| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Howard Hughes Medical Institute, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724
Submitted 23 June 2003; accepted in final form 12 February 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
). This developmental plasticity could involve many mechanisms, most of which we understand poorly (Crair and Malenka 1995; Fox 1995; Hensch et al. 1998; Huang et al. 1999; Kirkwood et al. 1995; Lu et al. 2001; Philpot et al. 2001; Quinlan et al. 1999). The barrel cortex of rodents is an ideal system for studying experience-dependent cortical development (Fox 1992; Simons and Land 1987). A topographical sensory map from the vibrissae (whisker) pad to primary somatosensory cortex (S1) is established during development. Barrels are discrete aggregates of neurons in layer 4 (Welker and Woolsey 1974; Woolsey and Van der Loos 1970) that correspond to representations of particular vibrissae: neurons in a barrel and the adjacent supra- and infragranular regions respond preferentially to a principal whisker (PW) (Armstrong-James and Fox 1987; Simons 1978; Welker 1971). Whiskers surrounding the PW provide weaker and slower inputs (Armstrong-James and Fox 1987; Armstrong-James et al. 1991; Fox 1994; Moore and Nelson 1998; Zhu and Connors 1999). Thus in animals raised with normal sensory experience, the electrophysiological map mirrors the anatomical map. Barrels form soon after birth (Schlaggar and O'Leary 1994) and are resistant to sensory manipulation after an early critical period [postnatal day (PND) 5] (Fox et al. 1996; Henderson et al. 1992; Rice and Van der Loos 1977). However, sensory deprivation by whisker clipping can continue to modify receptive fields in a layer-specific manner (Armstrong-James et al. 1994; Diamond et al. 1994; Fox 1992; Glazewski and Fox 1996; Hand 1982; Lendvai et al. 2000; Simons and Land 1987).
We recently analyzed layer 2/3 subthreshold whisker receptive fields using in vivo intracellular recordings from PND 14 to PND 20 (Stern et al. 2001). We found that trimming all contralateral whiskers from PND 9 disrupted receptive fields: responses to PW stimulation decreased significantly, while surround responses increased. This plasticity had a critical period: deprivation at PND 15 failed to disrupt receptive fields measured at PND 20. Layer 4 responses showed no deprivation-induced changes after clipping from PND 9, ruling out a prominent role for thalamocortical plasticity in layer 2/3 sensory map modification.
The mechanisms underlying this critical period for plasticity and its closure are unknown. Developmental changes in long-term potentiation (LTP) could be involved (Crair and Malenka 1995; Feldman et al. 1998; Isaac et al. 1997; Kirkwood et al. 1995). However, synaptic inputs to layer 2/3 are overwhelmingly intracortical (Armstrong-James et al. 1992; Feldmeyer et al. 2002; Laaris et al. 2000; Petersen and Sakmann 2001), and LTP at these synapses persists past the critical period (Allen et al. 2003; Feldman 2000; Maravall and Svoboda, unpublished observations). Intracortical synaptic density increases several-fold between PND 9 and 15 (De Felipe et al. 1997; Micheva and Beaulieu 1996), implying that synaptic circuits could change rapidly. Indeed, experience-dependent turnover of dendritic spines (Lendvai et al. 2000; Trachtenberg et al. 2002) and dendritic branches (Maravall et al. 2004) suggests that experience-dependent formation and elimination of synapses is involved in experience-dependent plasticity.
Another locus of plasticity could be the intrinsic excitability and input-output relationships of neurons (reviewed in Zhang and Linden 2003). Intrinsic properties including spike generation thresholds and various ionic conductances can be modified by experience (Aizenman et al. 2003; Disterhoft et al. 1986; Saar and Barkai 2003; Schreurs et al. 1998) and by manipulation of activity levels in vitro (Aizenman and Linden 2000; Armano et al. 2000; Daoudal et al. 2002; Desai et al. 1999; Franklin et al. 1992; Ganguly et al. 2000; Garcia et al. 1994; Li et al. 1996; Nelson et al. 2003; Nick and Ribera 2000; Sourdet et al. 2003; Turrigiano et al. 1994). It is unknown whether such activity-dependent modifications of intrinsic properties might contribute to the developmental plasticity of cortical somatosensory responses. We therefore measured the development of intrinsic excitability and spiking properties of excitatory neurons in layers 2/3 and 4 during the critical period mentioned (PND 12–17), using whole cell patch-clamp recordings in an acute barrel cortex (BC) slice preparation. Maturation of excitability and of spiking properties occurs in layer-specific fashion. Sensory deprivation induced by full contralateral whisker clipping delays maturation of spiking properties. These changes in spiking behavior correlate with changes in a Ca2+-dependent slow afterhyperpolarization (AHP). Deprivation does not affect intrinsic excitability as measured by I-f curves.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
All procedures were undertaken in accordance with animal care and use guidelines of Cold Spring Harbor Laboratory. Rat pups were deprived by clipping all whiskers on one side of the snout to <1 mm length, starting at age PND 9 and continuing at intervals of <48 h (recordings were made at PND 12, 14, and 17). Clipping was performed with no anesthesia and brief handling; control animals were also handled during clipping sessions. Deprived animals were kept in the same cages as their mothers and control littermates until the experiment. The experimenter was blind to animal deprivation history until after the experiment's analysis was completed.
