electrophysiological characterization of ca2 pyramidal cells from epileptic humans

HIPPOCAMPUS, VOL. 4, NO. 2, PAGES 226-237, APRIL 1994

Electrophysiological Characterization of CA2 Pyramidal Cells From Epileptic Humans

Anne Williamson and Dennis D. Spencer Section of Neurosurgery, Yale University School of Medicine, New Haven,

Connecticut. U.S.A.

ABSTRACT

The CA2 region of the hippocampus is more resistant to the principal cell loss seen in CAI and CA3 in both animal models of temporal lobe epilepsy and in medial temporal lobe sclerosis (MTS), a common neuropathological finding in human temporal lobe epilepsy. There is extensive synaptic reorganization in the MTS hippocampi that is not seen in the hippocampi of patients with tumor-associated temporal lobe epilepsy (TTLE). The authors examined the electrophysiological properties of CA2 pyramidal cells from these two types of human hippocampi. The two main findings are that most MTS cells do not have clear evidence for inhibition yet do not fire synaptically evoked bursts; and that mossy fiber stimulation could evoke excitatory postsynaptic potentials (EPSPs) in the MTS tissue, but not the TTLE cells. These data suggest that in MTS, CA2 cells are resistant to firing epileptiform bursts which may account for their survival. Moreover, the granule cell-CA2 cell connection represents a novel form of synaptic plasticity in this disease. 01994 Wiley-Liss. Inc.

Key words: hippocampus, brain slice, temporal lobe epilepsy, electrophysiology, mossy fibers

The CA2 region of the hippocampal formation is poorly defined in most species and, indeed, some researchers view it as an extension of CA3 rather than as a distinct functional entity in rodents (Amaral and Witter, 1989). Nevertheless, this region of the hippocampal formation is of interest because it appears to be a resistant zone in a number of animal models of epilepsy (Sloviter, 1991; Sutula, 1991; Cronin et al., 1992) and in certain forms of human temporal lobe epilepsy (Kim et al., 1990).

In order to analyze the properties of CA2 cells, we have obtained human hippocampi removed for treatment of medi- cally intractable seizures and have examined the electrophysiological characteristics of CA2 pyramidal cells from patients with two distinct types of temporal lobe epilepsy (TLE). In one group, the seizures were caused by extrahippocampal tumors (tumor-associated temporal lobe epilepsy, TTLE) and there were few anatomical changes in the hippocampus rela- tive to age-matched autopsy controls (Kim et al., 1990). In the other form of TLE, there was no structural lesion; however, there was extensive sclerosis of the hippocampal formation. This latter disease, medial temporal lobe sclerosis (MTS), is characterized by an extensive and consistent pattern of cellular loss and network reorganization throughout the hippocampus (de Lanerolle et al., 1992).

One of the best studied examples of the reorganization observed in MTS hippocampi is sprouting of the mossy fibers

Correspondence and reprint requests to Anne Williamson, Ph.D., Seclion of Neurosurgery. Yale University School of Medicine, 333 Cedar St LH122. New Haven, CT 06510 U.S.A.

01994 Wiley-Liss. Inc.

~ . ~ ~

that arise from the granule cells. These sprouted axons are believed to innervate both interneurons and granule cells in the inner molecular layer of the dentate gyrus (DG) (de Laner- olle et al., 1989; Sutula et al., 1989; Houser et al., 1990; Po- korny et al., 1991). In addition, in a number of MTS cases, there is anatomical evidence for extension of mossy fiber ter- minals into the CA2 region of the hippocampus (Houser et a]., 1990), suggesting that an additional form of synaptic plasticity involves the development of aberrant connections between the DG and CA2.

Based on these data we undertook to determine if there were any differences between the CA2 cells from these two distinct groups of hippocampi which would explain their resistance to destruction; and if the observed mossy fiber input into the CA2 region in the MTS tissue is physiologically significant.

The present data indicate that in MTS, the majority of the CA2 pyramidal cells did not receive powerful synaptic inhibition; however, unlike granule cells in this disease (Masukawa et al., 1989; Urban et al., 1990; Williamson et al., 1990; Maw- kawa et al., 1991), they did not fire multiple spikes in response to orthodromic stimulation. By contrast, all the TTLE CA2 cells studied showed clear biphasic inhibitory postsynaptic potentials (IPSPs) that resembled the GABAA and GABAB mediated events described in rodent pyramidal cells (Alger and Nicoll, 1982). Cells in the CAI region in MTS hippocampi were distinct from the CA2 cells as they could be induced to fire doublets and three spike bursts in response to stimulation in stratum radiatum of CA2-3, as has been shown previously (Knowles et al., 1992). Finally. in the majority of MTS CA2

CA2 PYRAMIDAL CELLS IN EPILEPTIC HUMANS / Williamson and Spencer 227

cells examined, stimulation of the hilus or the granule cells could evoke excitatory postsynaptic potentials (EPSPs), whereas this was not the case in the TTLE CA2 cells.

