Abstract

The heterogeneity of γ-aminobutyric acid interneurons in the rodent neocortex is well-established, but their classification into distinct subtypes remains a matter of debate. The classification of interneurons in the primate neocortex is further complicated by a less extensive database of the features of these neurons and by reported interspecies differences. Consequently, in this study we characterized 8 different morphological types of interneurons from monkey prefrontal cortex, 4 of which have not been previously classified. These interneuron types differed in their expression of molecular markers and clustered into 3 different electrophysiological classes. The first class consisted of fast-spiking parvalbumin-positive chandelier and linear arbor cells. The second class comprised 5 different morphological types of continuous-adapting calretinin- or calbindin-positive interneurons that had the lowest level of firing threshold. However, 2 of these morphological types had short spike duration, which is not typical for rodent adapting cells. Neurogliaform cells (NGFCs), which coexpressed calbindin and neuropeptide Y, formed the third class, characterized by strong initial adaptation. They did not exhibit the delayed spikes seen in rodent NGFCs. These results indicate that primate interneurons have some specific properties; consequently, direct translation of classification schemes developed from studies in rodents to primates might be inappropriate.

Introduction

Cortical γ-aminobutyric acid (GABA) interneurons are heterogeneous, with subpopulations distinguished by particular combinations of morphological, physiological, and molecular attributes (Cauli et al. 1997; Kawaguchi and Kubota 1997; McBain and Fisahn 2001) and with variance in these attributes within subpopulations (Soltesz 2006). In addition, abnormalities of specific types of cortical interneurons are thought to be critical components of the pathophysiological mechanisms underlying human brain disorders such as epilepsy and schizophrenia (DeFelipe 1999; Lewis et al. 2005). However, relatively few studies characterizing the properties of cortical GABA neurons have been performed in monkeys or humans (e.g., Krimer et al. 2005; Szabadics et al. 2006); for review of earlier studies, see Avoli and Williamson (1996). Consequently, the proper translation of the extensive and growing database of rodent cortical interneurons to an understanding of these human disorders requires the ability to identify similar types of interneurons across species.

Although homologous types of interneurons have been reported in multiple species, differences have been observed even between phylogenetically close species. For instance, in rat cortex parvalbumin (PV), somatostatin (SST), and calretinin (CR) interneurons constitute primarily nonoverlapping subpopulations (Gonchar and Burkhalter 1997; Kawaguchi and Kubota 1997), whereas in the mouse cortex, a large subpopulation of interneurons coexpresses CR and SST (Xu et al. 2006). Furthermore, compared with rodent neocortex, in the primate neocortex 1) the percentage of cortical neurons that are GABAergic is larger (Gabbott and Bacon 1996; Gabbott et al. 1997), 2) interneurons characterized by a vertical bundling of axons are much more common (Yanez et al. 2005), 3) the developmental origin of at least some interneurons appears to differ (Letinic et al. 2002; Molyneaux et al. 2007), and 4) the relative proportions of chemically identified subtypes of interneurons are dissimilar (Conde et al. 1994; Kawaguchi and Kubota 1997). Moreover, interneurons with firing properties unusual for rodents have been found in monkey prefrontal cortex (Krimer et al. 2005; Povysheva et al. 2007). Thus, a robust and reliable classification of different interneuron types in the primate neocortex is critically needed.

Recently, we attempted to functionally categorize interneurons in monkey dorsolateral prefrontal cortex (DLPFC) by correlating their electrophysiological properties either with morphological types (Krimer et al. 2005) or with calcium-binding protein (CaBP) content (Zaitsev et al. 2005). By using cluster analysis, we demonstrated that monkey interneurons form distinct physiological groupings. However, the physiological-based clusters obtained in these studies appeared to contain heterogeneous morphological and molecular types, and thus a different approach for classification is needed.

In order to address this issue, in this study we used morphological criteria as a starting point for identifying subsets of interneurons in monkey DLPFC. Of the 8 morphological types of layer 2–3 interneurons identified, the electrophysiological and molecular properties of 4 types have not been previously described in primates. We found that monkey interneurons of the same morphological type exhibited similar electrophysiological and molecular attributes; at least some morphological types of monkey interneurons demonstrated different membrane properties from those for homologous morphological types described in rat. We did not observe interneurons, exhibiting late-spiking, stuttering, or bursting firing patterns, which are typical for some types of rodent interneurons, while we detected some firing patterns that are unusual for rats. These findings indicate that direct translation of classification schemes developed from studies in rodents to primates might be inappropriate.