Electrophysiology
Neocortical brain slices were prepared from the hemisphere contralateral to the clipped whiskers. The brain was removed and blocked at an angle intermediate between coronal and sagittal (40-45° from the midline, anteromedial to posterolateral) and inclined 
10° from the vertical plane. This angle was approximately parallel to barrel arcs (Feldman 2000; Finnerty et al. 1999; Shepherd et al. 2003). The most posterior slices showing barrels under bright-field illumination were then those containing the large barrels of the posterior medial barrel subfield (PMBSF), corresponding to the longer vibrissae at the posterior edge of the whisker pad. Slices (300 µm thick) were cut on a VT-100 microtome (Leica, Wetzlar, Germany) with the brain submerged in a chilled (2-5°C) cutting solution bubbled with carbogen (95% O2-5% CO2). The solution contained (in mM) 110 choline-chloride, 25 NaHCO3, 25 D-glucose, 11.6 Na-ascorbate, 7 MgSO4, 3.1 Na-pyruvate, 2.5 KCl, 1.25 NaH2PO4, and 0.5 CaCl2. After cutting, slices were transferred to a submerged holding chamber containing artificial cerebrospinal fluid (ACSF) and incubated at 35°C for 25-50 min, and then held at room temperature until used. The composition of the normal ACSF was (in mM) 127 NaCl, 25 NaHCO3, 25 D-glucose, 2.5 KCl, 2 CaCl2, 1 MgSO4, and 1.25 NaH2PO4. All chemicals were from Sigma (St. Louis, MO) unless otherwise noted. Drugs (TTX, CdCl2, 4-aminopyridine) were routinely added to the bath perfusion.
Recordings were performed only within slice regions that contained barrels of the PMBSF identified under transillumination. Neurons were identified visually and selected for whole cell recording using IRDIC optics. Cells were selected according to morphology (pyramidal if in layer 2/3, stellate if in layer 4) and position. Layer 2/3 pyramids were 200-300 µm from the pia; this helped to limit the variability across cells at different depths (Peters and Jones 1984). Neurons with transversal apical arbors (usually adjacent to layer 1) were not used (van Brederode et al. 2000).
Experiments were performed at 34 ± 1°C (mean ± SD). Patch electrodes (3-6 M
) were filled with an internal solution containing (in mM) 135 K+-methylsulfonate, 10 HEPES, 10 Na-phosphocreatine, 4 MgCl2, 4 Na2-ATP, and 0.4 Na-GTP. For calcium imaging experiments, the 1,2-bis(2-aminophenoxy) ethane-N,N,N',N'-tetraacetic acid (BAPTA)–based dye Oregon Green BAPTA-1 (OGB-1; 40 or 100 µM, Molecular Probes, Eugene, OR) was added to the solution.
|
; Maravall et al. 2000). Acquisition protocols began 1-2 min after break-in to whole cell configuration. During calcium imaging experiments, immediately on breaking into whole cell mode, a position was found on the proximal apical dendrite. Laser scanning was then changed from frame-scan, two-dimensional mode to line-scan mode over a one-dimensional line across the dendrite. Calcium imaging began 2 min after break-in and ended when loading approached a steady state 
40 min later (Maravall et al. 2000).
Stimuli were delivered in current clamp mode. To construct current-frequency curves (Fig. 2), square pulses (500-ms duration) were given in increases of 50 pA; currents ranged from 50 to 700-1,000 pA. Individual action potentials (APs) were evoked using short (2-5 ms) current pulses: to characterize afterhyperpolarizing potentials, AHPs, APs were evoked either individually or as sequences of 10 pulses at 100 Hz (Fig. 3 A). To calibrate [Ca2+] from fluorescence recordings, sequences of 40-50 (83 Hz) or 48-60 (100 Hz) APs were interleaved with single APs (Maravall et al. 2000).
|
|
Electrophysiology and calcium-imaging analyses were carried out using custom software (written in Igor). Neurons were classified as phasic or regular-spiking depending on whether they could spike throughout the duration of 500-ms-duration depolarizing current pulses. If a neuron could spike throughout, it was scored as regular. If it never fired spikes in the last 200 ms of a 500-ms pulse of any current size, it was phasic. For further analysis of spiking behavior we used the "spike ratio, R," defined as the number of spikes fired during the second 100 ms of a stimulus, divided by the number of spikes fired during the first 100 ms. R was only measured for responses with two or more spikes. To compare R across experiments, we used the maximal R collected during the experiment. This was consistent with our scheme for scoring neurons as RS or PS.
Action potentials were characterized by their threshold as measured at the inflection point, by their height (from threshold to peak) and by their width measured at half-maximum height. Our results may underestimate the magnitude of age-related increases in AP amplitude and decreases in AP duration due to our use of the Axopatch 200B amplifier, which is not an optimal voltage follower.
