Electrophysiological properties of genetically identified subtypes of layer 5 neocortical pyramidal neurons: Ca²⁺ dependence and differential modulation by norepinephrine

Abstract

We studied neocortical pyramidal neurons from two lines of bacterial artificial chromosome mice (etv1 and glt; Gene Expression Nervous System Atlas: GENSAT project), each of which expresses enhanced green fluorescent protein (EGFP) in a different subpopulation of layer 5 pyramidal neurons. In barrel cortex, etv1 and glt pyramidal cells were previously reported to differ in terms of their laminar distribution, morphology, thalamic inputs, cellular targets, and receptive field size. In this study, we measured the laminar distribution of etv1 and glt cells. On average, glt cells were located more deeply; however, the distributions of etv1 and glt cells extensively overlap in layer 5. To test whether these two cell types differed in electrophysiological properties that influence firing behavior, we prepared acute brain slices from 2–4-wk-old mice, where EGFP-positive cells in somatosensory cortex were identified under epifluorescence and then studied using whole cell current- or voltage-clamp recordings. We studied the details of action potential parameters and repetitive firing, characterized by the larger slow afterhyperpolarizations (AHPs) in etv1 neurons and larger medium AHPs (mAHPS) in glt cells, and compared currents underlying the mAHP and slow AHP (sAHP) in etv1 and glt neurons. Etv1 cells exhibited lower dV/dt for spike polarization and repolarization and reduced direct current (DC) gain (lower f-I slope) for repetitive firing than glt cells. Most importantly, we found that 1) differences in the expression of Ca²⁺-dependent K⁺ conductances (small-conductance calcium-activated potassium channels and sAHP channels) determine major functional differences between etv1 and glt cells, and 2) there is differential modulation of etv1 and glt neurons by norepinephrine.

neocortical pyramidal neurons vary in morphology, physiology, molecular signature, laminar position, inputs, and projection targets, and they have been classified using each of these criteria. A focus of much present interest is whether types of pyramidal cells belonging to functional cortical circuits can be identified by other distinguishing characteristics. Two morphological classes of pyramidal cells have been identified in layer 5 (Kasper et al. 1994; Mason and Larkman 1990; Rumberger et al. 1998). One group includes large cells in deep layer 5 (layer 5B) that have a thick apical dendrite that reaches layer 1, where the dendrite terminates in an apical tuft (thick or thick-tufted cells) (Mason and Larkman 1990; Tsiola et al. 2003). Another group of cells is preferentially located in superficial layer 5 (layer 5A). They have slender dendrites with fewer oblique branches, and the apical dendrite often does not reach layer 1 (slender or slender-tufted cells) (Kasper et al. 1994). Passive and active electrical properties differ between these morphologically defined subpopulations of pyramidal cells (Kasper et al. 1994).

Pyramidal cells can also be classified by their projection patterns. Thick-tufted pyramidal neurons project subcortically to the pons, tectum, or pyramidal tract (PT) (Akintunde and Buxton 1992; Hattox and Nelson 2007; Kasper et al. 1994; Larkman and Mason 1990). Slender-tufted pyramidal neurons are heterogeneous, with corticocortical and corticofugal projecting subgroups (Hattox and Nelson 2007; Le Be et al. 2007; Mitchell and Macklis 2005). Slender-tufted neurons project to ipsilateral cortex, through the corpus callosum to contralateral cortex, and to both ipsilateral and contralateral striatum (Kasper et al. 1994; Larkman and Mason 1990). Electrical properties differ between subpopulations of pyramidal cells defined by their projections. For example, Le Be et al. (2007) found that callosally projecting neurons have a pronounced slow afterhyperpolarization (sAHP) and higher input resistance (R_in) and smaller and broader action potentials (APs) than noncallosally projecting layer 5 cells (Dembrow et al. 2010; Hattox and Nelson 2007).

Pyramidal cells with projections restricted to intratelencephalic (IT) targets have been classified as IT-type neurons and pyramidal neurons with projections beyond the telencephalon (i.e., corticospinal, corticotectal, corticopontine) have been classified as PT-type neurons (Catsman-Berrevoets et al. 1980; Kress et al. 2013; Reiner et al. 2003, 2010; Shepherd 2013). In rat somatosensory cortex, the vast majority of IT-type neurons are in layers 2/3 and layer 5A (Cowan and Wilson 1994; Molyneaux et al. 2009; Reiner et al. 2003, 2010; Shepherd 2013; Wilson 1987). In contrast, ∼90% of PT-type cells in somatosensory cortex are in layer 5B (Reiner et al. 2003, 2010). These data suggest a general relationship where slender-tufted pyramidal cells in layer 5A are IT-type and thick-tufted pyramidal cells in layer 5B are PT-type in motor, prefrontal, and somatosensory cortices although these relationships are far from perfect correlations (e.g., Reiner et al. 2003, 2010).

Pyramidal cells also have unique molecular signatures. Several genes are specific to pyramidal cells in particular layers (Greig et al. 2013; Molyneaux et al. 2009; Watakabe et al. 2007; Yoneshima et al. 2006) or those with defined projection patterns (Arlotta et al. 2005; Hevner et al. 2003; Molnar and Cheung 2006; Molyneaux et al. 2009). Subsets of pyramidal cells express signature genes including the transcription factor etv1, which is expressed in a subset of layer 5 neurons (Doyle et al. 2008). Etv1 has been shown to participate in neurogenesis in olfactory bulb (Stenman et al. 2003) and circuit formation in spinal cord (Arber et al. 2000). Another subset of layer 5 cells express glt25d2 (glt), a glycosyl transferase gene (Doyle et al. 2008; Gong et al. 2003, 2007). Differential functional involvement of glt cells in mice in vivo is exemplified by responses to antidepressants being mediated by IT-type pyramidal cells in layer 5A, whereas glt cells are not involved (Schmidt et al. 2012). Etv1- and glt-expressing neurons also differentially express many other genes (Doyle et al. 2008; Schmidt et al. 2012). Groh et al. (2010) reported that glt-enhanced green fluorescent protein (EGFP) cells are thick-tufted layer 5 neurons and that etv1-GFP cells correspond to slender-tufted layer 5 pyramidal neurons. In cortical barrel (somatosensory) cortex, glt-EGFP was reported to be expressed almost exclusively in layer 5B and etv1-EGFP in layer 5A, but there was more extensive intermingling in visual cortex (Groh et al. 2010). They also found that glt pyramids project to ipsilateral pons and thalamus but not to contralateral striatum and thus considered them to be a subset of PT-type neuron. They found that etv1 pyramids are IT-type, projecting callosally and to ipsilateral (and to some extent contralateral) striatum but not to pons or thalamus (Groh et al. 2010).

In the present study, our primary goal was to determine whether there are electrophysiological differences between etv1 and glt pyramidal cells that would shape their firing behavior. We used whole cell current- and voltage-clamp recordings in somatosensory cortex of acute brain slices to examine EGFP-expressing cells from the etv1 and glt mouse lines. We determined the laminar distribution and soma size of etv1 and glt cells. Also, because etv1 cells are reported to be a subset of IT-type neurons and glt cells a subset of PT-type neurons, we examined whether reported electrophysiological differences between PT and IT-type cells [in R_in, AP width, spike frequency adaptation (SFA), and hyperpolarization-activated cation current (I_H); e.g., Christophe et al. 2005; Dembrow et al. 2010; Gee et al. 2012; Hattox and Nelson 2007; Le Be et al. 2007; Sheets et al. 2011; Solomon et al. 1993] might also be characteristic of glt and etv1 cells.

Our principal finding was the differential expression of Ca²⁺-dependent K⁺ conductances responsible for dramatic differences in firing behavior between layer etv1 and glt pyramidal neurons and the differential modulation of these two cell types by norepinephrine (NE). In addition, we found that etv1 cells exhibited lower maximum dV/dt for spike polarization and repolarization, larger sAHPs, and reduced DC gain (lower f-I slope) during repetitive firing than glt cells. We confirmed that etv1 is expressed in pyramidal cells in layer 5A of somatosensory cortex, whereas on average glt cells are found deeper in layer 5 (extending into layer 6). There was, however, considerable overlap between the distribution of etv1 and glt cells. We found that soma size did not differ between etv1 and glt cells in somatosensory cortex. We also confirmed the findings of Groh et al. (2010) that etv1 pyramidal cells had significantly broader AP half-width and greater SFA compared with glt pyramidal cells.

MATERIALS AND METHODS

We studied layer 5 neurons from two bacterial artificial chromosome (BAC) lines of mice, each of which express EGFP in a different subpopulation of layer 5 pyramidal neurons (Gong et al. 2002, 2003, 2007). We maintain breeding colonies of both mouse lines (Swiss-Webster background), which were originally obtained from the Mutant Mouse Regional Resource Centers of the GENSAT project. The first line was Tg(Etvl-EGFP)BZ192Gsat/Mmucd (etv1). In the barrel cortex of these mice, EGFP was reported to be expressed in layer 5A pyramidal neurons that have slender apical dendrites, with more extensive intermingling in visual cortex (Groh et al. 2010). The second line was Tg(Glt25d2-EGFP)BN20Gsat/Mmnc (Glt). In somatosensory cortex of these mice, glt-EGFP was reported to be expressed in a subset of thick-tufted layer 5B pyramidal neurons (Groh et al. 2010). In addition, we recorded basic electrical properties from layer 2/3 pyramidal cells from both mouse lines.

These studies were performed on juvenile mice from 2–4 wk of age. All procedures were approved by the Animal Care and Use Committee, University of Tennessee Health Science Center. The animals were anesthetized with isoflurane until they were areflexive. Briefly, the animal was placed into a sealed plastic container into which gauze soaked with isoflurane was placed under a fiberglass screen floor. After anesthesia with isoflurane, the animal was decapitated, and the brain was removed and dropped into ice-cold cutting solution bubbled with O₂ for 30–60 s. This solution contained the following (in mM): 250 sucrose, 2.5 KCl, 1 NaH₂PO₄, 11 glucose, 4 MgSO₄, 0.1 CaCl₂, 0.4 ascorbate, 0.6 sodium pyruvate, and 15 HEPES (pH 7.3–7.4; 300 mOsm/l). The brain was then sliced into 300-μm-thick coronal sections using a vibrating tissue slicer (Vibroslice, Campden Instruments).

Histology and microscopy.

