0306-4522/91 $3.00+ 0.00

Neuroscience Vol. 42, No. 1, pp. 125-135, 1991 Printed in Great Britain

Pergamon Press plc 0 1991IBRO

APICAL DENDRITIC DEPOLARIZATIONS AND FIELD I~ERACTIONS EVOKED BY STIPULATION OF AFFERENT INPUTS TO RAT ~IPPOCAMPAL CA1 PYRAMIDAL CELLS R. W. TURNJB* and T. L. RICHARDSON~$ *Department of Anatomy, University of Ottawa, Ottawa, Ontario, Canada K1I-f 8M5 $School of Kinesiology, Simon Fraser University, Bumaby, British Columbia, Canada WA lS6 AIrstract-The relationship between orthodromic extracellular field potentials and intradendritic depolarizations in apical dendrites of CA1 pyramidal neurons was investigated using the in vitro slice preparation of rat hippocampus. Orthodromic synaptic field potentials evoked by stimulation of afferent inputs in stratum radiatum or stratum oriens were used to measure extracellular voltage gradients generated over the pyramidal cell axis. Extracellular gradients were of opposite polarity over the region of pyramidal cell apical dendrites in stratum radiatum. The stratum rad~atum~vok~ gradient was negative towards the apical dendrites and the stratum oriens-evoked gradient negative towards the cell body layer, with gradients reaching values of up to 50 mV/~ over the apical dendritic axis. Intradendritic recordings obtained > 150ym from stratum pyramidale directly measured the subthreshold apical dendritic excitatory postsynaptic potentials evoked by stratum radiatum or stratum oriens stimulation. These ground-referenced recordings were tben compared to the transmembrane potential calculated by subtraction of the corresponding extradendritic field potential. Both stratum radiatum and stratum oriens stimulation evoked graded excitatory postsynaptic potentials that could be recorded in apical dendritic impalements up to 265pm from stratum pyramidale. The calculated transmembrane potential of the stratum radiatum-evoked excitatory postsynaptic potential had a significantly greater rate of rise, peak amplitude, and rate of decay than that of the ground-referenced excitatory postsynaptic potential. In contrast, the rates of rise and decay of the transmembrane potential of the stratum oriens-evoked excitatory postsynaptic potential were reduced with respect to the ground-referenced recording. The peak amplitude of the stratum oriens-evoked transmembrane potential, however, varied according to the polarity of the corresponding extradendritic population spike response recorded in stratum radiatum. These data reveal that synaptic activation of either basal or apical dendrites of CA1 pyra~dal cells evokes a declaration that can be recorded over a subst~ti~ region of the apical dendritic arbor. Furthermore, extradendritic field potentials evoked by stim~ation of these inputs produce opposite effects on the transmembrane potential of apical dendrites. The magnitude of the accompanying extracellular voltage gradients suggest that these shifts in transmembrane potential reflect ephaptic interactions at the apical dendritic level of pyramidal cells.

Synchronous

neuronal

discharge

within

highly

laminar structures such as the hippocampus can generate substantial extracellular field potentials. In the live animal, these can be recorded in association with rhythmical slow activity {theta rhythm) and epileptifo~ discharge,“,“,‘*,~ or in response to !Several studies suggest electrical stimulation.*~16~z4~3s~s3 that field potentials generated under each of these conditions can intluence pyramidal cell discharge interactions Sa,21,29,33.34,3&43,49,53 through ephaptic whereby current flow associated with extracellular voltage gradients can alter the excitability of cells within the population (see Ref. 15 for review). In the hippocampus, ephaptic interactions have been shown to contribute to the synchronization of pyramidal cell

$To whom correspondence should be addressed. Abbreoiurions: CSD, current source-density; EPSP, excitatory postsynaptic potential; SO, stratum oriens; SR, stratum radiatum; TMP, transmembrane potential.

discharge. 2’Z%ss+W~L*s,rs Most of these studies fo_ cused upon the relationship between the cell body

population spike and the transmembrane potential (TMP) recorded at the somatic level Exceptions to this are reports that an extracellular antidro~~ population spike invading the region of stratum radiatum can also affect the apical dendritic TMP.4i*Sf Comparatively little attention has been paid to possible ephaptic interactions induced by orthodromic synaptic potentials. This may be particularly important at the dendritic level, where both synaptic depolarizations and voltage-dependent membrane conductances contribute to the integration of synapactivation tic input. 3,6,19,25,36,37,*7,*8,51 Since syn&ronous of pyramidal cells generates extracellular field potentials over the entire cell axis, it is possible that these fields also exert ephaptic influence on dendritic membrane. Orth~ro~c field potentials in dendritic regions of pyramidal cells have been described in the in vitro slice preparation using laminar profile and current 125

I26

R. W. TURNER and T. L. RICHARDSON

source-density (CSD) analysis.27,30,37.48Synaptic depolarization of pyramidal cell basal or apical dendrites generates a negative-going extracellular potentiai and current sink in the region of activated dendrites, and a positive potentiai and current source in the opposing dendritic arborization.31a37 Thus, field potential analyses suggest that direct synaptic depolarization of either pyramidal cell basal or apical dendrites evokes an eventual depolarization of apical dendritic membrane. Furthermore, the opposite polarity of extracellular potentials generated in stratum radiatum by either synaptic input suggests that field effects at the apical dendritic level may differ. As successful intradendritic recordings have been reit should be pass_ ported by several groups, 9X25+47,51,52 ible to test these hypotheses by monito~ng intra- and extradend~tic potentials of pyramidal cell apical dendrites during orthodromic depolarizations. The present study uses extracellular field potential analysis and intradendritic recordings at the apical dendritic level of in vitro hippocampal CA1 pyramidal cells to compare (1) the apical dendritic membrane potential response to simulation of afferent inputs in stratum radiatum (SR) and stratum oriens (SO), and (2) the influence of SR- and SOevoked field potentials on the apical dendritic TMP. Some of this work has been presented in abstract fom_28,3L32