Voltage deflections evoked by 500-ms current pulses were measured at plateau level (for positive currents) and at peak and plateau level (for negative currents). Sag potentials measured with negative current steps were computed as the difference between peak deflection and plateau deflection. Rebound potentials occurring on negative pulse offset were also analyzed and their comparison gave similar results to sag potentials. Current-voltage (I-V) curves were computed using subthreshold stimulation intensities. For negative currents, inward rectification was measured at peak deflection, giving an estimate of the relative strength of fast inwardly rectifying currents. For positive currents, outward rectification was measured only in experiments where APs were abolished by TTX (Fig. 4). In such cases, both early and late outward rectification were measured. Input impedances were measured in voltage clamp using single-exponential fits to a current step, and in current clamp by determining the slope of the current-voltage curve at small negative currents (–100-0 pA). Current-frequency (I-f) curves were constructed and smoothed (3-point boxcar).
|
). To enable calibration of fluorescence transients in terms of absolute [Ca2+], the high-affinity indicator OGB-1 was chosen. OGB-1 saturates at comparatively low [Ca2+] increases (KD 
205 nM). To measure resting calcium ([Ca2+]0), action potential-evoked Ca2+ transients (
[Ca2+] = [Ca2+] – [Ca2+]0), and the buffering capacity of internal calcium buffers (
), we used previously published methods (Maravall et al. 2000; Sabatini et al. 2002). |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
We compared functional properties of layer 2/3 pyramidal neurons and layer 4 spiny stellate neurons from the BC of control and deprived rats at PND 12, 14, and 17. Whole cell patch recordings in current-clamp mode (Fig. 1) were performed in pyramidal and spiny stellate cells from slices of BC. Characterization of spiking behavior was performed in a total of n = 162 neurons, all of which had resting potentials more negative than –55 mV, had overshooting APs and were able to fire repeatedly in response to depolarizing current pulses. Subgroups of these recordings were used for additional experimental protocols and analyses.
Spiking behavior
On stimulation with a square current pulse (500 ms) delivered through the patch pipette, most cells from older (
PND 14) animals responded with the regular-spiking (RS; Fig. 1A) patterns typical of mature supragranular pyramidal neurons (Amitai and Connors 1995; Connors and Gutnick 1990; McCormick et al. 1985). RS neurons have moderate spike-frequency adaptation (SFA): they continue to fire throughout the stimulation pulse. However, pyramidal cells from younger animals often showed phasic-spiking (PS; Fig. 1B) patterns with complete SFA. In these neurons, there were never any spikes in the last 200 ms of a 500-ms pulse. PS and RS spiking patterns are easily classified and completely distinct (see METHODS). Phasic spiking patterns have been observed previously in vitro in immature excitatory neurons from various cortical regions (Locke and Nerbonne 1997; Lorenzon and Foehring 1993; Metherate and Aramakis 1999) and also in vivo (Dégenètais et al. 2002). They differ from the stereotyped burst responses shown by intrinsically bursting pyramidal neurons in layer 5. Because the more mature RS neurons sustain spiking for longer, they can potentially generate a larger number of APs in response to a stimulus.
We scored cells according to their characteristic spike patterns as either regular or phasic spiking (RS or PS). The fraction of RS cells in layer 2/3 increased gradually with age both in control and deprived animals (Fig. 1C), progressing from 44% (n = 4 of 9) in both conditions at PND 12, through 65% of control cells (n = 15 of 23) and 48% of deprived cells (n = 12 of 25) at PND 14, to 91% of control cells (n = 20 of 22) and 85% (n = 17 of 20) of deprived cells at PND 17. Therefore expression of mature spiking patterns is essentially complete at PND 17, by the end of the critical period for sensory response plasticity in layer 2/3 (Stern et al. 2001).
Maturation of spiking patterns was significantly delayed by whisker deprivation (Fig. 1C; 3-way table, log-linear independence test, P < 0.004). Deprivation decreased the fraction of RS cells at ages PND 14 and 17 (Fig. 1C); its effect became significant when comparing across ages.
In layer 4, 100% of cells showed RS behavior at age PND 14 both in control and deprived conditions (n = 6 of 6 and 4 of 4, respectively; Fig. 1C). Therefore at this age, layer 4 spiny stellate spiking behavior was already as for mature spiny stellate neurons. This is consistent with the earlier critical period for response plasticity observed in this layer (Fox 1992).
We conducted further analysis of spiking behavior through the spike ratio R. Stimuli had 500-ms duration, and R was equal to the ratio of the numbers of spikes fired during the second and first 100 ms after stimulus onset (see METHODS). R correlated very well with spiking behavior as scored by the experimenter: RS neurons had a significantly higher R value than PS neurons (Fig. 1D; Mann-Whitney U test, P < 0.0001). Consistent with this, R values increased during development (Fig. 1E). Whisker deprivation also decreased R in a manner compatible with a deprivation-induced delay in maturation (Fig. 1E). Both age and deprivation had a statistically significant effect on R values (2-way ANOVA, after log-ANOVA test for homogeneity of variances; age, P < 0.02; deprivation, P < 0.05; sample sizes: n = 23 for PND 12 control, n = 21 for PND 12 deprived, n = 35 for PND 14 control, n = 32 for PND 14 deprived, n = 17 for PND 17 control, n = 22 for PND 17 deprived). R values of PND 14 spiny stellate neurons were close to 1 and were not significantly different for control and deprived neurons (control, n = 7; deprived, n = 5; Fig. 1E).
