The Journal of Neuroscience, December 1993, 13(12): 53015311

Apical of the : Correlation between Sodium- and Calcium-dependent Spiking and Morphology

Han G. Kim and Barry W. Connors Department of Neuroscience, Division of Biology and Medicine, Brown University, Providence, Rhode Island 02192

Apical dendrites and somata of layer V pyramidal Pyramidal cells of the neocortex may have apical dendrites over were recorded with tight-seal patch electrodes in a slice 1 mm long, from the in layer V to outermost layer I preparation of rat somatosensory cortex. Recording sites (Feldman, 1984). Apical dendritesallow a singlecell to combine were confirmed by measurements of the electrode location synaptic inputs from several cortical layers. The transformation and by staining with biocytin. Dendritic recordings were made applied by the apical to its inputs depends critically along the main trunk of the apical dendrite, usually within upon the dendrite’s biophysics, which are determined by its layer IV, at distances from 100 to 500 Am from the soma. morphology and membrane properties (Rall, 1977). Although Most cells recorded through the dendrite had a distinct en- the shapesof dendrites can often be measuredwith precision largement of the apical trunk around the presumed recording (Hillman, 1979) the electrical properties of their membranes site. have long been elusive. The electrical properties of apical dendrites were readily At least someapical dendrites in neocortex are probably elec- distinguishable from those of somata. Dendrites generated trically excitable. Small spike-like potentials recordedin somata two types of response when injected with depolarizing cur- of immature pyramidal tract cells were attributed to dendritic rent. Group I responses were relatively small and broad Na+- electrogenesis(Purpura and Shofer, 1964; Purpura et al., 1965; dependent action potentials whose amplitude and rate-of- Purpura, 1967), as were much slower somatic depolarizations rise were negatively correlated with recording distance from from mature cells (Arikuni and Ochs, 1973). Somatic recordings the soma. Group II responses were complex, clustered firing of thalamocortical synaptic potentials also included small all- patterns of Na+-dependent spikes together with higher- or-none depolarizing events that were not blocked in an anti- threshold slow spikes or plateaus; in these dendrites spike dromic collision test, suggestinga dendritic origin (Deschenes, parameters were not correlated with distance from the soma. 1981). Direct measurementsprovide the clearestevidence for These two response groups were correlated with dendritic active apical dendrites. Patch-clamp recordings from dissoci- morphology: group I had significantly fewer oblique branch- ated cells reveal Na+ channelsin the proximal apical dendrite es on the apical dendrite (5.5 vs 12.0) and a thinner apical of immature pyramidal neurons (Huguenard et al., 1989). In- trunk (2.0 vs 2.5 pm) than group II. TTX (1-2 PM) selectively tradendritic recordingswith conventional sharpmicroelectrodes blocked fast dendritic spikes, but not slow spikes and pla- in viva (Pockberger, 1991) showed complex active responses, teaus. Blocking Ca2+ currents reduced complex firing pat- and studiesin vitro suggestthat thesemay be mediated by both terns and suppressed high-threshold slow spikes. Physio- Na+- and Ca2+-dependentcurrents (Amitai et al., 1993). logical and pharmacological studies imply that slow spikes Active membrane currents have been demonstrated in the and plateau potentials were primarily generated by high- dendritesof many neurons; however, they are very variable and threshold Ca2+ channels in the apical dendrite. Stimulating cell type specific (Spencer and Kandel, 1961; Llinls and Nich- of layer I elicited EPSPs on distal apical dendrites of olson, 197 1; Wong et al., 1979; Llinas and Sugimori, 1981; layer V cells. Recordings from both groups of apical den- Turner et al., 1991). For example, dendrites seem drites revealed that EPSPs triggered a variety of distally to be dominated by active Ca2+ currents (Llinas and Sugimori, generated, all-or-nothing depolarizations. 1981; Ross et al., 1990; but see Regehr et al., 1992) while The results show that voltage-dependent Na+ and Ca2+ hippocampal dendrites can apparently sustainNa+- as well as currents are present in distal apical dendrites, in variable Ca2+-dependentregenerative events (Wong et al., 1979; Ben- densities. These currents significantly modify distal synaptic ardo et al., 1982). Pyramidal neuronsof neocortical layer V are events. The prevalence and character of active dendritic diverse in both structure and electrophysiology (Connors et al., spiking (and presumably of Na+ and Ca2+ channel densities) 1982; Stafstrom et al., 1984; McCormick et al., 1985; Chagnac- correlate with the morphology of the apical dendritic tree. Amitai et al., 1990; Connors and Gutnick, 1990; Mason and [Key words: dendrite, pyramidal cell, neocottex, action Larkman, 1990; Silva et al., 1991; Agmon and Connors, 1992). potentials, sodium currents, calcium currents] It is possiblethat different types of pyramidal cells exhibit dis- tinct forms of dendritic electrogenesis,and the variability of Received Jan. 15, 1993; revised June 9, 1993; accepted June 17, 1993. intradendritic recordings is consistent with this (Woody et al., We thank Cheryl Pulaski for excellent histological assistance. This work was supported by an NIMH postdoctoral traineeship (5T32-MH19118) to H.G.K., 1984; Pockberger, 199 1; Amitai et al., 1993). It is therefore and by grants from the NIH (NS25983) and ONR mOOO14-90-J-17011 to B.W.C. important to characterize the active conductancesin the apical Co&spondence should be‘addressei to Dr. Ba& W. Connors, Department of Neuroscience, Box G-M, Division of Biology and Medicine, Brown University, dendrites, and their variations among different classesof neu- Providence. RI 02 192. rons. Copyright 0 1993 Society for Neuroscience 0270-6474/93/135301-l 1$05.00/O Conventional intracellular microelectrodesare not well suited 5302 Kim and Connors l Excitability of Apical Dendrites in Neccortex

