Maturation of layer 5 neocortical pyramidal neurons: amplifying salient layer 1 and layer 4 inputs by Ca²⁺ action potentials in adult rat tuft dendrites

Slice preparation

Experiments were performed in somatosensory neocortical slices from P2-56 Wistar rats. The brain slices were prepared as previously described (Zhu & Lo, 1999). All animal procedures were carried out according to the animal welfare guidelines of the Max Planck Society. In brief, the rats were deeply anaesthetized with halothane and decapitated. Tissue blocks containing the somatosensory cortex were quickly removed into cold (0-4°C), oxygenated physiological solution containing (mM): NaCl 125, KCl 2.5, NaH₂PO₄ 1.25, NaHCO₃ 25, MgCl₂ 1, dextrose 25, CaCl₂ 2, at pH 7.4. The removal procedure was performed as quickly as possible (within 10 s), this appeared to be crucial for obtaining healthy slice tissues from the adult animals (>P42). Sagittal slices, 250-300 μm thick, were then cut from the tissue blocks in cold (0-4°C), oxygenated physiological solution, using a microslicer (Campden Instruments Ltd, UK) set at the slowest possible advancing speed. The slices were then kept in warm (37.0 ± 0.5°C), oxygenated physiological solution for about 1 h before recording. During recording, slices were submerged in a Plexiglass chamber and stabilized using a fine nylon net attached to a platinum ring, perfused with warmed and oxygenated physiological solution. The temperature of the bath solution in the chamber was kept at 35.0 ± 0.5°C.

Physiology

Simultaneous somatic and dendritic recording was performed on single, identified layer 5 pyramidal neurons using infrared differential interference contrast optics. Individual tuft branches (0.5-4 μm in diameter), located as deep as ∼75 μm from the slice surface, could be identified when the optical system was optimally adjusted. The dendritic recordings were made from a primary and/or a secondary tuft branch of the apical dendrite of the pyramidal neurons. Somatic (5-10 MΩ) and dendritic (10-25 MΩ) recording pipettes were filled with the standard intracellular solution containing (mM): potassium gluconate 115, Hepes 10, MgCl₂ 2, MgATP 2, Na₂ATP 2, GTP 0.3, KCl 20, and biocytin 0.25 %, at pH 7.3. Whole-cell recordings were made with two Axoclamp-2B amplifiers (Axon Instruments). Dual recordings from the soma and tuft were typically stable throughout experiments and the recordings were terminated if cells showed sustained change in resting membrane potential or input resistance. The hyperpolarization-activated inward rectification was quantified by measuring the amplitude of the depolarizing sag induced by hyperpolarizing membrane potential to -100 ± 2 mV. The attenuation of voltage responses between the tuft and soma was measured by hyperpolarizing either the tuft or soma to -100 ± 2 mV with a long current pulse. The somatofugal and somatopetal voltage attenuations were typically very similar and the two were averaged. The membrane potential right before the initiation of a regenerative potential was taken as the voltage threshold. An 11 mV liquid junction potential was subtracted from all membrane potentials.

Synaptic stimulation was made using a concentric bipolar electrode with single voltage pulses (200 μs, up to 40 V, 0.25 Hz). The stimulating electrode was typically placed at the border of layer 1 and layer 2, 1200-1500 μm lateral from the cells. A surgical cut was made between the stimulating electrode and the cells, from the middle layer 2 to the white matter, 400-600 μm lateral from the cells (see Fig. 14A). This arrangement allowed us to activate the layer 1/2 synaptic inputs more selectively (cf. Cauller & Connors, 1994; see also Reyes & Sakmann (1999) for the possible activation of the layer 2/3 neuron-relayed inputs).