Consistent with previous studies, we found that the amount of SFA of a cell and therefore its spiking pattern could vary during the course of experiments (Borde et al. 1995, 2000; Lorenzon and Foehring 1995; Zhang et al. 1995). Soon after breaking into whole cell configuration (sometimes as early as 2-3 min after break-in), RS cells often displayed increased adaptation, switching eventually (5-10 min) to PS behavior. For younger animals, the great majority of RS neurons underwent this change. These cells were healthy by all electrophysiological criteria including stability of resting membrane potential, input impedance, and spike parameters, and all experiments were performed using internal solution based on K+-methylsulfonate, which provides the most stable recording conditions (Velumian et al. 1997; Zhang et al. 1994). Furthermore, we excluded Ca2+ buffers from the pipette solution, so the change was not due to the effects of BAPTA or other buffers (Lancaster and Batchelor 2000; Lorenzon and Foehring 1995; Schwindt et al. 1992; Zhang et al. 1995). Finally, stability of spike patterns did not appear to depend on the history of activity: neurons switched spiking behavior even if kept silent during the first few minutes of a recording. The mechanisms determining regular or phasic spiking could therefore undergo rapid changes. Classifying RS cells into "regular" and "switching" categories according to whether they did or did not switch spiking patterns under these conditions revealed a significant difference in the distribution of categories between control and deprived groups (Fig. 1F; G-test, P < 0.025).
Membrane potential, passive properties, and spike properties
We measured basic intrinsic properties (resting membrane potential, input impedance, membrane time constant, spike height and width, absolute and relative spiking threshold: Tables 1 and 2). Developmental changes in layer 2/3 cells included a decrease in input impedance, a small hyperpolarization in resting potential, a faster membrane time constant, and larger and faster action potentials with a lower threshold. These changes agreed with those previously found in other cortical neurons (Kasper et al. 1994). We found no differences in any of these properties between cells from control and deprived animals at any age.
|
|
, layer 2/3: 213 ± 14 M; t-test, P < 0.05), as expected based on their small size. Current-frequency relationships
We next evaluated the relationship between stimulation current intensity and instantaneous spiking frequency (calculated for the 1st 2 APs fired; Fig. 2A). This provided a measure of neuronal excitability during the early part of responses, before the onset of SFA. Current-frequency (I-f) curves averaged over all PND 14 cells are plotted in Fig. 2B, showing the different responses of layer 4 and 2/3 neurons. Current-frequency relationships were not experience-dependent at any age. Layer 2/3 neurons were responsive at intensities as low as layer 4 spiny stellate cells, even though these have higher input impedances (Table 1) by virtue of their morphology.
We further characterized each cell's I-f curve using its maximum slope (fmax) and the current intensity at which the maximum slope was reached [I(fmax)]. By definition, I(fmax) is the current at which a cell is maximally responsive, i.e., maximally able to modulate its spike frequency in response to changes in input. More excitable cells are maximally responsive at smaller inputs [smaller I(fmax)] than less excitable cells. Maximum slope, fmax, was not affected by deprivation (Fig. 2C). Because the data did not vary significantly across control and deprived conditions for any of the age groups (t-test for means, F test for variances), we could pool data from control and deprived conditions for each age. This revealed a nonsignificant developmental increase in fmax for layer 2/3 cells at PND 17. The fmax of layer 4 cells and layer 2/3 cells differed significantly at PND 14 (t-test, P < 0.0001): this is explained simply by the ability of spiny stellate cells to fire at higher frequencies.
Deprivation had no effect on the current at which the maximum slope was reached, I(fmax; Fig. 2D). As for effects of development, although I(fmax) did not change from PND 12 to 14, it decreased significantly from PND 12 to 17 (Fig. 2D; t-test, P < 0.05; data pooled across conditions after satisfying the F test for homogeneity). The decrease in I(fmax) implies that older cells are more responsive to small inputs, in spite of their smaller input impedance (Table 1).
Because a cell's I(fmax) is affected by its passive properties (lower impedance implies that a higher current is necessary to achieve a given depolarization), we scaled I(fmax) by the cells' input resistances measured at resting potential. The resulting number V* is not a measure of true depolarization or excitation in response to realistic inputs, but allows comparison across groups (Fig. 2E). V* was again not affected by deprivation. Although V* did not change in layer 2/3 neurons from PND 12 to 14, it decreased significantly from PND 14 to 17 (t-test, P < 0.01 from either of the earlier ages to PND 17; data pooled across conditions after satisfying the F test for homogeneity). Therefore maturation of membrane properties shifts the I-f curve to the left (i.e., cells are responsive to smaller net inputs). Layer 2/3 excitability as measured by V* was higher than for layer 4 (Fig. 2E), allowing cells to be sensitive to weaker synaptic input (Feldmeyer et al. 2002).