EXPERIMENTAL PROCEDURES

Evoked extracellular field potentials were recorded in the CA1 region of the mid-dorsal hippocampus of male Wistar rats (150-250 g; Charles River) using the in vitro slice preparation. Details on the methods of slice preparation and maintenan~ have been reported elsewhere.M Briefly, 4OOpm slices were cut 10” oblique to the longitudinal axis of hippocampi dissected out in cold (4°C) oxygenated (95% 0,/S% CO,) Ringer solution consisting of (in mM): 124 NaCl, 3 KCl, 0.75 KH,PO,, 1.6 CaCl,, 1.2 MgSO,, 24 NaHCO,, and 10 D-@UCOSe. After transfer to an “interface” recording chamber, slices were perfused with the same solution maintained at 345°C and superfused with warmed, humi~fi~ 95% 0,/5% CO* gas. Slices were permitted at least 45 min to recover from this procedure before recordings were carried out. Extracellular recordings were made with glass microelectrodes (1.5mm o.d.) broken back under microscopic observation (l-2pm tips) and filled with 1 M NaCl. Final resistance ranged from 1 to 10 MQ. Intracellular recording electrodes were pulled from glass capillary tubing (1.5 mm o.d.) and filled with 1 M potassium acetate to yield a resistance most often between 50 and 100 MS2. Bipolar stimulating electrodes were constructed as twisted 6.5nrn insulated Nichrome wires connected to stimulus isolation units (Medical Systems; DS2) for delivery of square wave voltage pulses of I-SOV intensity. Synaptic depolarization of pyramidal cell basal dendrites was evoked by a stimulating electrode placed in the mid-stratum oriens (80-100 pm), while apical dendritic depolarization was evoked by a stimulating electrode in the mid-distal stratum radiatum (-250 pm; distance referenced to the boundaries of stratum pyramidale). Evoked potentials were recorded on one or two channels of a WPI Dual trace electrometer and photographed on a storage oscilloscope or led to a PDP 1l/23 computer for storage and subsequent off-line analysis.

All signals were low-pass filtered with a corner frequency of 10 kHz. Laminar profiles of extracellular population potentials over the pyramidal cell axis were constructed using a single recording electrode positioned sequentiaily at 25-60 nm intervals perpendicular to stratum pyramidale. The location of each recording site was measured with reference to the boundary between stratum pyramidale and stratum radiaturn. Potentials at each location were averaged over 3-5 single sweeps at depths between 100 and 200 nm deep to the surface of the slice. A second electrode was positioned in stratum pyramidale to monitor the cell body ~p~ation spike response to verify that the amplitude of evoked responses did not change while recording the laminar profile. Intracellular recordings were obtained from 44 apical dendrites of CA1 pyramidal cells at distances 150-265 pm from stratum pyramidale, corresponding to recordings from the mid-distal region of the dendritic arborization. Dendritic ~palements were achieved through use of a Burleigh Inchworm piezoelectric stepping device and brief bursts of capacitive feedback through the recording electrode. Recordings considered acceptable for data analysis had resting membrane potentials > -60 mV and input resistance > 18 MR (range of 1844 MD). In all cases, resting membrane potential and input resistance was stable throughout the period of recording (up to 3 h) with no evidence of spontaneous action potential discharge. Calculation of dendritic transmembrane potential Extracellular field potentials and intradendritic recordings were referenced to the Ag-AgCl ground of the recording chamber, and are designated as “ground-referenced” recordings. The dendritic TMP was calculated by subtracting the extracellular potential recorded in the immediate vicinity of the dendritic implement. This was accomplished by first recording intradendritic potentials over a range of stimulus intensities (3-5 sweeps/intensity). The electrode was then retracted 10-20 pm to withdraw from the impalement (indicated by a loss of resting potential) and repositioned to the original depth of recording (Burleigh Inchworm digital micrometer scale). Extradendritic field potentials were evoked over the same range of intensities and collected for averaging (3-5 sw~ps/intensity). Subtraction of extra- from intradendritic potentials yielded the TMP. In six cases, measurements were taken from enlarged photographs of stored oscilloscope traces recorded before and after withdrawing from the dendritic recording. Average values in this study are expressed as a mean with the standard error and number of values in parenthesis (S.E. = X; n = x). The P values shown for statistical significance were calculated using the paired Student’s t-test. RESULTS

Intracellular recordings from dendrites of CA1 pyramidal cells were obtained in the mid-distal region of stratum radiatum at distances > 150 pm from the border of stratum pyramidale and stratum radiatum. Electrophysiological identification of these impalements as corresponding to pyramidal cell apical dendrites was based upon the evoked characteristics of orthodromic dendritic potentials reported in Histological ve~fication previous investigations. 22+25,5i of the pyramidal cell origin of these recordings has also been obtained through the intradendritic injection of horseradish peroxidase (Turner R. W., Richardson T. L. and Meyers D. E. R., unpublished observation). The average input resistance was 27.0 Mfi (S.E. = 1.99; n = 14) while the average value

127

Dendritic field effects in hippocampus

St. Radiatum

St. Pyr.

B.

St. Oriens

15

_,

XI

+-...: 2

D.G

10 mV 2msec

Fig. 1. Orthodromic potentials recorded in apical dendrites of pyramidal cells in response to stimulation of afferent inputs in stratum radiatum (St. Radiatum) (A, C, Intradend.) or stratum oriens (St. Oriens) (B, D). Population field potentials at the level of pyramidal cell somata were recorded simultaneously with an extracellular recording electrode positioned in stratum pyramidale (A-D; St. Pyr.). Extradendritic field potentials in the immediate vicinity of intradendritic recordings were obtained after withdrawing from the impalement and repositioning the electrode to the original depth (A-D; Extradend.). (A), (B) Several superimposed sweeps near threshold for intradendritic spike discharge demonstrate that synaptic activation of pyramidal cell apical (A) or basal (B) dendrites evokes an excitatory postsynaptic potential (EPSP) and a single fast spike that can be recorded in apical dendrites in mid-distal stratum radiatum. (C), (D) Several superimposed averaged records (five sweeps) in another dendritic recording show that both synaptic potentials and associated field potentials are graded in amplitude with increments of stimulus intensity. Arrows (A-D) denote the approximate duration of dendritic population spike responses upon apical dendritic synaptic field potentials. 30*48 Note that the onset of intradendritic spikes aligns with the onset of dendritic population spikes. of resting potential obtained after withdrawing to the extracellular space was 64.9 mV (SE. = 0.85; n = 21). In the present study, most attention was focused upon synaptic potentials evoked by stimulation of afferent inputs in stratum radiatum or stratum oriens. A complete comparison or orthodromic spike discharge will be presented elsewhere. Orthodromic intradendritic potentials Orthodromic potentials recorded in pyramidal cell apical dendrites are shown in Fig. 1 following SR (A, C) and SO (B, D) stimulation. As previously reported, SR stimulation under normal conditions evokes an excitatory postsynaptic potential (EPSP) and a single all-or-none “fast” spike per stimu1us2s~5’ (Fig. 1A). Figure 1B illustrates that synaptic depolarization of pyramidal cell basal dendrites by SO stimulation also evoked an EPSP and a single fast spike that could be recorded in apical dendritic