For anatomical studies, four animals of each type were deeply anesthetized with sodium pentobarbital sodium (50 mg/kg, ip) and then perfused through the heart with 4% paraformaldehyde-0.1% picric acid in 0.15 M sodium phosphate buffer (pH 7.2–7.4). The brains were removed and stored in 4% paraformaldehyde-0.1% picric acid in 0.15 M sodium phosphate buffer for 48 h and then sectioned at 50 μM on a vibratome (Leica VT1000). Sections were matched for animal age (P18–22) and the anterior-posterior level of section and were counterstained with NEUROTRACE 530/615 (Life Technologies) to indicate the distribution of cells and allow the determination of cortical layers and sublayers. We rehydrated the sections for 40 min in 0.1 M PBS and then washed the sections for 10 min in PBS plus 0.1% Triton X-100. After washing the sections twice (5 min each) in PBS, we stained with 200 μl NEUROTRACE 530/615 at 1:200 for 20 min. We then washed the excess stain for 10 min in PBS plus 0.1% Triton X-100, followed by two 5-min washes in PBS. The sections were mounted and coverslipped with a solution containing 2.5% of the anti-fade reagent 1,4 diazabicyclo[2.2.2]octane that was dissolved in a 0.1 M Tris buffer (pH 8–8.5) containing 10% polyvinyl alcohol and 25% glycerin. All chemicals were obtained from Sigma-Aldrich.

For imaging and analysis, we used a Zeiss 710 confocal microscope attached to a Zeiss AxioObserver inverted microscope (University of Tennessee Neuroscience Institute Imaging Center) for the confocal images shown in Fig. 1. For the low-power images depicting expression pattern, images were taken with a ×10 objective and were tiled and stitched from several fields of view. Images shown are from individual optic sections (1-μm optical thickness) at the depth of clearest GFP expression. We examined the laminar distribution of cells from four animals of each type using higher magnification images (×20 objective, 0.8 NA) and Zeiss Zen software by measuring the intensity of fluorescence with a series of 1-μm-thick lines (1000 μm in length, 20–50 lines per animal), orthogonal to the pia and sampled through cells with EGFP fluorescence. We imported the data into IGOR, where each line was represented by a single wave. We determined an average wave for each group, smoothed the data (25-point rolling average), and fit the result with a single Gaussian function representing fluorescent fluorescence intensity as a function of distance from the pia. The fitting curves were scaled to the same baseline and maximum intensity to more easily compare the relative distribution of glt and etv1 cells. For measurements of soma area, we used a ×63 oil immersion lens (1.4 NA) and obtained a Z-stack (1-μm optical sections) through five to seven fields of view (in somatosensory cortex) per slice, one slice per animal (n = 4 etv1 and 4 glt animals). The Z-stack was read into Neurolucida (MBF Bioscience), and soma perimeters were outlined at the depth of greatest diameter for each cell.

Fig. 1.

Expression of enhanced green fluorescent protein (EGFP) in select populations of layer 5 pyramidal neurons in 2 bacterial artificial chromosome mouse lines: Tg(Etvl-EGFP)BZ192Gsat/Mmucd (etv1), and Tg(Glt25d2-EGFP)BN20Gsat/Mmnc (glt). Data were obtained from 4 animals from each mouse line. A: tiled and stitched low-power image of EGFP expression in superficial layer 5 of an etv1 animal. The slice was counterstained with NEUROTRACE 530/615 (Life Technologies) to reveal cells and laminae. B: tiled and stitched low-power image of EGFP expression in layers 5 and 6 in glt animal. The slice was counterstained with NEUROTRACE 530/615 (Life Technologies) to reveal cells and laminae. C: histogram indicating no differences in soma area between etv1 (n = 122 cells from 4 animals) and glt cells (n = 60 cells from 4 animals). Soma areas were measured with Neurolucida from high-power (×63) sections of cells at the level of the cell nucleus. D: comparison of laminar positions of EGFP+ cells in primary somatosensory cortex in etv1 and glt animals. The sections were counterstained with NEUROTRACE 530/615 (Life Technologies). Dashed lines indicate upper and lower boundaries of layer 5. Left: section of S1 cortex from A (etv1). Middle: section of S1 cortex from B (glt) rotated and aligned to show laminar expression of EGFP in the 2 animals. Right: neurotrace stain alone is shown (same glt section as center) to reveal cells and cortical laminae (indicated in white). E: depth profile for average EGFP fluorescence from 4 animals each from etv1 (blue) and glt (red). Data were smoothed (rolling average, 25 points) and fit with a single Gaussian in IGOR. F: Gaussian fits as in E, scaled to the same maximum intensity to show the relative depth distribution (peaks ∼625 μm for etv1 and 695 μm for glt).

Slice recordings.

Slices were placed in a recording chamber on the stage of an Olympus BX50WI upright microscope. Slices were bathed in artificial cerebrospinal fluid (aCSF) bubbled with 95% O₂-5% CO₂, delivered at 2 ml/min, and heated with an inline heater (Warner Instruments) to 33 ± 1°C (measured with a thermistor in the bath adjacent to the slice). The aCSF contained (in mM) 125 NaCl, 3 KCl, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 20 glucose (pH 7.4, 310 mOsm). Pharmacological agents were added directly to the aCSF. All slice recordings of repetitive firing were done in the presence of 6,7-dinitroquinoxaline-2,3-dione (DNQX; 20 μM) to block α-amino-3-hydroxy-5-methyl-4-isoxazoleproipionic acid receptors, 2-amino-5-phosphonovaleric acid (APV; 50 μM) or (α)-3-(2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP; 5 μM) to block N-methyl-d-aspartate receptors, and picrotoxin (100 μM) to block γ-aminobutyric acid type A receptors to ensure that studied effects of the pharmacological agents were exerted directly on the postsynaptic cell. NE was made up as a 20 mM stock in Millipore-filtered, distilled H₂O containing 20 mM ascorbic acid on the day of the experiment and then diluted to the final dose (10 μM) in aCSF. A 50 μM stock of the small-conductance calcium-activated potassium (SK) channel blocker apamin was dissolved in 5% acetic acid, frozen, and then thawed and diluted to 100 nM in aCSF on the day of the experiment.

Pyramidal neurons in layers 2/3 and 5 were visualized with infrared/differential interference contrast (IR/DIC) video microscopy (Dodt and Zieglgansberger 1990; Stuart et al. 1993) using a ×40 (0.8 NA) Olympus water-immersion objective and an IR-sensitive camera (Olympus OLY-150 or DAGE-MTI). Etv1 or glt pyramidal cells were visually identified by the presence of EGFP epifluorescence using an FITC filter. There was typically a main band of EGFP+ cells in layer 5 in each animal. Recordings were directed within this main band. We switched between IR/DIC and epifluorescence to determine cell type and to obtain a GΩ seal. Electrode position was controlled with Sutter ROE-200 manipulators and PC-200 controller or Luigs-Neumann manipulators and controller. Whole cell patch-clamp recordings were acquired using either an Axon Multiclamp 700A or Multiclamp 700B amplifier (Molecular Devices) and PClamp 9 or 10 software. For current-clamp recordings, the data were digitized at 20–50 kHz and filtered at 10 kHz. Voltage-clamp recordings of tail currents were digitized at 10 kHz and filtered at 2 kHz. We recorded with borosilicate electrodes (Warner G150F; 3–8 MΩ in the bath) produced with a horizontal electrode puller (Flaming-Brown P-87; Sutter Instruments). For current- and voltage-clamp recordings, electrodes were filled with an internal solution containing the following (in mM): 130.5 KMeSO₄, 10 KCl, 7.5 NaCl, 2 MgCl₂, 10 HEPES, 2 ATP, and 0.2 GTP. EGTA (100 μM) was added to the intracellular solution. Data were collected only from cells forming a 1-GΩ or tighter seal. All reported voltages were corrected by subtracting the measured liquid junction potential (8 mV).

An adaptation index was calculated for repetitive firing, as defined in Groh et al. (2010). We measured the time interval between successive spikes elicited using a 2-s current injection at 300 pA. We excluded the first two interspike intervals (ISIs) and then normalized each interval to the third ISI as a function of the ISI number. We then fitted the data with linear regression, and 100 × (the slope of the regression line) was the adaptation index. An adaptation index of 10 means that, on average, the time interval increases by 10% between successive spikes.

Statistics.

Prism (GraphPad Software) software was used to perform statistical tests. Student's paired t-test was used to compare sample population data between etv1 and glt cells, and summary data are presented as means ± SE, unless noted otherwise. Paired t-tests were used to compare control vs. drug effects. We used a one-way ANOVA to compare multiple experimental groups (etv1, glt, layer 2/3), with post-hoc Tukey's multiple-comparisons tests to determine which individual means differed. For all tests, P values <0.05 were considered significantly different. Sample population data are represented as scatterplots, histograms of means ± SE, or as box plots (Tukey 1977). Box plots indicate the upper and lower quartiles as edges of the box, with the median represented as a line crossing the box. The stems indicate the largest and smallest nonoutlying values. Outlying values are 1.5 times the quartile boundaries.

RESULTS

Anatomical measurements.

The detailed study of Groh et al. (2010) characterized etv1 and glt mice in terms of cell size, dendritic morphology, laminar position, proportion of excitatory cells accounted for by EGFP+ cells, and projection patterns (from retrograde labeling). In the present study, we concentrated on somatosensory cortex because Groh et al. (2010) found little overlap in distribution between these two cell lines in the somatosensory (barrel) cortex (in contrast to the substantial overlap in visual cortex). Groh et al. (2010) also reported that etv1 cells are a subset of slender pyramidal cells in layer 5A (∼55% of excitatory neurons in this layer; Groh et al. 2010) and that glt cells comprise a subset of thick-layer 5B neurons (∼55% of excitatory neurons in this layer). Figure 1 shows our findings for the relative laminar position of these two cell types in somatosensory cortex (aged P18–22: n = 4 animals for both etv1 and glt). On average, etv1 cells were located more superficially within layer 5 than glt cells (EGFP distribution was centered at ∼625 μm from the pial surface in etv1 cells and ∼695 μm in glt cells). Counterstaining with a cell marker (NEUROTRACE 530/615; Life Technologies) indicated that the etv1 cells are found within superficial layer 5 (layer 5A). We found that glt cells were distributed more deeply on average; however, there was considerable overlap with depths, where etv1 cells were expressed and glt cell distribution extended into layer 6 (Fig. 1). Our measurements also indicate that, on average, soma size did not differ between glt and etv1 cells (Fig. 1).

Electrical recordings.