the pia and the was measured using a calibrated eyepiece on the microscope. The slice was fixed with 4% paraformaldehyde and resectioned at 125 pm on a freezing microtome. Sections were processed with the avidin-biotin-peroxidase method to reveal the cell morphology (Horikawa and Armstrong, 1988). Cells were photographed and drawn with the aid of a computerized reconstruction system (Neurolucida). Dendritically recorded cells usually had a discrete swelling and/or a slight kink along the trunk of their apical dendrite (Fig. 1; n = 22 of 25 dendritic recordings); the position of this anomaly always corresponded well with the estimated position of the recording, as determined by measurement during the experiment. These dendritic structures were never observed in cells that were recorded from somata, nor have they been reported in Golgi studies (Feldman, 1984). Thus, we assume the swelling indicates the precise site of dendritic recording, and these were used to quantify distances between soma and recording site (e.g., see Fig. 4). Somatic membrane areas (SA) were estimated from the formula SA = rab, where a is the major diameter and b is the minor diameter of the soma (Larkman, 199 1b). Computational modeling of dendritic excitability. A computer model derived from a somatically recorded, biocytin-filled, layer V pyramidal cell was used to explore the relationship between dendritic spike shape and Na+ channel density. The details of this cell, the model, and its implementation have been described in detail (Cauller and Connors, 1992; Amitai et al., 1993). In brief, the was modeled by 120 2oum linked, isopotential cylindrical compartments, with the apical dendrite fully represented, the basal dendrites lumped into a single equivalent cylinder, and an initial segment appended to the soma. Simulations Figure 1. Dendritic recording site of a biocytin-tilled apical dendrite. were made with the program NEURON (Hines, 1989). Specific membrane Slight swelling along the apical trunk corresponded closely to the cal- capacitance (C,) was assumed to be 1 pF/cm2, specific membrane re- culated recording site. Scale bar, 20 pm. sistivity (It,) was 8014 Qcm2, and specific cytoplasmic resistivity (R,) was 69 R.cm2. Hodgkin and Huxley-type Na+ and K+ channels were added to the soma and initial segment compartments, with channel to the study of small cellular structures such as dendrites. Tight- densities and kinetics adjusted to generate a somatic seal methods, however, have made small cells much more ac- about 100 mV in amplitude, with a peak upward slope of 300 V/set, cessible(Sakmann and Neher, 1983). Using the whole-cell vari- and a duration of about 1 msec at the base. The principle variable in ation of this technique, we have routinely obtained recordings the simulations reported here was the density of Na+ and K+ channels in the apical dendrite; these varied from zero (i.e., passive dendritic-- from apical dendrites of layer V cells (Kim and Connors, 1992). membrane) to g,, = 0.096 S/cm2, keeping a constant ratio of g,,:g, = Our results show that the dendrites have active Na+ currents 6.67. Active currents were distributed uniformly throughout all apical and high-threshold Ca2+currents, and that dendritic morphol- dendrites in every case. As the density of active currents in the apical ogy and dendritic electrogenesisare correlated. dendrite was increased, the densities in the somatic compartment were reduced to maintain the amplitude and upward slope of the somatic Materials and Methods spike at constant levels (Amitai et al., 1993). Slice preparation. Animals were anesthetized by intraperitoneal injec- tion of nembutal(60 mg/kg). After decapitation, the brain was quickly removed into cold (6-8°C) physiological solution containing (mr.4) NaCl, Results 126; KCl, 3; NaH,PO,, 1.25; NaHCO,, 2.6; MgSO,, 1; dextrose, 20; and CaCl,, 2; solutions were continuously bubbled with 95% 0,, 5% Two groups of dendritic recordings CO,. A block of parietal neocortex was dissected and glued to the slicing Electrode tips placed into layer IV frequently recordedfrom the platform of a vibratome. Coronal slices 400 pm thick were cut from apical dendrites of pyramidal neurons whose somata were in the primary somatosensory area. These slices were kept at room tem- perature in oxygenated physiological solution until use. The recording layer V. All dendrites could generatediscrete, all-or-none events chamber was made from Teflon, surrounded by brass, and heated with when stimulated with step pulsesof current passedthrough the an element to 35°C. Slices were submerged and stabilized using a fine electrode. The patterns of electrogenesisvaried greatly between nylon net attached to a metal ring (Edwards et al., 1989). The chamber dendrites, however. About half of the dendrites (n = 13) gen- was perfused with oxygenated physiological solution; a peristaltic pump erated low-amplitude, relatively fast action potentials; strong removed the effluent and controlled the fluid level. The chamber volume was less than 1 ml, the flow rate was 2-4 ml/mitt, and the entire chamber stimulation yielded regular, adapting patterns of similarly sized solution was exchanged in about 3-4 min. Drug solutions flowed through spikes (Fig. 24. In contrast, the remaining dendrites (n = 12) a small mixing chamber with several taps. generated action potentials with very variable amplitudes and Whole-cell recording and staining methods. The methods for tight- durations and complex, often irregular patterns (Fig. 2B). Four seal patch recordings followed Hamill et al. (198 l), with modifications for brain slices (Blanton et al., 1989; Edwards et al., 1989). Patch elec- of the latter dendrites could produce depolarizing plateau po- trodes were made from borosilicate tubing (1.2 mm o.d., 0.9 mm i.d.) tentials. Plateau durations were variable, and sometimeslasted using a horizontal puller (Sutter Instruments). Electrode resistances were as long as the stimulus current (Fig. 2C). For purposesof com- 4-8 MQ. The standard intracellular solution contained (mM) KCH,O,S, parison, we will call dendrites of the first type (i.e., those with 120; HEPES, 10; EGTA, 5; MgCl,, 2; MgATP, 4; CaCl,, 0.5; KCl, 10; small, fast spikesof regular size and temporal pattern) group I, and biocytin, 1%, at pH 7.3. During recording, some laminar boundaries were visualized under the dissecting microscope using transillumination and dendrites of the secondtype (i.e., thosewith variable spike (Agmon and Connors, 1991). To obtain whole-cell recordings, elec- sizes,including plateaus,with complex temporal patterns)group trodes were advanced into the slice while pulsing with 0.1 nA current II. steps of 200 msec duration. When a significant increase in electrode Under normal recording conditions, apical dendrites of layer resistance was evident, gentle suction was applied to obtain a seal re- sistance of 1 GQ or greater. The patch of membrane was broken by V cells were readily distinguishable from somata by their re- applying more negative pressure to obtain a whole-cell configuration. sponsesto injected currents. Somata invariably generatedmuch At the end of each recording, the position of the electrode between faster and larger-amplitude spikes than did dendrites, with re- The Journal of Neuroscience, December 1993, 13(12) 5303