impalements. Both stimulus sites evoked EPSPs that were graded in amplitude and rate of rise with increasing stimulus intensities (Fig. lC, D). However, the characteristics of EPSPs evoked by either stimulus differed on the basis of several criteria. Using stimulus intensities just threshold for spike discharge for comparative purposes, EPSPs generated by basal dendritic activation (SO stimulation) had a significantly lower amplitude, rate of rise, and rate of decay than those evoked by SR stimulation (Table 1; P < 0.001). In contrast, intradendritic spikes evoked by either pathway were not significantly different in terms of amplitude or half-width (Turner R. W., Richardson T. L. and Meyers D. E. R., unpublished observation). Orthodromic extracellular potentials Population cell somata

potentials in the region of pyramidal were monitored during intradendt-itic

R. W. TURNERand T. L. RICHARDSON

128 recordings

by an extraeeliular

recording

electrode

positioned

in stratum pyramidale (Fig. 1). Extradendritic field potentials were recorded immediately after withdrawing from the dendrite in preparation for calculation of the TMP (Fig. 1; see Experimental Procedures). Although the characteristics of these evoked field potentials have been described in previous investigations,2*4~30~37*4* there are certain features particularly relevant to the analysis of dendritie field effects. For the case of SR stimulation, synaptic depolarization of apical dendrites is recorded as a negativegoing ~pulation EPSP in mid-stratum ~diatum (Fig. 1A). In contrast, the extradend~tic synaptic potential evoked by SO stimulation at this location is recorded as a positive-going potentia13’ (Fig. 13). Both fields were smoothly graded with increasing stimulus intensity (Fig. lC, D), with a close correspondence to the onset and time course of intradendritic EPSPs (Fig. lA-D). In stratum pyramidale, synaptic depolarization of apical or basal dendrites is recorded as a graded positive-going potential (Fig. IA-D) that gives rise to the pyramidal cell body population spike.4 At stimulus intensities suflicient to generate a cell body ~pulation spike, SR and SO stimulations evoke biphasic positive-negative potentials superimposed upon the apical dendritic EPSP (Fig. IA-D; approximate duration denoted by arrows). These represent a tetrodotoxin-sensitive population spike response that retrogradely conducts into the middendritic region from a site of origin in stratum A.

St. Radiatum

B.

St. Oriens

pyramidale or the proximal stratum radiatum.“*48 While both components of the population spike are obvious following SO stimulation, the negativegoing component of the SR-evoked response is often difficult to discern when superimposed on the negativity of the population EPSP’O (Fig. lA, C). A comparison of intra- and extradendritic spike potentials reveals that the onset of intradendritic spikes evoked at threshold intensity most often aligns temporally with the onset (positivity) of corresponding dendritic population spike responses (Fig. lA-D). Extra~ellu~ar voltage g~~d~~ts over the pyramidal cell axis In order to determine the possible role of field effects in shaping these depolarizations, we examined the nature and magnitude of extracellular voltage gradients generated in response to stimulation of either pathway. Voltage gradients were derived from laminar profiles of evoked potentials recorded at 60pm intervals along the somatodendritic axis of the cell population, providing an indirect estimate of current flow parallel to the pyramidal cell axis. An example of three ~pre~ntative potentials from laminar profiles of SR- and SO-evoked responses is shown in Fig. 2A, B, respectively. Voltage gradients were examined at three specific latencies associated with the most prominent features of dendritic field potentials (Fig. 2A, B; dashed lines, l-3). These were: (1) the negative peak of the SR-evoked apical dendritic EPSP in mid-distal stratum radiatum; (2) the

C.

.-.

S,.Rad,of (1) o---o st.orimr(2) v--v %Oriens (3) 2

y (1)

I (2)

I (3)

-200

-100

DISTANCE

0

100

200

300

FROM CELL BODY LAYER (pm)

Fig. 2. Extracellular voltage gradients over the pyramidal axis evoked by orthodromic depolarization of apical or basal dendritic arborizations of the pyramidal cell population. (A), (B) Representative extracellular responses taken from a laminar profile of orthodromic field potentials evoked by SR stim~ation (A) or SO stimulation (B) at varying distances from the border of stratum pyramidale and stratum radiatum (“0 rm”). Positive values correspond to recordings in the apicai dendritic region, and negative values to recordings in the basal dendritic region. Dashed lines (l-3) denote three latencies chosen to examine vokage gradients associated with generation of prominent extracehular potentials in the mid-distal stratum radiatum: (I) peak negativity of the apical dendritic synaptic depolarization evoked by SR stimulation; (2) peak negativity of the synaptic depolarization evoked in the basal dendrites by SO stimulation; (3) peak of the SO-evoked apical dendritic population spike negativity recorded in mid-stratum radiatum. (C) Plots of the voltage deviation (from baseline) of orthodromic potentials over the pyramidal cell axis at each of the three latencies (A, B, l-3; legend at top of C). Recordings in stratum pyramidale are represented at -3Ohm and the apical dendritic region is shown to the right. The voltage gradients evoked by SR and SO stinmlation are of opposite polarity and orientation through the region of pyramidal cell apical dendrites in stratum radiatum. The SO-evoked population spike gives rise to two voltage gradients oriented towards the mid-stratum radiatum.