We conclude that synaptic reorganization of the sclerotic hippocampus includes sprouting of the mossy fibers into the CA2 region where functional synapses are formed. Moreover, the CA2 pyramidal cells appear to be resistant to firing epileptiform bursts, suggesting that they are neurochemically and electrophysiologically diqtinct from cells in CA3 or CAI, and that these differences may limit excitotoxic damage to the CA2 region during seizures.

MATERIALS AND METHODS

Tissue preparation

The hippocampi of both types of patients were removed as described by Spencer (1991) and slices (5-10 mm thick) were cut perpendicular to the long axis of the hippocampus and placed in cold (4°C) artificial cerebrospinal fluid (ACSF) within 5 minutes. We usually obtained a section of the anterior body of the hippocampus. The tissue was transported to the laboratory where 400 pm thick slices were prepared using a Vibroslice (WPI) and placed in the recording chamber 15 to 20 minutes following the hippocampal dissection. The slices were maintained at the gas-liquid border in an interface type recording chamber (Fine Science Tools, city, state) at 35" +. 1°C and were allowed to recover for at least two hours prior to recording. The slices were constantly perfused at 1 mL/ minute. The ACSF was bubbled with 95% O2/5% C02 to main- tain a pH of 7.4 and contained (in mm): NaC1, 124.0; KC1, 3.5; MgS04, 2; NaH2P04, 1.2; NaHCO3, 26.0; CaC12, 2.0; and dextrose, 10.0.

Electrophysiology

Recordings were made with an Axoclamp I1 amplifier (Axon Instruments, Foster City, CA) using microelectrodes formed on a Brown-Flaming type electrode puller (Sutter In- struments). The electrodes were filled with either 4 M K- acetate or intracellular labels (see below) and had resistances of 40-80 MR. Only cells with membrane potentials more neg- ative than -55 mV and with input resistances greater than 20 MR were included in the analysis.

Unless noted, stimuli were delivered to the Schaffer collat- erals by placing a monopolar tungsten-stimulating electrode in stratum radiatum (SR) of CA3. Under these conditions, the distance between the recording electrode and the stimulating electrode ranged between 500 and 800 pm. This distance was the same for the recordings in CAI and in the granule cells described in Figure 5. Synaptic intensity ranged from 0.03 to 1.5 mA with a pulse duration of 5 ps.

Data analysis

Voltage versus current (V-I) relationships were measured at or near the resting potential in all cells studied using 100 ms-long square current pulses. The voltage measurement was taken at the point where the voltage had reached steady state, approximately 50 ms following the onset of the current pulse. The bridge balance was monitored at all times to ensure an accurate voltage. Inhibitory postsynaptic potentials were

measured at the peak of the fast and slow phases elicited at depolarized membrane potentials (approximately - 60 mV) and were measured at the same latency at all other membrane potentials. These measurements were only made when a biphasic IPSP was clearly visible. All statistical values are given as the mean 2 SD.

Morphological techniques

Cells were filled with either Lucifer Yellow dilithium salt (4% in 1 M LiCI) or with Neurobiotin (Vector Labs, Burl- ingame, CA) at 4% in 2 M K-acetate. After the end of the experiment, the slices were fixed in 4% paraformaldehyde between two sheets of filter paper for at least 6 hours and sectioned at 40 pm. For the Lucifer Yellow filled cells, the sections were dried on sub-bed slides, cleared with methyl salicylate and visualized using ultraviolet light. For the biocytin-filled cells, the tissue was treated as described by Hori- kawa and Armstrong (1988). Briefly, the sections were rinsed in phosphate buffer, incubated in Vectastain ABC reagent (Vector Laboratories) for several hours and then treated with diaminobenzidine and H202. The sections were then mounted on gelatin-coated slides, cleared, and intensified with osmium tetroxide.

RESULTS

Identification of CA2 pyramidal cells

We have recorded data from studying 20 MTS cells from 11 cases and eight TTLE CA2 pyramidal cells from five cases in this study. To reduce the variability in the data, only cells from MTS cases in which the full pattern of reorganization (mossy fiber spouting and cell loss in CA1 and CA3 with less than 20% loss in CA2) was present are included. Mossy fiber sprouting was assessed by dynorphin-like immunoreactivity and quantitative cell counts were performed on sections adja- cent to the piece used for the electrophysiological studies (see de Lanerolle et al., 1992, for details). In the TTLE cases, the lesions were all extrahippocampal and had not invaded the hippocampus. In these cases the hippocampi were removed to ensure tumor-free margins.