Histology

After recording, slices were fixed by immersion in 4 % paraformaldehyde in 0.1 M phosphate buffer. The tissue sections were processed with the avidin-biotin-peroxidase method to reveal cell morphology. Cells were then drawn with the aid of a computerized reconstruction system (Neurolucida). The diameters of the apical dendrite were measured at the midpoint between the centre of the soma and the main branch point. Somatic membrane areas (SAs) were estimated from the formula SA =πab, where a is the major diameter and b is the minor diameter of the soma. These values were obtained from the fixed slices after the cell morphology was recovered. Vertical distances of the dendritic recording from the soma and the lengths of the apical dendrite were measured during the recording and later from the fixed tissue. The values obtained from living slices were nearly identical to those measured from the fixed tissue, therefore, no shrinkage corrections were made for the diameters of the apical dendrite and SAs (cf. Stuart & Spruston, 1998).

Identification of pyramidal neurons at different postnatal ages

In slices of 2-day-old (P2) animals, whole-cell voltage recordings were made from pyramidal cells located in the cortical plate, where almost all cells at this developmental stage are destined to be layer 5 pyramidal cells (Berry & Rogers, 1965; Miller, 1981). In slices of 14- (P14), 28- (P28), 42- (P42) and 56-day-old (P56) animals the different cortical layers were readily distinguishable and recordings were made from identified, tufted layer 5 pyramidal cells.

Morphology and membrane properties of dendritic tuft

Maturation of shape and membrane properties

In summary the changes in morphological and passive membrane properties suggest that the apical dendrite reaches its final length during the first 4 postnatal weeks (Fig. 5A), whereas the soma area and diameter of the apical dendrite continue to increase over the next 2 weeks (Fig. 5B and C). The membrane properties mature during the first 6 postnatal weeks. Initially, the resting membrane potentials are similar for the dendritic tuft and soma, then the dendrite becomes gradually less negative and the soma gradually more negative (Fig. 6A). The input resistances of both the tuft and soma drop rapidly during the first 2 postnatal weeks (Fig. 6B). In the following 4 weeks, they slowly reach their final low values. The sag amplitude in the tuft increases gradually in the first 6 postnatal weeks while that in the soma changes little (Fig. 6C). The attenuation of voltage responses between the tuft and soma increase gradually, reaching about 85 % after the first 6 postnatal weeks (Fig. 6D).

Electrical excitability of dendritic tuft

Pharmacology of dendritic regenerative potentials

The increasing duration of the regenerative potentials in the tuft dendrites suggests an increased contribution of the Ca²⁺ conductance (Schiller et al. 1997) and prolongation by back-propagated action potentials. We tested the contribution of the Ca²⁺ conductance by applying sodium and calcium channel blockers TTX, Ni²⁺ and Cd²⁺ (Fig. 11). TTX (1 μm) blocked somatic action potentials at all developmental ages. In the same experiments, TTX also blocked regenerative potentials in the P2 tufts (n = 9), but was progressively less effective on regenerative potentials in the P14 (n = 6), P28 (n = 7), P42 (n = 9), and P56 (n = 5; not shown) tufts. Adding 100 μm Ni²⁺ and 200 μm Cd²⁺ to the bath solution had no further effect in the P2 tufts (n = 6). However, it blocked the residual potentials substantially in the P14 (n = 6), P28 (n = 6), P42 (n = 7), and P56 (n = 5; not shown) tufts.

Dendritic electrical excitability during maturation

The active properties of tufts at different developmental stages are summarized in Fig. 12. The current intensity for evoking dendritic regenerative potentials by current injection (50 ms) increases during the first 4 postnatal weeks (Fig. 12A; P < 0.001), and then remains constant (Fig. 12A; P > 0.05). Because the voltage threshold changes very little, the increase in the current intensity is due mainly to the decrease in input resistance (Fig. 6). The duration of the dendritic regenerative potentials decreases somewhat initially during the first 2 postnatal weeks and then increases in the following 4 weeks. The amplitude of regenerative potentials also increases slightly during development (Fig. 12C). The largest change occurs in the ionic dependence of the regenerative potentials, switching from being mainly Na⁺ dependent to mainly Ca²⁺ dependent (Fig. 12C). Finally, the reliability of the coupling between dendritic and axonal action potential initiation zones, measured by rectangular current injections, initially decreases between P14 and P28 and then increases again in >P42 rats (Fig. 12D; see also results below).

