- Email: [email protected]
Functional synapses in synchronized bursting of neocortical neurons in culture
Brain Research 795 Ž1998. 137–146
Research report
Functional synapses in synchronized bursting of neocortical neurons in culture Keiko Nakanishi a
a, )
, Fumio Kukita
b
Dept. of Physiology, Institute for DeÕelopmental Research, Aichi Human SerÕice Center, Kasugai, Aichi 480-0392, Japan b Laboratory of Membrane Biology, National Institute of Physiological Sciences, Okazaki, Aichi 444-8585, Japan Accepted 10 March 1998
Abstract Spontaneous electrical activities in pairs of neocortical neurons in culture were simultaneously recorded using a whole cell current clamp technique. Synchronous bursting activities were observed in all 59 pairs tested. In 52 pairs of neurons electrically stimulated, EPSPs were recorded in 20 pairs Ž39%., among which 3 pairs Ž6%. showed bidirectional coupling. The response latency observed was 4.05 " 0.61 ms Žmean" S.E.M... The synaptic delay was estimated at 1.5–1.9 ms, suggesting the response latency is derived from a polysynaptic connection. The burst latency which was defined as the time difference of the onset of bursting in each neuron was 5.87 " 0.47 ms Žmean " S.E.M.., and was weakly correlated with the spatial distance between the neurons Ž37.5–600 m m apart. Ž R s s 0.362, tied P value s 0.0065.. No synchronized bursting was observed in bathing solution with a low Ca2q concentration Ž0.4 mM. or in bathing solution containing 50 m M D-AP5 and 15 m M CNQX. No dye-coupling between bursting neurons was observed on injection of the small molecule dye Lucifer yellow or the neurotracer neurobiotin. Disrupting neural connections completely by cutting the cell layer, caused disappearance of synchronized bursting with each neuron bursting independently. In conclusion, these results are consistent with the hypothesis that synchronized bursting in cultured neocortical neurons is attributed to connections by way of several synapses rather than by way of gap junctions andror diffusible factors. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Functional synapse; Synchronized bursting; Neocortical neuron
1. Introduction Highly correlated, spontaneous neuronal bursting activities have been reported in many regions of developing mammalian brain, such as the visual system w22,37,42x, hippocampus w5,35,36,38x, locus coeruleus w16x, inferior olive w3x and neocortex w12,13x, suggesting important roles for them in the signal processing of central nervous systems. Furthermore, high-frequency network oscillations have been observed both in slice preparations w7x and in living animals such as rats and monkeys w6,27,28x, indicating that synchronized electrical activities in neurons are fundamental for integrated brain functions such as memory, learning, and recognition. However, the mechanism reported for oscillatory activities is complex because it is different in different systems and in different development stages.
)
Corresponding author. Fax: [email protected]
q 81-568-88-0829;
E-mail:
0006-8993r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 0 6 - 8 9 9 3 Ž 9 8 . 0 0 2 8 3 - 2
There have also been reports of synchronous intracellular Ca2q oscillations in hippocampal w31x and neocortical culture systems w18,25,26,29x. Furthermore, simultaneous measurement of intracellular Ca2q and electrical activity revealed that neuronal bursts are generated periodically and accompanied by slower Ca2q transients w34x, indicating a coupling of neuronal synchronized bursting and an intracellular signal transduction. Neuronal bursting and intracellular Ca2q oscillation in culture are believed to be synchronized via a neural network of synapses, since these phenomena are attenuated by the NMDA receptor antagonist APV w12,34x. Using electron microscopy, Ichikawa et al. w15x found a synapse formation in cultured neocortical neurons showing a synchronized intracellular Ca2q oscillation. However, it has not been demonstrated whether the functional synapses in these neurons are actually working. To demonstrate that synchronized bursting neurons have synaptic connections with each other, we recorded electrical activities simultaneously in a pair of neurons with a patch clamp method. The time difference between the
138
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
onset of bursting in each neuron Žburst latency. was analyzed and compared with a synaptic delay estimated from evoked synaptic response. Our results suggest that synchronized bursting can be attributed to connections by way of several synapses rather than by way of gap junctions andror diffusible factors.
2. Materials and methods 2.1. Cell culture Neocortical cells were prepared from the cerebral cortex of embryonic day 16–18 Wistar rats and cultured as described in the work of Nakanishi et al. w29x. Briefly, the cortices were digested with 0.02% papain and mechanically dissociated by trituration. The cells were plated on coverslips, which were coated with a 2-week old monolayer of astrocytes. The cell density was approximately 3 = 10 4 cellsrcm2 . The cells were maintained with Dulbecco’s modified Eagle’s medium ŽDMEM; Gibco. containing 10% heat-inactivated horse serum, 50 Urml penicillin, and 50 m grml streptomycin, in a 95% air–5% CO 2 , H 2 O-saturated atmosphere at 378C. Cells were treated with 2 m grml FdU Žfluorodeoxyuridine; Sigma. and 10 m grml Uridine ŽSigma. at 2 days in vitro ŽDIV. for 3 days, which suppressed further growth of glial cells derived from neuronal preparation. Half of the medium was replaced with fresh medium every 3 or 4 days. The cells were maintained for 12–25 days in vitro prior to bioelectric tests.
Spontaneous electrical activities were measured with amplifiers in the current clamp mode, and were digitized with an ArD converter ŽAPC204; Autonics, Japan. with a sampling time of 50 m s for bursting activity and 10 m s for synaptic response. Resting potential was measured without a correction for junction potential. To evoke an action potential ŽAP., a current pulse of 300–1000 pA for 10 or 20 ms was applied to either neuron, and the EPSP produced in the other neuron was recorded. In order to check electrical coupling, negative current pulses for 50 ms were injected to hyperpolarize one neuron by y80 mV from the resting potential. 2.3. Analysis of electrical recordings A spike burst consisted of more than one AP accompanying a large membrane depolarization. A burst latency of one episode was defined as the time between the onset of burst in each pair. The burst latency of each pair was determined by averaging the data from more than 10 episodes. The response latency between connective pairs of neurons was defined as the time between the peak of a differentiated AP in the stimulated neuron and the onset of the EPSP in the other neuron. Differentiation of the AP was achieved by computing a first derivative with respect to time. The response latency of each pair was determined by averaging more than five trials for each pair. The spatial distance between two neurons was measured as the distance between the center of each neuron’s somata. ŽFig. 7A.. Data are given as the mean " S.E.M. A P-value of less than 0.05 was considered significant.
2.2. Electrical measurements 2.4. Injection of dye or neurotracer Whole cell current clamp recordings were carried out from two neurons simultaneously by standard methods w14x or perforated patch clamp methods w1x with two patchclamp amplifiers Žmodel EPC-7; List Medical Electronic and Axopatch200A; Axon Instruments.. Both patch electrodes had a resistance of 5–10 M V with an internal solution containing Žin mM.: 145 KCl, 1 MgCl 2 P 6H 2 O, 1 ethylene glycol-bis Ž b-aminoethyl ether.-N, N, N X , N X-tetraacetic acid Ž EGTA . , 10 2-w4-Ž2-Hydroxyethyl.-1piperazinylx ethanesulfonic acid ŽHEPES. ŽpH 7.3., with or without Nystatin Ž0.3 mgrml; Sigma.. The bath solution contained Žin mM.: 135 NaCl, 5.4 KCl, 1.8 CaCl 2 P 2H 2 O, 0.8 MgCl 2 P 6H 2 O, 10 Glucose, 10 HEPES ŽpH 7.3.. The bath solution with a low Ca2q concentration was prepared by replacing MgCl 2 with the same molar of CaCl 2 . Electrical recordings were conducted at room temperature Ž20– 258C.. High-resistance seals Ž2–4 GV . were formed by application of a gentle negative pressure, and then the whole cell recording configuration was achieved by a further application of negative pressure or by waiting for a while when the pipet solution contained Nystatin.
To examine the possibility of gap junctional coupling, 1 mgrml Lucifer yellow ŽSigma. or 1–5 mgrml Neurobiotin ŽVector. was introduced into cells with a patch pipet. After 5–20 min in whole-cell recording mode, the pipet was carefully withdrawn to ensure a resealing of the membrane. After waiting for 30–60 min, stainings of Lucifer yellow were visualized by epifluorescence microscopy. To visualize Neurobiotin staining, cells were fixed in 4% paraformaldehyde in phosphate-buffer saline ŽPBS. for 20 min, incubated in 10 mgrml streptoavidin–conjugated horseradish peroxidase ŽVector. and 0.2% Triton X in PBS for 30 min, and reacted with diaminobenzidine and 0.004% hydrogen peroxide in PBS for 15 min. Stained neurons were observed by bright-field microscopy. 2.5. Physical sectioning by cutting the cell layer Cultured networks were disrupted by cutting the cell layer using needles or scalpels, and the electrical measurements were performed 30 min later. Physical sections were
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
about 100–300 m m in width and covered the full length of the coverslip. Cells in the physical section were completely removed ŽFig. 7A..
139
ade of synchronized bursting when the Ca2qrMg 2q ratio of the bathing solution was reduced is in good agreement with earlier studies of similar bursting activities recorded with pairs of extracellular microelectrodes in cultures of dissociated cerebral and spinal neurons w9x.