For these experiments, the CA2 region was taken as the region of the pyramidal cell body layer which ran approximately parallel to the terminal third of the short limb of the dentate granule cell layer (Duvernoy, 1988). Cells were posi- tively identified as being in this area following the experiment using intracellular fills with Neurobiotin. As the specimens were usually cut in similar fashions, the location of the CA2 region could be reliably identified between patients. An example of a filled CA2 cell from a patient diagnosed with MTS is shown in Figure 1. The location of this cell in the CA2 region is shown in Figure 1B.

Membrane properties

We noted no significant differences in the membrane properties of these two groups of human CA2 cells. The resting membrane potentia1,input resistance and time constant were measured for all cells studied. These data are given in Table 1 .

Figure 2 shows examples of the V-I relationships used to

228 HZPPOCAMPUS VOL. 4, NO. 2, APRIL 1994


Table 1 . Membrane Properties of Human CA2 Cells TTLE MTS

RMP (mV) -70.0 2 4.47 -73.4 t- 7.87 Input Resistance (Mn) 32.1 2 11.85 33.28 2 8.02 Time Constant (ms) 18.8 2 5.45 13.76 2 5.22

There were no significant differences in the resting membrane potential (RMP), input resistance or time constant between them from either TTLE (n = 12) or MTS (n = 8) CA2 cells. The data are shown as mean 5 SD.

measure the input resistance and time constant for these cells. While we did not observe the time-dependent rectification mediated by the mixed cationic current 10, we did note inward rectification in the depolarizing direction, which can be seen in the V-I plots in Figure 2.

All the cells studied from both groups of tissue fired rapidly adapting trains of action potentials in response to depolarizing current injections from 100 to 300 ms in duration. These trains of action potentials were followed by afterhyperpolarizations that lasted up to one second. Examples of the trains produced by 1 nA amplitude current pulses, 100 ms in duration, are shown in the lower panel of Figure 2 for both types of hippocampi.

Synaptic responses

Orthodromic EPSPs and action potentials could be evoked in all cells from both types of hippocampi by stimulating stratum radiatum of CA2 andlor the distal portions of CA3 (rela- tive to the DG). Synaptic responses with monosynaptic latencies (4-8 ms) could also be elicited by stimulation of stratum radiatum in CAI, possibly by back-firing the Schaffer collater- als. However, higher stimulus intensities were required at this site and the responses were less reliable than those seen with stimulation in CA2-3. Stimulation in CA1 at high stimulus intensities could also produce responses with antidromic latencies (1 5-2.0 ms) followed by orthodromic responses. However, we never were able to elicit a pure antidromic response in these cells. These responses were seen independent of which type of human tissue was being studied.

We did note an interesting difference in the voltage-dependence of the EPSPs between the MTS and the TTLE hippocampi. In all the TTLE CA2 pyramidal cells studied, the EPSP amplitude increased as the membrane potential was hyperpolarized. By contrast, in 10 out of 12 of the MTS cells, the EPSP amplitude did not vary significantly with membrane potential over a range of 20 to 30 mV, suggesting that responses to glutamate may be altered in these cells. Moreover, the duration of these events also did not change significantly over this range of voltages, suggesting that there was not a significant NMDA receptor-mediated contribution of the EPSPs. However, because action potential generation was not blocked by either intracellular QX-3 14 or extracellular

TTX in these studies, we could not clearly determine the reversal potentials for these EPSPs.

For cells from both types of hippocampi, orthodromic stimulation produced graded EPSPs that triggered action potentials. As the stimulus intensity was raised, the amplitude of the late phase of the EPSP increased, but we rarely elicited more than a single spike from an identified CA2 cell, as shown in Figure 3 . We never noted burst-firing activity in either the MTS or the TTLE CA2 cells. As can also be seen in Figure 3 , the primary difference between the TTLE and the MTS cells was the presence of a biphasic IPSP in the TTLE, but not in the MTS cell. This difference is described below. In 2 out of 20 CA2 cells from sclerotic hippocampi we were able to evoke doublets at very high stimulus intensities (>7.0 mA).

Thus, CA2 cells appear to be resistant to firing synaptically evoked burst potentials. By contrast, CA1 cells from MTS hippocampi could fire bursts in response to stimulation in stratum radiatum of CA2 (n = 7). Examples of burst firing in a CA1 cell compared to a CA2 and a dentate granule cell from MTS hippocampi are shown in Figure 4. Note that for both the granule cell and the CA1 pyramidal cell, increasing the stimulus intensity produced a graded excitatory response during which up to three action potentials could be triggered. We noted no burst-firing in CA1 cells from TTLE hippocampi (n = 5). Therefore, CA2 cells from MTS hippocampi appear to be distinct from other types of pyramidal cells in the hippocampus.