Spiking behavior and slow AHP
We next focused our analysis on neuronal properties that correlated with spiking behavior. Spike frequency adaptation appeared highly correlated with the magnitude of a slow AHP. However, prominent fast AHPs could be present in neurons regardless of age or spike frequency adaptation behavior. We therefore quantified AHPs using a standardized protocol that consisted of stimulating cells with 10 APs fired at 100 Hz (Fig. 3A). The overall AHP varied in magnitude, time to peak, and duration, indicating the varying contribution of a slow AHP with a time to peak of several hundred (500) ms and a time course of 1-1.5 s (Fig. 3A).
Grouping neurons by their spiking behavior (RS and PS groups; only PND 17 cells with stable spiking behavior were included in this analysis) revealed variations in slow AHP across spiking type. RS and PS cells did not differ significantly in the magnitude of the fast AHP peak. However, slow AHPs (time to peak: 500 ms) were significantly more prominent in PS neurons (Fig. 3A), where they dominated the overall AHP, as reflected in the time T elapsed from stimulation onset to overall AHP peak (Fig. 3, B and C). When T was plotted against the spike ratio R, data points clustered into groups corresponding to the PS and RS categories (Fig. 3B). In PS neurons at PND 17 (Fig. 3C) T = 291 ± 27 ms (n = 5) while in RS neurons T = 159 ± 8 ms (n = 6). This difference was significant (t-test, P < 0.01). Further, AHP amplitudes measured 400 ms after stimulation offset (a time when the slow AHP is the major contributor to the AHP waveform; Fig. 3D) were (PS neurons, n = 5) VAHP, 400 = 7.1 ± 1.1 mV, (RS neurons, n = 6) VAHP, 400 = 3.6 ± 1.0 mV. These values were significantly different (t-test, P < 0.05). In a group of neurons that switched from RS to PS behavior during recording (see preceding text), we tracked the evolution of the slow AHP's magnitude as the cells switched behavior. Slow AHP magnitude grew during the experiment in 14 of 16 neurons monitored. In n = 3 switching neurons, trials using AP trains (to measure the AHP) were interleaved with trials using single depolarizing pulses (to determine the spike ratio R): AHP magnitude measured at 7-12 min after break-in was 92 ± 42% larger than measured at 0-5 min after break-in. This change in the AHP occurred while the neurons' spiking was becoming more phasic: the relative decrease in R over the same period was 78 ± 22%. Thus spiking behavior was correlated with slow AHP magnitude.
We next grouped PND 17 neurons by control or deprived condition irrespective of spiking behavior. Deprivation had significant effects on the slow AHP (Fig. 3, E and F): the AHP magnitude measured 400 ms after stimulation offset (Fig. 3F) was significantly increased by deprivation (control neurons: VAHP, 400 = 3.2 ± 0.5 mV, n = 17; deprived neurons: VAHP, 400 = 5.0 ± 0.7 mV; t-test, P < 0.05). However, the time T to overall peak (Fig. 3E) did not differ significantly across groups (control neurons: T = 248 ± 25 ms, n = 17; deprived neurons: T = 255 ± 16 ms, n = 17): because RS and PS neuronal types were distributed across control and deprived conditions, differences in AHP parameters were diluted. Therefore the amplitude of the slow AHP correlates with spiking behavior and is regulated by sensory experience.
Spiking behavior and outward rectification
Comparison of PS versus RS patterns further revealed that PS behavior was correlated with a characteristic response waveform that combined an earlier, slow depolarization peak with a later, less depolarized plateau phase (compare voltage traces in Fig. 1, B and A). This behavior was present also during subthreshold responses (Fig. 4A). In PS neurons, APs were generated only during the early depolarization peak. In contrast, RS neurons never showed this characteristic behavior (Fig. 4B). We thus looked for a possible relationship to the slow AHP. We hypothesized that the end of the peak depolarization phase of PS responses might be the onset of a slow hyperpolarization, causing a late increase in outward rectification (OR). Because the putative OR occurred also during subthreshold responses, we characterized the phenomenon by depolarizing neurons in the presence of TTX (1 µM). Responses were measured on depolarization with 500-ms duration current pulses (Fig. 4, A and B). Measurements were taken at two times: the time of peak response, and late in a pulse (450-500 ms), when depolarization had reached steady state (Fig. 4, A and B). The difference between response magnitudes at these times was used as a measure of OR (Fig. 4C; 500-pA current pulses).
As expected, PS but not RS neurons had a substantial amount of OR that was maintained even after TTX had blocked spiking (PS, Fig. 4A; RS, Fig. 4B; collected data, Fig. 4C). We further quantified the amount of OR using I-V curves for depolarization starting from identical resting potentials. I-V curves were measured early in a pulse (at peak response, see Fig. 4A) and at steady state, late in a pulse (450-500 ms after pulse onset; Fig. 4A). We took the ratio of the I-V slope at rest (i.e., the input resistance) to the I-V slope at strongly depolarizing currents (data not shown). If a significant amount of late OR was present, this ratio was larger for I-V curves at steady state than for I-V curves at peak response. Dividing the late ratio by the peak ratio thus gave an index of late OR. PS cells had a significantly larger late OR index than RS cells (t-test, P < 0.02; n = 6 and n = 2, respectively).