129

Dendritic field effects in hippocampus

(S.E. = 2.7; n = 12) at threshold intensities for intradendritic spike discharge, while the distal gradient ranged from 6 to 3 1 mV/mm, average 17 mV/mm (S.E. = 2.9; n = 8) on extracellular laminar profiles. Therefore, the polarity of measured gradients indicates that current flow along the axis of apical dendrites was towards the mid-stratum radiatum at Latencies 1 and 3, but away from this region at Latency 2.

negative peak of the SO-evoked EPSP in the midregion of the basal dendrites; and (3) the negative peak of the SO-evoked dendritic population spike in mid-stratum radiatum. The distribution of extracellular voltage (deviation from baseline) along the somatodendritic axis of the cell population at each of these latencies is plotted in Fig. 2C. At Latency 1, the SR-evoked positive potential over the basal dendrites and negative potential over the apical dendrites gave rise to a voltage gradient sloping towards stratum radiatum (Fig. 2C; closed circles). This gradient ranged in magnitude from 11 to 50mV/mm with an average value of 26 mV/mm (S.E. = 3.1; n = 16) when evoked at an intensity just subthreshold for spike discharge during intradendritic recordings. At Latency 2, an opposite polarity in evoked dendritic field potentials gave rise to a voltage gradient of somewhat smaller magnitude oriented in the direction of the cell body layer (Fig. 2C; open circles). This gradient ranged from 4 to 39 mV/mm with an average of 18mV/mm (S.E. = 3.4; n = 12) at threshold intensities for intradendritic spike discharge. At Latency 3, the voltage gradient sloped from both the stratum pyramidale and distal stratum radiatum towards the relatively negative potential of the dendritic population spike in the mid-apical dendritic region (Fig. 2C; triangles). The proximal dendritic gradient ranged from 7.5 to 38 mV/mm with an average value of 17mV/mm

The relationship between field potentials and the intradendritic response was examined by comparing the ground-referenced potential to the TMP calculated by subtracting the extracellular potential recorded in the immediate vicinity of the dendritic recording29*39,40 (see Experimental Procedures). Stratum radiatum stimulation. Calculation of the TMP during SR stimulation revealed that the TMP was greater than that of the ground-referenced potential throughout the synaptic response (Fig. 3A, B). These changes were reflected as a statistically significant increase in the rate of rise, peak amplitude, and rate of decay of the TMP over corresponding groundreferenced records for stimulus intensities set to threshold for intradendritic spike discharge (Table 1). It is worth noting that spike discharge was not evoked from any particular component of the TMP waveform.

A.

1N

Relationship between orthodromic field potentials and the apical dendritic transmembrane potential

B.

8V

‘5msec

C.

.-.

GRREF.

O-0

TYP

v--o

FlELo

10 STIMULUS

INTENSITY

(V)

STIMULUS

12

14

INTENSITY

16 (V)

16

10 STIMULUS

12

14

INTENSITY

16

16

(V)

Fig. 3. SR-evoked dendritic field potentials alter the TMP of pyramidal cell apical dendrites. (A), (B) Ground-referenced intradendritic (1) and extradendritic (2) synaptic potentials and the calculated dendritic TMP (l-2) for an apical dendritic recording 230 pm from stratum pyramidale. Potentials were evoked at a low stimulus intensity (A; 8 V) and an intensity near threshold for spike discharge (B; 17 V). Sweeps 1 and l-2 are shown superimposed to aid comparison of ground-referenced and TMP recordings. Note that the TMP is depolarized with respect to the ground-referenced EPSP over the duration of the synaptic response (A, B, l-2). Intra- and extradendritic potentials are averaged records of five sweeps and extradendritic potentials were recorded after withdrawing from the impalement (see Experimental Procedures). (CHE) Plots of the absolute values of EPSP peak amplitude (C), rate of rise (D) and rate of decay of intra- and extradendritic synaptic potentials for this impalement over a range of stimulus intensities (E; legend at the top of C). The corresponding values for the TMP are greater than those of ground-referenced potentials at all stimulus intensities. Changes in the TMP reflect the amplitude of the extradendritic field potential (compare A and B).

R.

130

A.

W.

TURNER

B.

15~

STliiLUS

RICHARDSON

and T. L.

INTEiilTY

(V)

STliiLUS

INTENSITY

,~xw

STIMULUS

(V)

INTENSITY

(V)

Fig. 4. SO-evoked field potentials in stratum radiatum alter the TMP of pyramidal cell apical dendrites. (A), (B) Ground-referenced intradendritic (1) and extradendritic (2) synaptic potentials and the calculated dendritic TMP (l-2) for an apical dendritic recording 200 pm from stratum pyramidale. Potentials were evoked at a low stimulus intensity (A; 15 V) and an intensity near threshold for spike discharge (B; 38 V, spike superimposed). Sweeps 1 and 1-2 are shown superimposed to aid comparison of the groundreferenced and TMP recordings. Note the complex effects induced by the SO-evoked synaptic and population spike components on the dendritic TMP (A, B, l-2). Intra- and extradendritic potentials are averaged records of three sweeps, with a single sweep record of intradendritic spike discharge superimposed in (B). Extradendritic potentials were recorded after withdrawing from the impalement (see Experimental Procedures). (C)-(E) Plots of the absolute value of potentials at the latencies of EPSP peak amplitude (C, measured at the latency of open arrows in A, B). rate of rise (D) and rate of decay of synaptic potentials for this impalement over a range of stimulus intensities (E; legend at the top of C). The TMP rate of rise and rate of decay is reduced from that of ground-referenced recordings at all stimulus intensities. The peak of the dendritic population spike (C; open arrows in A, B) increased the peak amplitude of the TMP directly with stimulus intensity.

Plots of the absolute value of each of these parameters are shown over a range of stimulus intensities in Fig. 3C-E for the recording of Fig. 3A, B. These plots illustrate that the qualitative differences between TMP and ground-referenced responses were not dependent on stimulus intensity, increasing in conjunction with extracellular field potentials as stimulus intensity was increased. No consistent change in the percentage cont~bution of field potentials to the TMP was observed for any of these three parameters with increasing stimulus intensity. Stratum oriens stimulation. The relationship between SO-evoked field potentials and the TMP was complicated by the dominant biphasic population

spike response superimposed on the extradendritic field potential at higher stimulus intensities. Intradendritic recordings illustrate that both the dendritic population EPSP and population spike affected TMP measurements (Fig. 4A, B). Thus, for intensities near threshold for spike discharge, the rate of rise of the TMP was reduced with respect to the groundreferenced EPSP (Fig. 4B; Table 1). The rate of decay of the TMP was also reduced in 11 of 12 recordings, although this difference was not statistically significant (Table 1). In contrast, the peak amplitude of the TMP depended upon the polarity of the extradendritic population spike (Fig. 4A, B; open arrows). In the case shown in Fig. 4A, B, this potential reached

Table 1. The influence of field potentials upon the transmembrane potential of stratum radiatumand stratum oriens-evoked apical dendritic synaptic potentials at intensities near threshold for intradendritic spike discharge Site of stimulation

Recording configuration

Stratum Radiatum

Ground-~feren~d Transmembrane

Peak amplitude (mV) 21.9 (0.93) 26.2 (l.ls)***

Stratum Oriens

Ground-referenced Transmembrane

6.0 (1.24)7 7.4 (1.35)*

Rate of rise (mV/ms)

Rate of decay (mV/ms)

10.2 (0.66) 12.2 (0.83)***

2.7 (0.25) 3.2 (0.30)***

2.3 (0.51) 1.5 (oso)**

1.0 (0.18) 0.9 (0.20)

Mean values (S.E.) are shown for ground-referenced and TMP records. Significance values correspond to paired comparisons of ground-referenced and TMP records (*P < 0.01: **P < 0.005; ***P < 0.001). Stratum radiatum, n = 16; stratum oriens, n = 12 (tn = 9).