Figure 2. Examples of three different dendritic recordings and a somatic re- cording from layer V pyramidal cells. All cells were activated with rectangular current pulses incremented by 100 pA. A, Group I dendritic recording. Resting potential, -63 mV; time calibration, 90 msec. B, Group II dendritic record- ing with complex fast-spike patterns and long depolarizing plateaus. Resting po- tential, -70 mV; time calibration, 90 msec. C, Group II recording with a leading fast spike followed by variable duration plateaus. Resting potential, -63 mV; time cahbration, 50 msec. D, Somatic recording from a layer V py- ramidal neuron. Resting potential, - 7 1; time calibr:t ion, %O miec. petitive firing patterns more similar to group I dendrites than soma areasfor the two groups were also similar (group I = 856 group II (Fig. 20). The passive electrical properties of somatic f 278 pm2 and group II = 924 + 276 pm2). and dendritic recordings also differed. Dendritic recordings had more positive resting potentials, higher input resistances, and Physiology of group I dendritic recordings faster time constants (Table 1). Among the dendrites, the time Group I dendrites were similar to somata in their firing patterns, constantsfor group I were significantly fasterthan thoseof group but dendritic spike amplitudes were smaller and longer lasting II (Table 2), but input resistanceand resting potential did not than those of layer V somata (Fig. 5, control traces; Table 3). differ. Bath application of the Na+ channel blocker TTX (1 PM) elim- inated most spiking activity in the group I dendrites (n = 6), Correlations betweendendritic physiology and morphology leaving only a low-amplitude, regenerative seriesof spikesthat Apical dendrites of layer V pyramids vary greatly in diameter, was reversibly blocked by the substitution of MgZ+ for Ca2+in length, and the number and pattern of distal tuft and proximal the bathing medium (Fig. 5, left). The data imply that spikesin oblique branches(Larkman, 1991 a). In our sample of dendrit- group I dendrites(as in somata)are primarily mediatedby volt- ically filled cells it was not always possible to reconstruct the age-dependentNa+ current, with a significant contribution from apical tufts reliably, since in some casesit seemedthat portions Ca*+ currents. By contrast, in somatic recordings (n = 4) it was had been cut off either during the initial slicing procedure or not possible to evoke any regenerative events in the presence during histological processing. The rest of the cells were rea- of TTX (Fig. 5, right). Somata did generatea graded, transient, sonably intact, however, and amenableto quantification. TTX-insensitive depolarization that was blocked by Ca*+ re- When dendrites were grouped according to the physiological moval. criteria defined above, distinct morphologieswere evident. The most striking difference was the number of oblique branches Physiology of group II dendritic recordings from the apical trunks. Figure 3 showsexamples of biocytin- The most unique characteristic of group II dendrites was their stainedcells from group I (Fig. 3A) and group II (Fig. 3B). The ability to generate relatively large-amplitude, prolonged spikes approximate recording sites are marked by arrowheads, and or plateaus. Both Na+ and Ca2+ conductances seem to contribute each trunk with its oblique branchesis shown isolated and en- to this group II behavior. Figure 6A showsa group II dendritic larged. Group II dendrites had about twice as many primary recording in which a suprathreshold current pulse elicited a fast oblique branchesas group I, with little overlap between the two spike followed by a broad spike. The expanded trace below groups (Fig. 4, Table 2). Group I dendrites had thinner apical shows both the fast and slow spikes clearly. Addition of 1 pM trunks, overall, than group II (Table 2). This may account for TTX blocked the fast spikes, leaving a relatively large, slow, the fact that, on average, group II recording sites were signifi- high-threshold spike that was reversibly blocked by 2 mM Mg2+ cantly farther from the soma than those of group I, although and 0 Ca2+ (Fig. 6B). Notice that in the absence of the Na+ there was considerable overlap in the distributions (Table 2). spike the voltage threshold of the slow spike increased(Fig. 6B, The distancesbetween dendritic recordings sites and somata TTX). varied from about 100 to 470 pm. Figure 7A illustrates another group II dendritic recording un- The numbers of primary basal dendrites for group I (5.0 f der control conditions, showing both fast and slow spikes. Sup- 0.7) and group II (5.3 + 1.5) neuronswere the same. Estimated pression of its Ca2+ currents with 2 mM Co*+ (Fig. 7B) yielded

Table 1. Electrophysiology of dendritic and somatic recordings from layer V pyramidal cells

Dendrites Somata Resting potential (mV)** -60.3 f 5.3, n = 23 -69.3 + 3.8, n = 18 Input resistance (MQ)* 88.3 5 40, n = 24 68.8 + 27.3, n = 16 Time constant (msec)** 7.3 k 2.9, n = 25 16 + 5.3, n = 21 Data are mean + SD, *, p < 0.1, **, p i 0.01; two-tailed I test assuming unequal variances. 5304 Kim and Connors * Excitability of Apical Dendrites in Neocortex

Table 2. Electrophysiology of group I and group II dendrites

Group I Group II # of obliquebranches** 5.5 + 3.4, n = 13 12 + 4.5, n = 11 Apical trunk diameterbm)* 2.0 + 0.63, n = 13 2.5 f 0.45, n = 10 Distancefrom recordingsite to somakm)** 219 + 57, n = 13 336 IL 100, n= 11 Restingpotential (mV) -60.4 f 5.2, n = 12 -60.3 k 5.7, n = 11 Input resistance(Ma) 95.1 + 39.6,n = 13 80.3 t 41, n = 11 Time constant(msec)* 8.6 f 2.6, n = 12 6 k 2.6, n = 13

Data are mean + SD, *, p < 0.05, **, p < 0.01; two-tailed, t test assuming unequal variances. spike patterns similar to those of group I dendrites. Taken to- amples of proximal and distal group I recordings (a and b in gether, the data suggestthat group II cells have substantially Fig. 8AJ) are illustrated in Figure 9, A and B. larger Ca2+ currents in their apical dendrites than group I neu- Amplitudes and maximal slopesfor the fast spikesof group rons. II dendrites varied over wide ranges.For thesemeasurements, only the first fast spike in an evoked train was measured, to Spatial origins of Ca2+ and Na+ spikes avoid the confounding influencesof subsequentCa2+ -dependent The channels that generate dendritic Ca*+ spikes in group II spikes and depolarization. Unlike group I, group II dendritic cells must be located on the apical dendritic membrane, rather measurementsdid not correlate with the recordingdistance from than the soma or basaldendrites. This follows simply from the the soma (Fig. 8C,D). Examples of group II responses(c and d observation that dendrites, but not somata, displayed apparent in Fig. 8C,D) are shown in Figure 9, C and D. Ca2+-dependentspikes under normal recording conditions, that Resultsfrom computer simulations of Na+ spikesin an apical is, without blocking K+ currents (Fig. 6). The influence of Ca2+ dendritic trunk are illustrated in Figure 10, which plots the spike currents may also be stronger in the apical dendrites of group I height and maximal slope as functions of distance from the cells than in somata, as suggestedby the data shown in Figure soma, and dendritic Na+ channel density. The model predicts 5. Thus, it is likely that most layer V neurons have higher that spike amplitude and slope will fall precipitously along the densitiesof Ca*+ current in their apical dendrites than in their first 300 km of the apical trunk, if the dendritic membrane is somata. passive(Fig. lOA,& solid lines). This is similar to the relation- The spatial origin of Na+ -dependentdendritic events is more ship found for fast spikesin group I recordings (Figs. 8, 9). As difficult to infer. Becausegroup I dendritic spikesresemble trun- dendritic Na+ current density increasesin the model, spike am- cated somatic spikes, they may be due primarily to the passive plitude and slope become more complex functions of distance propagation of spikes generated by Na+ channels within the from the soma. Intermediate current densitiesdisplay a dip in perisomatic region (Turner et al., 1991; Amitai et al., 1993). spike size within the first 200 Km, probably becauseof the Consistent with this, acrossall sampled group I dendrites the impedance load of the apical oblique branches, followed by a spike height diminished asthe recording site moved farther from distal increasein current density (Fig. 10, broken lines). Higher the soma (Fig. 8A). The correlation is significant (r = -0.81). densitiesyield a dendrite that conducts like an axon (Fig. 10, The maximal positive slope of the spike showed a similar re- dotted line). The results suggestthat group I apical dendrites lationship with recording distance (Fig. 8B; r = -0.70). Ex- have a very low Na+ channel density, at least in their trunks. B .