Coupling of dendritic and axonal action potential initiation sites during maturation

The maturation of the electrical excitability of dendritic tufts is expected to affect forward signalling of apical inputs and backward signalling of basal inputs along the main apical dendrite. We attempted to delineate changes of signalling of distal dendritic inputs to the soma by current injection into the dendritic tuft and by selective layer 1 stimulation.

We first examined the effects of the regenerative potentials in the tuft evoked by injecting a depolarizing current at threshold intensity and measured the shape and time of occurrence of the regenerative potentials at the soma and dendrite (Fig. 13). At P2 regenerative potentials were recorded almost simultaneously in the tuft and soma. At P14 and P28 the regenerative potentials in about half the tufts (4 of 7 cells at P14, Fig. 13A; 9 of 19 cells at P28) were not accompanied by somatic action potentials (Fig. 13B). Suprathreshold current injections did, on the other hand, frequently, but not always, elicit somatic action potentials (see also Schiller et al. 1997). In the tufts of the adult cortex (20 of 20 cells, P42 and P56), current injections in the tuft evoked a complex regenerative potential in the tuft and one to four action potentials in the soma (Fig. 13C). When a larger sample of cells were examined, however, we did observe a small population of adult neurons generating isolated dendritic regenerative potentials (2 of 56 cells; authors’ unpublished data). In older animals, the dendritic recordings were often made more distally to the soma and the electrotonic attenuation of steady-state signals was larger. Therefore the higher probability with which threshold current injection into the dendritic tuft evoked both dendritic and somatic action potentials was not due to a shorter electrotonic distance between the site of current injection and the axon. Rather this reflects the fact that more regenerative conductances are activated by dendritic current injection.

The dendritic regenerative potentials and the concomitant somatic action potential pattern evoked by layer 1 stimulation in slices which had a longitudinal cut also changed with development (Fig. 14). Stimulation of the distal layer 1 synaptic inputs could evoke regenerative potentials in the tuft of both young and adult neurons. For P14 (4 of 4 cells) and P28 cells (3 of 4 cells), the regenerative potential at threshold stimulation remained localized to the dendrite (Fig. 14B and E). Suprathreshold stimulation did, in some neurons, evoke somatic action potentials. In contrast, in P42 and P56 cells (n = 8), threshold stimulation elicited a complex regenerative potential in the tuft and concomitantly a single action potential or, more frequently, with a burst of two to three action potentials in the soma (Fig. 14C-E). The pattern of somatic action potentials changed with stimulus strength. The number of somatic action potentials could either increase or decrease with increasing stimulus strength. In addition, the sequence of dendritic and somatic action potentials could vary. At threshold the first somatic action potential could either precede (2 of 8 cells) or follow (6 of 8 cells) the dendritic regenerative potential. The following somatic action potentials (i.e. the second and third action potential) always preceded the respective peaks of the dendritic complex regenerative potential, suggesting that back-propagating axonal action potentials might further broaden the dendritic complex potential.

Morphological maturation and passive electrical properties

The morphological properties of pyramidal cells mature within the first 6 postnatal weeks, consistent with the previous anatomical findings (Petit et al. 1988; Kasper et al. 1994). The development of the electrophysiological properties in the soma has been examined previously using the sharp electrode recording technique (McCormick & Prince, 1987; Franceschetti et al. 1998). Using whole-cell voltage recording with patch pipettes, we found a considerably higher input resistance and a significantly more negative resting membrane potential in the soma of young pyramidal neurons (557 vs. 150 MΩ, respectively; see also Kim et al. 1995). These differences probably result from the fact that sharp intracellular electrodes used previously may introduce a non-selective leak, particularly in the small neurons of young animals (Staley et al. 1992). By recording directly from the dendritic tuft of layer 5 pyramidal neurons, we revealed that the input resistance in the tuft decreases during development, similar to the soma. The large decrease in the input resistance may result from the increasing expression of the conductances that open at the resting membrane potential in layer 5 pyramidal neurons (e.g. I_h, I_M and other potassium conductances; see Hamill et al. 1991; Kang et al. 1996; Zhu et al. 1999).