Dendritic field effects in hippocampus

a negative value at the higher stimulus intensity (Fig. 4B), resulting in a peak depolarization of the TMP. This result was statistically significant for the nine out of 12 cases in which the dendritic population spike was negative in polarity at intensities near threshold for spike discharge (Table 1). In contrast, the TMP amplitude was significantly decreased for three of 12 cases in which the peak of the population spike did not attain a negative value (P < 0.01; n = 3). The TMP calculated over a range of SO stimulus intensities is shown in Fig. 4C-E. These plots illustrate graphically the dependence of TMP peak amplitude on the dendritic population spike (Fig. 4C), and a decrease in TMP rate of rise and decay at all stimulus intensities (Fig. 4D, E). As found for the case of SR stimulation, the percentage contribition of field potentials to each of these parameters did not change consistently with the intensity of stimulation.

A

Stim. Site - St. Radiatum

131 DISCUSSION

Apical dendritic synaptic depolarizations

The present study demonstrates that activation of aflerent inputs in stratum radiatum and stratum oriens of the CA1 region results in a depolarization of pyramids cell apical dendritic membrane. The intradend~tic response to direct synaptic activation of apical dendrites by SR stimulation was a smoothly graded EPSP of up to 30mV amplitude and discharge of a single fast action potential. These characteristics are consistent with previous reports of intradendritic potentials of CA1 pyramidal cells.22,25,5’However, the recording of an EPSP in apical dendrites following SO stimulation has not been previously described. This result is particularly striking in that dendritic recordings were ‘obtained in locations > 150 pm from stratum pyramidale, a region corresponding to the mid-distal region of apical dendrites.=

B

- St. Oriens

St.Oriens _____ St.Pyr.

-____

St.Redietum -

1 bpulation

Cell

(+ ) Fig. 5. Schematic diagram of extracellular current flow underlying putative ephaptic interactions associated with synaptic potentials evoked by SR (A) and SO (B) stimulation. The population of pyramidal cells generating field potentials is shown schematically by the large cell (Population) with soma in stratum pyramidale (St. Pyr.) and basal dendritic (St. Oriens) and apical dendritic (St. Radiatum) extensions. Smaller cells of similar construction represent single cells within the population (Cell) affected by extracellular current flow. For simplicity, synaptic potentials within single cells are not considered, and synaptic current sinks (termination site of shaded afferent fibres) and sources in the population are localized to restricted regions of dendritic membrane. Synchronous synaptic activation evokes a current sink in dendritic regions, giving rise to intracellular current flow along the axis of the population (large solid arrows) that sources in the region of the opposing dendritic arborization. The return current path is through the extraceilular space (open arrows) along an extracellular voltage gradient (+ , -) generated by extracellular field potentials associated with current sources and sinks, respectively. A small proportion of current enters indi~du~ cells within the population (small solid arrows) to exit in the region of the ~pulation synaptic current sink. Passive current Row during entry of this current into the single cell leads to a TMP hy~~la~~tion, and current exit a TMP de~la~~tion, evident following subtraction of extra- from intradendritic potentials. The opposite polarities of extracellular voltage gradients generated by SR and SO stimulation thus give rise to differential effects on the TMP of apical dendritic membrane.

132

R. W. TURNERand T. L. RICHARDSON

Several lines of evidence support the conclusion that the SO-evoked EPSP recorded in apical dendrites originates as a synaptic depolarization of basal dendritic membrane. First, current source-density (CSD) analysis demonstrates that SO stimulation generates a current sink restricted to the region of the basal dendrites and a current source extending over the apical dendritic arborization.24vM Thus, the SO-evoked EPSP does not result from spread of stimulus current and direct activation of afferent inputs in stratum radiatum. Second, this pattern of CSD suggests that current entering the basal dendrites spreads electrotonically past the cell bodies of pyramidal neurons and into the apical dendritic arborization. In this regard, the onset and duration of the SO-evoked intradendritic EPSP coincided with that of the positive-going extradendritic potential recorded in stratum radiatum (Fig. 1). Third, EPSPs evoked by SO stimulation had a lower peak amplitude and slower rate of rise than SR-evoked EPSPs that result from direct depolarization of apical dendritic membrane. This finding is consistent with an electrotonic propagation of EPSPs evoked by SO stimulation. A similar explanation might account for the slower rate of decay of the SO-evoked EPSP, although the difference in this parameter will also reflect the activation of feed-forward or recurrent inhibitory synaptic inputs invoked by stimulation of either pathway.‘*5~7~22~45 Therefore, it is reasonable to conclude that the EPSP evoked by SO stimulation was the result of direct synaptic depolarization of the basal dendrites. These data thus indicate that an EPSP generated in basal dendritic membrane conducts over a substantial portion of the pyramidal cell apical dendritic arborization. This result is not entirely unexpected given the short electrotonic length of pyramidal cells’2*“,46and the long time course of an EPSP. It is important to note that conduction of an EPSP might also be enhanced by the activation of voltage-dependent processes, including the slow prepotential proposed to occur in dendritic regions.3,6*26*48 Conduction of an EPSP over the opposing dendritic arborization has important implications for spatial and temporal processing of synaptic inputs upon the pyramidal cell. For instance, these data provide intracellular evidence as to the membrane declaration underlying long-term potentiation evoked by associative stimulation of SR and SO afferent inputs.23 Identification of the intradendritic results of such paired stimulation awaits further analysis. The relationship between extracellular voltage gradients and dendritic transmembrane potential Extracellular current flow along the somatodendritic axis of a cell is a basic requirement for ephaptic alterations of neuronal excitability.” In the present study, the relative magnitude and direction of extracellular current flow during orthodromic depolarizations was estimated from the spatial distri-