Our primary goal was to determine whether there are electrophysiological differences between etv1 and glt pyramidal cells that would shape their firing behavior. All recordings were from etv1-expressing or glt- expressing pyramidal neurons from 2–4-wk-old BAC mice. We used these relatively young ages because, at >4 wk, we were unable to consistently detect the fluorescent cells. Initial comparisons were made between etv1 and glt neurons in layer 5. We also recorded basic electrical properties from layer 2/3 pyramidal neurons from both mouse lines.

Passive properties, sag, and rheobase.

We first examined resting membrane potential (RMP) to compare these cell types in the absence of significant synaptic input, similar to “down states” observed during slow-wave sleep in vivo (Steriade et al. 1993; Stern et al. 1997). RMP was measured immediately after entering whole cell mode. We found that the average RMP of layer 2/3 pyramidal cells was significantly more negative than either of the layer 5 subtypes (see also Breton and Stuart 2009; Etherington and Williams 2011; Higgs et al. 2006; Larkman and Mason 1990), and etv1 cells had significantly more negative RMPs than glt cells (Fig. 2B, inset). We also measured R_in (measured at peak voltage deflection to a −50-pA current injection) as an index of how the cells would passively respond to current inputs. R_in did not differ significantly between the two subtypes of layer 5 pyramidal cells or for layer 5 cells vs. layer 2/3 pyramidal cells (Fig. 2D).

Fig. 2.

Comparison of passive properties, voltage sag in response to hyperpolarization, and rheobase among 3 subpopulations of pyramidal neurons. A: representative traces for a layer 3 pyramidal neuron in response to a −50-pA hyperpolarizing current injection (used to determine input resistance, R_in) and a +150-pA depolarizing current injection that elicited 3 action potentials (APs). Inset: scale bars for A–C. B: representative traces for an etv1 pyramidal neuron in response to a −50-pA hyperpolarizing current injection and a +200-pA depolarizing current injection that elicited a single AP (rheobase). Inset: population summary for resting membrane potential (RMP) in layer 2/3, etv1, and glt neurons. Layer 2/3 neurons had significantly more negative RMPs than layer 5 cells. C: representative traces for a glt pyramidal neuron in response to a −50-pA hyperpolarizing current injection and a +150-pA depolarizing current injection that elicited several APs. Note the spike doublet at the onset of firing. Similar initial doublets were observed in ∼13% of glt neurons. Inset: current injection protocol for A–C. D: summary population data for peak R_in in layer 2/3, etv1, and glt neurons. Peak R_in was determined as peak voltage change/injected current (−50 pA). E: summary population data for percentage of sag in layer 2/3, etv1, and glt neurons. Percentage of sag was determined as 100 × (peak change − steady-state voltage change)/(peak change). Sag was measured from a −150-pA current injection from RMP. Both layer 5 subtypes had significantly larger sag than layer 2/3 neurons, and glt neurons had significantly larger sag than etv1 neurons. F: summary population data for rheobase, the minimal current (500-ms current injection) required to elicit a single AP. Both layer 5 subtypes had significantly lower rheobase than layer 2/3 neurons, and glt neurons had significantly lower rheobase than etv1 neurons. *Significant difference from etv1 neurons; #significant difference from layer 2/3 (L2/3) neurons.

Many pyramidal cells exhibit a time-dependent relaxation of the voltage response to hyperpolarizing current injections (“sag”, measured in response to a −150-pA current injection). This sag is considered an indicator of the kinetics and amplitude of I_H relative to the membrane time constant (Schwindt et al. 1988c). We measured % sag as 100 × [(peak response) − (steady state response)]/(peak response). All three types of pyramidal neurons had a sag percentage that differed significantly from each other, with glt neurons having the greatest sag (Table 1, Fig. 2E). The time to the peak negative voltage deflection in the sag protocol was significantly shorter for glt neurons vs. etv1 or layer 2/3, suggesting more rapid kinetics of I_H in glt neurons (Table 1). We tested the response of several cells of each type to the blocker of I_H, ZD-7288 (ZD, 50 μM). Application of ZD-7288 significantly reduced voltage sag in layer 2/3 pyramidal cells (P < 0.004; 5.7 ± 1.1 control vs. 1.26 ± 0.7% ZD; n = 8), etv1 cells (P < 0.0001; 21.1 ± 1.9 control vs. 0.3 ± 0.2% ZD; n = 6), and glt cells (P < 0.0001; 28.5 ± 1.4 control vs. 1.4 ± 0.7% ZD; n = 6) (data not shown).

Table 1.

RMP, R_in, and rheobase

	Etv1	Glt	Layer 2/3
RMP, mV	−75 ± 0.4 (102)†	−73 ± 0.4 (67)*†	−82 ± 0.8 (35)
Rin, MΩ	138 ± 6 (102)	138 ± 5.6 (68)	139 ± 13 (35)
Sag, %	19 ± 1.0 (69)†	25 ± 1.2 (37)*†	6.5 ± 0.9 (28)*
Sag TTP, ms	58 ± 2.0 (59)	44 ± 1.9 (37)*†	57 ± 5.4 (20)
Rheobase, pA	172 ± 12 (76)	97 ± 8 (62)*†	246 ± 28 (25)

Data are presented as means ± SE (n, number of cells).

^*Glt significantly different from etv1 (P < 0.05, unpaired t-test).

^†Significantly different from layer 2/3 (ANOVA with Tukey's post hoc test). RMP, resting membrane potential; R_in, peak input resistance; Sag TTP, time to peak voltage deflection.

We also measured rheobase, the current required to reach AP threshold during a long current injection (500 ms), as an indicator of neuronal excitability (lower rheobase = greater excitability). We found that rheobase was significantly lower in glt vs. etv1 cells (Table 1, Fig. 2F), indicating that glt cells are more easily brought to threshold by DC input. The rheobase of layer 2/3 pyramidal cells was significantly higher than that of glt cells but was not significantly different from etv1 cells (Table 1, Fig. 2F).

APs.

We next examined the response of etv1 and glt neurons to a single 10-ms, suprathreshold current injection. To characterize the AP, we measured AP amplitude, voltage threshold (V_th, measured as the voltage corresponding to the abrupt increase in dV/dt), width at half amplitude (relative to RMP), and the maximum rate of change (dV/dt) for both the upstroke and downstroke (Table 2, Fig. 3). AP amplitude was significantly larger for layer 2/3 pyramidal cells than either of the layer 5 subtypes. V_th was lower in glt vs. etv1 cells, suggesting that glt neurons may reach threshold more readily than etv1 neurons. We found that spikes were narrower in glt cells vs. etv1 cells (Groh et al. 2010), and both layer 5 subtypes have narrower spikes than layer 2/3 cells (Fig. 3D). Narrow spikes often correlate with high firing rates in various cell types (Bean 2007). The maximal rates (dV/dt) of AP depolarization and repolarization (indicators of Na⁺ current density and repolarizing current density, respectively) were both significantly greater in glt pyramidal neurons than in etv1 cells (Fig. 3E). Layer 2/3 pyramidal cells had lower maximum dV/dt for spike repolarization than glt or etv1 cells.

Table 2.

Single AP parameters

	AP, mV	Vth, mV	HW, ms	dV/dt up, V/s	dV/dt down, V/s
Etv1	100 ± 2 (71)†	−49 ± 1 (69)	1.0 ± 0.03 (71)†	310 ± 10 (71)	105 ± 3 (71)†
Glt25	98 ± 2 (57)†	−52 ± 1 (57)*	0.9 ± 0.02 (57)*†	352 ± 12 (57)*	125 ± 7 (35)*†
Layer 2/3	112 ± 2 (23)	−52 ± 1 (23)	1.2 ± 0.04 (23)	331 ± 13 (23)	89 ± 4 (23)

Data are presented as means ± SE (n, number of cells).

^*Glt significantly different from etv1 (P < 0.05, unpaired t-test).

^†Significantly different from layer 2/3 (ANOVA plus post hoc test). AP, action potential amplitude; V_th, voltage threshold; HW, AP width at half amplitude (from resting potential); dV/dt up, maximum rate of AP rise; dV/dt down, maximum rate of AP repolarization.

Fig. 3.

Single APs. A: layer 2/3 neuron. Inset: same AP at expanded time base to show spike width. Bottom: current stimulus protocol. B: etv1 neuron (scale bars apply to A–C). Inset: same AP at expanded time base to show spike width. mAHP, medium afterhyperpolarization; fAHP, fast AHP. Bottom: current stimulus protocol. C: glt neuron. Inset: same AP at expanded time base to show spike width. Bottom: current stimulus protocol (gain for insets shown in B). D: summary data for half-width (width at half-amplitude from RMP). Glt neurons had narrower APs than etv1 neurons. Layer 2/3 neurons had significantly broader APs than both layer 5 cells. E: summary data for maximum dV/dt for AP polarization (upstroke). Glt neurons had significantly greater peak dV/dt than etv1 cells. F: summary data for the mAHP following a single AP. Layer 2/3 neurons had significantly smaller mAHPs than layer 5 cells (although layer 2/3 RMP was 5 mV hyperpolarized on average). The mAHP was significantly larger in glt cells than etv1 cells. *Significant difference from etv1 neurons. #Significant difference from layer 2/3 neurons.

AHPs.

AHPs reflect conductances activated during spiking. Three AHPs have been described in neocortical pyramidal neurons (Abel et al. 2004; Andrade et al. 2012; Lorenzon and Foehring 1992, 1993; Schwindt et al. 1988a, 1988b). The fast AHP reflects repolarization mechanisms for the AP, as well as residual inward currents and electrotonic redistribution of charge (Schwindt et al. 1988b). The medium AHP (mAHP) can also be activated by a single AP and is partially mediated by Ca²⁺-dependent and apamin-sensitive SK channels (Abel et al. 2004; Lorenzon and Foehring 1993; Pineda et al. 1998; Schwindt et al. 1988b; Spain et al. 1991). The mAHP is also influenced by voltage-gated K⁺ conductances, as well as I_H (Gulledge et al. 2013; Schwindt et al. 1988b, 1988c; Storm 1989). The sAHP has Ca²⁺- and Na⁺-dependent parts (Foehring et al. 1989; Gulledge et al. 2013; Schwindt et al. 1988b), requires multiple APs to activate, and is insensitive to apamin. The initial Ca²⁺- sensitive portion of the sAHP lasts ∼1–2 s and is mediated by unknown channels that are negatively modulated by transmitters that activate PKA (e.g., NE through β-receptors; Andrade et al. 2012; Foehring et al. 1989).