Our results also indicate that neonatal (P2) pyramidal cells are electrotonically compact because the voltage responses at the soma and tuft are comparable. In contrast, in adult (> P42) pyramidal cells, the dendritic tuft is electrotonically isolated from the soma and axon. This difference appears to result from the alterations in both morphological and physiological properties since both properties change significantly during development. In addition, there is also an intriguing change in resting membrane potential in the soma and the tuft during development. Instead of becoming more hyperpolarized like the soma, the resting membrane potential in the tuft becomes more depolarized. Thus, a prominent potential difference between the soma and the tuft is gradually established during postnatal development. The potential difference probably results from two changes that occur during development. First, there is an increase in electrotonic isolation between the soma and the tuft, which makes the potential difference possible. Second, selectively expressing more depolarizing I_h in the tuft (this study) and more hyperpolarizing potassium conductances in the soma (Kang et al. 1996) charges the potential difference.

Active membrane properties during maturation

The soma of layer 5 pyramidal cells can generate action potentials at all developmental stages (McCormick & Prince, 1987; Kim et al. 1995; Franceschetti et al. 1998). The dendritic tuft of layer 5 pyramidal cells also supports the regenerative potential at all developmental stages, indicating that active conductances in the dendrite are expressed at all stages. The duration of the regenerative potentials evoked by current injection in the tuft increases during development, presumably reflecting an increased contribution of Ca²⁺ conductance. At P14 the duration of dendritic regenerative potentials is slightly shorter than at P2, most probably resulting from a rapid increase in K⁺ conductance during the first 2 postnatal weeks (Hamill et al. 1991; Kang et al. 1996).

The dendritic regenerative potentials switch from being mainly Na⁺ dependent to being dependent on both Na⁺ and Ca²⁺ during development. Their voltage threshold remains quite constant at a low value. On the other hand, the intensity of depolarizing current needed for evoking regenerative potentials in the tuft of older animals increases due to the reduced input resistance. In adult tufts, the average threshold intensity for action potentials is about 1.1 nA, higher than the soma (0.7 nA).

In adult cells, Ca²⁺ channels mediate a substantial part of the regenerative potentials in the tuft. The contribution of Na⁺ channels in these potentials is relatively small and is typically ‘masked’ by the large Ca²⁺-dependent potentials. Clearly, this does not imply that Na⁺ channel density in the dendritic tuft decreases during development since the input resistance of the tuft drops sharply in the same period and a much smaller depolarization will result from the activation of the same number of Na⁺ channels. Because the density of K⁺ channels in neocortical pyramidal neurons increases during development (Hamill et al. 1991; Kang et al. 1996), the broadening of Na⁺- and Ca²⁺-dependent regenerative potentials in the tuft implies a substantial increase in Ca²⁺ channel density in the tuft during the postnatal development. At P42, the regenerative potentials in the tuft reach their final amplitude and duration, suggesting that Ca²⁺ channels may have reached their peak density. The developmental increase in the density of Ca²⁺ channels in the apical dendrite thus parallels a similar increase of Ca²⁺ currents in the soma of cortical cells (Lorenzon & Foehring, 1995).

Using a brief EPSC waveform injection, a small locally restricted regenerative potential can be isolated in the tuft (our unpublished data). The potential is mainly Na⁺ dependent, due presumably to the decoupling of the activation of Na⁺ and Ca²⁺ conductances and failure of generating the complex dendritic potential. This suggests that Na⁺ conductance is essential in initiating the tuft regenerative potentials at low threshold. In addition, T- and R- type Ca²⁺ channels are present in the apical dendrite and have a relatively low threshold for activation (Magee & Johnston, 1995; Markram et al. 1995; Kavalali et al. 1997). These conductances may help to lower the threshold of the dendritic regenerative potentials. Because these Ca²⁺ and Na⁺ channels show prolonged inactivation (Kavalali et al. 1997; Colbert et al. 1997; Jung et al. 1997), they may account for the long refractory period of the dendritic regenerative potentials (our unpublished data; Chang, 1951). N- and P-/Q- channels are the main types of Ca²⁺ channels present in the dendrite (Kavalali et al. 1997). Since these channels inactivate slowly, they may contribute to the prolonged time course of the distal regenerative potentials.