3. Results 3.1. Bursting in cultured neocortical neurons is synchronized
3.2. Functional synapses among synchronized bursting neurons.
Spontaneous electrical activities of two randomly chosen neurons in culture Ž12–25 DIV; mean " S.E.M.s 16.1 " 0.31 DIV. were simultaneously recorded using two patch-clamp amplifiers in current clamp mode. The distance between two neurons ranged from 37.5 m m to 600 m m Žmean " S.E.M.s 159.3 " 16.8 m m.. The mean resting potential was y61.4 " 0.7 mV Ž n s 118., and the mean size of somata was 18.1 " 0.3 m m Ž n s 98.. The neuronal pairs showed synchronously a periodic bursting accompanying a membrane depolarization. Typical results are shown in Fig. 1. The same response was observed in all pairs tested Ž59 pairs.. The frequency of bursts Ž0.45 " 0.06 Hz; n s 11 pairs. was close to the frequency of neuronal intracellular Ca2q oscillation, which was measured with an optical Ca2q analyzer w29x. No synchronized bursting was observed in bathing solution with a low Ca2q concentration Ž0.4 mM; Fig. 2A. and in bathing solution containing 50 m M D-AP5 ŽD-Žy.-2amino-5-phosphonopentanoic acid; NMDA receptor antagonist; Tocris. and 15 m M CNQX Ž6-Cyano-7nitroquinoxaline-2,3-dione; non-NMDA receptor antagonist; Tocris. ŽFig. 2B., indicating that the synchronized bursting requires synaptic transmission. These features of synchronized bursting were similar to those of neuronal intracellular Ca2q oscillation w31,34x. Moreover, the block-
To test whether synchronized bursting neurons have a synaptic connection, single presynaptic action potentials were evoked in either neuron after dual whole-cell voltage recordings were established. Recordings from 52 pairs of neurons were obtained, of which 17 pairs Ž33%. were unidirectionally coupled ŽFig. 3A. and three pairs Ž6%. were connected bidirectionally ŽFig. 3B.. We observed EPSPs but not IPSPs in these pairs. The mean of response latencies Žsee Section 2. was 4.05 " 0.61 ms Ž n s 23; Three bidirectional pairs were counted twice.. The average distance between two neurons was 150.5 " 24.5 m m Ž n s 20.. The relationship between the observed response latency and the spatial distance is shown in Fig. 4A. A response latency was not linearly correlated to the spatial distance between neurons, indicating that these response latencies might be attributed to a polysynaptic connection. Assuming that the AP conduction velocity in an axon is about 0.3 mrs w10,32x and the axons extend straight to neighboring neurons, the cumulative synaptic delay Ž t SD ; ms. was calculated by subtracting conduction time Ždistance m mr300. from observed response latency Žms.. A histogram of t SD is shown in Fig. 4B. There are peaks at 1.5–1.9 ms and 3–3.4 ms in this histogram Žasterisks.: the estimated synaptic delay in our culture system was 1.5–1.9
Fig. 1. Two neocortical neurons in culture synchronously bursting. Spontaneous electrical activities of two randomly chosen neurons in culture were simultaneously recorded using two patch-clamp amplifiers in current clamp mode. Both neurons showed synchronously periodic bursting accompanying a membrane depolarization. Resting potential: cell 1, y57 mV; cell 2, y58 mV.
140
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 2. Synchronized bursting required synaptic transmission. ŽA. The synchronized bursting of two neurons disappeared after perfusing with solution containing 0.4 mM Ca2q. Only APs and EPSPs were observed independently in each neuron. The synchronized bursting recovered in normal solution Ždata not shown.. Resting potential: cell 1, y55 mV; cell 2, y56 mV. ŽB. No synchronized bursting observed in the solution containing 50 m M D-AP5 and 15 m M CNQX. The synchronized bursting recovered in normal external solution Ždata not shown.. Resting potential: cell 1, y59 mV; cell 2, y59 mV.
ms Žthe mean of the first peak in the histogram; 1.77 ms.. In the histogram, the response latency of one pair is excluded, because two distinct latencies, 2.0 ms and 7.0 ms, were obtained for this pair. The neurons in this pair might be connected through two different pathways, strongly suggesting a polysynaptic connection of these neurons. 3.3. Burst latency has a linear relationship to distance We also analyzed the relationship between burst latency and spatial distance Ž37.5–600 m m apart; mean " S.E.M.s 159 " 24 m m.. The burst latency Žsee Section 2.
ranged from 0.95 ms to 17.6 ms Žmean " S.E.M.; 5.87 " 0.47 ms.. Fig. 5A shows that burst latency weakly correlated with distance. The correlation between burst latency and distance was significant, and the coefficient of Spearman’s rank correlation Ž R s . was 0.362 Žtied P value s 0.0065, z s 2.733.. The average of the burst latency in the connective pairs’ group Ž n s 20. was 4.36 " 0.52 ms, and in the non-connective pairs’ group Ž n s 32. was 6.67 " 0.73 ms. An analysis using Welch’s t-test showed that there was a significant difference between the two groups ŽFig. 5B; t s 2.578, P s 0.0129; F s 3.195., although the distance did not significantly differ between the groups. These results indicate that bursting occurred more coincidentally
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
141
Fig. 3. Synchronized bursting neurons have synaptic connections. ŽA. Unidirectional coupling. Left trace; Current injection to cell 1 to evoke action potential ŽAP. leads to EPSP in cell 2. Right trace; AP evoked by current injection in cell 2 did not lead to any response in cell 1. Resting potential: cell 1, y60 mV; cell 2, y69 mV. ŽB. bidirectional coupling. Both APs evoked by current injection in cell 1 Žleft trace. and cell 2 Žright trace. lead to EPSP in the other neuron. Resting potential: cell 1, y55 mV; cell 2, y52 mV. Current pulse: 20 ms.
in pairs with a direct synaptic connection than those without. Moreover, assuming the AP conduction velocity in an axon is 0.3 mrs and the axons extend straight to neighboring neurons, t BL Žms. was defined as the observed burst latency Žms. minus the conduction time Ždistance m mr300.. t BL was also plotted in a histogram ŽFig. 5C.. There are three peaks at 2 ms, 4 ms, and 6.5 ms. The first peak of t BL was close to the estimated synaptic delay described above Ž1.5–1.9 ms, Fig. 4B.. Our results suggest that the delay in bursting in each pair is attributable to the
difference in the number of synapses in the polysynaptic pathway. 3.4. Dye coupling was not obserÕed among neurons bursting synchronously To examine the contribution of gap junctions to synchronized bursting, the small molecule dye Lucifer yellow or neurotracer neurobiotin was introduced to bursting neurons through patch pipets. No dye coupling between neigh-
Fig. 4. Analysis of response latency. ŽA. Relationship between response latency and spatial distance. Data are mean Žclosed circles. and S.E.M. Žerror bars. in each pair Ž n s 23.. Dotted line indicates conduction time assuming the conduction velocity was 0.3 mrs. ŽB. Histogram of total synaptic delay Ž t SD . which was calculated as observed response latency minus conduction time in each pair Žsee text.. One datum Žwhich has long error bar in Fig. 4A. was excluded from this histogram as explained in the text Ž n s 22.. Asterisks indicate the peak of histogram.
142
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 5. Analysis of burst latency. ŽA. Relationship between burst latency and spatial distance. There was a weak but statistically significant correlation between burst latency and spatial distance. Statistical analysis was done using Spearman’s correlation coefficient by rank because the distribution of the data was far from normal Ž R s s 0.362, tied P - 0.01.. Data Žopen circles. show the burst latency of each pair Žn s 59.. ŽB. The average burst latency in the connected pairs’ group Žopen bar; n s 20. was significantly shorter than that in the unconnected pairs’ group Žhatched bar; n s 32.. ŽC. Histogram of burst latency minus conduction time Ž t BL . which was calculated as described in the text Ž n s 59.. Asterisks indicate the peak of histogram.
boring neurons was observed in any cell checked ŽLucifer yellow; n s 9, neurobiotin; n s 3. ŽFig. 6A.. No electrical coupling among bursting neurons was observed when either neuron was hyperpolarized by an injection of a negative current ŽFig. 6B.. These findings suggest that gap junctions among neurons were unlikely to contribute to a synchronized bursting. 3.5. Bursting of two neurons separated by physical sectioning was not synchronized To examine whether synchronized bursting requires direct cell–cell interaction, we performed the experiments after a physical sectioning by needles or fine scalpels ŽFig. 7A.. Fig. 7B shows a typical trace of records obtained in two neurons separated by physical sectioning. Spontaneous electrical activities of neurons on both sides after sectioning were not synchronous, with bursting occurring independently. Burst latency obtained in neurons 475–650 m m apart ranged from 43.9 ms to 233 ms Žmean " S.E.M.; 155.4 " 22.5 ms.. On the other hand, two neurons on the same side after sectioning were still synchronized Ždata not shown.. Compared with a non-sectioned pair between which the spatial distance is similar to that of a sectioned pair Žthe distance is longer than 400 m m, n s 5., the burst latency of a sectioned pair was apparently longer ŽFig. 7C,D.. There was a significant difference in burst latency between the two groups. ŽFig. 7D; non-sectioned pairs;
7.67 " 1.23 ms Ž n s 5. vs. sectioned pairs; 155.34 " 22.50 ms Ž n s 8., t s 6.554, P - 0.001, F s 532.6, Welch’s t-test., although there was no significant difference in distance between the two groups. The burst latency of the pair sectioned is considered to be underestimated, because we did not measure latencies longer than 400 ms. These results suggest that a synchronization of bursting requires direct cell–cell interactions such as chemical synapses but not diffusible factors.