In addition, the late OR correlated strongly with the slow AHP because neurons with a large slow AHP component (as measured by a long time to peak AHP) also had large late OR voltage drops (Fig. 4D). Plotting the OR voltage drop against the spike ratio R showed clustering corresponding to the RS and PS categories, showing that both the slow AHP and the OR covaried with spiking behavior (Fig. 4E). This correlation between the slow AHP and the delayed OR suggested the possibility that they might share an underlying slow hyperpolarizing current that would contribute to terminating spiking.
Regulation of adaptation and spiking behavior
Delayed hyperpolarization mechanisms clearly could play an important role in determining spiking patterns. One candidate mechanism for PS behavior was the comparatively slow and low-threshold "ID" K+ current (Locke and Nerbonne 1997). However, we found that blocking ID with 4-aminopyridine (50 µM; after TTX application) (Fig. 4A, right; repeated in n = 2 experiments) did not reduce OR, indicating that ID did not play a significant role in the determination of PS or RS behavior.
Changes in slow AHPs (Schwindt et al. 1988a, b, 1992) have been shown to be involved in spiking behavior maturation in cortical pyramidal cells (Lorenzon and Foehring 1993). We evaluated whether differences in slow AHPs were involved in shaping the depolarization waveforms distinguishing PS cells from RS cells, and whether these differences were dependent on Ca2+ influx. To do so, we examined whether PS patterns depended on calcium entry by using CdCl2 (50 µM) to block influx through Ca2+ channels. In all cells tested (n = 5), CdCl2 induced RS behavior (Fig. 5 A) and reduced the size of the slow AHP (VAHP, 400 reduced to 0.29 ± 0.09 of baseline value; Fig. 5, B and C). Both the reduction in the slow AHP and the increase in spike ratio R induced by CdCl2 were significant (paired t-test; AHP: P < 0.025; R: P < 0.02; Fig. 5C): PS behavior and the slow AHP were jointly dependent on Ca2+.
|
; Crill 1996).
We next tested for a contribution of changes in Ca2+-mediated depolarizations or in Ca2+ entry to the developmental process leading from PS to RS behavior. First, we tested for participation of T-type Ca2+ currents in the enhanced slow AHP and late OR characteristic of PS behavior by raising the holding membrane potential to between –55 and –50 mV with a positive current, enough to inactivate T-type currents (Hille 2001). Depolarized holding potentials did not change spiking waveforms from PS- to RS-like (Fig. 5D) or reduce slow AHPs (Fig. 5, D and E): instead, raising the holding voltage linearly increased the magnitude of the slow AHP (Fig. 5E; repeated n = 5 times), consistent with a dependence only on the reversal potential of K+. This ruled out a significant effect of T-type Ca2+ influx on spiking behavior and the slow AHP.
We searched for developmental changes in Ca2+ fluxes and handling. [Ca2+] imaging experiments were performed at PND 12-17, using the BAPTA-based dye OGB-1 (see METHODS). Because BAPTA alters Ca2+ chelation and interferes with intrinsic Ca2+ buffers including Ca2+-activated K+ channels (Lancaster and Batchelor 2000), we expected dye perfusion to make immature neurons highly susceptible to switching their spiking behavior. Indeed, moderate concentrations of OGB-1 BAPTA (50-100 µM) always induced switching to PS behavior and enhanced the slow AHP (n = 16; data not shown), consistent with previous reports (Borde et al. 1995; Jahromi et al. 1999; Lorenzon and Foehring 1995; Schwindt et al. 1992; Velumian and Carlen 1999; Zhang et al. 1995) and with the relationship between spiking behavior and the Ca2+-dependent slow AHP (Figs. 3 and 5). For this reason, we did not divide the data into PS and RS cell types. Instead, we classified our [Ca2+] recordings by age group (Table 3) with more immature neurons corresponding to more phasic responses. We found no significant developmental changes in AP-evoked (high-threshold) [Ca2+] increases or in Ca2+ buffer capacity in the proximal apical dendrite (Table 3).