We found that the mAHP after a single AP (Fig. 3F) was significantly larger in glt cells (4.1 ± 0.3 mV, n = 56) compared with etv1 cells (3.1 ± 0.23 mV, n = 70) or layer 2/3 cells (1.3 ± 0.2, n = 23; which is also significantly different vs. etv1). The amplitude of the peak AHP and sAHP are known to depend on the number and frequency of APs used to elicit them (Abel et al. 2004; Schwindt et al. 1988a). Therefore, to compare the amplitude of the peak AHP and sAHP between cell types, we used a standard protocol with ten suprathreshold current injections (5 ms), repeated at 50 Hz. Under these conditions, etv1 cells had a significantly larger sAHP (measured at 500 ms after the last AP) and larger peak AHP than glt cells (Table 3, Fig. 4). The peak AHP after multiple APs includes a contribution from the sAHP conductance as well as that from the mAHP conductance (Abel et al. 2004; Kaczorowski et al. 2007). Layer 2/3 cells had smaller peak AHPs than either layer 5 cell subtype and smaller sAHP vs. etv1 cells although the RMP when this protocol was run was significantly more negative in the layer 2/3 cells (−78 ± 1 mV) vs. the layer 5 cells (−74 ± 0.5 mV for etv1 and −74 ± 0.4 mV for glt), which would have reduced the driving force and influenced AHP amplitude.

Table 3.

Peak and slow AHPs

	Peak AHP, mV	sAHP, mV
Etv1	6.6 ± 0.3 (77)†	3.2 ± 0.2 (77)†
Glt	5.5 ± 0.2 (62)*†	0.8 ± 0.2 (62)*
Layer 2/3	3.7 ± 0.4 (21)	1.3 ± 0.2 (21)

Data are presented as mean ± SE (n, number of cells). Afterhyperpolarizations (AHPs) were elicited in response to 10 suprathreshold, 5-ms current injections at 50 Hz. Note that the RMP when this protocol was run was significantly more negative in the layer 2/3 cells (−78 ± 1 mV) vs. the layer 5 cells (−74 ± 0.5 for etv1 and −74 ± 0.4 mV for glt), which likely contributed to the smaller AHP amplitudes in these cells.

^*Significant difference, glt from etv1 (P < 0.05, unpaired t-test).

^†Significant difference vs. layer 2/3 (ANOVA with Tukey's post hoc test). sAHP, slow AHP.

Fig. 4.

Comparison of AHPs after 10 APs elicited by suprathreshold, 5-ms current injections. A: Etv1 cell. Note presence of large slow AHP (sAHP) as well as large peak AHP. B: Glt cell. Note lack of a sAHP (the total AHP duration is <500 ms). Inset: stimulation protocol was 10 5-ms suprathreshold current injections at 50 Hz. C: summary data for peak AHP. Both layer 5 cell types had larger peak AHPs than layer 2/3 cells; however, layer 2/3 cells were ∼5 mV hyperpolarized on average from the layer 5 cells. D: summary data for the sAHP (measured 500 ms after the last AP). Etv1 neurons had larger sAHPs than layer 2/3 cells; however, layer 2/3 cells were ∼5 mV hyperpolarized on average from the layer 5 cells. Glt neurons had significantly smaller sAHP vs. etv1 cells. *Significant difference from etv1 neurons; #significant difference from layer 2/3 neurons.

Repetitive firing.

Most neocortical pyramidal neurons fire rhythmically in response to a DC stimulus and are referred to as regular-spiking (RS) cells (Agmon and Connors 1992; Chagnac-Amitai et al. 1990; Mason and Larkman 1990; McCormick et al. 1985). RS cells initially fire at a higher rate and exhibit SFA to a constant stimulus (Chagnac-Amitai et al. 1990; Lorenzon and Foehring 1993; McCormick et al. 1985). All of the pyramidal cells in this study were of the RS type (Fig. 5A) (McCormick et al. 1985) although some glt cells exhibited an initial spike doublet before RS firing (6/63 cells = 13%; Fig. 2C). RS firing can be characterized by the relationship between average firing frequency and the injected current (f-I curve) and the relationship between instantaneous firing rate and time (f-t plot).

Fig. 5.

Average firing frequency vs. current relationships (f-I curves) for pyramidal cell subtypes. Data are presented as means ± SE for the first 100 ms of firing (black), the first 200 ms (red), the first 500 ms (green), and the entire 2-s firing epoch (blue). A: representative repetitive firing traces in response to 2-s current injections at 100 pA. A1: layer 2/3 neuron. A2: etv1 neuron. A3: glt neuron. B: etv1 cells. f-I plots are shown for averaged data from 22 cells. C: layer 2/3 cells. Average f-I plots are shown for averaged data from 16 cells. D: glt cells. Average f-I plots are shown for averaged data from 32 cells. Note the more curved relationship and steeper initial slope compared with that of etv1 cells in B. E: summary data for all cell populations for the first 200 ms and entire 2 s of firing (means ± SE). f-I slopes were obtained by fitting the initial slope (currents up to ∼3× rheobase) where f-I relationships were relatively linear. Glt neurons had significantly steeper f-I slope for the 2-s data than etv1 neurons. *Significant difference vs. etv1 cells; #significant difference vs. layer 2/3 cells.

The overall f-I relationship was more linear in etv1 cells and bilinear or curvilinear in glt neurons (Fig. 5). The f-I slopes for the first 100–200 ms of firing were similar for all three pyramidal cell subtypes (Table 4, Fig. 5), but the relationships were significantly different between layer 5 cell types for 500-ms or 2-s firing (Fig. 5). Glt cells have a steeper initial f-I slope than etv1 cells (i.e., higher gain; Table 4, Fig. 5). In both cell types, f-I slopes were higher for the initial 100–200 ms of firing compared with 500 ms or 2 s of firing although this was much more dramatic for etv1 neurons (Table 4, Fig. 5). For the full 2 s, the f-I slope of layer 2/3 cells was intermediate between etv1 and glt cells and was significantly greater than etv1 cells (Table 4).

Table 4.

Repetitive firing: slopes of the relationship between average firing frequency vs. current (f-I) and SFA

	f-I Slope, 200 ms, Hz/nA	f-I Slope, 2 s, Hz/nA	% Adaptation	Adaptation Index
Etv1	79 ± 1 (21)	30 ± 1 (22)†‡	76 ± 2 (21)‡	39 ± 7 (21)
Glt	89 ± 1 (30)	72 ± 1 (30)*	17 ± 2 (27)*‡	1.1 ± 0.3 (27)*
Layer 2/3	92 ± 10 (15)	55 ± 7 (15)†	52 ± 4 (15)	24 ± 14 (15)

Data are presented as means ± SE (n, number of cells). Firing was elicited in response to 2-s current injections at different amplitudes. Percentage of adaptation (% Adaptation) = 100 × (frequency of 3rd ISI) − (frequency of final ISI)/(frequency of 3rd ISI). Adaptation index = 100 × the slope of a linear regression line fit to 1/ISI vs. ISI number (Groh et al. 2010). Percent adaptation and adaptation index were for firing at 300 pA.

^*Significant difference, glt from etv1 (P < 0.05, t-test).

^†Significant difference from same cell type (etv1 or glt) vs. 200 ms f-I slope (P < 0.05, t-test).

^‡Significant difference vs. layer 2/3 (ANOVA with Tukey's post hoc test). SFA, spike frequency adaptation; ISI, interspike interval.

The greater divergence in f-I slopes after the first 200 ms of firing suggests greater SFA in etv1 neurons. This is shown directly by the f-t relationships of these two cell types (in response to a 2-s, 300-pA current injection; Fig. 6, A vs. B). All of the cell types differed in % adaptation [100 × (frequency of 3rd ISI − frequency of final ISI)/frequency of 3rd ISI], with adaptation greatest in etv1 cells, almost nonexistent in glt cells, and intermediate in layer 2/3 cells (Fig. 6D). We also found a significant difference in the adaptation index (as defined by Groh et al. 2010; see materials and methods). Etv1 neurons (39 ± 7, n = 21) had a much greater adaptation index than glt neurons (1.1 ± 0.3, n = 27), with layer 2/3 pyramidal cells intermediate (24 ± 14; Fig. 6D, Table 4). The third ISI was chosen for normalization (for both adaptation index and % adaptation) to emphasize the prolonged time-dependent changes evident after the steep decline in firing rate in the first two intervals in both cell types and the presence of an initial AP doublet in some glt cells (Guan et al. 2013; Hattox and Nelson 2007). Etv1 neurons exhibit a clear biexponential decay of firing rate during a 2-s current injection, with fast (τ ∼18 ms: 89% of decay) and slow (τ∼240 ms; Fig. 6A) components. In contrast, glt cells exhibit very little slow adaptation (∼5% of total; Fig. 6B). Layer 2/3 pyramidal cells also exhibited two time constants for adaptation, with one τ ∼16 ms (95%) and the other τ ∼435 ms (5%; Fig. 6C). We did not observe a time-dependent acceleration of firing during any of the DC injections, as previously shown in a subset of layer 5 pyramidal cells in mouse motor cortex (Miller et al. 2008) and prefrontal cortex (Dembrow et al. 2010) but not somatosensory cortex (Miller et al. 2008).

Fig. 6.

Relationships for instantaneous firing frequency (1/ISI, where ISI is interspike interval) vs. time (f-t curves). A: etv1 cells exhibited fast and slow phases of spike frequency adaptation (SFA). The slow phase made up ∼11% of the total SFA. B: glt cells also exhibited fast and slow phases of SFA, but the slow phase made up only ∼5% of the total SFA. C: layer 2/3 cells exhibited fast and slow phases of SFA. On average, the slow phase made up ∼5% of the total SFA. D: summary data for percentage of adaptation = 100 × [(frequency of 3rd ISI − frequency of final ISI)/(frequency of 3rd ISI)]. Etv1 cells had significantly greater percentage of adaptation than glt cells and layer 2/3 pyramidal cells. *Significant difference from etv1 neurons; #significant difference from layer 2/3 neurons.

Voltage clamp of AHP currents.