bution of extracellular voltage over the pyramidal cell axis. Intradendritic recordings and measurements of the TMP suggest that these gradients are sufficient to induce dendritic ephaptic interactions during synchronous activation of the cell pop~ation. A model outlining the somatodendritic current flow expected to be generated by SR and SO stimulation is shown in Fig. 5. The large diameter cylinders represent a population of neurons undergoing synchronous synaptic activation of either the apical (A; St. Radiatum) or basal (B; St. Oriens) dendrites, while the smaller cylinders represent single neurons under the influence of field potentials generated by the activity of the population. The large solid arrows represent the electrotonic conduction of synaptic current throughout the cell and the open arrows indicate the return path for this current in the extracellular space. The extracellular currents flow down the voltage gradient generated by the negative potential at the site of synaptic current (Fig. 5A,B ; -) and the relatively positive potential at sites of leakage current throughout the cell axis (Fig. 5A, B; +). The directions of current flow induced by SR and SO stimulation are thus opposite in polarity. The small solid arrows indicate that a fraction of the total current flowing down these voltage gradients will travel within the intracellular space of individual neurons. The passage of current across the membranes of these cells will alter their TMP. Note, however, that the influence on the TMP will vary along the cell axis. A passive outward flow of current will be associated with a depolarization of the TMP whereas the passive inward flow will be associated with h~r~la~~tion. The TMP of the single cell will thus be depolarized at the synaptic site and hyperpolarized at the opposite pole of the cell, with a gradual transition of ephaptic influence between these two extremes. Therefore, the fields generated by SR stimulation should, on theoretical grounds, lead to an ephaptic depolarization of the apical dendrite of pyramidal cells. Extracellular fields associated with SO stimulation should lead to hyperpolarization of the apical dendrite. Our results support these predictions, as the amplitude of the calculated TMP of apical dendrites was increased by SR-evoked field potentials and decreased by SO-evoked potentials with respect to ground-referenced recordings. Considering the positive polarity of extracellular synaptic potentials in stratum pyramidale (Fig. l), TMP calculations shouid also uncover a membrane h~~l~~tion at this location. In fact, this result can be found in previously published records29*” and simulations.‘* An important consideration, however, is the degree to which a measured change in TMP reflects a true effect upon voltage-dependent ion channels and membrane excitability. At the somatic level, the cell body population spike has been shown to produce ephaptic interactions by increasing the probability for spike generation? Unfo~unately, similar tests

Dendritic field effects in hippocampus cannot be carried out at the dendritic level as other work indicates that fast Na+ spikes recorded in dendrites can be initiated in distant regions of the cell structure,47.48 perhaps resulting in part from ephaptic interactions in other locations. Thus, a direct demonstration of a change in dendritic excitability with respect to TMP measurements cannot be strictly satisfied for our recordings. Nevertheless, the voltage gradients recorded here (4-50 mV/mm at threshold for spike discharge) fall well within the range of previous reports, indicating a change in cell excitability induced by artificial voltage gradients as small as >7 mV/mm in CA18*31and dentate gyrus,20 and 10-l 5 mV/mm in cerebellum;‘3~‘4the latter being sufficient to induce regenerative activity in dendritic membrane. Therefore, both the magnitude of voltage gradients and the agreement between measured and predicted changes in TMP suggest that results of the present study reflect local ephaptic interactions capable of altering voltage-dependent processes in pyramidal cell apical dendrites. Further work will be required to assess the effects of these gradients upon slow Na+- or Ca2+-dependent dendritic conductances3*6,9,48Xs1*52 or voltage-dependent N-methyl-naspartate responses in the subsynaptic region.‘&J9 Interactions between synaptic potentials and ephaptic interactions

It is clear that the example of Fig. 5 is highly schematic, and that an overall representation of current flow must include current sourcing from all regions of the cell. The extent of ephaptic interactions will thus vary along the pyramidal cell axis according to the relative magnitude and location of population current sinks and sources contributing to the level of extracellular current flow. Another important factor is that dendritic membrane of the individual cells shown in Fig. 5 may undergo a simultaneous synaptic conductance change by some fraction of the activated afferent inputs. These effects cannot be ignored, as the flow of synaptic current along the dendrosomatic axis for the case of SR stimulation is in a direction opposite to that induced by the extracellular voltage gradient (Fig. 5). Thus, ephaptic interactions will also vary according to the relative number of activated afferents synapsing upon the single cell as compared to the cell population generating the field potential. Realistic predictions are further complicated when considerations are made for the time course of synaptic conductance changes” and the charging rate of pyramidal cell membrane.if44*46 A complete analysis of the interactions between synaptic and ephaptic currents will thus require suitable computer models incorporating synaptic conductance parameters and extracellular voltage gradients.42 The contribution of population spike potentials to ephaptic interactions

Synaptic activation of the basal dendrites by SO stimulus intensities sufficient to evoke a popu-

133

lation spike in stratum pyramidale also generates a biphasic population response in stratum radiatum (Fig. lB, D). Previous studies have interpreted this potential as representing an active (tetrodotoxinsensitive) invasion of apical dendrites by a population spike that originates in the region of stratum pyramidale.“s” Comparison of intra- and extradendritic potentials now provides strong evidence that this population response reflects the summed discharge of all-or-none dendritic spikes (Fig. lB, D). Furthermore, subtraction of this potential from intradendritic recordings revealed a measurable effect on the apical dendritic TMP. The SO-evoked dendritic population spike gave rise to two extracellular voltage gradients over the apical dendrites; each oriented towards the negativity in mid-stratum radiatum from the relatively positive potentials in distal stratum radiatum and the cell body layer (Fig. 2C). Ephaptic interactions induced by current flowing down these gradients should be expressed as a depolarization of dendritic TMP through a process equivalent to that described for the population spike in stratum pyramidale.29Z9*40*s3 Thus, passive current flow exciting single cells in the region of the extracellular current sink will give rise to a net TMP depolarization. In accordance with this, the SO-evoked dendritic population spike was associated with a depolarizing wave on TMP records when evoked at intensities sufficient to generate an extracellular negativity in mid-stratum radiatum (Fig. 4B, C). However, unlike the TMP depolarization induced by the cell body population spike in pyramidal cell somata,29*33*449the onset of intradendritic spikes did not align with the peak of the TMP depolarization (cf. Fig. 4B). Instead, the spike discharged prior to the TMP depolarization at a latency coinciding, in general, with the preceding positivity of the extradendritic population spike. It therefore appears that the sharp TMP depolarization is not directly associated with generation of the SO-evoked spike in mid-distal apical dendrites, at least for intensities near threshold for spike discharge. In a similar fashion, SR-evoked intradendritic spikes evoked at threshold intensities did not arise from any prominent depolarization of the TMP. Nevertheless, an ephaptic influence upon dendritic spike discharge at higher stimulus intensities cannot be ruled out, either for SO or SR stimulation.47 Ephaptic interactions in relation to pyramidal cell activity