MTS CA2 pyramidal cells lack inhibition

The most consistent difference between the MTS and the TTLE CA2 pyramidal cells was the absence of inhibitory potentials in the majority of cells from MTS tissue, whereas all cells studied from TTLE hippocampi had robust biphasic IPSPs. Examples of these IPSPs are shown in the top panel of Figure 5. The reversal potentials for these IPSPs suggest that they are mediated by GABAA- and GABAB-type receptors. The reversal potential for the fast IPSP was -63.9 &

1.36 mV and -83.6 t 4.71 for the slow IPSP. By contrast, the lower panel shows results from cells from

two different patients diagnosed with MTS. Note that at all membrane potentials, there was no evidence for either fast or slow IPSPs following either an action potential or an EPSP. The absence of inhibitory potentials in these cells was quite consistent between cells. Only 3 out of 9 MTS CA2 cells studied exhibited any type of inhibitory potential following orthodromic stimulation. In only one of these cells did we see the fast and slow inhibitory potentials similar to those shown in the upper panel of Figure 5. In the remaining two cells, we noted a slow hyperpolarization that resembled a GABAB-type IPSP.

In five MTS cells from three patients we examined the responses to Schaffer collateral stimulation from three different sites in the SR of CA3 ranging from locations very close to

Fig. 1. Anatomy of localization of human CA2 cells. (A) A human CA2 cell labeled with intracellular biocytin. Note the extensive spiny apical and basal dendrites. Calibration bar, 100 km. (B) The location of this cell in the CA2 region. The arrowhead points to the filled cell; the CA3 region and the granule cell layer are also labeled. Calibration bar, 1 mm. DG, dentate gyrus.

230 HIPPOCAMPUS VOL. 4, NO. 2, APRIL 1994

A TTLE S c lero t i e

V 1 50 msec

T

t -30 -I

A Vm (mV) A

A Vrn (mV)

B

IL ~~

50 msee - - Fig. 2. Membrane properties of human CA2 neurons. (A) The voltage responses to 100 ms duration injected current pulses in human CA2 neurons are shown in the upper traces. The data are plotted below. There were no significant differences in any of the membrane properties for these two classes of neurons. Membrane potentials (M.P.s): TTLE = -74 mV; Sclerotic = -69 mV. (B) Firing properties of human CA2 neurons. The responses to 1 nA depolarizing current pulses are shown for both sclerotic and TTLE CA2 cells. Note that both types of cells fire adapting trains of action potentials. M.Ps sclerotic = -72 mV; TTLE = -59 mV.

TTLE

Sc

lero

tic

0.4 Jr

---------

0.58

1.2

0.76

t t

Fig.

3.

Nei

ther

type

of C

A2

cell

coul

d be

indu

ced

to fi

re m

ore

than

one

spi

ke in

resp

onse

to o

rtho

drom

ic st

imul

atio

n (f

rom

str

atum

radi

atum

of

CA

3). N

ote

that

ther

e w

as n

o ev

iden

ce fo

r an

IPSP

in th

e ce

ll fr

om a

scl

erot

ic h

ippo

cam

pus.

The

stim

ulus

inte

nsiti

es a

re s

how

n on

the

left

in m

A. M

.P.s

: T

TL

E =

-78

m

V;

Scle

rotic

=

-73

mV

.

A

B

CA

1 Py

ram

idal

Cel

ls

TTIJE

L

I 0.

21

Den

tate

Gra

nule

Cel

l

Scle

rotic

I

Fig.

4.

CA

I py

ram

idal

cel

ls f

rom

scl

erot

ic h

ippo

cam

pi f

ire

grad

ed e

pile

ptifo

rm b

urst

s in

res

pons

e to

stim

ulat

ion

in s

trat

um ra

diat

um o

f C

A2.

By

cont

rast

, as

show

n in

the

left

pane

l, bu

rsts

cou

ld n

ot b

e el

icite

d in

a C

AI

pyra

mid

al c

ell f

rom

a tu

mor

-ass

ocia

ted

hipp

ocam

pus.

T

he s

timul

us in

tens

ities

(in

mA

) are

sho

wn

to th

e le

ft of

the

trac

es. M

.P.s

: TT

LE

=

-69

mV

; Scl

erot

ic =

-6

7 m

V. B

sho

ws a

n ex

ampl

e of

the

syn

aptic

ally

evo

ked

burs

ts s

een

in a

den

tate

gra

nule

cel

l fro

m a

MTS

hip

poca

mpu

s. M

.P.

=

-62

mV

.