The dramatic differences in firing behavior (f-I relationships and SFA) between etv1 and glt cells suggested that these two populations of layer 5 cells may have differential expression of the Ca²⁺-dependent conductances that underlie the I_mAHP and I_sAHP (Abel et al. 2004; Schwindt et al. 1988a). As noted above, the peak AHP after multiple APs is influenced by both the mAHP and sAHP conductances (Abel et al. 2004; Schwindt et al. 1988a). Thus, to obtain clearer separation between the underlying conductances, we used whole cell voltage-clamp to reveal the currents underlying the mAHP (I_mAHP) and sAHP (I_sAHP). We elicited the AHP currents with a voltage step from −70 mV to +30 mV (for 200 ms). Space clamp is poor in these dendritic neurons, but we focused on the underlying small and very slowly deactivating tail currents upon return to −60 mV. Tetrodotoxin was not present, so typically a few unclamped APs occurred in all cell types at the beginning of the step. Tail currents were obtained from 14 etv1 neurons and 14 glt neurons (Fig. 7). Most of the tail currents were well fit as the sum of two exponentials, corresponding to I_mAHP and I_sAHP, respectively. In etv1 neurons, both I_mAHP and I_sAHP were usually of similar amplitude (estimated by extrapolation of the exponential fit of the tail current to time zero; Table 5, Fig. 7D) although in some cells a large slow component (I_sAHP) obscured I_mAHP. The slower I_sAHP had a τ of ∼900 ms, and the faster I_mAHP had a τ of ∼66 ms (Table 5). In contrast, whereas similar fast and slow components could be seen in glt neurons, the amplitude of I_mAHP was much larger than in etv1 neurons and much larger than I_sAHP, which only made up ∼18% of the tail current amplitude in glt neurons (Table 5). Two time constants were also evident in layer 2/3 pyramidal neurons (n = 21) although the time constants for both I_mAHP and I_sAHP were significantly slower in layer 2/3 cells vs. layer 5 cells (Table 5). The proportion of the current attributable to I_sAHP was similar in layer 2/3 and glt cells.

Fig. 7.

Voltage clamp of I_AHPs. Currents were elicited by a step from −70 mV to +30 mV. We measured tail currents upon return to −60 mV (see inset in C). A: example from etv1 neuron. The tail current was fit as the sum of 2 exponential functions (I_mAHP, fast; I_sAHP, slow). Inset: summary data for time constants (τs) for 14 etv1 cells. The mean is indicated by a horizontal line. B: example from glt neuron. The tail current was fit as the sum of 2 exponential functions. Inset: summary data for τs for 14 glt cells. The mean is indicated by a horizontal line. C: representative traces from an etv1 neuron, showing the Ca²⁺ dependence of both the slow and fast decay phase. D: summary data for tail current amplitude. Etv1 cells had significantly larger I_sAHP than layer 2/3 cells or glt cells. *Significant difference from etv1 neurons; #significant difference from layer 2/3 neurons.

Table 5.

Voltage clamp of the mAHP and sAHP currents

	Tail, pA	τmedium	τslow	% Slow τ
Etv1	259 ± 17 (14)	66 ± 10 (14)†	933 ± 101 (14)†	52 ± 4 (14)†
Glt	372 ± 30 (14)*†	74 ± 11 (14)†	791 ± 75 (14)†	18 ± 4 (14)*
Layer 2/3	227 ± 26 (21)	149 ± 18 (21)	2611 ± 550 (21)	16 ± 2 (21)

Data are presented as means ± SE (n, number of cells). Currents were elicited by a 200-ms voltage step from −70 mV to +30 mV. Tail currents were measured upon a step to −60 mV. Tail is the difference between the initial and final amplitude of the tail current at −60 mV; tail currents were fit with the sum of 2 time (τ) constants (τ_medium and τ_slow). τ_medium is the faster τ component corresponding to the medium AHP (mAHP) current (I_mAHP). τ_slow is τ for the sAHP current (I_sAHP). % Slow τ is the percentage of tail amplitude accounted for by τ_slow.

^*Significant difference glt from etv1 (P < 0.05, t-test).

^†Significant difference vs. layer 2/3 (ANOVA with Tukey's post hoc test).

Calcium dependence.

To further test for cell-type differences in Ca²⁺-dependent mechanisms, we compared cells before and after bath application of the nonspecific inorganic Ca²⁺ channel blocker Cd²⁺ (400 μM). In those experiments, we did not add NaH₂PO₄ to the aCSF (to prevent precipitation). Cd²⁺ did not cause significant changes in RMP or rheobase in etv1 (n = 4) or glt cells (n = 5; data not shown). We also found no statistical differences for any parameters of a single AP in either cell type (data not shown). We also elicited AHPs with 10 APs at 50 Hz to test for Ca²⁺ dependence of the AHPs. In etv1 cells (n = 5), both the peak AHP (control 6.4 ± 1.9 mV; Cd²⁺ 2.9 ± 1.1 mV) and the sAHP (at 500 ms after the last AP; control 2.6 ± 1.7 mV; Cd²⁺ 0.6 ± 0.5 mV) were significantly reduced by Cd²⁺. In glt cells (n = 5), both the peak (control 4.8 ± 0.7 mV; Cd²⁺ 2.4 ± 1.0 mV) and sAHP (control 0.6 ± 0.4 mV; Cd²⁺ 0.2 ± 0.2 mV) were reduced although this was only significant for the peak AHP. In voltage-clamped etv1 cells (n = 3), both I_mAHP and I_sAHP were reduced by Cd², indicating that they were Ca²⁺ dependent (Fig. 7C). For the four glt cells tested, there were no clear slow tail components present initially, but Cd²⁺ reduced the remaining faster time constant in all cases (data not shown). Our data indicate differential expression of Ca²⁺-dependent mechanisms between etv1 and glt cells. We next used a pharmacological approach to examine specific Ca²⁺-dependent conductances.

Role of SK channels.

In neocortical pyramidal cells, a large proportion of the mAHP and underlying current is mediated by apamin-sensitive, SK-type K⁺ channels (Abel et al. 2004; Jones and Stuart 2013; Lorenzon and Foehring 1993; Pineda et al. 1998; Schwindt et al. 1988a, 1992a). We tested whether SK channels differentially contribute to the properties of etv1 vs. glt cells by application of the SK channel blocker, apamin (100 nM). We examined AHP currents in voltage-clamp mode, the AHPs after a single AP and after 10 APs (at 50 Hz), and repetitive firing in response to 2-s current injections.

In voltage clamp, we elicited tail currents (at −60 mV) with a 200-ms step from −70 mV to +30 mV. In etv1 cells (n = 9), the tail current was dominated by a single exponential component with a τ_decay of 1,258 ± 184 ms and amplitude of 228 ± 32 pA (extrapolated from the exponential fit). Apamin (100 nM) blocked current in five out of nine etv1 cells (with no effect in the remaining 4 cells). The apamin-sensitive component in the five apamin-sensitive cells (obtained by subtraction, 84 ± 31 pA) had a decay τ of 200 ± 65 ms. In glt cells, the tail current exhibited only a single decay time constant in three of six cells tested (Fig. 8D, inset). The other three cells had an additional slow component (decay τ of 1,071 ± 307 ms). The apamin-sensitive current in these six cells had a decay τ of 158 ± 71 ms and an amplitude of 344 ± 124 pA. Thus apamin blocked a current consistent with I_mAHP in both cell types, but this current was much larger in glt cells. Given the microdomain interaction between intracellular Ca²⁺ and SK channels (Abel et al. 2004; Andrade et al. 2012; Fakler and Adelman 2008; Jones and Stuart 2013), it is likely that either Ca²⁺ entry is greater, or there are more SK channels expressed in glt cells vs. etv1 cells.

Fig. 8.

Actions of small-conductance calcium-activated potassium (SK) channels were tested by their sensitivity to 100 nM apamin. A: single AP (10-ms current injection) in etv1 neuron. Apamin blocked the AHP after the spike (mAHP) in this cell, but this did not reach statistical significance across cells. Insets: scale bars and box plot of AHP amplitude (n = 8 cells). B: single AP in glt neuron (10-ms current injection). Apamin significantly reduced the mAHP (inset: n = 8). C: 10 APs at 50 Hz in etv1 neuron. Apamin had no significant effect on the peak AHP. Inset: summary data for peak AHP amplitude (n = 8 etv1 cells and 8 glt neurons). D: 10 APs at 50 Hz in glt neuron. Apamin reduced the peak AHP in glt neurons only (n = 8 etv1 cells and 8 glt neurons). Inset: apamin-sensitive tail current had fast decay (τ ∼200 ms) in voltage clamp (protocol as in Fig. 7). Control and apamin-sensitive traces are the average of 8–12 individual traces. *Significant difference from etv1 neurons.

Apamin had no effect on RMP, peak or steady-state R_in, or rheobase in etv1 (n = 15) or glt (n = 11) neurons. Apamin also caused no significant change in the AHP following a single AP in etv1 neurons (control 2.4 ± 0.5 mV; apamin 1.5 ± 0.3 mV; n = 13; Fig. 8A). In contrast, in glt neurons, apamin significantly reduced the single AP mAHP (control 2.3 ± 0.3; apamin 1.1 ± 0.2; n = 13; Fig. 8B). Following 10 APs at 50 Hz, apamin again significantly reduced the peak AHP in glt neurons (control 4.8 ± 0.7 mV; apamin 3.2 ± 0.7 mV; n = 10; Fig. 8D) but not etv1 neurons (control 6.2 ± 0.8 mV; apamin 5.2 ± 0.8 mV; n = 14; Fig. 8C). The sAHP (measured at 500 ms after the last AP) was not affected by apamin in etv1 (control 3.6 ± 0.6 mV; apamin 3.2 ± 0.6 mV; n = 14) or glt (control 1.5 ± 0.7 mV apamin: 1.3 ± 0.6 mV; n = 10) cells (Fig. 8). Together with the tail current data, these data suggest relatively greater expression of SK channels (or greater Ca²⁺ entry or Ca²⁺ coupling to SK) in glt vs. etv1 cells and a much greater sAHP in etv1 cells.

To examine the functional consequences of differential expression of SK channels, we studied the effects of apamin on repetitive firing (at 3× rheobase) in eight etv1 cells and eight glt cells (Fig. 9). Neither firing rate nor f-I slope (for 200 ms or 2 s) was significantly affected by apamin in etv1 cells (Table 6, Fig. 9A). Furthermore, apamin did not affect slow SFA (Table 6, Fig. 9C), and the instantaneous firing frequency for the first ISI did not change significantly in etv1 cells (% of control 93 ± 15 Hz; % of apamin 101 ± 14 Hz) (Fig. 9C). Percentage of fast adaptation [100 * (frequency of 1st interval − frequency of 3rd interval)/frequency of 1st interval] was not significantly decreased by apamin in etv1 cells (% of control 45 ± 3%; % of apamin 40 ± 4%; P < 0.06).