The functional consequence of dendritic field effects is simplest to interpret for the case of SR stimulation, where extradendritic field potentials may depolarize dendritic membrane during synchronous orthodromic activation. Although a direct role for this process in eliciting dendritic spike discharge has not been shown, ephaptic interactions could

134

R. W. TURNERand T. L. RICHARDSON

enhance slower voltage-de~ndent dendritic condnCtances,3Y,‘y,26,37,48 particularly those evoked under conditions of reduced inhibitory function.25,36,48*52 It is important to note that the extracellular voltage gradients shown here to affect dendritic TMP are within the range of those observed in vivo. Rhythmic slow activity in the hippocampus gives rise to extracellular voltage gradients in the order of 4mV/mm,‘73B while epileptiform discharge can generate gradients of up to 20 mV/mm.‘“‘8 Thus, ephaptic interactions at the dendritic level

may contribute to depolari~tion and discharge of pyramidal celis during naturally occurring spontaneous activity and epileptiform discharge in the live animal. Acknowledgements-R. W. Turner was supported by a Canada Medical Research Council Studentship and T. L. Richardson by the British Columbia Health Care Foundation. The authors wish to thank Dr J. J. Miller for support and laboratory facilities, funded by the Canada Medical Research Council, and Dr L. Maler for comments on the manuscript.

REFERENCE

1. Alger B. E. and Nicoll R. A. (1982) Feedforward dendritic inhibition in rat hipp~mpal pyramidal cells studied in i&o. .I. Physic& Lend. 28, 105-123. 2. Andersen P. (1960) Interhippocampal impulses. II. Apical dendritic activation of CA1 neurons. Acta ph,vsiol. mad. 48, 178-208. 3. Andersen P. (1986) Factors influencing the efficiency of dendritic synapses on hippocampal pyramidal cells. Neurosci. Res. 3, 521-530. 4. Andersen P., Bliss T. V. P. and Skrede K. K. (1971) Unit analysis of hippocampal population spikes. Expl Brain Res. 13, 208-221. 5. Andersen P., Gross G. N., Lomo T. and Sveen 0. (1969) Participation of inhibitory and excitatory interneurons in the control of hippocampal cortical output. In The Interneuron (ed. Brazier M. A. B.), pp. 415-465. University of California Press, Berkeley. 6. Andersen P., Lacaille J.-C., Hvalby O., Hu G.-Y., Piercey B. W. and Ostberg T. (1986) Synaptic pre-potentials-an intermediary process in synaptic excitation. Acfa physiof. stand. 128, 17A. 6a. Arvanataki A. (1942) Effects evoked in an axon by the activity of a contiguous one. J. ~europhys~I. 5, 89-108. 7. Ashwood T. J., Lancaster B. and Wheal H. V. (1984) In uivo and in t;itro studies on intemeurons in the rat hip~campus: possible mediators of feed-forward inhibition. Brain Res. 293, 279-291. 8. Bawin S. M., Sheppard A. R., Mahoney M. D., Abu-Assal M. and Adey W. R. (19861 Comuarison between the effects of extracellular d&t and sinusoidal currents on excitability in hipp&al slices. -BrainRes. 362, 350-354. 9. Benardo L. S., Masukawa L. M. and Prince D. A. (19821 Electronhvsiolozv of isolated hinuocamnal dendrites. J. Neurosci. 2, 1614-1622. 10. Bland B. H. (1986) The physiology and pharmacology of hippocampal formation theta rhythms. Prog. Neurobiof. 26, l-54.

11. Brown T. H. and Johnston D. (1983) Voltage-clamp analysis of mossy fiber synaptic input to hippocampal neurons. J. Neurophysiof. 50, 487-507. 12. Brown T. H., Fricke R. A. and Perkel D. H. (1981) Passive electrical constants in three classes of hippocampal neurons. J. Neurophysiol. 46, 812-827. 13. Chan C. Y. and Nicholson C. (1986) Modulation by applied electric fields of Purkinje and stellate cell activity in the isolated turtle cerebellum. J. Physiof., Lund. 371, 89-l 14.

14. Chan C. Y., Hounsgaard J. and Nicholson C. (1988) Effects of electric fields on transmembrane potential and excitability of turtle cerebellar Purkinje cells in t&o. J. Physiof., Lond. 402, 751-771. 14a. Col~n~dge G. L., Herron C. E. and Lester R. A. J. (1988) frequency-de~nden~ N-methyl-D-aspa~ate receptor mediated synaptic transmission in rat hippocampus. J. Physicd., Lond. 399, 301-312. 15. Faber D. S. and Kom H. (1989) Electrical field effects: their relevance in central neural networks. Physiol. Rev. 69, 821-863. 16. Gloor P., Vera C. L. and Sperti L. (1963) Investigations on the mechanism of epileptic discharge in the hippocampus. Epilepsia (Series 4) 2, 42-63. 17. Green J. D. and Petsche H. (1961) Hippocampal electrical activity. II.Virtual generators. Efectroenceph. clin. Neurophysiol. 13, 868-879. 18. Green J. D. and Petsche H. (1961) Hippocampal electrical activity. IV. Abnormal electrical activity. Efectroenceph. cfin. Neurophysiol. 13, 868-879. 19. He&n S., Nicoll R. A., Perkel D. J. and Sah P. (1990) Analysis of excitatory synaptic action in pyramidal cells using whole-cell recording from rat hippocampal slices. J. Physiof., Land. 422, 203-225. 20. Jefferys J. G. R. (1981) Influence of electric fields on the excitability of granule cells in guinea-pig hippocampal slices. J. Physiof., Lond. 319, 143-152. 21. Jefferys J. G. R. and Haas H. L. (1982) Synchronized bursting of CA1 hipp~m~l pyramidal cells in the absence of synaptic transmission. Nature 300, 448-450. 22. Lacaille J.-C. and Schwartzkroin P. A. (1988) Stratum lacunosum-moleculare intemeurons of hip~~mpal CA1 region. II. Intrasomatic and intradendritic recordings of locaf circuit synaptic interactions. J. Neurosci. 8, 141l-1424.