Materials and Methods

Slice Preparation

Seventeen experimentally naive young adult (3.5–6 kg, 3.5–4 years old) male long-tailed macaque monkeys (Macaca fascicularis) were used in this study. Animals were treated according to the guidelines outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals, as approved by the University of Pittsburgh Institutional Animal Care and Use Committee. The procedure used to obtain tissue from the DLPFC has been previously described in detail (Gonzalez-Burgos et al. 2004). Briefly, animals were treated with ketamine hydrochloride (25 mg/kg, intramuscular [im]), dexamethasone phosphate (0.5 mg/kg, im), and atropine sulfate (0.05 mg/kg, subcutaneous); an endotracheal tube was inserted, and anesthesia was maintained with 1% halothane in a 28% O2–air mixture. Monkeys were placed in a stereotaxic apparatus, and a craniotomy was performed over the DLPFC in one hemisphere. The dura was removed in a location determined by stereotaxic coordinates and by the position of relevant sulcal landmarks, and a small block of tissue was excised containing both the medial and lateral banks of the principal sulcus (area 46) as well as a small adjacent portion of dorsal area 9. After the surgery, the animals were treated with an antibiotic (chloramphenicol, 15 mg/kg, im) and an analgesic (hydromorphone, 0.02 mg/kg, im) 3 times a day for 3 days. All animals recovered quickly with no impairments in eating or drinking and no overt behavioral deficits. In most cases, the animals underwent the same procedure 2–4 weeks later to obtain tissue from the opposite hemisphere. During the second procedure, after the craniotomy, the animal was given an overdose of pentobarbital (30 mg/kg) and was perfused through the heart with ice-cold modified artificial cerebrospinal fluid. A tissue block containing portions of areas 46 and 9 from a nonhomotopic portion of the contralateral hemisphere was quickly excised. Subsequent treatment of the tissue was the same for both procedures.

The tissue blocks were placed in ice-cold Ringer solution, containing (in mM): NaCl 126, KCl 2.5, NaH2PO4 1.25, CaCl2 2, MgSO4 1, NaHCO3 26, and dextrose 10, pH 7.4, perfused with a 95%O2/5%CO2 gas mixture. Coronal 350-μm-thick slices were cut from each block using a Vibratome (VT 1000S, Leica, Germany) and incubated for 1 h at 36 °C, and at room temperature thereafter, or at room temperature from the beginning. For recordings, slices were submerged in a chamber mounted on the microscope and perfused with Ringer solution at 32 °C. These brain slices had been also used in other studies (Gonzalez-Burgos et al. 2005; Krimer et al. 2005; Zaitsev et al. 2005; Povysheva et al. 2007),

Electrophysiological Recordings

Interneurons in layers 2–3 were visualized using infrared differential interference contrast videomicroscopy and distinguished from pyramidal cells based on their small, round, or oval soma and the absence of an apical dendrite. Patch electrodes with open-tip resistances of 5–10 MΩ were filled with a solution containing (in mM): potassium gluconate 114, KCl 6, ATP-Mg 4, GTP 0.3, Hepes 10 (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), 0.5% biocytin, and pH 7.25 adjusted with KOH. Whole-cell current clamp recordings were performed after reaching seal resistance of at least 4–5 GΩ. Voltages were amplified using Intracellular Electrometers IE-210 (Warner Instrument Corporation, Hamden, CT) or MultiClamp 700A (Axon Instruments, Union City, CA), operating in a bridge-balance mode, filtered on line at 4–5 kHz and acquired on a personal computer at a sampling rate of 20 kHz using Power 1401 interface and Signal 2 or Signal 3 software program (CED, Cambridge, UK). To characterize the intrinsic membrane properties of neurons, hyper- and depolarizing current steps of 500 ms duration were applied in 5–10 pA increments at 0.2 Hz with 2 repetitions.

Electrophysiological Analysis

Twenty-one electrophysiological parameters were measured as follows:

  • 1. RMP (in mV): resting membrane potential, the stable membrane potential (no holding current applied) reached a few minutes after breaking the membrane.

Subthreshold membrane properties

  • 2. Rin (in MΩ): input resistance, the slope of the regression line fitted to the I–V curve (usually between –50 and –10 pA), as measured at the end of the 500-ms voltage responses.

  • 3. τ (in ms): membrane time constant, determined from the monoexponential curve best fitting to the average voltage response to hyperpolarizing current steps of –10 to –30 pA.

  • 4. Rb (in pA): rheobase, the intercept of the extrapolated F – I curve with the current axis.

  • 5. Sag (dimensionless): sag being the difference between the most negative membrane potential during a 500-ms hyperpolarizing current step and the membrane potential at the end of the step. For analysis, we graded sag into 3 intensities: “0,” if sag amplitude has been less than 20% of total voltage step at the end of 500 ms hyperpolarizing current step; “1,” if more than 20%, but less than 50%; “2,” if more than 50%.