Fig. 9.

Effects of apamin (100 nM) on repetitive firing. A: plot of f-I curve in etv1 neurons (n = 8). Apamin had no effect on f-I relationships in etv1 cells. B: f-I curve for glt neurons (n = 8). Glt cells fired faster in the presence of apamin, but apamin did not significantly affect f-I slope. C: plot of f-t for an exemplar etv1 cell (at 3× rheobase). Apamin had no effect on SFA in etv1 cells. Inset: same data expanded to emphasize the first few ISIs. In this cell, firing rate was increased for the first ISI, resulting in a slightly higher firing rate for the rest of the 2 s. On average, this effect was not statistically significant in etv1 cells. D: f-t plot for an exemplar glt cell (at 3× rheobase). Apamin increased firing rate in glt cells but had no effect on SFA. Inset: same data expanded to emphasize the first few ISIs. In this cell, firing rate was increased for the first ISI, resulting in a slightly higher firing rate for the rest of the 2 s. Similar data were obtained in 8/8 glt cells.

Table 6.

Effects of apamin and NE on repetitive firing

	Firing Rate, 2 s, Hz	f-I Slope, 200 ms, Hz/nA	f-I Slope, 2 s, Hz/nA	% Adaptation	Adaptation Index
Etv1
Control	15 ± 3 (8)	101 ± 9 (10)	40 ± 12 (10)	82 ± 3 (10)	35 ± 11 (10)
Apamin	12 ± 2 (8)	108 ± 14 (10)	45 ± 13 (10)	82 ± 5 (10)	39 ± 7 (10)
Control (NE)	9 ± 2 (8)	49 ± 8 (8)	15 ± 5 (8)	83 ± 2 (8)	55 ± 11 (8)
NE	26 ± 4 (8)*	81 ± 5 (8)*	47 ± 3 (8)*	59 ± 5 (8)*	5 ± 2 (8)*
Glt
Control	26 ± 3 (8)	160 ± 20 (10)	140 ± 20 (10)	18 ± 5 (10)	0.6 ± 0.2 (10)
Apamin	37 ± 16 (8)*	170 ± 20 (10)	150 ± 20 (10)	26 ± 5 (10)	0.5 ± 0.1 (10)
Control (NE)	23 ± 2 (9)	150 ± 7 (9)	120 ± 10 (9)	17 ± 3 (9)	0.6 ± 0.2 (9)
NE	24 ± 2 (9)	160 ± 7 (9)	130 ± 7 (9)	23 ± 5 (9)	0.7 ± 0.2 (9)

Data are presented as means ± SE (n, number of cells). Firing was elicited in response to 2-s current injections at different amplitudes. Firing rate (2 s) is the average firing frequency over the entire 2-s current injection (in Hz). Percentage of adaptation = 100 × (frequency at 3rd ISI − frequency at steady state ISI/frequency at 3rd ISI). Adaptation index was for firing at 3× rheobase.

^*Significant difference vs. control (P < 0.05, paired t-test). NE, norepinephrine.

In contrast, apamin did cause a significant increase in firing rate in glt cells (Table 6). This was due to a parallel (additive) shift in the f-I relationships (Fig. 9B), with no statistically significant effect on f-I slopes for either the first 200 ms or for the entire 2 s of firing (Table 6). Adaptation index and percentage of adaptation were also unchanged in glt cells by apamin (Table 6, Fig. 9D). The firing frequency for the first ISI was significantly increased by apamin in glt cells (control 70 ± 20 Hz; apamin 91 ± 22 Hz) (Fig. 9D). Two of the eight glt cells fired with an initial doublet after apamin. Percentage of fast adaptation was not significantly decreased by apamin in glt cells (control 37 ± 11%; apamin 40 ± 7%). Thus apamin increased firing rate in glt cells starting from the first ISI but did not alter fast or slow SFA.

Role of sAHP conductance.

The sAHP (and I_sAHP) in neocortical pyramidal cells is Ca²⁺ dependent and K⁺ selective, but the underlying channel type is not known (Abel et al. 2004; Andrade et al. 2012; Schwindt et al. 1988a). In many cell types, the sAHP is negatively modulated by neurotransmitters that activate PKA (reviewed in Andrade et al. 2012). A classic example is the sAHP reduction by NE, acting through β-receptors (Foehring et al. 1989; Madison and Nicoll 1982). We applied NE (10 μM) in the bath and examined R_in, sag, rheobase, AHPs, and repetitive firing.

The effects of NE were dramatically different between etv1 and glt cells for AHPs, AHP currents, and repetitive firing. Peak and slow AHPs were elicited by 10 5-ms current injections at 50 Hz. We tested 12 glt cells with 10 μM NE (Fig. 10B) and found that the effects of NE application were not statistically significant for the sAHP (control 0.9 ± 0.4 mV; NE 0.4 ± 0.1 mV) or the peak AHP (control 6.1 ± 0.6 mV; NE 5.8 ± 0.7 mV; n = 12 cells; Fig. 10, D and E). The effects of NE on the sAHP were much greater in etv1 cells, consistent with the greater expression of the sAHP (and I_sAHP) in this cell type. For all etv1 cells tested (n = 21), application of 10 μM NE greatly reduced the sAHP (measured at 500 ms after the last AP; Fig. 10A). The block was 40–100% (mean 74 ± 4.5%) from 3.5 ± 0.3 mV in control and 0.9 ± 0.2 mV in NE (P < 0.001; Fig. 10E). The peak AHP was also significantly reduced by 10 μM NE in etv1 cells (∼20% on average; control 6.5 ± 0.5 mV; NE 5.4 ± 0.5 mV; Fig. 10D), reflecting the contribution of the sAHP conductance to the peak AHP after 10 spikes (see above). The effects of NE on the sAHP and repetitive firing were mimicked by 100 μM isoproterenol (a β-agonist) in three etv1 cells.

Fig. 10.

Effects of norepinephrine (NE, 10 μM) on AHPs and tail currents. A: 10-AP protocol (50 Hz) for etv1 neurons. Note large sAHP in control trace (black) and reduction of peak and sAHP (measured at 500 ms) by NE (red trace). Top inset: average depolarizing RMP response of 10 cells to NE as a function of time. Bottom inset: scale bar for insets in A and B. B: 10-AP protocol for glt neuron. Note lack of sAHP in control trace (black) and lack of effect of NE on the AHP (red trace). Top inset: average depolarizing RMP response of 10 cells to NE as a function of time. Bottom inset: scale bars for voltage traces in A and B. C: tail currents measured at −60 mV after step from −70 mV to +30 mV in an etv1 neuron. The black trace is the control current. Note selective block of the slow component of decay in NE (blue trace). The NE-sensitive current (green trace) rose slowly to its peak and then decayed slowly. D: summary data for peak AHP. NE significantly reduced the peak AHP in etv1 cells (n = 21) but not glt cells (n = 9). Note that the sAHP conductance also contributes to the peak AHP. E: NE significantly reduced the sAHP (measured at 500 ms) in etv1 neurons (n = 21) but not glt neurons (n = 9). F: NE significantly reduced I_sAHP in etv1 cells (n = 6) but not glt neurons (n = 4). *Significant difference from etv1 neurons.

In voltage clamp, NE was tested on AHP tail currents (Fig. 10C). In etv1 cells (n = 6), the tail current could be well fit by a sum of two exponential components: 79 ± 27 ms (43% of total) and 1,058 ± 181 ms (57% of total). NE (10 μM) selectively reduced the slow component; the NE-sensitive current (obtained by subtraction) had a decay τ of 718 ± 173 ms (amplitude 152 ± 31 pA, 54% of control; Fig. 10, C and F). Glt cells (n = 4) exhibited only a single component to the tail current, with a decay τ of 75 ± 15 ms (n = 4; data not shown). NE did not significantly affect the amplitude of this tail current in glt cells (Fig. 10F).

To examine the functional consequences of differential expression of the sAHP, we next examined the effects of NE on repetitive firing of etv1 and glt cells (at 3× rheobase; Fig. 11). NE is known to reduce SFA in pyramidal cells, as well as increase f-I slope (Foehring et al. 1989; Madison and Nicoll 1982; Schwindt et al. 1988a). Firing (at 3× rheobase) was faster in NE for etv1 cells (Table 6). The f-I slopes for the first 200 ms of firing, as well as for the entire 2 s firing, were significantly increased in etv1 cells by NE (n = 9; Fig. 11A, Table 6). In contrast, in glt cells (n = 9), NE did not significantly affect firing rate (Table 6) or f-I gain over the first 200 ms or at 2 s (Table 6, Fig. 11B). NE also had dramatically different effects on SFA in etv1 and glt cells. NE greatly reduced SFA measured as a percentage of adaptation (n = 9) or adaptation index (n = 8) in etv1 cells but not glt cells (Fig. 11, Table 6; both at 3× rheobase). This confirms the previously determined mechanistic link between the sAHP conductance and slow SFA in pyramidal neurons (Foehring et al. 1989; Madison and Nicoll 1982; Schwindt et al. 1988b).

Fig. 11.

Effects of 10 μM NE on repetitive firing of etv1 and glt neurons. A: average f-I curve in etv1 neurons (n = 10). Data are presented as means ± SE for 10 cells tested before and after NE. NE significantly increased firing rate and f-I slope in etv1 neurons. B: average f-I plot for glt neurons (n = 9). Data are presented as means ± SE for 9 cells tested before and after NE. NE did not significantly affect firing rate or f-I slope in glt neurons. C: instantaneous firing frequency (1/ISI) vs. time (f-t) in an etv1 neuron (I = 3× rheobase). NE increased firing rate by blocking SFA in this cell. Inset: box plot showing significant reduction in percentage of adaptation by NE in etv1 cells (n = 10). D: f-t plot for a glt neuron (I = 3× rheobase). NE did not affect SFA in this cell. Inset: box plot showing no change in percentage of adaptation by NE in glt cells (n = 9). *Significant difference from etv1 neurons.

Other effects of NE.