Y

Figure 3. Morphology of group I (A) andII (B) cells.Group I cellshad thin- ner apicaldendrites and fewer oblique branchesthan group II cells.The en- largedfigure on the right of eachpane1 showsthe samecell without the basal anddistal apical dendrites. Arrowheads indicatethe presumedrecording sites. kb Physiologyof the cell in A is shownin Figure 9A. and the cell in B, in Fig- ure 7. The Journal of Neuroscience. December 1993, 13(12) 5305

Table 3. Comparison between fast action potentials of group I dendrites and layer V somata n Group I 0 Group II Dendrites Somata Action potential half-width (msec)** 6.3 + 1.5, n = 13 2.6 +- 0.86, n = 21 Action potential height(mV)** 23.7 k 8.7, n = 13 40.4 + 14.2, n = 18 Action potential height is measured from threshold voltage to peak depolarization; half-width is the duration at 50% of peak height. Data are mean + SD, **, p c 0.01; two-tailed, t test assuming unequal variances.

1-2 3-4 6-6 78 9-10 1112 13-14d-J1516 17-16 19-20 # apical oblique dendrites rents. Ni*+ at a concentration of 100 I.LM has a strong depressive action on low-threshold Ca2+channels of rat dorsal root ganglia Figure4. Plot of the numberof the obliquebranches from the apical cells (Carbone et al., 1990). However, even 200 PM NiZ+ did trunk, for groupI and II dendrites. not have any noticeable effect on the Ca2+ spikes of group I dendrites, when Na+ and K+ currents were suppressedwith while group II dendrites are more variable, with at least some TTX (2 PM) and tetraethylammonium (TEA; 15 mM), respec- having Na+ channel densities high enough to influence fast- tively (Fig. 1 lA,B; n = 3). Only at 500 FM did Ni2+ begin to spiking ability strongly as far as 500 pm from the soma(compare have a noticeable suppressiveeffect (Fig. 1 lC), and even at 3 Figs. SC,D; 9C,D). mM it did not block the dendritic Ca2+ spikes completely (not shown). Characteristicsof Ca2+ conductances Unlike Ni*+, Cd2+ suppresseshigh-threshold Ca*+ channels Central neuronshave a variety of active Ca*+currents. Our data at concentrations of 2-20 PM, while having no effect on low- suggestthat most of the Ca2+ current in the apical dendrite is threshold channels(Fox et al., 1987). Figure 12A showsa group of the high-threshold type. Ca2+-dependent spikes were acti- II dendrite with complex firing patterns riding on a plateau vated only by large depolarizations, either on the trailing limbs potential. The plateau was reversibly suppressedby 10 I.LM Cd2+ ofNa+ spikes,or with injected currents following TTX blockade (Fig. 12B; n = 2), suggestingan involvement of high-threshold ofNa+ spikes.The long durations of many dendritic Ca2+spikes, Ca2+ currents. Interestingly, 10 PM CdZ+ did not abolish all even in the absenceof K+ channel blockers, also suggestthat broad spikes in the illustrated case (Fig. 12B), although it did the underlying Ca2+currents did not inactive rapidly (Figs. 2C, in another dendrite. 7,9C,D). Voltage plateaus often did not repolarize until the end of a stimulus current pulse (up to 300 msecduration). Synaptic activation of dendritic spikes Voltage clamping of apical dendrites was not feasiblebecause Activation of excitatory synapsesin layer I can generate large, of inadequate spaceclamp and the relatively high seriesresis- fast EPSPs in the somata of layer V pyramidal cells (Cauller tance of the electrodes.However, pharmacologicalstudies were and Connors, 1992). Stimulation of layer I while recording from also consistent with the presenceof high-threshold Ca2+ cur- the apical dendrites of layer V cellsgenerated EPSPs that always

Group I Dendrite Layer V Soma

Figure 5. Effectsof currentblockade control on recordingsfrom a group I dendrite and a layer V soma. Under control con- ditions, the dendrite resembled the soma in generalfiring pattern,but the den- dritic spikes were broader and smaller. 1UMll-X TTX (1 WM) blocked fast spikes in both dendrite and soma. However, small, higher-threshold, repetitive spikes re- mained in the dendrite but not the soma. Blockade of Ca2+ currents by addition of extra Mg2+ (2 mM) and the removal of extracellular Ca2+ abolished the re- maining small dendritic spikes, and suppressed the initial, graded depolar- ization in the soma. Resting potential of the dendrite was -52 mV, current 2thIVL steps incremented by 100 pA; resting 9OlIl.WC potential ofthe soma was -68 mV, cur- rent steps incremented by 50 pA. 5305 Kim and Connors - Excitability of Apical Dendrites in Neocortex