4. Discussion 4.1. Synchronized bursting neurons haÕe functional synapses We have presented evidence of functional synapses in pairs of spontaneously synchronized bursting neurons in culture. Furthermore, we found direct Žmono andror poly. synaptic responses in about 39% of pairs, although we observed synchronized bursting in all pairs. This means that direct synaptic connections in all neurons are not necessary for synchronization of bursting, and that direct synaptic connections in one-third of neurons are sufficient for synchronization. The synaptic delay in our culture system was estimated to be 1.5–1.9 ms Žthe mean of the first peak in the histogram; 1.77 ms.. This is much greater than those
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
143
Fig. 6. Gap junctions are unlikely to contribute to neuronal bursting. ŽA. Neurobiotin introduced to one bursting neuron did not spread to other neurons Žbar: 50 m m.. ŽB. There was no electrical coupling between bursting neurons. Negative current injection to hyperpolarize either neuron did not cause a change in the membrane potential of the other neuron.
reported in other studies in vivo Ž0.62 " 0.18 ms in cat visual cortex w39x, 0.3 ms in spinal motoneurons w2,11x, 0.3 ms in rat neocortical slices w21x.. We assumed the AP conduction velocity ŽC.V.. to be 0.3 mrs w10,32x and that axons extend in a straight line to the neurons, and calculated the synaptic delay. The C.V.s of APs in the neurites of cultured neurons may have been considerably lower than 0.3 mrs w10x. For example, if we used a C.V. value of 0.2 mrs, the estimated synaptic delay derived from the mean of the first peak in the histogram was 1.44 ms. Also, the C.V.s of APs in our study may have been attenuated by the lower Žroom. temperature at which these tests were carried out. Correction for temperature by assuming that Q10 is 2.0 w17x and thus C.V. is 0.15 mrs leads to an estimated synaptic delay of 1.2 ms. However, C.V. in our culture system can not be smaller than 0.1 mrs since that would result in conduction times that are in many cases longer than the observed response latencies. Furthermore, actual axons do not always extend in a straight line and the actual conduction time might be longer than our estimate.
If we calculated the conduction time by assuming an axon trajectory 1.5 times the distance of a straight line Žcorresponding to a half circle between the two neurons Žnote that pr2 is about 1.5.., the estimated synaptic delay was 1.44 ms. By not considering temperature and a winding axon trajectory the synaptic delay in our data may have been overestimated, but neither of these factors should affect our results in a significant way. Moreover, the greater estimates of synaptic delay may also be explained by the immaturity of synapses in developing neurons andror by an involvement of dendritic delay, which may be unexpectedly longer than that in vivo Ž0.3 ms in rat neocortical slices w21x.. In cerebral slice preparations the probability of synaptic responses was reported to be 0.1 or lower w21x, while our probability of 0.36 is higher. Mennerlick et al. w23x reported that synaptic properties of neurons grown in microculture are different from those in mass culture and that synaptic probability in microculture Ž0.85. is higher than that in mass culture Ž0.42., which is similar to our data.
144
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 7. Synchronized bursting requires direct cell–cell contact. ŽA. Experimental arrangement. ŽB. The neurons separated by cutting the cell layer showed independent but not synchronous bursting. ŽC. Burst latencies of sectioned pairs Žclosed circles. were much longer than those of unsectioned pairs Žopen circles.. Note the scale of the ordinate is different from that of Fig. 5A. ŽD. The average of burst latencies in sectioned pairs Ž n s 8. was significantly longer than that in unsectioned pairs Ž n s 5; Only those pairs between which the spatial distance was longer than 400 m m were used in this analysis..
These results suggest that synaptic probability decreases as neuronal density increases andror when the overall system becomes organized. 4.2. Synchronized bursting of neurons may be attributed to polysynaptic connections Even neurons that did not have a direct synaptic connection, exhibited synchronous bursting at a burst latency of 5.87 " 0.47 ms. These results are consistent with our previous finding that most of the neuronal intracellular Ca2q, measured by optical imaging, oscillated synchronously w29x. The burst latency Žmean " S.E.M.; 5.87 " 0.47 ms. is much greater than the estimated synaptic delay Ž1.5–1.9 ms., suggesting that synchronized bursting is attributable to polysynaptic connections. That the average burst latency in the connected pairs’ group was shorter than that in the unconnected pairs’ group ŽFig. 5B. is consistent with this hypothesis. Also, disappearance of synchronized bursting in bathing solution containing D-AP5 and CNQX ŽFig. 2B. strongly supports the contention that synchronized bursting requires synaptic connections in the network. The contribution of gap junctions and diffusible factors to neuronal synchronized bursting does not seem to be significant even if they existed, because the small molecule dye Lucifer yellow or the neurotracer neurobiotin introduced to one neuron did not stain neighboring neurons ŽFig. 6A. and
physical sectioning blocked synchronized bursting ŽFig. 7B,C,D.. Similar synchronized neuronal activities in neocortical or other CNS explants and in dissociated cultures have been reported w8,9,12,13,19,20,30x. The features of these synchronized activities are consistent with those in our neocortical culture system. First, no synchronized bursting was observed when the Ca2qrMg 2q ratio in the bathing solution was reduced w9,20x. Second, the frequency of spontaneous depolarizations was markedly reduced by the NMDA receptor antagonist, AP5 w12,20x. Third, synchronous discharge was observed in two neurons that showed no evidence of direct synaptic connection w12x. Synchronized neuronal activities may be intrinsic properties of neurons which connect via a polysynaptic network. Traub et al. w40,41x extended their elaborate model of the epileptic hippocampal slice induced by low Mg 2q, and in their computer simulation they included the contribution of polysynaptic connections. Their simulation suggests that no synchronized activity occurs when wMg 2q x is 2.0 mM. In our experiments, no synchronized bursting was observed in bathing solution containing AP5 and CNQX or in bathing solution with a low Ca2q Ž0.4 mM. and a wMg 2q x of 2.2 mM ŽFig. 2.. The synchronized bursting we observed has similar characteristics to these previous experimental studies, and may possibly be explained by Traub’s model. On the other hand, many investigators have suggested that the synchronous activity of neurons is due to electrical
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
interactions between dendrites Žin locus coeruleus neurons w16x, and in hippocampal slices w38x.. Furthermore, a role for gap junctions or dye coupling is suggested in columnar domains in developing neocortical slices w33,43x and in rat neocortical interneurons w4x. It is possible that generator cells Žsuch as interneurons in hippocampus. are present in our culture system and are connected to each other via gap junctions, thus mediating synchronized bursting of larger neurons, although the slice preparations used by the above authors differed from our culture system and dye-coupling in neonatal cortical slices is regulated developmentally w33,43x. Muller et al. w24x recently speculated that the ratio of inhibitory and excitatory neurons affects the spontaneous activity in cultured hypothalamic neuronal networks. In our culture system, excitatory and inhibitory neurons, such as glutaminergic and GABAergic neurons should exist, and inhibitory connections may affect the synchronized bursting activity in neocortical neurons as well. We did not observe an IPSP response, possibly because of the experimental conditions. We can not rule out the presence of IPSPs in our system and their contribution to synchronized bursting activity in other neurons via the neural pathway. Our results indicate that synchronized bursting of neurons in culture requires synaptic connection and that gap junctions andror diffusible factors by themselves cannot explain the phenomenon of synchronized bursting. However, further study is required to clarify completely the mechanism of synchronized bursting between neurons. 4.3. Comparison among other measurements The frequency of synchronized bursting is close to that of intracellular Ca2q oscillation in the same culture system w29x. Simultaneous measurement of electrical bursting and Ca2q oscillation suggested that these two phenomena are coupled via a polysynaptic network w34x. Comparison with the data for intracellular Ca2q oscillation shows the synchronization observed with two patch clamp electrodes to be accurate with regard to time resolution Ž50–100 m s sampling time., and that the synaptic delay of 1–2 ms was also accurate. Extracellular recordings using an array electrode w19x are useful for observing the overall synchronization of many neurons and provide better time resolution than optical methods, but they give no information on the shape of the AP and bursting. Analysis using the intracellular recording of two neurons give better time resolution and much more information about synchronized bursting for further analyses, such as the shape of the burst and synaptic responses. 4.4. Neuron–glia coculture may be a good model of network formation Our culture system may be too complicated to analyze simple synaptic responses, but may be suitable for a study
145
of the neural network, which is important for elucidating complicated brain functions. Since our culture model is similar to the epileptiform of slice preparations, it may also be suitable for studying epileptogenesis.
Acknowledgements We are grateful to Dr. Muneyuki Ito of the Department of Physiology, Institute for Developmental Research, Aichi Human Service Center, for helpful discussions. We thank Dr. Taiji Kato of the Department of Bioregulation Research, Nagoya City Univ. Med. Sch. and Dr. Shunichi Yamagishi of NIPS for their kind support of this work. We also thank the members of the Lab. of Membrane Biology at NIPS for their technical support. This research was supported in part by the Aichi Cancer Research Foundation.