TTLE

110 mV

100 msec

t t 1

IPSP Amp (mV)

-67 -67 u -78 -7 2 .- -86 - -81 _)n Fig. 5. Some MTS CA2 cells lack synaptic inhibition. In the TTLE CA2 cell shown here, biphasic IPSPs were elicited by stimuli subthreshold to action potential generation. The reversal potentials of the fast and slow IPSPs were investigated by shifting the membrane potential with injected D.C. current. The data are plotted on the right. The reversal potentials for the IPSPs in this cell were -71 mV for the fast IPSP and -91 mV for the slow IPSP. By contrast, in the two cells from sclerotic hippocampi shown in the lower panel, clear IPSPs could not be elicited at any membrane potential.

the recording electrode (250 pm) to a distance of 1.25 mm. As the stimulator was moved away from the cell, the latency to the onset of the excitatory response and the stimulus intensity needed to trigger an action potential both increased. We were unable to evoke biphasic IPSPs at any of these sites (data not shown).

MTS CA2 cells receive mossy fiber innervation

The connectivity of CA2 cells from these two groups of human hippocampi was investigated by moving the stimulating electrode to different sites within the hippocampal formation. We delivered a range of stimulus intensities at each site

and were able to stimulate between three and five sites in each slice.

In the TTLE cells, synaptic stimuli could only be evoked by stimuli within CA2 or CA3. EPSPs with monosynaptic latencies (4-8 ms) could be evoked by stimulating in ether stratum radiatum or in the pyramidal cell layer itself. How- ever, there was no response in the CA2 cells when the stimulating electrode was in the dentate (n = 3) slices from three patients. In the majority of these experiments the stimuli were applied both in the granule cell layer of the inferior and supe- rior limbs, and in the hilus. Examples of the EPSPs elicited at different locations in a TTLE hippocampus are shown in the top panel of Figure 6.

TTLE

CA

3 St

ratu

m R

adia

tum

CA

3 St

ratu

m R

adia

tum

1

CA

3/C

A4

Bor

der

Scle

rotic

Hilu

s

CA

3/C

A4

Bor

der

Gra

nule

Cel

l Lay

er

3 1 10m

V

40 m

sec

Fig.

6.

The

hyp

othe

sis

that

mos

sy f

iber

s in

nerv

ated

CA

2 by

mov

ing

the

stim

ulat

ing

elec

trod

e to

dif

fere

nt s

ites t

hrou

ghou

t th

e hi

ppoc

ampa

l fo

rmat

ion.

In

the

tum

or-a

ssoc

iate

d ce

ll sh

own

in t

he t

op p

ane!

, on

ly a

sm

all r

espo

nse

coul

d be

see

n w

hen

the

stim

ulat

ing

elec

trod

e w

as

plac

ed n

ear t

he C

A3-

4 bo

rder

. No

resp

onse

was

see

n w

hen

the

elec

trod

e w

as p

lace

d in

the

hilu

s. B

y co

ntra

st, i

n th

e sc

lero

tic c

ell s

how

n he

re,

a sy

napt

ic re

spon

se c

ould

be

elic

ited

whe

n th

e st

imul

atin

g el

ectr

ode

was

bot

h at

the

CA

3-4

bord

er a

nd i

n th

e gr

anul

e ce

ll la

yer.

The

loca

tions

of t

he e

lect

rode

pla

cem

ents

are

indi

cate

d on

the

sche

mat

ic d

iagr

ams

show

n on

the

righ

t. T

hese

dat

a su

gges

t tha

t th

e sp

rout

ed

mos

sy f

iber

s ca

n fo

rm a

berr

ant c

onta

cts

with

CA

2 ce

lls i

n sc

lero

tic h

ippo

cam

pi.

M.P

.s: T

TL

E =

-7

8 m

V:

Scle

rotic

=

-71

mV

.


By contrast, we were able to elicit EPSPs in CA2 cells from sclerotic hippocampi from a number of sites throughout the hippocampus and the dentate as shown in the lower panel of Figure 6. These experiments were carried out in eight cells from five patients. For each cell, we were able to study at least three different sites. In the example shown here, EPSPs could be elicited from the CA3-CA4 border, in the hilus itself as well as in granule cell layer of the infrapyramidal blade. We were able to elicit EPSPs with stimulation of the dentate gyrus in all MTS cells studied (n = 8). It is important to note, however, that there was variation in the areas of the dentate that were connected to the CA2 pyramidal cells being studied. For example, stimulation of the infrapyramidal granule cell layer produced EPSPs in 7 out of 8 cells examined, while stimulation of the hilus in the region where the CA4 pyramidal cells spread out always produced an EPSP. It was more difficult to obtain CA2 responses when the stimulating electrode was located on the opposite side of the dentate gyrus (n = 3 out of 8 cells). Stimulation of the subgranular zone on the suprapyramidal limb could produce EPSPs in only 2 out of 8 cells and we were never able to produce a response in CA2 to stimulation of the granule cells on that side. It was difficult to accurately measure the true distance between the recording and stimulating electrodes in these experiments as the exact pathways are not clear and there was a great deal of variation in the size of the hippocampi being studied.