A B A control

A

1OmVL 90 mxc

B 2m~c43++

Figure 6. Effects of current blockade in a group II dendritic recording. A, Injected currents evoked an initial fast spike, followed by a long- lasting spike under control conditions. The lower record shows an ex- Figure 7. Blocking Ca2+ currents with Co2+ (2 mM) converted group panded trace of the spikes. B, Fast spikes were blocked by 1 PM TTX, II dendritic responses to group I-like responses. leaving a long-lasting spike with high threshold. The long-lasting spikes were reversibly blocked by additional 2 mM Mg2+ in the absence of Ca*+. Resting potential was -56 mV. Current steps were incremented by 100 pA. Recordings from the somata of layer V pyramidal cells show a variety of intrinsic spiking patterns, including different rates initiated all-or-none, fast and slow spikes. Figure 13A shows of spike frequency adaptation and several forms of intrinsic recordings from a group II dendrite as stimulus intensity was bursting (Connors et al., 1982; Stafstrom et al., 1984; McCor- increased;several discreteall-or-none events with different am- mick et al., 1985; Silva et al., 1991). Until somataand dendrites plitudes and durations could be recruited, suggestingmultiple have been recorded simultaneously in single cells, we can only locations for dendritic spike generation. Similar results were speculate about how the different types of each correspond. obtained in five of six group II dendrites tested. Figure 13B However, morphology can guide our conjecture. Somatic re- illustrates a different group II dendrite, in which a layer I EPSP cordings from layer V cells have shown that intrinsically burst- triggered two all-or-none events with different amplitudes and ing neurons tend to be relatively large, with thicker and more durations when recorded at the resting potential of -56 mV. arborizing apical dendrites than regular-spiking neurons(Chag- When the dendrite was hyperpolarized to -13 mV the slower, nac-Amitai et al., 1990; Larkman and Mason, 1990). Judging higher-threshold spike was blocked. Surprisingly, even group I from their morphology and complex spiking, it seems likely that dendrites reliably generated both fast and slow spikes when at least some group II apical dendrites are attached to intrin- activated by distal EPSPs (six of six group I dendrites tested; sically bursting somata. Group I dendrites are most similar to Fig. 13C,D). regular-spiking somata in their firing patterns and morphology, It is likely that many of the layer I-evoked spikes recorded and may correspond. in dendritic trunks were actually initiated in the dendrites. In There is evidence that the apical dendrite does not signifi- three of four group I dendrites, and two of three group II den- cantly influence the firing patterns of the soma (Telfeian et al., drites, the voltage threshold for fast-spikegeneration by a layer 1991). The membrane properties of the apical dendrite may I EPSP was lower than the threshold when current was injected thus be disassociatedfrom those of the soma, and the attributes through the recording electrode. This implies that distal syn- of one may not fully predict the attributes of the other. A full apsesmay commonly trigger regenerative events within their characterization of a pyramidal cell will require separateiden- local region of dendrite, in most types of layer V neuron. tification of its somatic, basal, and apical dendritic physiology (P. A. Rhodes and C. M. Gray, unpublished observations). Discussion Physiological and morphological diversity among apical Origin of dendritic spikes dendrites Ca*+-dependent dendritic spikesare probably generatedby Ca*+ This study confirms that the apical dendrites of many layer V channels in the dendrites, for the following reasons.First, it is pyramidal cells have complex, nonlinear membrane properties not normally possible to observe slow CaZ+ spikes from the that often allow the generation of Na+- and Ca2+-dependent soma under normal recording conditions, but in dendritic re- action potentials (Amitai et al., 1993). In addition, we found cordings Ca*+ spikes are often prominent. Some intrinsically that the spiking characteristicsof apical dendritesvaried widely, bursting cells generatea slow, Ca2+-dependent wave underlying and that the complexity of spiking correlated with dendritic the burst of fast spikes(Connors et al., 1982; McCormick et al., morphology. In general, apical dendrites with prominent Ca2+- 1985) but its amplitude and duration are much smaller than is dependentspikes and irregular patterns of Na+-dependentspikes common in group II dendrites. Second, activation of layer I (group II) had the most side branches and the largest trunk synapsesoften elicited slow, all-or-nothing events that could be diameters. Dendrites with primarily fast, Na+-dependent spikes selectively blocked by hyperpolarization, suggestinga distal site and more regular, soma-like temporal patterns (group I) tended of slow spike generation. Third, if the sourcesof CaZ+currents to have fewer primary branchesand narrower trunks. are in distal dendrites, it might be expected that the number The Journal of Neuroscience, December 1993, 13(12) 5307

Group I Group II

C. l * d.- . l . . .

.I a . . l

w . . . . b Figure 8. Plots of fast-spike height and maximal upstroke slope versus the dis- ml 100 500 tance between the soma and the re- cf~tance from~ma (urn) distance fro?wma (urn). , cording site for group I and II dendrites. The group I spike height (A) and max- imal slope (B) decreased as the record- Group I Group II ing distance increased. Both correla- tions are significant (for spike height, r = -0.8 1, p < 0.01; for maximal slope, C9 r = -0.7, p < 0.01). Neither group II . da. spike height (C) nor maximal slope (D) . was correlated with distance from the . soma. In each case, spike heights were a’< .b . measured from the threshold voltage. ‘. l . . Onlv the first fast soike for each cell was . : used. Examples of proximal and distal loo So0 100 So0 group I and II recordings, labeled ad, distance froiioma (um) distance fro?mma (um) are shown in Figure 9A-D, respectively.

and dimension of dendritic brancheswould correlate with the Our data do not clearly addressthe distribution of Na+ and prevalence of Ca*+ spikes. Indeed group II recordings, which Ca*+ channels within the apical dendrites. Becausethe preva- showed more prominent Ca2+ spikes than group I recordings, lence of Ca2+-dependentpotentials seemsto correlate with the had larger apical dendritesand more oblique branches.Because number of oblique dendritic branches,it may be that CaZ+chan- of their proximity to the recording sites, the oblique branches nels are enriched in more distal dendritic branches. Several would be the ones most likely to contribute significant current recent studiesbolster this possibility. Reuveni et al. (1993) have to the apical trunk. shown that Ca*+-dependent plateau potentials recorded in so- Some apical dendrites may also support at least limited Na+ mata of pyramidal cells often decay stepwise rather than spiking, as suggestedby Huguenard et al. (1989). Models of smoothly. Stepsare reproduced in their computational model passiveapical dendrites predict that spike size and rate-of-rise if high-threshold Ca2+ channels are distributed nonhomoge- should fall precipitously with distance from a somaticgeneration neously through the dendritic tree. Some of our dendrite re- site (Fig. 10). However, the size and rate-of-rise of Na+ spikes cordings in the presenceof TTX and TEA (Fig. 11) also showed from group II recordings did not correlate well with distance stepwisedecay of plateau potentials. Jaffe et al. (1992) using from the soma. Somedendritic sites300-400 pm from the soma intracellular indicators of Na+ and Ca2+ in hippocampal py- generatedrelatively largefast spikes,implying that the Na+ spike ramidal cells, concluded that active Na+ channels are located is not simply passively conducted from the soma. It is conceiv- primarily on the somaand most proximal dendrites,while CaZ+ able that the spike height may be somewhatinfluenced by sub- channelsare spreadalong the entire dendritic tree. Finally, mod- sequentCa2+ spikes (Fig. 7). However, since Ca*+spikes always eling resultsof Mel (in press)suggest that the apical trunk of a followed the initial fast spike it seemsunlikely that it might pyramidal cell is likely to be lessactive compared to more distal have boosted the fast-spike amplitude. It is even less likely to have influenced the slopeof the spike upstroke, since the rising phaseof the first spike substantially precededthe following Ca*+ A spike (Fig. 64. In addition, layer I EPSPs often triggered low- C threshold, fast spikesin apical dendrites. Thus, our data imply that group II dendritescan generateNa+ as well as CaZ+spikes. It is lesslikely that group I dendrites have relatively high Na+ channel densities. Their spikes fell in size and speed as the recording site was moved away from the soma (Fig. 8). The repetitive firing patterns of group I recordings also resembled somatic Na+ spike patterns (Fig. 5). Unlike spike amplitudes, the dendritic spike half-widths were not correlated with distance from the somata (not shown). However, spike duration is sen- sitive to many things besidespassive electrotonic propagation. Dendritic Ca*+ currents, variations in K+ currents or specific Figure 9. Examples of group I (A and B) and II (C and D) recordings, marked in Figure 8 as a&. A, Resting potential, -52 mV; 100 pA membraneresistivity, or small numbers of dendritic Na+ chan- current; time calibration, 90 msec. B, Resting potential, -58 mV, 100 nels themselvescould affect duration. It is notable that layer I pA steps; time calibration, 40 msec. C, Resting potential, -64 mV, 100 EPSPs triggered fast spikes in group I dendrites, but their site pA steps; time calibration, 50 msec. D, Resting potential, -65 mV, 300 of origin is ambiguous. and 400 pA steps, time calibration, 50 msec. 5308 Kim and Connors - Excitability of Apical Dendrites in Neocortex