In addition to its effects on the sAHP and repetitive firing, NE elicited a depolarization that lasted several minutes in both etv1 (n = 10) and glt cells (n = 9; Fig. 10, A and B). R_in was not significantly changed by NE in etv1 cells (n = 10; control 119 ± 8 MΩ; NE 125 ± 10 MΩ) or glt cells (n = 9; control 135 ± 13 MΩ; NE 145 ± 14 MΩ). Sag was not changed by NE in etv1 cells (control 16 ± 2%; NE 14 ± 2%; n = 9) but was significantly reduced in glt cells (control 26 ± 2%; NE 22 ± 2%; n = 8). Rheobase was unchanged in glt neurons by NE (control 65 ± 12 pA; NE 65 ± 19 pA; n = 10) but was significantly reduced in etv1 cells (control 179 ± 23 pA; NE 138 ± 21 pA; n = 12).

DISCUSSION

We used whole cell recordings from mouse neocortical pyramidal cells in acute brain slices from BAC mice to test for electrophysiological differences between etv1 and glt cells. We found clear differences between etv1 and glt cells in APs, AHPs, and firing properties. Compared with glt cells, etv1 cells were generally less excitable in that they had higher rheobase and V_th. Etv1 cells also had broader APs, less sag, and larger sAHPs, and they fired slower, with lower gain and greater SFA. In contrast, the more excitable Glt cells had lower V_th, narrower APs, greater sag, larger mAHPs, and very little sAHP. The narrower APs in glt cells reflected the higher peak dV/dt for both spike polarization and repolarization, suggesting differences between layer etv1 and glt cells in both depolarizing Na⁺ currents and repolarizing K⁺ currents (Sheets et al. 2011). Glt cells also fired faster and with higher gain and exhibited almost no SFA. We confirmed the findings of Groh et al. (2010) that etv1 pyramidal cells differed from glt pyramidal cells in AP width and SFA. A notable difference in our study compared with Groh et al. (2010) was that they found R_in to be lower in glt cells, but R_in did not differ significantly between pyramidal cell types in our sample.

Ca²⁺-dependent AHPs and associated currents.

The most dramatic differences between etv1 and glt neurons were in the expression of Ca²⁺-dependent mechanisms: the mAHP, sAHP, and repetitive firing (especially SFA). Whereas glt neurons had larger apamin-sensitive mAHP and I_mAHP, etv1 neurons had larger sAHP and I_sAHP. This suggests differential expression of channel types, with greater relative importance of SK channels in glt cells and greater relative importance of sAHP channels in etv1 and layer 2/3 pyramids. In CA1 pyramidal cells (Storm 1987), Ca²⁺-dependent mechanisms also contribute to spike repolarization. In contrast, only the last phase of spike repolarization is Ca²⁺ dependent in neocortical pyramidal cells (Abel et al. 2004; Andrade et al. 2012; Higgs et al. 2006; Pineda et al. 1998; Schwindt et al. 1988a; Sun et al. 2003; Traub et al. 2003). Our Cd²⁺ data suggest that the more rapid repolarization of the AP in glt neurons is unlikely to be attributable to differential expression of somatic BK channels compared with etv1 cells. Rather, these cell types may differ in voltage-gated potassium channel expression (Sugino et al. 2006; Toledo-Rodriguez et al. 2004).

Differences in repetitive firing.

The dramatic differences in the relative importance of SK and sAHP channels underlie distinctive patterns of repetitive firing in etv1 and glt cells. Glt cells had greater expression of SK conductance. These channels had a subtractive effect on firing such that apamin significantly increased the instantaneous firing rate starting from the first ISI in glt cells but had no effects on firing rate in etv1 cells. Our apamin data also indicated that SK channels do not affect gain of firing or SFA in either etv1 or glt neurons.

In contrast, etv1 cells had greater expression of the sAHP conductance. This conductance regulates f-I slope and SFA in neocortical pyramidal neurons (Foehring et al. 1989; Lorenzon and Foehring 1993; Schwindt et al. 1988b). Consequently, the most dramatic firing difference between pyramidal cell types was for SFA. Whereas most glt cells did not display SFA after the third spike, etv1 neurons exhibited a strong slow phase of SFA. Also consistent with the greater sAHP expression in etv1 neurons, we found that firing rates to given current injections and the gain of firing (f-I slope) in response to DC inputs were lower in etv1 neurons vs. glt neurons. Thus firing rate would be expected to change less with increased synaptic drive in etv1 cells compared with glt cells.

Also striking was the differential modulation of firing in etv1 and glt neurons by NE. NE selectively reduced the sAHP conductance. Because most glt neurons had very small sAHPs (and I_sAHP), NE had little effect on sAHP amplitude in these cells. The more substantial sAHP expression in etv1 cells provides a greater substrate for modulation by NE acting through β-receptors (Foehring et al. 1989; Lorenzon and Foehring 1993; Madison and Nicoll 1982). Consequently, NE had much greater effects on repetitive firing behavior in etv1 cells than glt cells. By blocking the conductance underlying the sAHP, NE increased gain and decreased SFA in etv1 cells, eliminating the most dramatic differences in firing behavior between etv1 and glt neurons. Thus firing in etv1 cells has a temporal structure (larger SFA) that is highly sensitive to transmitter modulation. In contrast, NE had no significant effects on f-I slope or SFA in glt cells.

Morphological characterization of etv1 and glt cells.

The morphology, laminar location, and projections of EGFP-positive cells in etv1 and glt mice were characterized in detail by Groh et al. (2010). They found that, in barrel cortex, etv1 cells were restricted to layer 5A and glt cells to layer 5B (see also Sharifullina 2011). GFP-positive cells accounted for ∼55% of excitatory neurons in layers 5A (etv1) and 5B (glt). Furthermore, etv1 cells were reported to be slender-tufted (Larkman and Mason 1990), having significantly smaller soma and apical dendrite diameters, as well as narrower apical tufts in layer 1, compared with glt cells (Groh et al. 2010). Groh et al. (2010) also used retrograde labeling to show that glt pyramids project to ipsilateral pons and thalamus but not to contralateral striatum, and thus they are a subset of PT-type cells (Catsman-Berrevoets et al. 1980; Reiner et al. 2003, 2010; Shepherd 2013). Etv1 pyramidal cells were found to be a subset of IT-type neurons; they project callosally and to ipsilateral striatum and to some extent contralateral striatum but not to pons or thalamus (Groh et al. 2010). In our study, we did not examine projection patterns; however, we confirmed the findings of Groh et al. (2010) that etv1 cells were mostly located in superficial layer 5 in somatosensory cortex and that, on average, glt cells had a deeper distribution. In contrast to their findings, we found extensive overlap between the distribution of glt and etv1 cells in layer 5 and no differences in their soma areas. One possible reason for these differences between studies is that we examined S1 somatosensory cortex in general, and they restricted their study to barrel cortex. Another possibility is genetic differences in our glt animals vs. those used by Groh et al. (2010) despite both being derived from the same BAC lines.

Comparisons with the electrical properties of pyramidal cells identified by other means.

Layer 5 pyramidal cells have been classified many ways, including dendritic morphology, laminar location, projections to specific targets, projection pattern (IT-type vs. PT-type), or possession of specific electrophysiological characteristics. Numerous studies have revealed correlations between classification schemes, but their concordance is not perfect. There have also been several attempts to correlate gene expression with pyramidal cell projection pattern (Christophe et al. 2005; Groh et al. 2010; Sugino et al. 2006), but it is not clear how precisely these groupings correspond to other types of classification.

In somatosensory (but not visual) cortex, etv1 cells were reported to be a subset of IT-type, slender-tufted pyramidal cells and glt neurons a subset of the PT-type, thick-tufted pyramidal cells (Groh et al. 2010). How do our results for etv1 and glt cells compare to the electrophysiological properties of pyramidal cells defined by dendritic morphology (slender- vs. thick-tufted), laminar location (layer 5A vs. 5B), or projection pattern (IT-type vs. PT-type)?

Slender- vs. thick-tufted cells.

Sharp electrode recordings from rat cortex were originally used to correlate morphology with electrophysiology in layer 5 (Larkman and Mason 1990; Mason and Larkman 1990). They found that slender-tufted layer 5 cells had higher R_in, less sag, a lower rate of rise and fall of the AP, broader APs, and greater SFA vs. thick-tufted cells (Kasper et al. 1994; Larkman and Mason 1990). AP thresholds were similar between these cell types. The mAHP after a single AP was larger in the slender-tufted layer 5 cells vs. the thick-tufted cells. Our findings that etv1 cells had less sag, lower rate of rise and fall or the AP, broader APs, and greater SFA than glt cells are similar to differences between thin- and thick-tufted cells, respectively. Our data differed in that AP thresholds were higher and the mAHP smaller in etv1 cells compared with thin-tufted cells.

Firing pattern.

In the original sharp electrode recordings (Kasper et al. 1994; Mason and Larkman 1990), layer 2/3 pyramidal cells, slender-tufted layer 5 cells, and ∼1/3 of the thick-tufted cells fired in an RS pattern (Connors et al. 1982; McCormick et al. 1985; but see Schubert et al. 2006). The rest of the thick-tufted cells in layer 5 fired in an intrinsic bursting pattern, with 3–5 APs of descending amplitude and high frequency in response to a just suprathreshold current injection (Connors et al. 1982; McCormick et al. 1985). Burst firing in neocortical pyramidal cells is labile, however. Layer 5B pyramidal cells have a higher propensity for burst firing in response to dendritic stimulation vs. somatic current injection (Larkum et al. 1999, 2001, 2004, 2007). Also, prolonged anesthesia with ketamine or pentobarbital elicits burst firing in vivo in layer 5 neurons projecting to superior colliculus (PT-type) but not in cells projecting to the contralateral hemisphere (IT-type; Christophe et al. 2005). All of the cells we tested fired in an RS pattern, i.e., no true burst-firing cells were observed. True burst firing (as defined by McCormick et al. 1985) is rarely seen in our hands with somatic whole cell recordings from rodent cortex (with somatic current injection). Some of the burst firing observed in sharp electrode studies may reflect dendritic recordings and dendritic current injection (Schwindt et al. 1997). Burst firing is sensitive to animal age (Franceschetti et al. 1998; Kasper et al. 1994), bath temperature (Hedrick and Waters 2012), internal Ca²⁺ chelation (Lorenzon and Foehring 1995; Schwindt et al. 1992b), ionic composition of internal and external solutions, and perhaps species and strain differences.

Most neocortical neurons are, however, capable of two-spike bursts, and Groh et al. (2010) observed that ∼50% of glt cells fired with two-spike intrinsic bursts. Under our recording conditions, two-spike bursts were restricted to glt cells (∼13%). Bursts, by means of synaptic facilitation, are thought to make signaling more reliable compared with single APs (Lisman 1997) and can be regulated by K⁺ currents with properties similar to SK channels (Kepecs et al. 2002). Perhaps the relatively large SK conductance in glt cells (vs. etv1) is related to the need for regulation and termination of facultative burst firing in these cells (in response to dendritic excitation).