  • 6. Hump (dimensionless): hump being the difference between the most positive membrane potential during a 500-ms depolarizing current step and the membrane potential at the end of the step. We used the same grades as for Sag.

  • 7. RD (dimensionless): rebound depolarization (or spike/s), measured as positive voltage deflection above RMP after the offset of hyperpolarizing current. For analysis, we distinguished 3 grades of RD intensity: “0,” if RD amplitude has been less than 20% of total voltage step at the end of 500 ms hyperpolarizing current step; “1,” if more than 20%, but did not evoke rebound spike; “2,” if evoked rebound spikes.

  • 8. Sum (dimensionless): the sum of grades for Sag, RD, and Hump.

Action potential (AP) properties (minimal suprathreshold current were applied)

  • 9. DAP (in ms): time to first spike from the beginning of stimulation.

  • 10. APT (in mV): action potential threshold, the membrane potential at the point at which the interpolated rate of voltage rise (dV/dt) reached >10 mV/1 ms.

  • 11. APA (in mV): action potential amplitude, measured from the threshold to the peak.

  • 12. APD (in ms): action potential duration, the spike width at its half-amplitude.

  • 13. AHPAF (in mV): amplitude of the fast component of the afterhyperpolarization (AHP), measured from the APT downward the fast voltage drop until the point of a marked slowing in the voltage drop to <5 mV/1 ms.

  • 14. AHPAM (in mV): amplitude of medium component of the AHP was measured from the end of fast component of the AHP to the most negative membrane potential after the spike.

  • 15. tAHP (in ms): AHP latency, the time interval between onset of AHP and the hyperpolarization peak.

Firing pattern properties (2× Rb current was applied)

  • 16. Fr (in Hz): steady-state frequency, the reciprocal of the average of the 4–9 interspike intervals (ISIs), measured within the last 250 ms of the response to depolarizing current pulses where firing frequency remained relatively stable.

  • 17. kISIs (dimensionless): coefficient of variance of ISIs measured within the last 250 ms of the response to depolarizing current pulses.

  • 18. FrAL1 (in %): late frequency adaptation, percentage of decrease in the frequency from onset (reciprocal to the first ISI) to steady-state frequency.

  • 18a. FrAL2 (in %): late frequency adaptation, percentage of decrease in the frequency from onset (reciprocal to the second ISI) to steady-state frequency.

  • 19. FrAI (in %): initial frequency adaptation, percentage of decrease in frequency, measured as the reciprocal of first and second ISIs.

  • 20. ΔAPA (in %): percentage change in AP amplitude between the first and last APs.

  • 21. ΔAHPA (in %): percentage change in total AHP amplitude between the first and last APs.

Principal Component Factor Analysis of Electrophysiological Parameters

To compress the variability of these electrophysiological parameters into a smaller number of variables, principal component factor analysis was conducted using multiple R-square algorithms. In this method, prior to factoring the diagonal of the correlation matrix (communalities) is computed as the multiple R-square of the respective variable with all other variables. According to the Kaiser criterion, we retained 3 factors with eigenvalues greater than 1. Then a “varimax” rotation of the factor loadings was performed. This rotation is aimed at maximizing the variances of the squared raw factor loadings across variables for each factor; this is equivalent to maximizing the variances in the columns of the matrix of the squared raw factor loadings. Interpretation of factors was done according to their factor loadings.

“Phase Plot” Representation of AP

Phase plots show the rate of change in the membrane potential dV/dt (velocity) against the instantaneous membrane potential V(t). Differentiation of the membrane potential was done with Signal 3 software; each data point was replaced with the difference between that point and the previous point and divided the result by the sample interval; the first data point was set to zero. Sample interval (Δt) was equal to 0.05 μs.

These plots provide additional information about ionic currents during AP (Bean 2007). The net ionic current is proportional to velocity (Iionic = –CdV/dt, where Iionic is the net ionic current, C is the cell capacitance). Positive velocity means inward current and the negative velocity outward current. In phase plots, an AP is represented by a loop (Fig. 8). Start point of the loop represents the APT. From this point, the velocity rapidly increases during depolarizing phase of AP from about 0 Vs−1 to 200–500 Vs−1 and then decreases, crossing 0 Vs−1 at the AP peak values. During repolarizing, phase velocity has negative values and gets 0 Vs−1 at the peak of fast hyperpolarization. Shift in voltage between start point and end of the loop represents fast component of AHPA.