A Na current density=

$ ,EJ 80

% 60 8 B 200uM Ni++ ?!i 40 rR 20 0.000 0 I 1 I 1 0 200 400 600 Distance from soma (urn) C 500 uM Ni++ B 5001

Figure 11. Effects of Ni*+ on dendritic Caz+ spikes. A, Dendrite was bathed in 15 mM TEA and 2 PM TTX to isolate Ca2+ -dependent events. B, At 200 PM, Ni*+ had no effect. C, At 500 FM, Ni2+ increased the threshold of Ca2+ spikes. At 3 mM Ni 2+ spikes were blocked (not shown). Resting potential was -43 mV, current increments were 100 pA, be- ginning at 500 pA. 0 200 400 600 Distance from soma (urn) be several varieties of high-threshold Ca2+ currents (i.e., N, L, Figure 10. Modeling results showing computed relationship between and P) present in a single cortical neuron (Brown et al., 1992), fast-spike height (A) and maximal slope (B) as a function of distance but the relative densitieshave not been determined. along the apical trunk and Na+ channel density. The neuron modeled The Ca2+ currents present in apical dendrites seem to be was a large layer V pyramidal cell similar to those studied here (see generatedprimarily by high-threshold channels.Dendritic CaZ+ Cauller and Connors, 1992). spikes were always activated at high voltage levels, and they often did not inactivate over several hundred milliseconds(Figs. branches;if slow, active conductancesare too prominent in the 2,6). Pharmacologicalstudies also agreewith this interpretation. proximal trunk, the soma behaves unrealistically. Only more In most neurons, a relatively low dose(20 PM) of Cd2+ blocks refined measurementswill illuminate the channel distribution high-threshold Ca*+ currents (i.e., L- and N-type) while not on apical dendrites. affecting low-threshold currents (i.e., T-type) (Fox et al., 1987). The partial effect of Cd2+ that we observed here (Fig. 12) may reflect the presenceof some relatively Cd2+-insensitive Ca2+ Properties of dendritic Ca2+ currents currents, or it may simply be that the concentration of Cd*+ The characterization of Ca2+currents in neocortical neurons is within the slice did not completely equilibrate with the bath very incomplete. In general, two classesof neuronal Ca2+chan- level. Ni*+ (< 100 PM) preferentially blocks low-threshold CaZ+ nels can be distinguished by kinetic criteria: a slowly inactivat- currents (Carbone et al., 1990). In the dendrites, NiZ+ at rela- ing, high voltage-threshold variety, and a rapidly inactivating, tively high concentrations (200 PM) did not block the Ca2+cur- low voltage-threshold type (Tsien et al., 1988; Carbone and rents, while Cd2+ at relatively low concentrations reduced the Swandulla, 1989; Llinds et al., 1989). Somatic recordings pro- potential plateau. Consistent with the physiological data, im- vide evidence for both low-threshold (Friedman and Gutnick, munocytochemical evidence implies that there are N-type cal- 1987; Sutor and Zieglgansberger,1987; Sayer et al., 1990; Giffin cium channels along apical dendrites of the neocortex (Westen- et al., 1991) and high-threshold (Stafstrom et al., 1985; Franz broek et al., 1992). Rigorous characterization of dendritic Ca2+ et al., 1986; Sayer et al., 1990; Giffin et al., 1991; Hamill et al., currents will require voltage-clamp measurementsand phar- 1991) Ca2+currents in pyramidal cells of neocortex. There may macological studies. The Journal of Neuroscience, December 1993, 73(12) 5309