Thick- and slender-tufted cells vs. projection pattern.

Projection pattern and slender- vs. thick-tufted morphology are strongly correlated in several cortical areas (Avesar and Gulledge 2012; Dembrow et al. 2010; Gee et al. 2012; Hattox and Nelson 2007; Morishima et al. 2011). Whereas thick-tufted pyramidal cells projecting to brainstem (PT-type) are restricted to deep layer 5 (layer 5B), slender-tufted neurons projecting callosally (IT-type) are more widespread but are more frequently observed in the more superficial layer 5A (Christophe et al. 2005; Larkman and Mason 1990; Mason and Larkman 1990; Wise and Jones 1976). Connectivity between pyramidal cells also correlates with subcortical projection pattern (Brown and Hestrin 2009; Morishima et al. 2011; Ueta et al. 2013).

Several studies have examined differences in electrophysiological properties depending on projection targets. These data can be summarized as follows: thick-tufted neurons projecting to the brainstem or spinal cord (PT-type) have low R_in and V_th, narrow APs, small sAHPs, rapid firing with high gain, and little SFA (Avesar and Gulledge 2012; Brown and Hestrin 2009; Dembrow et al. 2010; Hattox and Nelson 2007; Le Be et al. 2007; Schwindt et al. 1997; Sheets et al. 2011; Suter et al. 2013). Slender-tufted callosal-projecting neurons (IT-type) have high R_in and V_th, broad APs, pronounced sAHP, slower firing with lower gain, and pronounced SFA (Avesar and Gulledge 2012; Brown and Hestrin 2009; Dembrow et al. 2010; Hattox and Nelson 2007; Le Be et al. 2007; Schwindt et al. 1997; Sheets et al. 2011; Suter et al. 2013). Thick-tufted neurons projecting to the brainstem or spinal cord also have larger and faster I_H (and correspondingly more pronounced voltage sag in current clamp) than callosal-projecting (or contralateral striatal-projecting) neurons (Brown and Hestrin 2009; Christophe et al. 2005; Dembrow et al. 2010; Gee et al. 2012; Solomon et al. 1993; Sheets et al. 2011).

Despite the lack of clear differences in laminar distribution or soma size, our findings for the electrical properties of etv1 vs. glt cells were compatible with their representing subsets of IT-type vs. PT-type neurons, respectively (as reported by Groh et al. 2010). Consistent with the IT-type pattern described above, etv1 cells had higher V_th, broader APs, less sag, and larger sAHPs, and they fired slower, with lower gain and greater SFA. Glt cells were like PT-type neurons in that they had lower V_th, narrower APs, greater sag, and larger mAHPs with very little sAHP. Glt cells fired faster and with higher gain and exhibited almost no SFA. Glt neurons also had greater sag in response to hyperpolarization, consistent with increased I_H in these cells compared with etv1 neurons. Compared with layer 5 cells, layer 2/3 pyramidal cells were more hyperpolarized, with intermediate R_in, almost no sag, intermediate SFA, RS, smaller mAHP, similar V_th, intermediate spike width, faster rate of rise, and intermediate rate of fall.

Differences between studies.

In addition to the overlapping laminar distribution of etv1 and glt cells, some of our electrophysiological measurements were not consistent with studies of IT-type vs. PT-type neurons. In motor cortex, IT-type corticostriatal neurons had lower rheobase than corticospinal neurons (Sheets et al. 2011), whereas we found that rheobase was lower in glt neurons compared with etv1. Although our data showed that etv1 cells had less sag than glt cells, our sag values for etv1 cells were higher than typically seen for callosal-projecting IT-type cells (Avesar and Gulledge 2012; Dembrow et al. 2010; Sheets et al. 2011). Possible reasons for these discrepancies include diversity within PT-type and IT-type pyramidal cells, species and age differences, differences in internal and external solutions, differences in protocols between labs, statistical differences between small data samples, or genetic drift within mouse lines. A notable difference in our study compared with Groh et al. (2010) was that they found R_in to be lower in glt cells, but R_in did not differ between pyramidal cell types in our sample. This difference is consistent with our finding that etv1 and glt cells did not differ in soma area, in contrast to differences in cell size reported by Groh et al. (2010).

Functional consequences.

In summary, the following suite of characters appears to be generally characteristic of etv1 cells (and IT-type cells): higher V_th, broader APs, less sag, larger sAHPs, and slower firing with lower gain and greater SFA. It has been suggested that SFA in IT-type neurons may be required to stabilize cortical networks (Shepherd 2013). Glt cells (and PT-type cells) had the following characteristics: lower V_th, narrower APs, greater sag, very little sAHP, and faster firing with higher gain and almost no SFA. With these properties, glt cells (and PT-type cells) may be specialized for reliable coding of inputs, especially at higher frequencies (Schwindt et al. 1997; Shepherd 2013). Pyramidal cells with small sAHPs (e.g., glt) are the closest pyramidal cell type to being ideal temporal integrators (Higgs et al. 2006). On the other hand, adapting cells (e.g., like etv1) can accurately follow rapidly changing input at any amplitude (Schwindt et al. 1997).

Consistent with possible roles in processing temporal aspects of sensory input, layer 2/3 pyramidal cells and layer 5 cells with large sAHPs (of which etv1 cells are an example) would have more sensitivity to fluctuations in voltage and greater capabilities for coincidence detection and help to maintain the signal-to-background ratio during strong background excitation (Higgs et al. 2006). In vivo, pyramidal cells are embedded in the cortical network and subjected to synaptic bombardment that has temporal structure. Such noisy “background” synaptic input may alter the sensitivity (modulate gain) of cortical neurons (Chance et al. 2002). Similar gain changes have been proposed as a potential mechanism for multiplicative gain control, which occurs during cognitive processes such as focused attention (McAdams and Maunsell 1999; Salinas and Thier 2000; Treue and Martinez-Trujillo 1999). In slice recordings from pyramidal cells with a large sAHP (as found in etv1 cells), suprathreshold coincidence detection is promoted, and gain is selectively increased by injected background noise that mimics in vivo synaptic input (Higgs et al. 2006). In contrast, pyramidal cells with little sAHP (like glt cells) show decreased gain in the presence of noise (Higgs et al. 2006). These results predict that large gain increases would occur with synaptic bombardment in etv1 cells in an awake behaving animal but not in glt cells.

Effects of NE.

Noradrenergic projections from the locus coeruleus (LC) are thought to control vigilance via cortical actions (Aston-Jones et al. 1999; Berridge and Waterhouse 2003). Low LC activity is correlated with low arousal and drowsiness, optimal performance is associated with intermediate levels of LC activity that may promote focused or selective attention, and high LC activity may produce a state of high behavioral flexibility or scanning attentiveness that leads to loss of attention and reduced performance on tasks requiring attention or vigilance (Aston-Jones et al. 1999).

In vivo, there is a continuum of effects of NE on pyramidal cells in layer 5 (Waterhouse et al. 2000). In slices, NE depolarized PT-type layer 5 pyramidal neurons, via α1-adrenergic action (Wang and McCormick 1993). In the present study, we observed depolarizations in both etv1 and glt neurons in response to NE application. We also found that NE induced a reduction in voltage sag to hyperpolarization in glt cells but not etv1 cells. This finding is similar to previous studies where NE was found to modulate I_H through α2 inhibition of PKA (Wang et al. 2007) selectively in PT-type layer 5B pyramidal cells, resulting in increased excitability (Dembrow et al. 2010; Sheets et al. 2011; Shepherd 2013). Our findings suggest that NE also has additional profound effects on the gain and time-dependent properties of repetitive firing specifically in etv1 neurons but had little effect on repetitive firing of glt neurons. The effects of NE on the sAHP and firing in pyramidal neurons are mediated by β-receptor activation of PKA (Foehring et al. 1989; Lorenzon and Foehring 1993; Madison and Nicoll 1982). Thus different effects of NE are exerted on glt (or PT-type) cells vs. etv1 (or IT-type) cells. It is presently unknown how these mechanisms interact at the level of neocortical circuits.

Summary.

We found that, although glt cells were found deeper than etv1 cells, on average, there was extensive overlap in the distribution of these cell types in layer 5 of somatosensory cortex and that soma size did not differ between them. There were several electrophysiological differences between etv1 and glt pyramidal neurons. The repetitive firing behavior of these two cell types was distinctive, with etv1 cells firing slower, with less gain and strong SFA. Etv1 cells also have higher AP threshold, broader APs, and higher rheobase. These electrophysiological properties are similar to those of IT-type pyramidal neurons in other studies. In contrast, glt cells fired faster, with higher gain and almost no SFA. Glt neurons also have lower AP threshold, narrower APs, and lower rheobase. These properties are similar to those of PT-type pyramidal neurons. A prominent difference was that differential expression of Ca²⁺-dependent SK and sAHP conductances underlie much of the differences in firing behavior between etv1 and glt cells. The low gain and high SFA of etv1 cells is controlled by a much larger sAHP conductance, which also manifests as larger sAHPs and larger peak AHPs in response to multiple APs at high frequency. The large sAHP would be predicted to increase the capacity of etv1 neurons to detect coincidence between inputs, especially in the face of noisy background inputs. In contrast, glt neurons have very little sAHP or I_sAHP and fire with high gain and almost no SFA. In glt neurons, the effective SK expression is greater, and these channels have a much greater effect controlling firing rate in glt than etv1 cells. Glt neurons may more reliably code inputs and respond to noisy background inputs with reduced gain. The larger sAHP mechanism in etv1 cells provides a greater substrate for NE modulation of firing behavior. In the presence of NE, etv1 cells fire faster, and SFA is reduced, eliminating much of the difference in firing behavior with glt cells.

GRANTS

This work was supported by USPHS Grant NS-044163-10.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: D.G., W.E.A., and R.C.F. conception and design of research; D.G. and R.C.F. performed experiments; D.G., W.E.A., and R.C.F. analyzed data; D.G., W.E.A., and R.C.F. interpreted results of experiments; D.G., W.E.A., and R.C.F. prepared figures; D.G., W.E.A., and R.C.F. edited and revised manuscript; D.G., W.E.A., and R.C.F. approved final version of manuscript; R.C.F. drafted manuscript.