References w1x N. Akaike, N. Harata, Nystatin perforated patch recording and its applications to analyses of intracellular mechanisms, Jpn. J. Physiol. 44 Ž1994. 433–473. w2x T. Araki, J.C. Eccles, M. Ito, Correlation of the inhibitory postsynaptic potential of motoneurones with latency and time course of inhibition of monosynaptic reflexes, J. Physiol. ŽLondon. 154 Ž1960. 354–377. w3x T. Bal, D.A. McCormick, Synchronized oscillations in the inferior olive are controlled by the hyperpolarization-activated cation current Ih, J. Neurophysiol. 77 Ž1997. 3145–3156. w4x L.S. Benardo, Recruitment of GABAergic inhibition and synchronization of inhibitory interneurons in rat neocortex, J. Neurophysiol. 77 Ž1997. 3134–3144. w5x Y. Ben Ari, E. Cherubini, R. Corradetti, J.L. Gaiarsa, Giant synaptic potentials in immature rat CA3 hippocampal neurons, J. Physiol. ŽLondon. 416 Ž1989. 303–325. w6x A. Bragin, G. Jando, Z. Nadasdy, J. Hetke, K. Wise, G. Buzsaki, Gamma Ž40–100 Hz. oscillation in the hippocampus of the behaving rat, J. Neurosci. 15 Ž1995. 47–60. w7x G. Buzsaki, Z. Horvath, R. Urioste, J. Hetke, K. Wise, Highfrequency network oscillation in the hippocampus, Science 256 Ž1992. 1025–1027. w8x B.W. Connors, M.J. Gutnick, D.A. Prince, Electrophysiological properties of neocortical neurons in vitro, J. Neurophysiol. 48 Ž1982. 1302–1320. w9x S.M. Crain, M.B. Bornstein, Organotypic bioelectric activity in cultured reaggregates of dissociated rodent brain cells, Science 176 Ž1972. 182–184. w10x S.M. Crain, Neurophysiologic Studies in Tissue Culture. Raven Press, New York, 1976, pp. 43–152. w11x J.C. Eccles ŽEd.. The Physiology of Synapses. Springer, Berlin, 1964, pp. 49–53 and pp. 152–172. w12x M.J. Gutnick, B. Wolfson, F. Baldino Jr., Synchronized neuronal activities in neocortical explant cultures, Exp. Brain Res. 76 Ž1989. 131–140. w13x M.J. Gutnick, B.W. Connors, D.A. Prince, Mechanisms of neocortical epileptogenesis in vitro, J. Neurophysiol. 48 Ž1982. 1321–1335. w14x O.P. Hamill, A. Marty, E. Neher, B. Sakmann, F.J. Sigworth, Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches, Pflug. Arch. 391 Ž1981. 85–100.
146
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
w15x M. Ichikawa, K. Muramoto, K. Kobayashi, M. Kawahara, Y. Kuroda, Formation and maturation of synapses in primary cultures of rat cerebral cortical cells: an electron microscopic study, Neurosci. Res. 16 Ž1993. 95–103. w16x M. Ishimatsu, J.T. Williams, Synchronous activity in locus coeruleus results from dendritic interactions in pericoerulear regions, J. Neurosci. 16 Ž1996. 5196–5204. w17x F. Kukita, Properties of sodium and potassium channels of the squid giant axon far below 08C, J. Membr. Biol. 68 Ž1982. 151–160. w18x Y. Kuroda, M. Ichikawa, K. Muramoto, K. Kobayashi, Y. Matsuda, A. Ogura, Y. Kudo, Block of synapse formation between cerebral cortical neurons by a protein kinase inhibitor, Neurosci. Lett. 135 Ž1992. 255–258. w19x E. Maeda, H.P.C. Robinson, A. Kawana, The mechanisms of generation and propagation of synchronized bursting in developing networks of cortical neurons, J. Neurosci. 15 Ž1995. 6834–6845. w20x E. Maeda, H.P.C. Robinson, Y. Kuroda, A. Kawana, Transient changes of network activity in cultured cortical neurons induced by Mg 2q and APV Žabstract., Neurosci. Res. suppl. 18 Ž1993. S233. w21x H. Markram, J. Lubke, M. Frotscher, A. Roth, B. Sakmann, Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex, J. Physiol. ŽLondon. 500 Ž1997. 409–440. w22x M. Meister, R.O.L. Wong, D.A. Baylor, C.J. Shatz, Synchronous bursts of action potentials in ganglion cells of the developing mammalian retina, Science 252 Ž1991. 939–943. w23x S. Mennerick, J. Que, A. Benz, C.F. Zorumski, Passive and synaptic properties of hippocampal neurons grown in microcultures and in mass cultures, J. Neurophysiol. 73 Ž1995. 320–332. w24x T.H. Muller, D. Swandulla, H.U. Zeilhofer, Synaptic connectivity in cultured hypothalamic neuronal networks, J. Neurophysiol. 77 Ž1997. 3208–3225. w25x K. Muramoto, M. Ichikawa, M. Kawahara, K. Kobayashi, Y. Kuroda, Frequency of synchronous oscillations of neuronal activity increases during development and is correlated to the number of synapses in cultured cortical neuron networks, Neurosci. Lett. 163 Ž1993. 163– 165. w26x T.H. Murphy, L.A. Blatter, W.G. Wier, J.M. Baraban, Spontaneous synchronous synaptic calcium transients in cultured cortical neurons, J. Neurosci. 12 Ž1992. 4834–4845. w27x V.N. Murthy, E.E. Fetz, Oscillatory activity in sensorimotor cortex of awake monkeys: synchronization of local field potentials and relation to behavior, J. Neurophysiol. 76 Ž1996. 3949–3967. w28x V.N. Murthy, E.E. Fetz, Synchronization of neurons during local field potential oscillations in sensorimotor cortex of awake monkeys, J. Neurophysiol. 76 Ž1996. 3968–3982. w29x K. Nakanishi, Y. Okouchi, T. Ueki, K. Asai, I. Isobe, Y. Eksioglu,
w30x
w31x
w32x
w33x
w34x
w35x w36x
w37x w38x
w39x
w40x
w41x
w42x
w43x
T. Kato, Y. Hasegawa, Y. Kuroda, Astrocytic contribution to functional synapse formation estimated by spontaneous neuronal intracellular Ca2q oscillations, Brain Res. 659 Ž1994. 169–178. K. Nakanishi, F. Kukita, Functional synapses in synchronized bursting neurons in culture, Neurosci. Res. Suppl. 21 Ž1997. S64, Žabstract.. A. Ogura, T. Iijima, T. Amano, Y. Kudo, Optical monitoring of excitatory synaptic activity between cultured hippocampal neurons by a multi-site Ca2q fluorometry, Neurosci. Lett. 78 Ž1987. 69–74. H.D. Patton, Special properties of nerve trunks and tracts, in: T.C. Ruch, J.F. Fulton ŽEds.., Medical Physiology and Biophysics, Saunders, Philadelphia, 1960, pp. 66–95. A. Peinado, R. Yuste, L.C. Katz, Extensive dye coupling between rat neocortical neurons during the period of circuit formation, Neuron 10 Ž1993. 103–114. H.P.C. Robinson, M. Kawahara, Y. Jimbo, K. Torimitsu, Y. Kuroda, A. Kawana, Periodic synchronized bursting and intracellular calcium transients elicited by low magnesium in cultured cortical neurons, J. Neurophysiol. 70 Ž1993. 1606–1616. M. Segal, Rat hippocampal neurons in culture: responses to electrical and chemical stimuli, J. Neurophysiol. 50 Ž1983. 1249–1264. M. Segal, J.L. Barker, Rat hippocampal neurons in culture: voltageclamp analysis of inhibitory synaptic connections, J. Neurophysiol. 52 Ž1984. 469–487. C.J. Shatz, Impulse activity and the patterning of connections during CNS development, Neuron 5 Ž1990. 745–756. F. Strata, M. Atzori, M. Molnar, G. Ugolini, F. Tempia, E. Cherubini, A pacemaker current in dye-coupled hilar interneurons contributes to the generation of giant GABAergic potentials in developing hippocampus, J. Neurosci. 17 Ž1997. 1435–1446. K. Toyama, K. Matsunami, T. Ohno, S. Tokashiki, An intracellular study of neuronal organization in the visual cortex, Exp. Brain Res. 21 Ž1974. 45–66. R.D. Traub, R. Miles, J.R. Jefferys, Synaptic and intrinsic conductances shape picrotoxin-induced synchronized after-discharges in the guinea-pig hippocampal slice, J. Physiol. ŽLondon. 461 Ž1993. 525–547. R.D. Traub, J.G.R. Jefferys, M.A. Whttington, Enhanced NMDA conductance can account for epileptiform activity induced by low Mg 2q in the rat hippocampal slice, J. Physiol. ŽLondon. 478 Ž1994. 379–393. R.O.L. Wong, M. Meister, C.J. Shatz, Transient period of correlated bursting activity during development of mammalian retina, Neuron 11 Ž1993. 923–938. R. Yuste, D.A. Nelson, W.W. Rubin, L.C. Katz, Neuronal domains in developing neocortex: mechanisms of coactivation, Neuron 14 Ž1995. 7–17.