B

10 uM Cd++ 1OmVl -73 mV 4 n, 2OUlSCC

recovery

Figure 13. Fast and slow dendritic spikes evoked by layer I synaptic potentials. A, EPSPs generated by increasing stimulus levels of stimulus intensity while recording from a group II dendrite. As the stimulus intensity increased, multiple all-or-none events with different ampli- tudes and durations were generated. Resting potential was -63 mV. B, 10mVI A different group II dendrite produced two all-or-none events with 50mscc different amplitude and duration when activated from layer I at a resting potential of -56 mV. Hyperpolarizing the cell (-73 mV) blocked the Figure 12. Effects of Cd2+ on dendritic Ca2+ spikes. A relatively low slow spike component but not the fast spike. This dendrite’s response concentration of Cd2+ (10 PM) reduced the plateau potential, but did to injected currents is shown in Figure 7A. C and D) illustrate two not completely block slow spikes. Resting potential was -62 mV. different group J dendritic recordings while activating layer I at increasing intensities. Each generated all-or-none fast- and slow-spike events at relatively low thresholds. Functional implications of excitable dendrites The precise functions of active dendritic conductancesare un- currents are necessaryboth for efficient communication of distal known. One ofthe most straightforward possibilitiesis that they synaptic inputs, and to allow richer computational capabilities. facilitate the otherwise small effectsof very distal synapses(e.g., The efficacy of distal synapseswill be strongly influenced by Spencerand Kandel, 1961; Deschenes,198 1; Amitai et al., 1993). any change in dendritic membrane properties, including the In this scenariothe depolarizations produced by EPSPs would, selective modulation of dendritic ion channels. For example, by activating voltage-dependentinward currents, generatemuch synaptic inhibition reduces the effect of layer I EPSPs on the larger potentials over a longer distancethan ispossible in passive somata of layer V neurons (Cauller and Connors, in press), dendrites. It is very likely that active currents do amplify EPSPs presumably by shunting synaptic and active currents of the api- in the apical dendritesof neocortical layer V cells. Purely passive cal dendrite. Since most (or all) membrane ion channels are cable models of these cells invariably predict that excitatory modifiable by a variety of secondmessenger-mediated processes synapsesin layer I, even if they are very strong, could generate (McCormick, 199l), it is possiblethat neurotransmitter-medi- only very small, slow EPSPs within the soma (Stratford et al., ated alterations in Na+, Ca2+,or K+ currents could changethe 1989; Cauller and Connors, 1992). However, experimental re- effective gain of distal inputs. cordings show that layer I synapsescan be very powerful at the level of the layer V soma (Cauller and Connors, 1992, in press). Additional, more exotic, functions for active dendrites have References beenproposed. Excitability in dendritic membranegreatly alters Agmon A, Connors BW (199 1) Thalamocortical responses of mouse the rules by which sets of synapseson a single neuron interact. somatosensory (barrel) cortex in vitro. Neuroscience 41:365-380. Several modeling studies have noted that excitable dendrites Agmon A, Connors BW (1992) Correlation between intrinsic firing could have a tendency to generatemultiplicative computations patterns and thalamocortical responses of mouse barrel cortex neu- rons. J Neurosci 12:3 19-330. (e.g., Shepherd and Brayton, 1987; Segevand Rall, 1988; Koch Amitai Y, Friedman A, Connors BW, Gutnick MJ (1993) Regener- and Poggio, 1992). Under certain conditions, active dendritic ative activity in apical dendrites ofpyramidal cells in neocortex. Cereb branches could act as coincidence detectors (Jaslove, 1992). Cortex 3:26-38. With strong,voltage-dependent inward currents, dendritesmight Arikuni T, Ochs S (1973) Slow depolarizing potentials and spike gen- eration in pyramidal tract cells. J Neurophysiol 36: l-l 2. also favor the coactivation of clustered synapsesover more Benardo L, Masukawa L, Prince DA (1982) Electrophysiology of iso- spatially distributed synapsesand improve the capacity for in- lated hippocampal pyramidal dendrites. J Neurosci 2: 16 14-1622. formation processing(Mel, 1992). It may be that active dendritic Blanton MG, LoTurco JJ, Kriegstein AR (1989) Whole cell recording 5310 Kim and Connors * Excitability of Apical Dendrites in Neocortex