Functional significance of layer 5 pyramid maturation

In addition to the layer 4 inputs received at the basal dendrites, the distal apical dendrites of layer 5 pyramidal neurons are densely innervated by the layer 1 inputs, coming from the cortical feedback and ascending regulatory neural systems (e.g. Herkenham, 1980; Delima & Singer, 1986; Zeki & Shipp, 1988). They may contribute to complex behaviour such as attention and perception (Cauller & Kulics, 1988; Squire & Zola-Morgan, 1991). Previous studies have shown that the regenerative conductances in the apical dendritic trunk can amplify the distal synaptic inputs in pyramidal cells (Kim & Connors, 1993; Magee & Johnston, 1995; Schwindt & Crill, 1995; Gillessen & Alzheimer, 1997). When these tuft conductances are activated, for example by synaptic inputs in vitro (Schiller et al. 1997) or natural stimuli in vivo (Zhu & Connors, 1999), a regenerative dendritic potential is generated which integrates synaptic inputs locally. Therefore, the very distal synaptic inputs can be amplified as early as the signals reach the dendritic arbor.

In this study, we showed that the long dendritic trunk and low input resistance make the tuft dendrite of adult layer 5 pyramidal cells not only distally located, but also electrotonically isolated from the basal dendrite, soma and axon. Correspondingly, in adult pyramidal neurons, large, suprathreshold synaptic inputs from layer 1 can be non-linearly amplified and can trigger axonal action potentials. In contrast, small synaptic inputs, without being amplified by the dendritic regenerative potentials, give only very small depolarizations in the soma. Thus, the isolated distal tuft, capable of generating the dendritic regenerative potentials, can boost selectively the salient distal inputs.

The distal action potential initiation zone can interplay with the axonal action potential initiation zone. This makes the dynamic interactions between the distal and proximal synaptic inputs possible. Indeed, our recent study has shown that coincident apical and basal inputs enhance the axon's ability to detect and amplify the small distal dendritic inputs (Larkum et al. 1999). Namely, the axon initiation site can generate a burst of action potentials when a subthreshold distal dendritic input and a suprathreshold basal dendritic input occur within a short time window of a few milliseconds.

Conclusions

During the early developmental period (≤ P14), new synapses form and mature rapidly at the dendritic arbor of pyramidal cells (Miller & Peters, 1981; Harris et al. 1992; see also Isaac et al. 1997). The relative electrotonic compactness, as well as the reliable back-propagating action potentials (Stuart & Sakmann, 1994), may be important for the dendrite to modulate synaptic efficacy according to the cells’ input and output patterns (Markram et al. 1997). At the later stages, when most cortical synapses have been formed (> P28), the tuft dendrite becomes electrotonically isolated from the soma. The development of the slow and Na⁺- and Ca²⁺-dependent regenerative potentials in the adult tuft may be crucial for its mature functions. For example, the Ca²⁺ action potentials are important for detecting the salient distal synaptic inputs and bridging the electrotonic distance between the tuft dendrite and soma. In addition, they are involved in detecting the coincident synaptic inputs arriving at the apical and basal dendritic arbors (Larkum et al. 1999). Finally, the slow time course allows the Ca²⁺ action potentials to enhance the efficiency of the salient distal inputs and coincident inputs by promoting burst firing, which increases the reliability of the synaptic transmission between layer 5 pyramidal neurons and their target cells (Lisman, 1997, Williams & Stuart, 1999).