomerulus, it is unlikely that inhibition takes place along the shaft. Evidence exists of inhibitory input at the cell body v~,> due to granule cell dendrites. However, this input would be activated by L O T stimulation as well as PON stimulation. Therefore, deep inhibition along the mitral shaft does not appear to be the generator of the secondary PON source-sink pair. Conversely, superficial excitation of the mitral apical dendrite may occur in several ways. (1) The PON terminals may excite the apical dendrite within the glomerulus, and depending on the time course of the resulting EPSP, may account for the long-lasting (30-50 ms) secondary source-sink. Slow EPSPs have been described in cells of mammalian sympathetic ganglia le, but their properties include long synaptic delays (200-300 ms) and durations of many seconds. (2) Superficial excitation occurring only on PON stimulation could also be due to activation of an excitatory periglomerular cell population. Periglomerular cells make synaptic contact with mitral cells within the glomerulus (dendrodendritic) and just outside the glomerulus (axodendritic) on the apical shaft. The PSTH of periglomerular cells in response to a PON stimulus shows a monotonic increase in unit activity lasting up to 100 ms, long enough to last through several oscillations of the A E P 5. (3) A third possibility for producing a superficial excitation is the accumulation of a substance in the glomerulus 5. This process has been suggested to result in partially depolarized axon terminals, reduction of the output of transmitter from PON axons, causing attenuation of transmission through glomeruli on paired shock stimulation *-7. The attenuation appears to be a secondary effect of synaptic transmission in the glomerulus, since its decay rate is an order of magnitude slower than the decay rate of induced firing probability of glomerular units as seen in PSTHs. It appears then that the most likely generator of the PON secondary source-sink is either a slow EPSP due to the afferent volley, or excitatory PG synaptic activity onto the mitral cell. Recent work r with intracellular mitral cell recording in the isolated turtle olfactory bulb has shown the existence of a slow, excitatory potential produced only by PON
I
I
G.L
M
I[Icl
EP,
I G,
I"oNI
-1.0 .,"°x
•-..~ ,,.
0
i.'xl
}///%" iI [ \,
"V'\..J\// Y/
1.C
"
SINK
loo,u
+
SOURCE
Fig. 8. PON-evoked depth profiles using the results from one experiment (see Fig. 5 for the mean of 4 experiments), constructed at two times following the stimulus artifact: at time T1 (solid line), 2 ms before the start of the N1 peak and at time T2 (dashed line) at the start of the N1 peak. Thc source-sink pair at T1 is much more superficial than the pair at T2, which is similar in distribution to the combined N1 + P1 profiles with PON stimulation. CSD is in arbitrary units. stimulation and generated by synaptic activity at the glomerular level. This potential differed from an earlier and more superficially generated brief EPSP, presumably due to PON terminals. These findings and the locations of synaptic interactions on the apical dendrite suggest that the PON-generated EPSP should result in a dipole field located more superficially than a dipole due to periglomerular cell activity {activity which extends deep to the glomerulus on the apical shaft). In an attempt to verify this, the CSD profiles of Figs. 5 and 8 were examined. Fig. 5a, at time T1, should show early EPSP activity due to afferent input while Fig. 5b, at time T2, should show the activity of periglomerular cells becoming prominent, before the granule cell dipole becomes dominant. The CSD configuration at time T2 is very similar to that obtained when N1 + P1 or PI + N2 profiles are combined, i.e. when the granule cell dipole is minimized. However, Fig. 5b shows a definite current sink in the superficial third of the EPL, where PON afferents do not extend. The CSD profiles at these two times for one experiment are shown in Fig. 8. In this data there was little detectable PON action potential, unlike the other PON A E P sets. The reversal level of the secondary source-sink for this experiment at T1 was clearly more superficial than the reversal level 2 ms later at time 1"2. Although the results of CSD profiles made at very short latencies are not conclusive, they do suggest
232 that the P O N EPSP dipole and P G EPSP dipole are distinguishable with this m e t h o d , as well as by intracellular recording tT. It is also possible that a longlasting P O N EPSP could mask a dipole field generated by mitral apical dendrites in response to periglomerular cell activity, at least near the surface of the bulb. The extension of the dipole sink well into thc upper E P L during the initial slope of N1 and with N I + P1 or P1 + N2 CSD profiles indicates that some activity may be taking place in the u p p e r E P L that is not due to granule cell activity. In o r d e r for this secondary dipole to be due to the p e r i g l o m e r u l a r cell population, the p e r i g l o m e r u l a r cells must be excitatory to the mitral cells. There does not a p p e a r to be direct inhibition (excluding non-synaptic attenuation) taking place at the glomerular layer, although if a subpopulation of p e r i g l o m e r u l a r cells (e.g. the d o p a m i n e r gic cells) were inhibitory, their action may be overshadowed by excitation of mitral cells due to the PON or another larger subpopulation of periglomerular cells (e.g. the G A B A e r g i c cells). This latter hypothesis is compatible with G A D - c o n t a i n i n g periglomerular cells, since G A B A is thought to be responsible for presynaptic inhibition in the spinal cord by depolarizing primary afferent terminals l,~'. L O T - e v o k e d s o u r c e - s i n k pair