from neurons in slices of reptilian and mammalian . Larkman AU (199 1b) Dendritic morphology of pyramidal neurons in J Neurosci Methods 30:203-2 10. the visual cortex of the rat: II. Parameter correlations. J Comp Neurol Brown AM, Schwindt PC, Crill WE (1992) Kinetics and voltage de- 306:320-33 1. pendence of the high threshold calcium current in rat neocortical Larkman AU, Mason A (1990) Correlations between morphology and neurons. Sot Neurosci Abstr 18:430. electrophysiology of pyramidal neurons in slices of rat visual cortex. Carbone E, Swandulla D (1989) Neuronal calcium channels: kinetics, I. Establishment of cell classes. J Neurosci 10:1407-1414. blockade and modulation. Prog Biophys Mel Biol 54:3 l-58. Llinas R, Nicholson C (197 1) Electrophysiological properties of den- Carbone E, Sher E, Clementi F (1990) Ca currents in human neuroblas- drites and somata in alligator Purkinje cells. J Neurophysiol 34:534- toma IMR32 cells: kinetics, permeability and pharmacology. Pflue- 551. gers Arch 416:170-179. Llinas R, Sugimori M (198 1) Electrophysiological properties of in vitro Cauller LJ, Connors BW (1992) Functions of very distal dendrites: Purkinje cell dendrites in mammalian cerebellar slices. J Physiol (Lond) experimental and computational studies of layer I synapses on neo- 305:197-213. cortical pyramidal cells. In: Single neuron computation (McKenna T, Llinas R, Sugimori M, Lin J-W, Cherksey B (1989) Blocking and Davis J, Zometzer SF, eds), pp 199-229. Boston: Academic. isolation of a from neurons in mammals and cenh- Cauller LJ, Connors BW (in press) Synaptic physiology of long hori- alopods utilizing a toxin fraction (FTX) from funnel-web spider poi- zontal afferents to layer I of primary somatosensory cortex in rats. J son. Proc Nat1 Acad Sci USA 86:1689-1693. Neurosci, in press. Mason A, Larkman A (1990) Correlations between morphology and Chagnac-Amitai Y, Luhmann HJ, Prince DA (1990) Burst generating electrophysiology of pyramidal neurons in slices of rat visual cortex. and regular spiking layer 5 pyramidal neurons of rat neocortex have II. Electrophysiology. J Neurosci 10: 1415-1428. different morphological features. J Comp Neurol 296:598-6 13. McCormick DA (199 1) Neurotransmitter actions in the thalamus and Connors BW, Gutnick MJ (1990) Intrinsic firing patterns of diverse cerebral cortex and their role in neuromodulation of thalamocortical neocortical neurons. Trends Neurosci 13:99-104. activity. Prog Brain Res 39:337-388. Connors BW, Gutnick MJ, Prince DA (1982) Electrophysiological McCormick DA, Connors BW, Lighthall JW, Prince DA (1985) Com- properties of neocortical neurons in vitro. J Neurophysio148: 1302- parative electrophysiology of pyramidal and sparsely spiny neurons 1320. of the neocortex. J Neuriphysibl54:782-806.- - - Deschenes M (198 1) Dendritic spikes induced in fast pyramidal tract Mel BW f 1992) NMDA-based nattem discrimination in a modeled neurons by thalamic stimulation. Exp Brain Res 43:3&t-308. cortical neuron. Neural Comp 4:502-5 16. Edwards FA. Konnerth A. Sakmann B. Takahashi T f 1989) A thin Mel BW (in press) Synaptic integration in an excitable dendritic tree. slice preparation for patch clamp recordings from nkuronks of the J Neurophysiol, in press. mammalian central . Pfluegers Arch 4 14:6OC-6 12. Pockberger H (199 1) Electrophysiological and morphological prop- Feldman ML (1984) Morphology of the neocortical pyramidal neuron. erties of rat motor cortex neurons in vivo. Brain Res 539: 18 l-l 90. In: Cerebral cortex, Vol 1, Cellular components of the cerebral cortex Purpura DP (1967) Comparative physiology of dendrites. In: The (Peters A, Jones EC, eds), pp 123-200. New York: Plenum. neurosciences (Quarton C. Melnechuck T, Schmitt FO, eds). __pp 372- Fox AP, Nowycky MC, Tsien RW (1987) Kinetic and pharmacological 393. New York:-Rockefeller UP. properties distinguishing three types of calcium currents in chick sen- Purnura DP. Shofer RJ f 1964) Cortical intracellular ootentials durina sory neurones. J Physiol (Lond) 394: 149-l 72. augmenting and recruiting responses. I. Effects of injected hyperpo- Franz P, Galvan M, Constanti A (1986) Calcium-dependent action larizing currents on evoked membrane potential changes. J Neuro- potentials and associated inward currents in guinea-pig neocortical physiol 27: 117-132. neurons in vitro. Brain Res 3661262-27 1. Purpura DP, Shofer RJ, Scarff T (1965) Properties of synaptic activ- Friedman A, Gutnick MJ (1987) Low-threshold calcium electroge- ities and spike potentials of neurons in immature neocortex. J Neu- nesis in neocortical neurons. Neurosci Lett 8 I : 117-l 22. rophysiol 28:925-942. Giffin K, Solomon JS, Burkhalter A, Nerbonne JM (I 99 1) Differential Rail W (1977) Core conductor theory and cable properties of neurons. expression of voltage-gated calcium currents in identified visual cor- In: Handbook of physiology, Set 1, The nervous system, Vol 1 (Kan- tical neurons. Neuron 6:321-332. de1 ER, ed), pp 39-98. Bethesda: American Physiological Society. Hamill OP, Marty A, Neher E, Sakmann B, Sigworth FJ (1981) Im- Reaehr WC. Konnerth A. Armstrona CM (1992) Sodium action DO- proved patch-clamp techniques for high-resolution current recording t&tials in’ the dendrites of cerebeliar Purkinje cells. Proc Nat1 Acad from cells and cell-free membrane patches. Pfluegers Arch 391:85- Sci USA 89~5492-5496. 100. Reuveni I, Friedman A, Amitai Y, Gutnick MJ (1993) Stepwise re- Hamill OP, Huguenard JR, Prince DA (199 1) Patch-clamp studies of polarization from Ca*+ plateaus in neocortical pyramidal cells: evi- voltage-gated currents in identified neurons of the rat cerebral cortex. dence for nonhomogeneous distribution of HVA Ca2+ channels in Cereb Cortex 1:48-6 1. dendrites. J Neurosci 13:4609-462 1. Hillman DE (1979) Neuronal shape parameters and substructures as Ross WN, Lasser-Ross N, Werman R (1990) Spatial and temporal a basis of neuronal form. In: The neurosciences. Fourth study section analysis’ of calcium-dependent electrical activity in guinea pig-Pur- (Schmitt FO, Worden FG, eds), pp 477498. Cambridge, MA: MIT kinje cell dendrites. Proc R Sot Lond [Biol] 240: 173-185. Press. Sakmann B, Neher E (1983) Single channel recording. New York: Hines M (1989) A program for simulation of nerve equations with Plenum. branching geometries. Int J Biomed Comput 24:55-68. Sayer RJ, Schwindt PC, Grill WE (1990) High-threshold and low- Horikawa K, Armstrong WE (1988) A versatile means of intracellular threshold calcium currents in neurons acutely isolated from rat sen- labelling: injection of biocytin and its detection with avidin conju- sorimotor cortex. Neurosci Lett 120: 175-I 78. gates. J Neurosci Methods 25: l-l 1. Segev I, Rall W (1988) Computational study of an excitable dendritic Huguenard JR, Hamill 0, Prince DA (1989) Sodium channels in spine. J Neurophysiol 60:499-523. dendrites of rat cortical pyramidal cells. Proc Nat1 Acad Sci USA 86: Shepherd GM, Brayton RK (1987) Logic operations are properties of 2473-2477. computer-simulated interactions between excitable dendritic spines. Jaffe DB, Johnston D, Lasser-Ross N, Lisman JE, Miyakawa H, Ross Neuroscience 2 1: 15 l-l 66. WN (1992) The spread of Na+ spikes determines the pattern of Silva LR, Amitai A, Connors BW (199 1) Intrinsic oscillations of neo- dendritic Ca2+ entry into hippocampal neurons. Nature 357!244-246. cortex generated by layer 5 pyramidal neurons. Science 25 1:432-435. Jaslove SW (1992) The integrative properties of spiny distal dendrites. Spencer WA, Kandel ER (1961) Electrophysiology of hippocampal Neuroscience 47:495-5 19. neurons. IV. Fast prepotentials. J Neurophysiol 24:272-285. Kim HG, Connors BW (1992) Calcium currents in the apical dendrites Stafstrom CE, Schwindt PC, Crill WE (1984) Repetitive firing in layer of neocortical pyramidal neurons. Sot Neurosci Abstr 18:2 17. V neurons from cat neocortex in vitro. J Neurophysiol 521264-277. Koch C, Poggio T (1992) Multiplying with synapses and neurons. In: Stafstrom CE, Schwindt PC, Chubb MC, Grill WE (1985) Properties Single neuron computation (McKenna T, Davis J, Zometzer SF, eds), of persistent sodium conductance and calcium conductance of layer pp 3 15-345. Boston: Academic. V neurons from cat sensorimotor cortex in vitro. J Neurophysiol 53: Larkman AU (199 1a) Dendritic morphology of pyramidal neurons in 153-170. the visual cortex of the rat: I. Branching patterns. J Comp Neurol Stratford K, Mason A, Larkman A, Major G, Jack J (1989) The mod- 306:306-3 19. eling of pyramidal neurones in the visual cortex. In: The computing The Journal of Neuroscience, December 1993, 73(12) 5311

neuron (Durbin R, Mial C, Mitchison G, eds), pp 296-321. Menlo Turner RW, Meyers DE, Richardson TL, Barker JL (1991) The site Park, CA: Addison-Wesley. for initiation of action uotential discharee over the somatodendritic Sutor B, Zieglg&tsberger W (1987) A low-voltage activated, fully in- axis ofrat hippocampalCA1 pyramidal &rons. J Neurosci 11:2270- activated, transient calcium current is responsible for the time-de- 2280. pendent depolarizing inward rectification or rat neocortical neurons Westenbroek RE, Hell JW, Warner C, Dubel SJ, Snutch TP, Catterall in vitro. Pfluegers Arch 4 10: 102-l 11. WA (1992) Biochemical properties and subcellular distribution of Telfeian AE, Cauller LJ, Connors BW (199 1) Contribution of apical an N-type calcium channel cul subunit. Neuron 9: 1099-l 115. dendrites to somatic membrane properties of layer V pyramidal cells Wong RKS, Prince DA, Basbaum AI (1979) Intradendritic recordings in neocortex. Sot Neurosci Abstr 17:3 11. from hippocampal neurons. Proc Nat1 Acad Sci USA 76:986-990. Tsien RW, Lipscombe D, Madison DV, Bley KR, Fox AP (1988) Woody CD, Gruen E, McCarley K (1984) Intradendritic recordings Multiple types of neuronal calcium channels and their selective mod- from neurons of motor cortex of cats. J Neurophysiol 5 1:925-938. ulation. Trends Neurosci 11:431-438.