DISCUSSION

This study is the first to focus on the physiological properties of human CA2 cells. Our two main findings are first, that CA2 pyramidal cells in MTS hippocampi are resistant to firing synaptically evoked bursts even though there is no clear evidence for synaptic inhibition in the majority of cells studied and second, that they receive excitatory input from the mossy fibers which is not seen in the TTLE hippocampi.

CA2 pyramidal cells lack synaptic inhibition

One of the clearest differences between the TTLE and MTS CA2 cells was the absence of IPSPs in the latter tissue. This lack of inhibition cannot be attributed simply to a loss of interneurons as anatomical findings have demonstrated that GABAergic cells in the hippocampus are not significantly reduced in MTS (Babb et al., 1989). This in contrast to the dentate hilus in which there is extensive interneuronal loss (de Lanerolle et al., 1989). One hypothesis to explain these findings is that the synaptic connectivity between the principal cells and the interneurons is disrupted based on our ability to elicit either feedforward or feedback IPSPs in the MTS CA2 cells. This dormant basket cell hypothesis proposed by Sloviter et al. (1991) argues that in certain forms of kindling, the connection between afferent cells and the interneurons is disrupted, but that the interneuronal-target cell connection is intact. This hypothesis was recently supported by data from CAI pyramidal cells (Bekenstein and Lothman, 1993).

However, we were unable to elicit IPSPs in response to Schaffer collateral stimulation at different points in SR of CA3. Thus, we could not clearly evoke IPSPs by directly stimulating interneurons directly as would be predicted by the dormant basket cell hypothesis. However, a disruption

between pyramidal cells and interneurons cannot be com- pletely ruled out, as these experiments were not carried out in APV and CNQX to block excitatory transmission. In addition, we applied stimulation primarily in the stratum radiatum so that only interneurons in this region could have been stimulated directly in our experiments. Alternately, as in the kindling model developed by Sloviter et al. (1991), the CA2 interneurons may receive their primary input contralaterally. If this were the case, it would be difficult to selectively activate these cells in the slice preparation.

CA2 pyramidal cells from sclerotic hippocampi do not fire bursts

The CA2 cells appear to be unique in their resistance to firing synaptically evoked bursts in the absence of clear inhibition. Granule cells (Masukawa et al., 1989; Urban et al., 1990; Masukawa et al., 1991; Williamsonet al., 1991) and CAI pyramidal cells (Fig. 4; Knowles et al., 1992) from MTS hippocampi are clearly hyperexcitable. One hypothesis that might serve to explain this difference lies in the complement of Ca2 + -binding proteins found in different cell populations in the human hippocampus. Both granule cells and CA2 pyramidal cells, along with specific populations of interneurons, contain the CA2 + -binding proteins parvalbumin and calbindin-128D. Populations of cells that are lost in MTS, such as CAI and CA3 pyramidal cells and the hilar mossy cells, have very low levels of these proteins (Sloviter et al., 1991). This pattern of cell death has suggested that these Ca” binding proteins are protective against excitotoxic damage. Scharf- man and Schwartzkroin ( 1990) demonstrated that excitotoxic damage could be prevented by injecting the Ca2+ chelator BAPTA into a single hilar interneuron.

However, the levels of Ca2+ entry during a single synaptic input and during the long trains used to produce cell death are probably very different. Therefore, variations in the level of Ca2+-binding proteins between CAI and CA2 pyramidal cells probably cannot fully account for the differences in their burst-firing behavior. This difference is not due to an inability to fire multiple action potentials to a given input, as the CA2 cells will fire trains of spikes in response to direct depolariza- tion, as shown in Figure 2.

Alternative possibilities include, first, variations in the density of Ca2+ channels between the CA1 and CA2 pyramidal cells. The “t” type Ca channel would probably be involved as this has been shown to underlie intrinsic burst firing in a number of neuronal types. As Ca2+ entry appears to be essen- tial to the generation of either intrinsic or synaptically evoked bursts, a small change in the density of Ca2+ channels on the dendrites of these cells could have profound effects on the duration and amplitude of the envelope that underlies the action potentials. Second, an enhancement of the inactivation of the AMPA- and/or NMDA-type glutamate receptors could also affect the ability of the CA2 cells to fire synaptic bursts.

Differences in EPSP characteristics

Another interesting difference between the two populations of CA2 cells was that the amplitude of the EPSPs in the MTS tissue did not vary with membrane potential. These data suggest that either the voltage dependence of either the AMPA-

236 HIPPOCAMPUS VOL. 4, NO. 2, APRIL 1994

or NMDA-type glutamate receptors was altered or that voltage-dependent membrane conductances are activated that either shunt or augment the EPSP conductance.