The secondary L O T - e v o k e d s o u r c e - s i n k pair revealed in Fig. 7b, d after minimization of the granule cell dipole is not as p r o m i n e n t as the P O N - e v o k e d s o u r c e - s i n k pair discussed above. The pair consists of a source in the middle E P L and a sink in the 1 P L - M C L - d e e p EPL. This pair appears to occur later during the A E P oscillation and therefore makes
REFERENCES 1 Curtis, D. R., Pre- and non-synaptic activities of GABA and related amino acids in the mammalian nervous system. In F. Fonnum (Ed.), Amino Acids as Chemical Transmitters, Plenum Press, New York, 1978, pp. 55-86. 2 Freeman, J. A. and Nicho!son, C., Experimental optimization of current source-density technique for anuran cerebellum, J. Neurophysiot., 38 (1975) 369-382. 3 Freeman, W. J., Spatial divergence and temporal dispersion in primary olfactory nerve of cat, J. Neurophysiok, 35 (1972) 733-744. 4 Freeman, W. J., Attenuation of transmission through glomeruli of olfactory bulb on paired shock stimulation, Brain Research, 65 (1974) 77-90. 5 Freeman, W. J., Relation of glomerular neuronal activity
more likely the involvement o[ siow-conductmg (~ slow-acting synaptic pathways. Atso, the tocatiot~ o! the source and sink m a k e it possible for the granuk. cell to generate this dipole. MitJal cells are excluded from this configuration by the ,:mal~,sis of Rail md Shepherd2~L The granule cell population is known to recetvc thc majority of centrifugal input to the bulb. These centrifugal fibers terminate on several parts of the granule cell. One mechanism by which granule cells might give the observed late response is, for e x a m p l e , exci~ tation of the anterior olfactory nucleus by L O T antidromic stimulation of en passant synapses_ resulting in o r t h o d r o m i c excitation to the basal dendrites of granule cells by ipsilateral collaterals of commissural axons. It is unlikely that some intrinsic inputs to the granule cell (e.g. mitral collaterals, short-axon cells) were responsible for this L O T - e ~ o k e d secondary dipole, because this pair was not observed on PON stimulation of mitral cells. Its failure to a p p e a r (if it is based on a neural loop through lhe forebrain) can bc ascribed to the greater temporal dispersion ol I~(Yt activity after P O N stimulation as c o m p a r e d with the L O T stimulation. Since the superficial source---sink pair did not a p p e a r on I,OT stimulation, there is no reason to suppose that centrifugal axons to the (;1~ were responsible for its a p p e a r a n c e following PON stimulation. ACKNOWI,EDGEMENT This study was s u p p o r t e d by G r a n t MH06686 from the National Institute of M e n t a l Health.
to glomerular transmission attenuauon, Brain Research~ ~5 (1974) 91-107. 6 Freeman. W. J.. A model for mutual excitanon in a neuron population in olfactory bulb. IEEE Trans. biomed. Eng., 21 (1974) 350-358. 7 Freeman. W I . Stability characteristics of positive feedback in a neural population, IEEE Trans. biomed. En~.. 21 (1974) 358-364. 8 Freeman. W. J.. Mass Action in ttze Nervous System, Academic Press. New York, 1975. 9 Getchell. T. V and Shepherd. G. M., Short-axon cells in the otfactor~ bulb: dendrodendritic synaptie interaction. J. Physiol. (Lond.) 251 (1975) 523-548. 10 Gonzalez-Estrada. M. T. and Freeman- W. J.. Effects ot carnosine on olfactory bulb EEG. evoked potentials and DC potentials, Brain Research, 202 f1980~ 373- 386.
233 ll Jahr, C. E. and Nicoll, R. A., Primary afferent depolarization in the in vitro frog olfactory bulb, J. Physiol. (Lond.), 318 (1981) 375-384. 12 Libet, S. B. and Tosaka, T., Slow inhibitory and excitatory postsynaptic responses in single cells of mammalian sympathetic ganglia, J. Neurophysiol., 32 (1969) 43-50. 13 Martinez, D. P., Current source density analysis of olfactory bulb evoked potentials, Ph. D. Thesis, University of California, Berkeley, 1982. 14 Mori. K. and Takagi, S. F., An intracellular study of dendrodendritic inhibitory synapses on mitral cells in the rabbit olfactory bulb, J. Physiol. (Lond.), 279 (1978) 569-588. 15 Nicholson, C., Generation and analysis of extracellular field potentials. In Electrophysiological Techniques, 1979 Short Course. Society for Neuroscience, Bethesda, Maryland, 1979. 16 Nishi, S., Minota, S. and Karczman, A. G., Primary afferent neurons. The ionic mechanism of GABA-mediated depolarization, Neuropharm., 13 (1974) 215-219. 17 Nowycky, M. C., Mori, K. and Shepherd, G. M., Blockade of synaptic inhibition reveals long-lasting synaptic excitation in isolated turtle olfactory bulb, J. Neurophysiol., 46 (1981) 649-658.
18 Phillips, C. G., Powell, T. P. S. and Shepherd, G. M., Responses of mitral cells to the stimulation of the lateral olfactory tract in the rabbit, J. Physiol. (Lond.), 168 (1963) 65-88. 19 Price, J. L. and Powell, T. P. S., The mitral and short axon cells of the olfactory bulb, J. Cell Sci., 7 (197(I) 631-651. 20 Rall, W. and Shepherd, G. M., Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in the olfactory bulb, J. Neurophysiol., 31 (1968) 884-915. 21 Shepherd, G. M., Synaptic organization of the mammalian olfactory bulb, Physiol. Rev., 52 (1972) 864-917. 22 Shepherd, G. M., The Synaptic Organization of the Brain, Oxford University Press, New York, 1974. 23 Sotelo, C., General features of the synaptic organization in the central nervous system. In R. Paoletti and A. N. Davison (Eds.), Advances in Experimental Medicine and Biology, Vol. 13, Chemistry and Brain Development, Plenum Press, New York, 1971, pp. 239-280. 24 Voronkov, G. S. and Gusel'nikova, K. G., Presynaptic inhibition in the frog olfactory bulb, Neurosci. Transl., 7 (1969) 775-777.