23. Larson J. and Lynch G. (1986) Induction of synaptic potentiation in hippocampus by patterned stimulation involves two events. Science, Wash. 232, 985-988. 24. Leung L. S. (1979) Orthodromic activation of hippocampal CA1 region of the rat. Brain Res. 176, 49-63. 25. Masukawa L. M. and Prince D. A. (1984) Synaptic control of excitability in isolated dendrites of hippocampal neurons. J. Neurosci. 4, 217-227. 26. Miles R. and Wong R. K. S. (1986) Excitatory synaptic interactions between CA3 neurons in the guinea-pig hippocampus. J. Physiol., Land. 373, 397-418. 27. Miyakawa H. and Kato H. (1986) Active properties of dendritic membrane examined by current source-density analysis in hippocampal CA1 pyramidal neurons. Brain Res. 399, 303-309.

Dendritic field effects in hippocampus

135

28. Richardson T. L., Turner R. W. and Miller J. J. (1984) Extracellular voltage gradients and ephaptic interactions in the hipp~mp~ formation. Sot. Neurosei. Abstr. 10, 873. 29. Richardson T. L., Turner R. W. and Miller J. J. (1984) Extracellular fields influence transmembrane potentials and synchronization of hippocampal neuronal activity. Bruin Res. 294, 255-262. 30. Richardson T. L., Turner R. W. and Miller J. J. (1987) Action potential discharge in hippocampal CA1 pyramidal neurons: current source-density analysis. J. Neurophysiol. 58, 981-996. 31. Richardson T. L., Turner R. W. and Miller J. J. (1989) The influence of applied field potentials on the excitability of hippocampal neurons. Can. J, Physiol. Pharmac. 67, axxxi. 32. Richardson T. L., Turner R. W. and Miller J. J. (1989) Voltage gradient analysis of evoked potentials in the hippocampus. Can. J. Physiol. Pharmac. 67, axxxi. 33. Snow R. W. and Dudek F. E. (1984) Electrical fields directlv contribute to action uotential svnchronization during convulsant-induced epileptiform’bursts. Brain Res. 323, 114-i 18. 34. Snow R. W. and Dudek F. E. (1986) Evidence for neuronal interactions by electrical field effects in the CA3 and dentate regions of rat hippocampal slices, Brain Res. 367, 292-295. 35. Sperti L., Gessi T. and Volta F. (1967) Extracellular potential tield of antidromi~lly activated CA1 pyramidal neurons. Bruin Res. 3, 343-361. 36. Swann J. W., Brady R. J., Friedman R. J. and Smith E. J. (1986) The dendritic origins of ~niciliin-indu~ epileptogenesis in CA3 hippocampal pyramidal cells. J. N~rophysjol. 56, 1718-1738. 37. Tat&e J. S. and Schwartzkroin P. A. (1988) Mechanisms of long-term potentiation: a current source-density analysis. J. Neurosci. 8, 164.5-1655. 38. Taylor C. P. and Dudek F. E. (1982) Synchronous chemical synapses. Science, Wash. 218, 810-812.

neural afterdischarges in rat hippocampal slices without active

39. Taylor C. P. and Dudek F. E. (1984) Synchronization without active chemical synapses during hippocampal afterdischarges. J. Neurophysiol. 52, 143-155. 40. Taylor C. P. and Dudek F. E. (1984) Excitation of hippocampal pyramidal cells by an electrical field effect. J. Neurophysiol. 52, 126-142. 41. Taylor C. P., Kmjevic K. and Ropert N. (1984) Facilitation of hippocampal

CA3 pyramidal cell firing by electrical fields generated antidromically. Neuroscience 11, 101-109. 42. Traub R. D., Dudek F. E., Snow R. W. and Knowles W. D. (1985) Computer simulations indicate that electrical field effects contribute to the shape of the epiieptiform field potential. Natroscience 15, 947-958. 43. Traub R. D., Dudek F. E., Taylor C. P. and Knowles W. D. (1985) Simulation of hip~mpal afterdi~harges synchronize by electrical interactions. ~euroseience 14, 1033-1038. 44. Turner D. A. (1984) Segmental cable evaluation of somatic transients in hipp~ampal neurons (CAL CA3 and dentate). Biophys. J. 46, 73-84. 45. Turner D. A. (1990) Feed-forward inhibitory potentials and excitatory interactions in guinea-pig hippocampal pyramidal cells. J. Physiol., Lond. 422, 333-350. 46. Turner D. A. and Schwartzkroin P. A. (1980) Steady-state electrotonic analysis of intracellularly stained hippocampal neurons. J. Neurophysiol. 44, 184-199. 47. Turner R. W., Meyers D. E. R. and Barker J. L. (1988) Intensity-dependent shift in the site of origin of orthodromic action potential discharge in CA1 hippocampal pyramidal neurons. Sot. Neurosci. Abstr. 14, 836.

48. Turner R. W., Meyers D. E. R. and Barker J. L. (1989) Localization of tetrodotoxin-sensitive field potentials of CA1 pyramidal cells in the rat hippocampus. J. Neurophysiol. 62, 1375-1387. 49. Turner R. W., Richardson T. L. and Miller J. J. (1984) Ephaptic interactions contribute to paired pulse and frequency potentiation of hippocampai field potentials. Expl Brain Res. 54, 567-570. 50. Winson J. (1974) Patterns of hippocampal theta rhythm in the freely moving rat. E~ec~roenceph. clin. ~~ropkysioi. 36, 291-301. 51. Wong R. K. S., Prince D. A. and Basbaum A. I. (1979) Intradend~tic recordings from ~p~ampal neurons. Proc. natn. Acad. Sci. U.S.A. 76, 986-990. 52. Wong R. K. S. and Prince D. A. (1979) Dendritic mechanisms underlying ~nicillin-indu~ epileptiform activity. Science 204, 1228-i 23 1.

53. Yim C. C., Krnjevic K. and Dalkara T. (1986) Ephaptically generated potentials in CA1 neurons of rat’s hippocampus in situ. J. Neurophysiol. 56, 99-122. (Accepted 24 October 1990)