The changes seen in granule cells from amygdala-kindled rats are potentially similar. When studied with sharp electrodes, the slope conductance of these cells decreased (Mody, 1990). In addition, in these cells a population of NMDA-activated channels with very long open times was seen in the kindled, but not the control animals (Kohr et al., 1993). Thus epileptiform activity can change both membrane and synaptic conductances. It is unclear which of these hypotheses pertain to the human CA2 cells. We noted no differences in either the slope conductance or the input resistance of these cells as was noted by Mody et al. (1990) and the I-V curves showed no obvious rectification in the voltage range used in these studies. These experiments, however, would not detect slowly activating conductances well.

Therefore, changes in the glutamate receptor responses are probably involved. As suggested above, variations in either the activation and inactivation kinetics of the AMPA- or NMDA-type glutamate-activated channels may be altered in the MTS CA2 cells.

Reorganization of mossy fibers The second major finding in this study is that CA2 cells

can receive mossy fiber input in sclerotic hippocampi. These physiological data corroborate the anatomical findings of Houser et al. (1988) who noted Timm stain in the CA2 region of sclerotic hippocampi in a percentage of the cases studied.

The data indicate that the granule cells from the infrapyramidal blade of the dentate provide the bulk of the input to CA2, as it was difficult to elicit EPSPs from the suprapyramidal blade. Alternatively, in rodents, the mossy fibers from the infrapyramidal blade course towards CA3 in a tight bundle, while those from the suprapyramidal blade are more diffuse (Claiborne et al., 1990). Thus, our electrode placements near the infrapyramidal blade may have preferentially activated the large number of mossy fibers necessary to generate EPSPs in CA2.

It is unlikely that these results are due to current spread from the stimulating electrode. The stimulating electrode was closest to the recording electrode when the granule cells were stimulated. If the EPSPs generated by stimulation in the dentate gyrus were due primarily to passive current spread across the hippocampal fissure we would predict the following: the synaptic delay should be shorter and the rise time of the EPSP should be faster than when the stimulating electrode was physically further away. We consistently found, as shown in Figure 6 that the EPSPs produced by granule cell stimulation had longer latencies and slower rise times than those seen following stimulation of the SR of CA3.

A persistent question in MTS has been the precise location of ictal onset and pathways of seizure spread out of the medial temporal lobe. Depth electrode studies in these patients have demonstrated that seizures often originate in the hippocampus and then spread rapidly to the ipsilateral temporal neocortex and/or contralateral hippocampus (Spencer et al., 1987; Spen- cer et al., 1990). Moreover, there is evidence for a loop in this system, as seizures which start in the hippocampus propagate rapidly to the entorhinal cortex (Spencer and Spencer, 1993).

Therefore, seizures can both begin in and propagate through reorganized hippocampi in which the number of primary out- put cells (CA1 and CA3 pyramidal cells) has been reduced by up to 80%. The present results suggest, therefore, that the CA2 pyramidal cells may provide a critical relay in the spread of activity through the reorganized hippocampus. We are cur- rently investigating the hypothesis that the CA2 axons them- selves reorganize and extend to innervate the subiculum and the distal regions of CA1.

ACKNOWLEDGMENTS

We are indebted to the patients without whose informed consent we could not have performed these experiments. We thank Dr. N. C. de Lanerolle and Dr. Matthew Philips for histological assistance and Dr. Gordon M. Shepherd for read- ing the manuscript. Supported by NINDS grants to A.W.

REFERENCES

Alger BE, Nicoll RA (1982) Feed-forward dendritic in rat hippocampal pyramidal cells studied in vitro. J Physiol 328:105-123.

Amaral DG, Witter MP (1989) The three-dimensional organization of the hippocampal formation: a review of anatomical data. Neurosci 3 1 : 57 1-59 1.

Babb TL, Pretorius JK, Kupfer WR, Crandall PH (1989) Glutamate decarboxylase-immunoreactive neurons are preserved in human epileptic hippocampus. J Neurosci 9:2562-2574.

Bekenstein JW, Lothman EW (1993) Dormancy of inhibitory interneurons in a model of temporal lobe epilepsy. Science 359: 97- 100.

Claiborne BJ, Amaral DG, Cowan WM (1990) Quantitative, three- dimensional analysis of granule cell dendrites in the rat dentate gyrus. J Comp Neurol 302:206-219.

Cronin J , Obenhaus A, Houser CR, Dudek FE (1992) Electrophysiol- ogy of dentate granule cells after kainate-induced synaptic reorganization of the mossy fibers. Brain Res 573:305-310.

de Lanerolle NC, Kim JH, Robbins RJ, Spencer DD (1989) Hippo- campal interneuron loss and plasticity in human temporal lobe epilepsy. Brain Res 495:387-395.

de Lanerolle NC, Brines ML, Williamson A, Kim JH, Spencer DD (1992) Neurotransmitters and their receptors in human temporal lobe epilepsy. In: The dentate gyrus and its role in seizures (Ribak CE, Gall CM, Mody I, eds), pp 235-250. The Netherlands: El- sevier.