|
). T-type channels were, in addition, ruled out because, similarly to the slow AHP (Fig. 5E), subthreshold [Ca2+] transients did not decrease at elevated holding potentials (Fig. 5F). L-type channels are likely involved because application of nimodipine (10 µM) appeared to eliminate subthreshold Ca2+ entry (n = 3; data not shown) and reduced slow AHP magnitude in PS neurons (by 35.5 ± 1.0%, n = 4; data not shown). Facilitated entry through L-type channels is consistent with evidence showing that L-type channels in pyramidal neurons can contribute to Ca2+ entry at low voltages (Avery and Johnston 1996; Magee et al. 1996) and undergo voltage-dependent facilitation (Dolphin 1996). Facilitation can occur even with moderate depolarizing voltages (Svirskis and Hounsgaard 1997). Short AP trains can generate prolonged enhancement of D class L-type channel activation at membrane potentials negative to –50 mV (Cloues et al. 1997). In hippocampal pyramidal neurons, this delayed facilitation of L-type channels can directly evoke slow AHP waveforms (Bowden et al. 2001), and L-type activation underlies potentiation of the slow AHP during experiments (Borde et al. 2000). Consistent with this, in our experiments nimodipine had negligible effects on AHP magnitude in neurons that remained RS (data not shown). However, nimodipine did not systematically switch spiking behavior back from PS to RS or significantly increase R (44 ± 29%, n = 3). Inward rectification and sag potentials
Other intrinsic properties measured included fast and slow inward rectification under hyperpolarization. Under hyperpolarizing stimulation, cortical pyramidal cells display fast inward rectification dependent on a K+ current (Hille 2001) as well as a depolarizing voltage "sag" due to activation of Ih, a slowly activating, [Ca2+]-regulated nonspecific cation current with important effects on excitability (Luthi and McCormick 1998; Pape 1996). Ih plays an important role in the activation and maintenance of spontaneous rhythmic patterns in the CNS and elsewhere (Luthi and McCormick 1998; McCormick and Pape 1990).
Sag depolarization decreased with age while early inward rectification grew. PND 12 and 14 neurons had significantly larger sags than PND 17 cells (data not shown; t-test significance levels: PND 12 vs. 17, P < 0.0001; PND 14 vs. 17, P < 0.001) and significantly smaller rectification (t-test; PND 12 vs. 17, P < 0.005; PND 14 vs. 17, P < 0.01). This developmental effect on sag depolarization is in the same direction as has been observed in layer 1 neocortical cells (Zhou and Hablitz 1996), but in the opposite direction to changes observed in other systems (Reece and Schwartzkroin 1991; Viana et al. 1994). There was no effect of sensory deprivation or of cell layer location.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Layer 2/3 responses change from being mostly phasic to being mostly regular during the period PND 12-17 in synchrony with intracortical synaptogenesis (Micheva and Beaulieu 1996) and with extensive experience-dependent synaptic (Lendvai et al. 2000) and dendritic (Maravall et al. 2004) rearrangement. This period immediately follows the beginning of active whisker exploration (Welker 1964). In contrast, layer 4 spiny stellate cells, with earlier critical periods for synaptic input plasticity (before PND 7) (Crair and Malenka 1995; Feldman et al. 1998) and subthreshold receptive fields (Stern et al. 2001), have mature response properties by PND 14. Maturation of layer 2/3 spiking behavior and therefore of neuronal responsiveness is delayed by sensory deprivation (Fig. 1C).
Because more mature RS neurons could sustain spiking throughout a 500-ms stimulus pulse, their excitability measured in terms of number of APs fired was greater than that of PS neurons. RS neurons can respond to stronger depolarization with an increasing number of spikes. In contrast, because PS neurons stop spiking before stimulation ends, the overall number of spikes generated by these neurons is not highly sensitive to increases in depolarization. PS responses of immature cells resemble bursts and could help promote synchrony of cortical responses. Correlated activity in the immature cortex is well known (Garaschuk et al. 2000; Katz and Shatz 1996; Mao et al. 2001; Schwartz et al. 1998) and could enhance correlation-based synaptic plasticity (Katz and Shatz 1996).
Our data suggest that layer 2/3 PS neurons are an immature form of RS neurons (Montoro et al. 1988) rather than a distinct category of intrinsically bursting neurons (Zhu and Connors 1999) present only in the developing animal (Metherate and Aramakis 1999). This idea is supported by the finding that the short spike sequences fired by PS cells were not stereotyped AP bursts (Nowak et al. 2003), according to several criteria. First, minimal spike frequencies varied across PS cells as much as they did over RS cells. Second, spike frequencies over the first one or two interspike intervals were as tunable by stimulus magnitude in PS cells as they were in RS cells. Third, PS and RS cells did not have noticeable differences in spike threshold or in the ratio between thresholds for single spikes and for spike pairs. Neurons showing RS and PS behavior differed only in the mechanisms specifically responsible for determining SFA.
What mechanisms participate in the changes observed in spiking behavior? A variety of ionic currents are capable of contributing to PS or RS behaviors and their developmental regulation (see e.g., Amitai and Connors 1995). In several classes of neurons (Wang et al. 1997), phasic spiking can be promoted by slow recovery from inactivation of Na+ currents, which is a mark of immaturity. For instance, immature retinal ganglion cells are usually rapidly adapting and often show single-spike discharges. Na+ currents of single-spike neurons have significantly slower speeds of recovery from inactivation than those of rapidly adapting cells, which in turn recover significantly slower than Na+ currents of RS cells. These differences are due to a developmental change in Na+ channel subtype expression at the initial axonal segment of ganglion cells (G. Matthews, personal communication) like that in optic nerve nodes of Ranvier (Boiko et al. 2001). At the youngest age recorded in the present study (PND 12), we found a subgroup of PS neurons that had difficulty firing more than one spike in response to depolarization, suggesting slow recovery from inactivation of Na+ channels. However, we found that PS neurons differed from RS neurons in that they always showed a large late OR. This OR was not abolished by TTX (Fig. 4) and therefore could not be explained by differences in expression of inactivating Na+ channels. Another candidate mechanism for promoting PS behavior is the ID K+ current (Locke and Nerbonne 1997). Our experiments ruled out a role of ID in shaping PS spiking behavior: blocking ID did not induce reductions in the OR that would make cells more RS-like (Fig. 4).