Research report
Functional synapses in synchronized bursting of neocortical neurons in culture Keiko Nakanishi a
a, )
, Fumio Kukita
b
Dept. of Physiology, Institute for DeÕelopmental Research, Aichi Human SerÕice Center, Kasugai, Aichi 480-0392, Japan b Laboratory of Membrane Biology, National Institute of Physiological Sciences, Okazaki, Aichi 444-8585, Japan Accepted 10 March 1998
Abstract Spontaneous electrical activities in pairs of neocortical neurons in culture were simultaneously recorded using a whole cell current clamp technique. Synchronous bursting activities were observed in all 59 pairs tested. In 52 pairs of neurons electrically stimulated, EPSPs were recorded in 20 pairs Ž39%., among which 3 pairs Ž6%. showed bidirectional coupling. The response latency observed was 4.05 " 0.61 ms Žmean" S.E.M... The synaptic delay was estimated at 1.5–1.9 ms, suggesting the response latency is derived from a polysynaptic connection. The burst latency which was defined as the time difference of the onset of bursting in each neuron was 5.87 " 0.47 ms Žmean " S.E.M.., and was weakly correlated with the spatial distance between the neurons Ž37.5–600 m m apart. Ž R s s 0.362, tied P value s 0.0065.. No synchronized bursting was observed in bathing solution with a low Ca2q concentration Ž0.4 mM. or in bathing solution containing 50 m M D-AP5 and 15 m M CNQX. No dye-coupling between bursting neurons was observed on injection of the small molecule dye Lucifer yellow or the neurotracer neurobiotin. Disrupting neural connections completely by cutting the cell layer, caused disappearance of synchronized bursting with each neuron bursting independently. In conclusion, these results are consistent with the hypothesis that synchronized bursting in cultured neocortical neurons is attributed to connections by way of several synapses rather than by way of gap junctions andror diffusible factors. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Functional synapse; Synchronized bursting; Neocortical neuron
1. Introduction Highly correlated, spontaneous neuronal bursting activities have been reported in many regions of developing mammalian brain, such as the visual system w22,37,42x, hippocampus w5,35,36,38x, locus coeruleus w16x, inferior olive w3x and neocortex w12,13x, suggesting important roles for them in the signal processing of central nervous systems. Furthermore, high-frequency network oscillations have been observed both in slice preparations w7x and in living animals such as rats and monkeys w6,27,28x, indicating that synchronized electrical activities in neurons are fundamental for integrated brain functions such as memory, learning, and recognition. However, the mechanism reported for oscillatory activities is complex because it is different in different systems and in different development stages.
)
Corresponding author. Fax: [email protected]
q 81-568-88-0829;
E-mail:
0006-8993r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 0 6 - 8 9 9 3 Ž 9 8 . 0 0 2 8 3 - 2
There have also been reports of synchronous intracellular Ca2q oscillations in hippocampal w31x and neocortical culture systems w18,25,26,29x. Furthermore, simultaneous measurement of intracellular Ca2q and electrical activity revealed that neuronal bursts are generated periodically and accompanied by slower Ca2q transients w34x, indicating a coupling of neuronal synchronized bursting and an intracellular signal transduction. Neuronal bursting and intracellular Ca2q oscillation in culture are believed to be synchronized via a neural network of synapses, since these phenomena are attenuated by the NMDA receptor antagonist APV w12,34x. Using electron microscopy, Ichikawa et al. w15x found a synapse formation in cultured neocortical neurons showing a synchronized intracellular Ca2q oscillation. However, it has not been demonstrated whether the functional synapses in these neurons are actually working. To demonstrate that synchronized bursting neurons have synaptic connections with each other, we recorded electrical activities simultaneously in a pair of neurons with a patch clamp method. The time difference between the
138
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
onset of bursting in each neuron Žburst latency. was analyzed and compared with a synaptic delay estimated from evoked synaptic response. Our results suggest that synchronized bursting can be attributed to connections by way of several synapses rather than by way of gap junctions andror diffusible factors.
2. Materials and methods 2.1. Cell culture Neocortical cells were prepared from the cerebral cortex of embryonic day 16–18 Wistar rats and cultured as described in the work of Nakanishi et al. w29x. Briefly, the cortices were digested with 0.02% papain and mechanically dissociated by trituration. The cells were plated on coverslips, which were coated with a 2-week old monolayer of astrocytes. The cell density was approximately 3 = 10 4 cellsrcm2 . The cells were maintained with Dulbecco’s modified Eagle’s medium ŽDMEM; Gibco. containing 10% heat-inactivated horse serum, 50 Urml penicillin, and 50 m grml streptomycin, in a 95% air–5% CO 2 , H 2 O-saturated atmosphere at 378C. Cells were treated with 2 m grml FdU Žfluorodeoxyuridine; Sigma. and 10 m grml Uridine ŽSigma. at 2 days in vitro ŽDIV. for 3 days, which suppressed further growth of glial cells derived from neuronal preparation. Half of the medium was replaced with fresh medium every 3 or 4 days. The cells were maintained for 12–25 days in vitro prior to bioelectric tests.
Spontaneous electrical activities were measured with amplifiers in the current clamp mode, and were digitized with an ArD converter ŽAPC204; Autonics, Japan. with a sampling time of 50 m s for bursting activity and 10 m s for synaptic response. Resting potential was measured without a correction for junction potential. To evoke an action potential ŽAP., a current pulse of 300–1000 pA for 10 or 20 ms was applied to either neuron, and the EPSP produced in the other neuron was recorded. In order to check electrical coupling, negative current pulses for 50 ms were injected to hyperpolarize one neuron by y80 mV from the resting potential. 2.3. Analysis of electrical recordings A spike burst consisted of more than one AP accompanying a large membrane depolarization. A burst latency of one episode was defined as the time between the onset of burst in each pair. The burst latency of each pair was determined by averaging the data from more than 10 episodes. The response latency between connective pairs of neurons was defined as the time between the peak of a differentiated AP in the stimulated neuron and the onset of the EPSP in the other neuron. Differentiation of the AP was achieved by computing a first derivative with respect to time. The response latency of each pair was determined by averaging more than five trials for each pair. The spatial distance between two neurons was measured as the distance between the center of each neuron’s somata. ŽFig. 7A.. Data are given as the mean " S.E.M. A P-value of less than 0.05 was considered significant.
2.2. Electrical measurements 2.4. Injection of dye or neurotracer Whole cell current clamp recordings were carried out from two neurons simultaneously by standard methods w14x or perforated patch clamp methods w1x with two patchclamp amplifiers Žmodel EPC-7; List Medical Electronic and Axopatch200A; Axon Instruments.. Both patch electrodes had a resistance of 5–10 M V with an internal solution containing Žin mM.: 145 KCl, 1 MgCl 2 P 6H 2 O, 1 ethylene glycol-bis Ž b-aminoethyl ether.-N, N, N X , N X-tetraacetic acid Ž EGTA . , 10 2-w4-Ž2-Hydroxyethyl.-1piperazinylx ethanesulfonic acid ŽHEPES. ŽpH 7.3., with or without Nystatin Ž0.3 mgrml; Sigma.. The bath solution contained Žin mM.: 135 NaCl, 5.4 KCl, 1.8 CaCl 2 P 2H 2 O, 0.8 MgCl 2 P 6H 2 O, 10 Glucose, 10 HEPES ŽpH 7.3.. The bath solution with a low Ca2q concentration was prepared by replacing MgCl 2 with the same molar of CaCl 2 . Electrical recordings were conducted at room temperature Ž20– 258C.. High-resistance seals Ž2–4 GV . were formed by application of a gentle negative pressure, and then the whole cell recording configuration was achieved by a further application of negative pressure or by waiting for a while when the pipet solution contained Nystatin.
To examine the possibility of gap junctional coupling, 1 mgrml Lucifer yellow ŽSigma. or 1–5 mgrml Neurobiotin ŽVector. was introduced into cells with a patch pipet. After 5–20 min in whole-cell recording mode, the pipet was carefully withdrawn to ensure a resealing of the membrane. After waiting for 30–60 min, stainings of Lucifer yellow were visualized by epifluorescence microscopy. To visualize Neurobiotin staining, cells were fixed in 4% paraformaldehyde in phosphate-buffer saline ŽPBS. for 20 min, incubated in 10 mgrml streptoavidin–conjugated horseradish peroxidase ŽVector. and 0.2% Triton X in PBS for 30 min, and reacted with diaminobenzidine and 0.004% hydrogen peroxide in PBS for 15 min. Stained neurons were observed by bright-field microscopy. 2.5. Physical sectioning by cutting the cell layer Cultured networks were disrupted by cutting the cell layer using needles or scalpels, and the electrical measurements were performed 30 min later. Physical sections were
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
about 100–300 m m in width and covered the full length of the coverslip. Cells in the physical section were completely removed ŽFig. 7A..
139
ade of synchronized bursting when the Ca2qrMg 2q ratio of the bathing solution was reduced is in good agreement with earlier studies of similar bursting activities recorded with pairs of extracellular microelectrodes in cultures of dissociated cerebral and spinal neurons w9x.