Duvernoy HM (1988) The human hippocampus. Munich: J . F. Berg- mann Verlag.

Horikawa KA, Armstrong WE (1988) A versatile means of intracellular labeling: injection ofbiocytin and its detection with avidin conju- gates. J Neurosci Meth 25:l-11.

Houser CR, Miyashiro JE, Swartz BE, Walsh GO, Rich JR, Delgado- Escueta AV (1990) Altered patterns of dynorphin immunoreactivity suggest mossy fiber reorganization in human hippocampal epilepsy. J Neurosci 10:267-282.

Kim JH, Guimaraes PO, Shen M-Y, Masukawa LM, Spencer DD (1990) Hippocampal neuronal density in temporal lobe epilepsy with and without gliomas. Acta Neuropathol 80:41-45.

Knowles WD, Awad IA, Nayel MH (1992) Difference of in vitro electrophysiology of hippocampal neurons from epileptic patients with mesiotemporal sclerosis versus structural lesions. Epilepsia 33:

Kohr G, De Koninck Y, Mody I(1993) Properties of NMDA receptor channels in neurons acutely isolated from epileptic (kindled) rats. J Neurosci 13:3612-3627.

Masukawa LM, Higashima M, Kim JH, Spencer DD (1989) Epilepti-

601-609.


form discharges evoked in hippocampal brain slices from epileptic patients. Brain Res 493:168-174.

Masukawa LM, Higashima M, Hart GJ, Spencer DD, O’Connor MJ (1991) NMDA receptor activation during epileptiform responses in the dentate gyrus of epileptic patients. Brain Res 562: 176-180.

Mody I , Reynolds JN, Salter MW, Carlen PL, MacDonald JF (1990) Kindling-induced epilepsy alters calcium currents in granule cells of rat hippocampal slices. Brain Res 531:88-94.

Pokorny J , Schwartzkroin PA, Franck JE (1991) Physiologic and mor- phologic characteristics of granule cells from the human epileptic hippocampus. Epilepsia 32 Suppl 3:46.

Scharfman HE, Schwartzkroin PA (1990) Responses of cells of the rat fascia dentata to prolonged stimulation of the perforant path: sensitivity of hilar cells and changes in granule cell excitability. Neurosci 35:491-504.

Sloviter RS (1991) Permanently altered hippocampal structure, excitability, and inhibition after experimental status epilepticus in the rat: the “dormant basket cell” hypothesis and its possible rele- vance to temporal lobe epilepsy. Hippocampus 1:41-66.

Sloviter RS, Sollas AL, Barbaro NM, Laxer KD (1991) Calcium- binding protein (calbindin-D28) and parvalbumin in the normal and epileptic human hippocampus. J Comp Neurol 308:381-396.

Spencer DD (1991) Anteriomedial temporal lobectomy: directing the surgical approach to the pathological substrate. In: Surgery for epi-

lepsy (Spencer DD, Spencer SS, eds) pp 129-137. Boston: Black- well Scientific Publications.

Spencer SS, Spencer DD (1993) Entorhinal hippocampal interactions in medial temporal lobe epilepsy. Epilepsia 33:59.

Spencer SS, Williamson PD, Spencer DD, Mattson RH (1987) Human hippocampal seizure spread studied by depth and subdural recording: the hippocampal commisure. Epilepsia 28:479-489.

Spencer SS, Spencer DD, Williamson PD, Mattson RH (1990) Com- bined depth and subdural electrode investigation in uncontrolled epilepsy. Neurology 40:74-79.

Sutula TP (1991) Reactive changes in epilepsy: cell death and axon sprouting induced by kindling. Epil Res 10:62-70.

Sutula TP, Cavazos J, Parada I, Ramirez L (1989) Mossy fiber synaptic reorganization in the epileptic human temporal lobe. Ann Neurol

Urban L, Aitken PG, Friedman A, Somjen GG (1990) An NMDA- mediated component of excitatory synaptic input to dentate granule cells in “epileptic” human hippocampus studied in vitro. Brain Res 515:319-322.

Williamson A, McCorrnick DA, Shepherd GM, Spencer DD (1990) Intracellular recordings from epileptic human dentate granule cells show evidence of hyperexcitability. Epilepsia 3 I :625.

Williamson A, Shepherd GM, Spencer DD (1991) Evidence for hyperexcitability near neocortical lesions in epileptic patients. Epilepsia 32:40.

26: 32 1-330.

electrophysiological characterization of ca2 pyramidal cells from epileptic humans

Documents