Our results were consistent with a role of a slow AHP (Schwindt et al. 1988a, b, 1992) in maturation of spiking behavior (Lorenzon and Foehring 1993) (Fig. 3). Both spiking behavior and the slow AHP were modified by CdCl2, implying that they depended on Ca2+ influx (Fig. 5). Changes in AHP size and in SFA could also be induced by BAPTA perfusion and could occur rapidly, on the time scale of a whole cell experiment (within 10–20 min). This suggested that modulation of Ca2+ entry and handling was a more likely reason for the developmental changes in the AHP than alterations in the expression of the AHP channels themselves. Our imaging experiments did not detect any developmental changes in either AP-evoked [Ca2+] increases or in Ca2+ buffer capacity measured in the proximal apical dendrite (Table 3) but did reveal the presence of slow subthreshold [Ca2+] transients in PS but not RS neurons (Fig. 5F). This supports the possibility of developmental adjustments in Ca2+ entry mechanisms. Previous studies (Lorenzon and Foehring 1995; Schwindt et al. 1992) have suggested that modifications in the dominant sources of Ca2+ influx are responsible for changes in slow AHP magnitude and spiking behavior. Slow Ca2+-dependent AHP currents are a target of experience- and training-dependent plasticity in several brain regions (Coulter et al. 1989; Disterhoft et al. 1986, 1988; Saar and Barkai 2003; Schreurs et al. 1998). Increases in intrinsic excitability due to a reduction in AHP may have a permissive function in circuit plasticity and memory (Moyer et al. 1996; Thompson et al. 1996; Zhang and Linden 2003).
Layer 2/3 subthreshold sensory maps are affected by deprivation started around PND 9-12 but not later (Stern et al. 2001). Layer 4–layer 2/3 synaptic plasticity can still be evoked at later ages (Allen et al. 2003; Feldman 2000). Additional plasticity mechanisms (Maravall et al. 2004) are required to explain the critical period for layer 2/3 subthreshold sensory maps.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Present address and address for reprint requests and other correspondence: M. Maravall, Instituto de Neurociencias CSIC-UMH, Campus de San Juan, Apartado 18, 03550 Sant Joan d'Alacant, Spain (E-mail: mmaravall{at}umh.es).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Aizenman CD and Linden DJ. Rapid, synaptically driven increases in the intrinsic excitability of cerebellar deep nuclei neurons. Nat Neurosci 3: 109–111, 2000.[CrossRef][ISI][Medline]
Allen CB, Celikel T, and Feldman DE. Long-term depression induced by sensory deprivation during cortical map plasticity in vivo. Nat Neurosci 6: 291–299, 2003.[CrossRef][ISI][Medline]
Amitai Y and Connors BW. Intrinsic physiology and morphology of single neurons in neocortex. In: The Barrel Cortex of Rodents, edited by Jones EG and Diamond IT. New York: Plenum, 1995, p. 299-324.
Armano S, Rossi P, Taglietti V, and D'Angelo E. Long-term potentiation of intrinsic excitability at the mossy fiber-granule cell synapse of rat cerebellum. J Neurosci 20: 5208–5216, 2000.
Armstrong-James M, Callahan CA, and Friedman MA. Thalamo-cortical processing of vibrissal information in the rat. I. Intracortical origins of surround but not center-receptive fields of layer IV neurones in the rat S1 barrel field cortex. J Comp Neurol 303: 193–210, 1991.[CrossRef][ISI][Medline]
Armstrong-James M, Diamond ME, and Ebner FF. An innocuous bias in whisker use in adult rats modifies receptive fields of barrel cortex neurons. J Neurosci 14: 6978–6991, 1994.[Abstract]
Armstrong-James M and Fox K. Spatiotemporal convergence and divergence in the rat S1 "barrel" cortex. J Comp Neurol 263: 265–281, 1987.[CrossRef][ISI][Medline]
Armstrong-James M, Fox K, and Das-Gupta A. Flow of excitation within rat barrel cortex on striking a single vibrissa. J Neurosci 68: 1345–1354, 1992.
Avery RB and Johnston D. Multiple channel types contribute to the low-voltage-activated calcium current in hippocampal CA3 pyramidal neurons. J Neurosci 16: 5567–5582, 1996.
Boiko T, Rasband MN, Levinson SR, Caldwell JH, Mandel G, Trimmer JS, and Matthews G. Compact myelin dictates the differential targeting of two sodium channel isoforms in the same axon. Neuron 30: 91–104, 2001.[CrossRef][ISI][Medline]