3. Results 3.1. Bursting in cultured neocortical neurons is synchronized
3.2. Functional synapses among synchronized bursting neurons.
Spontaneous electrical activities of two randomly chosen neurons in culture Ž12–25 DIV; mean " S.E.M.s 16.1 " 0.31 DIV. were simultaneously recorded using two patch-clamp amplifiers in current clamp mode. The distance between two neurons ranged from 37.5 m m to 600 m m Žmean " S.E.M.s 159.3 " 16.8 m m.. The mean resting potential was y61.4 " 0.7 mV Ž n s 118., and the mean size of somata was 18.1 " 0.3 m m Ž n s 98.. The neuronal pairs showed synchronously a periodic bursting accompanying a membrane depolarization. Typical results are shown in Fig. 1. The same response was observed in all pairs tested Ž59 pairs.. The frequency of bursts Ž0.45 " 0.06 Hz; n s 11 pairs. was close to the frequency of neuronal intracellular Ca2q oscillation, which was measured with an optical Ca2q analyzer w29x. No synchronized bursting was observed in bathing solution with a low Ca2q concentration Ž0.4 mM; Fig. 2A. and in bathing solution containing 50 m M D-AP5 ŽD-Žy.-2amino-5-phosphonopentanoic acid; NMDA receptor antagonist; Tocris. and 15 m M CNQX Ž6-Cyano-7nitroquinoxaline-2,3-dione; non-NMDA receptor antagonist; Tocris. ŽFig. 2B., indicating that the synchronized bursting requires synaptic transmission. These features of synchronized bursting were similar to those of neuronal intracellular Ca2q oscillation w31,34x. Moreover, the block-
To test whether synchronized bursting neurons have a synaptic connection, single presynaptic action potentials were evoked in either neuron after dual whole-cell voltage recordings were established. Recordings from 52 pairs of neurons were obtained, of which 17 pairs Ž33%. were unidirectionally coupled ŽFig. 3A. and three pairs Ž6%. were connected bidirectionally ŽFig. 3B.. We observed EPSPs but not IPSPs in these pairs. The mean of response latencies Žsee Section 2. was 4.05 " 0.61 ms Ž n s 23; Three bidirectional pairs were counted twice.. The average distance between two neurons was 150.5 " 24.5 m m Ž n s 20.. The relationship between the observed response latency and the spatial distance is shown in Fig. 4A. A response latency was not linearly correlated to the spatial distance between neurons, indicating that these response latencies might be attributed to a polysynaptic connection. Assuming that the AP conduction velocity in an axon is about 0.3 mrs w10,32x and the axons extend straight to neighboring neurons, the cumulative synaptic delay Ž t SD ; ms. was calculated by subtracting conduction time Ždistance m mr300. from observed response latency Žms.. A histogram of t SD is shown in Fig. 4B. There are peaks at 1.5–1.9 ms and 3–3.4 ms in this histogram Žasterisks.: the estimated synaptic delay in our culture system was 1.5–1.9
Fig. 1. Two neocortical neurons in culture synchronously bursting. Spontaneous electrical activities of two randomly chosen neurons in culture were simultaneously recorded using two patch-clamp amplifiers in current clamp mode. Both neurons showed synchronously periodic bursting accompanying a membrane depolarization. Resting potential: cell 1, y57 mV; cell 2, y58 mV.
140
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 2. Synchronized bursting required synaptic transmission. ŽA. The synchronized bursting of two neurons disappeared after perfusing with solution containing 0.4 mM Ca2q. Only APs and EPSPs were observed independently in each neuron. The synchronized bursting recovered in normal solution Ždata not shown.. Resting potential: cell 1, y55 mV; cell 2, y56 mV. ŽB. No synchronized bursting observed in the solution containing 50 m M D-AP5 and 15 m M CNQX. The synchronized bursting recovered in normal external solution Ždata not shown.. Resting potential: cell 1, y59 mV; cell 2, y59 mV.
ms Žthe mean of the first peak in the histogram; 1.77 ms.. In the histogram, the response latency of one pair is excluded, because two distinct latencies, 2.0 ms and 7.0 ms, were obtained for this pair. The neurons in this pair might be connected through two different pathways, strongly suggesting a polysynaptic connection of these neurons. 3.3. Burst latency has a linear relationship to distance We also analyzed the relationship between burst latency and spatial distance Ž37.5–600 m m apart; mean " S.E.M.s 159 " 24 m m.. The burst latency Žsee Section 2.
ranged from 0.95 ms to 17.6 ms Žmean " S.E.M.; 5.87 " 0.47 ms.. Fig. 5A shows that burst latency weakly correlated with distance. The correlation between burst latency and distance was significant, and the coefficient of Spearman’s rank correlation Ž R s . was 0.362 Žtied P value s 0.0065, z s 2.733.. The average of the burst latency in the connective pairs’ group Ž n s 20. was 4.36 " 0.52 ms, and in the non-connective pairs’ group Ž n s 32. was 6.67 " 0.73 ms. An analysis using Welch’s t-test showed that there was a significant difference between the two groups ŽFig. 5B; t s 2.578, P s 0.0129; F s 3.195., although the distance did not significantly differ between the groups. These results indicate that bursting occurred more coincidentally
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
141
Fig. 3. Synchronized bursting neurons have synaptic connections. ŽA. Unidirectional coupling. Left trace; Current injection to cell 1 to evoke action potential ŽAP. leads to EPSP in cell 2. Right trace; AP evoked by current injection in cell 2 did not lead to any response in cell 1. Resting potential: cell 1, y60 mV; cell 2, y69 mV. ŽB. bidirectional coupling. Both APs evoked by current injection in cell 1 Žleft trace. and cell 2 Žright trace. lead to EPSP in the other neuron. Resting potential: cell 1, y55 mV; cell 2, y52 mV. Current pulse: 20 ms.
in pairs with a direct synaptic connection than those without. Moreover, assuming the AP conduction velocity in an axon is 0.3 mrs and the axons extend straight to neighboring neurons, t BL Žms. was defined as the observed burst latency Žms. minus the conduction time Ždistance m mr300.. t BL was also plotted in a histogram ŽFig. 5C.. There are three peaks at 2 ms, 4 ms, and 6.5 ms. The first peak of t BL was close to the estimated synaptic delay described above Ž1.5–1.9 ms, Fig. 4B.. Our results suggest that the delay in bursting in each pair is attributable to the
difference in the number of synapses in the polysynaptic pathway. 3.4. Dye coupling was not obserÕed among neurons bursting synchronously To examine the contribution of gap junctions to synchronized bursting, the small molecule dye Lucifer yellow or neurotracer neurobiotin was introduced to bursting neurons through patch pipets. No dye coupling between neigh-
Fig. 4. Analysis of response latency. ŽA. Relationship between response latency and spatial distance. Data are mean Žclosed circles. and S.E.M. Žerror bars. in each pair Ž n s 23.. Dotted line indicates conduction time assuming the conduction velocity was 0.3 mrs. ŽB. Histogram of total synaptic delay Ž t SD . which was calculated as observed response latency minus conduction time in each pair Žsee text.. One datum Žwhich has long error bar in Fig. 4A. was excluded from this histogram as explained in the text Ž n s 22.. Asterisks indicate the peak of histogram.
142
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 5. Analysis of burst latency. ŽA. Relationship between burst latency and spatial distance. There was a weak but statistically significant correlation between burst latency and spatial distance. Statistical analysis was done using Spearman’s correlation coefficient by rank because the distribution of the data was far from normal Ž R s s 0.362, tied P - 0.01.. Data Žopen circles. show the burst latency of each pair Žn s 59.. ŽB. The average burst latency in the connected pairs’ group Žopen bar; n s 20. was significantly shorter than that in the unconnected pairs’ group Žhatched bar; n s 32.. ŽC. Histogram of burst latency minus conduction time Ž t BL . which was calculated as described in the text Ž n s 59.. Asterisks indicate the peak of histogram.
boring neurons was observed in any cell checked ŽLucifer yellow; n s 9, neurobiotin; n s 3. ŽFig. 6A.. No electrical coupling among bursting neurons was observed when either neuron was hyperpolarized by an injection of a negative current ŽFig. 6B.. These findings suggest that gap junctions among neurons were unlikely to contribute to a synchronized bursting. 3.5. Bursting of two neurons separated by physical sectioning was not synchronized To examine whether synchronized bursting requires direct cell–cell interaction, we performed the experiments after a physical sectioning by needles or fine scalpels ŽFig. 7A.. Fig. 7B shows a typical trace of records obtained in two neurons separated by physical sectioning. Spontaneous electrical activities of neurons on both sides after sectioning were not synchronous, with bursting occurring independently. Burst latency obtained in neurons 475–650 m m apart ranged from 43.9 ms to 233 ms Žmean " S.E.M.; 155.4 " 22.5 ms.. On the other hand, two neurons on the same side after sectioning were still synchronized Ždata not shown.. Compared with a non-sectioned pair between which the spatial distance is similar to that of a sectioned pair Žthe distance is longer than 400 m m, n s 5., the burst latency of a sectioned pair was apparently longer ŽFig. 7C,D.. There was a significant difference in burst latency between the two groups. ŽFig. 7D; non-sectioned pairs;
7.67 " 1.23 ms Ž n s 5. vs. sectioned pairs; 155.34 " 22.50 ms Ž n s 8., t s 6.554, P - 0.001, F s 532.6, Welch’s t-test., although there was no significant difference in distance between the two groups. The burst latency of the pair sectioned is considered to be underestimated, because we did not measure latencies longer than 400 ms. These results suggest that a synchronization of bursting requires direct cell–cell interactions such as chemical synapses but not diffusible factors.
4. Discussion 4.1. Synchronized bursting neurons haÕe functional synapses We have presented evidence of functional synapses in pairs of spontaneously synchronized bursting neurons in culture. Furthermore, we found direct Žmono andror poly. synaptic responses in about 39% of pairs, although we observed synchronized bursting in all pairs. This means that direct synaptic connections in all neurons are not necessary for synchronization of bursting, and that direct synaptic connections in one-third of neurons are sufficient for synchronization. The synaptic delay in our culture system was estimated to be 1.5–1.9 ms Žthe mean of the first peak in the histogram; 1.77 ms.. This is much greater than those
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
143
Fig. 6. Gap junctions are unlikely to contribute to neuronal bursting. ŽA. Neurobiotin introduced to one bursting neuron did not spread to other neurons Žbar: 50 m m.. ŽB. There was no electrical coupling between bursting neurons. Negative current injection to hyperpolarize either neuron did not cause a change in the membrane potential of the other neuron.
reported in other studies in vivo Ž0.62 " 0.18 ms in cat visual cortex w39x, 0.3 ms in spinal motoneurons w2,11x, 0.3 ms in rat neocortical slices w21x.. We assumed the AP conduction velocity ŽC.V.. to be 0.3 mrs w10,32x and that axons extend in a straight line to the neurons, and calculated the synaptic delay. The C.V.s of APs in the neurites of cultured neurons may have been considerably lower than 0.3 mrs w10x. For example, if we used a C.V. value of 0.2 mrs, the estimated synaptic delay derived from the mean of the first peak in the histogram was 1.44 ms. Also, the C.V.s of APs in our study may have been attenuated by the lower Žroom. temperature at which these tests were carried out. Correction for temperature by assuming that Q10 is 2.0 w17x and thus C.V. is 0.15 mrs leads to an estimated synaptic delay of 1.2 ms. However, C.V. in our culture system can not be smaller than 0.1 mrs since that would result in conduction times that are in many cases longer than the observed response latencies. Furthermore, actual axons do not always extend in a straight line and the actual conduction time might be longer than our estimate.
If we calculated the conduction time by assuming an axon trajectory 1.5 times the distance of a straight line Žcorresponding to a half circle between the two neurons Žnote that pr2 is about 1.5.., the estimated synaptic delay was 1.44 ms. By not considering temperature and a winding axon trajectory the synaptic delay in our data may have been overestimated, but neither of these factors should affect our results in a significant way. Moreover, the greater estimates of synaptic delay may also be explained by the immaturity of synapses in developing neurons andror by an involvement of dendritic delay, which may be unexpectedly longer than that in vivo Ž0.3 ms in rat neocortical slices w21x.. In cerebral slice preparations the probability of synaptic responses was reported to be 0.1 or lower w21x, while our probability of 0.36 is higher. Mennerlick et al. w23x reported that synaptic properties of neurons grown in microculture are different from those in mass culture and that synaptic probability in microculture Ž0.85. is higher than that in mass culture Ž0.42., which is similar to our data.
144
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
Fig. 7. Synchronized bursting requires direct cell–cell contact. ŽA. Experimental arrangement. ŽB. The neurons separated by cutting the cell layer showed independent but not synchronous bursting. ŽC. Burst latencies of sectioned pairs Žclosed circles. were much longer than those of unsectioned pairs Žopen circles.. Note the scale of the ordinate is different from that of Fig. 5A. ŽD. The average of burst latencies in sectioned pairs Ž n s 8. was significantly longer than that in unsectioned pairs Ž n s 5; Only those pairs between which the spatial distance was longer than 400 m m were used in this analysis..
These results suggest that synaptic probability decreases as neuronal density increases andror when the overall system becomes organized. 4.2. Synchronized bursting of neurons may be attributed to polysynaptic connections Even neurons that did not have a direct synaptic connection, exhibited synchronous bursting at a burst latency of 5.87 " 0.47 ms. These results are consistent with our previous finding that most of the neuronal intracellular Ca2q, measured by optical imaging, oscillated synchronously w29x. The burst latency Žmean " S.E.M.; 5.87 " 0.47 ms. is much greater than the estimated synaptic delay Ž1.5–1.9 ms., suggesting that synchronized bursting is attributable to polysynaptic connections. That the average burst latency in the connected pairs’ group was shorter than that in the unconnected pairs’ group ŽFig. 5B. is consistent with this hypothesis. Also, disappearance of synchronized bursting in bathing solution containing D-AP5 and CNQX ŽFig. 2B. strongly supports the contention that synchronized bursting requires synaptic connections in the network. The contribution of gap junctions and diffusible factors to neuronal synchronized bursting does not seem to be significant even if they existed, because the small molecule dye Lucifer yellow or the neurotracer neurobiotin introduced to one neuron did not stain neighboring neurons ŽFig. 6A. and
physical sectioning blocked synchronized bursting ŽFig. 7B,C,D.. Similar synchronized neuronal activities in neocortical or other CNS explants and in dissociated cultures have been reported w8,9,12,13,19,20,30x. The features of these synchronized activities are consistent with those in our neocortical culture system. First, no synchronized bursting was observed when the Ca2qrMg 2q ratio in the bathing solution was reduced w9,20x. Second, the frequency of spontaneous depolarizations was markedly reduced by the NMDA receptor antagonist, AP5 w12,20x. Third, synchronous discharge was observed in two neurons that showed no evidence of direct synaptic connection w12x. Synchronized neuronal activities may be intrinsic properties of neurons which connect via a polysynaptic network. Traub et al. w40,41x extended their elaborate model of the epileptic hippocampal slice induced by low Mg 2q, and in their computer simulation they included the contribution of polysynaptic connections. Their simulation suggests that no synchronized activity occurs when wMg 2q x is 2.0 mM. In our experiments, no synchronized bursting was observed in bathing solution containing AP5 and CNQX or in bathing solution with a low Ca2q Ž0.4 mM. and a wMg 2q x of 2.2 mM ŽFig. 2.. The synchronized bursting we observed has similar characteristics to these previous experimental studies, and may possibly be explained by Traub’s model. On the other hand, many investigators have suggested that the synchronous activity of neurons is due to electrical
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
interactions between dendrites Žin locus coeruleus neurons w16x, and in hippocampal slices w38x.. Furthermore, a role for gap junctions or dye coupling is suggested in columnar domains in developing neocortical slices w33,43x and in rat neocortical interneurons w4x. It is possible that generator cells Žsuch as interneurons in hippocampus. are present in our culture system and are connected to each other via gap junctions, thus mediating synchronized bursting of larger neurons, although the slice preparations used by the above authors differed from our culture system and dye-coupling in neonatal cortical slices is regulated developmentally w33,43x. Muller et al. w24x recently speculated that the ratio of inhibitory and excitatory neurons affects the spontaneous activity in cultured hypothalamic neuronal networks. In our culture system, excitatory and inhibitory neurons, such as glutaminergic and GABAergic neurons should exist, and inhibitory connections may affect the synchronized bursting activity in neocortical neurons as well. We did not observe an IPSP response, possibly because of the experimental conditions. We can not rule out the presence of IPSPs in our system and their contribution to synchronized bursting activity in other neurons via the neural pathway. Our results indicate that synchronized bursting of neurons in culture requires synaptic connection and that gap junctions andror diffusible factors by themselves cannot explain the phenomenon of synchronized bursting. However, further study is required to clarify completely the mechanism of synchronized bursting between neurons. 4.3. Comparison among other measurements The frequency of synchronized bursting is close to that of intracellular Ca2q oscillation in the same culture system w29x. Simultaneous measurement of electrical bursting and Ca2q oscillation suggested that these two phenomena are coupled via a polysynaptic network w34x. Comparison with the data for intracellular Ca2q oscillation shows the synchronization observed with two patch clamp electrodes to be accurate with regard to time resolution Ž50–100 m s sampling time., and that the synaptic delay of 1–2 ms was also accurate. Extracellular recordings using an array electrode w19x are useful for observing the overall synchronization of many neurons and provide better time resolution than optical methods, but they give no information on the shape of the AP and bursting. Analysis using the intracellular recording of two neurons give better time resolution and much more information about synchronized bursting for further analyses, such as the shape of the burst and synaptic responses. 4.4. Neuron–glia coculture may be a good model of network formation Our culture system may be too complicated to analyze simple synaptic responses, but may be suitable for a study
145
of the neural network, which is important for elucidating complicated brain functions. Since our culture model is similar to the epileptiform of slice preparations, it may also be suitable for studying epileptogenesis.
Acknowledgements We are grateful to Dr. Muneyuki Ito of the Department of Physiology, Institute for Developmental Research, Aichi Human Service Center, for helpful discussions. We thank Dr. Taiji Kato of the Department of Bioregulation Research, Nagoya City Univ. Med. Sch. and Dr. Shunichi Yamagishi of NIPS for their kind support of this work. We also thank the members of the Lab. of Membrane Biology at NIPS for their technical support. This research was supported in part by the Aichi Cancer Research Foundation.
References w1x N. Akaike, N. Harata, Nystatin perforated patch recording and its applications to analyses of intracellular mechanisms, Jpn. J. Physiol. 44 Ž1994. 433–473. w2x T. Araki, J.C. Eccles, M. Ito, Correlation of the inhibitory postsynaptic potential of motoneurones with latency and time course of inhibition of monosynaptic reflexes, J. Physiol. ŽLondon. 154 Ž1960. 354–377. w3x T. Bal, D.A. McCormick, Synchronized oscillations in the inferior olive are controlled by the hyperpolarization-activated cation current Ih, J. Neurophysiol. 77 Ž1997. 3145–3156. w4x L.S. Benardo, Recruitment of GABAergic inhibition and synchronization of inhibitory interneurons in rat neocortex, J. Neurophysiol. 77 Ž1997. 3134–3144. w5x Y. Ben Ari, E. Cherubini, R. Corradetti, J.L. Gaiarsa, Giant synaptic potentials in immature rat CA3 hippocampal neurons, J. Physiol. ŽLondon. 416 Ž1989. 303–325. w6x A. Bragin, G. Jando, Z. Nadasdy, J. Hetke, K. Wise, G. Buzsaki, Gamma Ž40–100 Hz. oscillation in the hippocampus of the behaving rat, J. Neurosci. 15 Ž1995. 47–60. w7x G. Buzsaki, Z. Horvath, R. Urioste, J. Hetke, K. Wise, Highfrequency network oscillation in the hippocampus, Science 256 Ž1992. 1025–1027. w8x B.W. Connors, M.J. Gutnick, D.A. Prince, Electrophysiological properties of neocortical neurons in vitro, J. Neurophysiol. 48 Ž1982. 1302–1320. w9x S.M. Crain, M.B. Bornstein, Organotypic bioelectric activity in cultured reaggregates of dissociated rodent brain cells, Science 176 Ž1972. 182–184. w10x S.M. Crain, Neurophysiologic Studies in Tissue Culture. Raven Press, New York, 1976, pp. 43–152. w11x J.C. Eccles ŽEd.. The Physiology of Synapses. Springer, Berlin, 1964, pp. 49–53 and pp. 152–172. w12x M.J. Gutnick, B. Wolfson, F. Baldino Jr., Synchronized neuronal activities in neocortical explant cultures, Exp. Brain Res. 76 Ž1989. 131–140. w13x M.J. Gutnick, B.W. Connors, D.A. Prince, Mechanisms of neocortical epileptogenesis in vitro, J. Neurophysiol. 48 Ž1982. 1321–1335. w14x O.P. Hamill, A. Marty, E. Neher, B. Sakmann, F.J. Sigworth, Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches, Pflug. Arch. 391 Ž1981. 85–100.
146
K. Nakanishi, F. Kukitar Brain Research 795 (1998) 137–146
w15x M. Ichikawa, K. Muramoto, K. Kobayashi, M. Kawahara, Y. Kuroda, Formation and maturation of synapses in primary cultures of rat cerebral cortical cells: an electron microscopic study, Neurosci. Res. 16 Ž1993. 95–103. w16x M. Ishimatsu, J.T. Williams, Synchronous activity in locus coeruleus results from dendritic interactions in pericoerulear regions, J. Neurosci. 16 Ž1996. 5196–5204. w17x F. Kukita, Properties of sodium and potassium channels of the squid giant axon far below 08C, J. Membr. Biol. 68 Ž1982. 151–160. w18x Y. Kuroda, M. Ichikawa, K. Muramoto, K. Kobayashi, Y. Matsuda, A. Ogura, Y. Kudo, Block of synapse formation between cerebral cortical neurons by a protein kinase inhibitor, Neurosci. Lett. 135 Ž1992. 255–258. w19x E. Maeda, H.P.C. Robinson, A. Kawana, The mechanisms of generation and propagation of synchronized bursting in developing networks of cortical neurons, J. Neurosci. 15 Ž1995. 6834–6845. w20x E. Maeda, H.P.C. Robinson, Y. Kuroda, A. Kawana, Transient changes of network activity in cultured cortical neurons induced by Mg 2q and APV Žabstract., Neurosci. Res. suppl. 18 Ž1993. S233. w21x H. Markram, J. Lubke, M. Frotscher, A. Roth, B. Sakmann, Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex, J. Physiol. ŽLondon. 500 Ž1997. 409–440. w22x M. Meister, R.O.L. Wong, D.A. Baylor, C.J. Shatz, Synchronous bursts of action potentials in ganglion cells of the developing mammalian retina, Science 252 Ž1991. 939–943. w23x S. Mennerick, J. Que, A. Benz, C.F. Zorumski, Passive and synaptic properties of hippocampal neurons grown in microcultures and in mass cultures, J. Neurophysiol. 73 Ž1995. 320–332. w24x T.H. Muller, D. Swandulla, H.U. Zeilhofer, Synaptic connectivity in cultured hypothalamic neuronal networks, J. Neurophysiol. 77 Ž1997. 3208–3225. w25x K. Muramoto, M. Ichikawa, M. Kawahara, K. Kobayashi, Y. Kuroda, Frequency of synchronous oscillations of neuronal activity increases during development and is correlated to the number of synapses in cultured cortical neuron networks, Neurosci. Lett. 163 Ž1993. 163– 165. w26x T.H. Murphy, L.A. Blatter, W.G. Wier, J.M. Baraban, Spontaneous synchronous synaptic calcium transients in cultured cortical neurons, J. Neurosci. 12 Ž1992. 4834–4845. w27x V.N. Murthy, E.E. Fetz, Oscillatory activity in sensorimotor cortex of awake monkeys: synchronization of local field potentials and relation to behavior, J. Neurophysiol. 76 Ž1996. 3949–3967. w28x V.N. Murthy, E.E. Fetz, Synchronization of neurons during local field potential oscillations in sensorimotor cortex of awake monkeys, J. Neurophysiol. 76 Ž1996. 3968–3982. w29x K. Nakanishi, Y. Okouchi, T. Ueki, K. Asai, I. Isobe, Y. Eksioglu,
w30x
w31x
w32x
w33x
w34x
w35x w36x
w37x w38x
w39x
w40x
w41x
w42x
w43x
T. Kato, Y. Hasegawa, Y. Kuroda, Astrocytic contribution to functional synapse formation estimated by spontaneous neuronal intracellular Ca2q oscillations, Brain Res. 659 Ž1994. 169–178. K. Nakanishi, F. Kukita, Functional synapses in synchronized bursting neurons in culture, Neurosci. Res. Suppl. 21 Ž1997. S64, Žabstract.. A. Ogura, T. Iijima, T. Amano, Y. Kudo, Optical monitoring of excitatory synaptic activity between cultured hippocampal neurons by a multi-site Ca2q fluorometry, Neurosci. Lett. 78 Ž1987. 69–74. H.D. Patton, Special properties of nerve trunks and tracts, in: T.C. Ruch, J.F. Fulton ŽEds.., Medical Physiology and Biophysics, Saunders, Philadelphia, 1960, pp. 66–95. A. Peinado, R. Yuste, L.C. Katz, Extensive dye coupling between rat neocortical neurons during the period of circuit formation, Neuron 10 Ž1993. 103–114. H.P.C. Robinson, M. Kawahara, Y. Jimbo, K. Torimitsu, Y. Kuroda, A. Kawana, Periodic synchronized bursting and intracellular calcium transients elicited by low magnesium in cultured cortical neurons, J. Neurophysiol. 70 Ž1993. 1606–1616. M. Segal, Rat hippocampal neurons in culture: responses to electrical and chemical stimuli, J. Neurophysiol. 50 Ž1983. 1249–1264. M. Segal, J.L. Barker, Rat hippocampal neurons in culture: voltageclamp analysis of inhibitory synaptic connections, J. Neurophysiol. 52 Ž1984. 469–487. C.J. Shatz, Impulse activity and the patterning of connections during CNS development, Neuron 5 Ž1990. 745–756. F. Strata, M. Atzori, M. Molnar, G. Ugolini, F. Tempia, E. Cherubini, A pacemaker current in dye-coupled hilar interneurons contributes to the generation of giant GABAergic potentials in developing hippocampus, J. Neurosci. 17 Ž1997. 1435–1446. K. Toyama, K. Matsunami, T. Ohno, S. Tokashiki, An intracellular study of neuronal organization in the visual cortex, Exp. Brain Res. 21 Ž1974. 45–66. R.D. Traub, R. Miles, J.R. Jefferys, Synaptic and intrinsic conductances shape picrotoxin-induced synchronized after-discharges in the guinea-pig hippocampal slice, J. Physiol. ŽLondon. 461 Ž1993. 525–547. R.D. Traub, J.G.R. Jefferys, M.A. Whttington, Enhanced NMDA conductance can account for epileptiform activity induced by low Mg 2q in the rat hippocampal slice, J. Physiol. ŽLondon. 478 Ž1994. 379–393. R.O.L. Wong, M. Meister, C.J. Shatz, Transient period of correlated bursting activity during development of mammalian retina, Neuron 11 Ž1993. 923–938. R. Yuste, D.A. Nelson, W.W. Rubin, L.C. Katz, Neuronal domains in developing neocortex: mechanisms of coactivation, Neuron 14 Ž1995. 7–17.