Histological Processing and Morphological Analysis

After recordings, slices were immersed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) for 24–72 h at 4 °C and then cryoprotected (33% glycerol, 33% ethylene glycol, in 0.1 M phosphate-buffered saline [PBS]) and stored at −80 °C. To visualize biocytin, slices were incubated with streptavidin-Alexa Fluor 633 conjugate (Invitrogen, Carlsbad, CA, dilution 1:500) for 24–48 h at 4 °C in PB, containing 0.4% Triton X-100. Interneurons were completely reconstructed in 3 dimensions using an Olympus Fluoview 500 confocal laser scanning microscope (Olympus America Inc., Center Valley, PA) equipped with a ×20/0.80 N.A. oil immersion objective. After the confocal reconstruction of the recorded interneurons, 106 with sufficiently extensive axonal arbors were processed for visualization of combinations PV/CR/calbindin D28k (CB)/neuropeptide Y (NPY)/SST immunoreactivity using triple immunofluorescent labeling (Zaitsev et al. 2005). Slices were serially resectioned at 40–50 μm and then incubated for 2–3 days at 4 °C in blocking serum (10% normal goat serum, 2% bovine serum albumin in PB) containing a mixture of 2 antibodies raised in the different hosts (Table 1). After thorough rinsing, the sections were incubated in a mixture of 2 corresponding secondary antibodies with the fluorescent tags: Alexa Fluor 546-labeled goat anti-mouse IgG, Alexa Fluor 488–conjugated goat anti-rabbit IgG, and Alexa Fluor 488–conjugated goat anti-rat IgG 1:500 (Invitrogen). This procedure yielded differential fluorescent covisualization of Alexa Fluor 633-biocytin–filled interneurons and Alexa Fluor 488- and 546-labeled molecules. Localization of neuropeptides and CaBPs was analysed with a ×40/1.3 N.A. oil immersion objective on the same confocal microscope.

Table 1

Primary antisera used in this study

Antigen Host Dilution Mono/polyclonal Source of antibody and its number 
CB Mouse 1:1000 Monoclonal Swant (300) 
CR Mouse 1:1000 Monoclonal Chemicon (MAB1568) 
CR Rabbit 1:1500 Polyclonal Swant (7699/4) 
NPY Rabbit 1:500 Polyclonal Sigma (N9528) 
PV Mouse 1:2000 Monoclonal Sigma (P3088) 
PV Rabbit 1:2000 Polyclonal Swant (PV 28) 
Somatostatin-14 Rat 1:200 Monoclonal Chemicon (MAB354) 
Antigen Host Dilution Mono/polyclonal Source of antibody and its number 
CB Mouse 1:1000 Monoclonal Swant (300) 
CR Mouse 1:1000 Monoclonal Chemicon (MAB1568) 
CR Rabbit 1:1500 Polyclonal Swant (7699/4) 
NPY Rabbit 1:500 Polyclonal Sigma (N9528) 
PV Mouse 1:2000 Monoclonal Sigma (P3088) 
PV Rabbit 1:2000 Polyclonal Swant (PV 28) 
Somatostatin-14 Rat 1:200 Monoclonal Chemicon (MAB354) 

After analysis of the fluorescent signals, the sections were treated with 1% H2O2 for 2–3 h at room temperature, rinsed, and incubated with the avidin–biotin–peroxidase complex (1:100; Vector Laboratories, Burlingame, CA) in PBS for 4 h at room temperature. Sections were rinsed, stained with 3,3′-diaminobenzidine (DAB), mounted on gelatin coated glass slides, dehydrated, and coverslipped.

In addition to this, new set of 106 confocally reconstructed interneurons, 88 physiologically and morphologically characterized interneurons from our previous studies (Krimer et al. 2005; Zaitsev et al. 2005) were included; 50 of these cells were tested for CaBPs (Zaitsev et al. 2005), whereas the remaining 38 interneurons were not tested for any molecular markers. Biocytin was visualized with DAB chromogen as described above. Some (n = 16) of these 88 neurons were digitally reconstructed using the Neurolucida tracing system (MicroBrightField, Williston, VT). Their morphological and physiological properties were reanalysed for the present study.

The horizontal and vertical extent of axons was measured as the mean distance between the 3 most distal axonal endings on each side from the soma of individual interneurons. Somal sizes were estimated from confocal images; cell bodies were approximated by an ellipse, and the area was calculated by an equation;

forumla, where Drd and Dtng are the radial and tangential axes, respectively, of the somata.

Statistical Analysis

All statistical tests were performed using Statistica 6.1 software (Statsoft Inc., Tulsa, OK). Unless otherwise stated, all data are reported as means and standard deviations. The statistical significance between group means was tested using analysis of variance (ANOVA), followed by Fisher’s Least Significant Difference (LSD) post hoc tests (multiple comparison tests).

Results

Morphological Diversity of the Recorded Interneurons and Their Classification

Layer 2–3 interneurons (n = 194) that retained a sufficiently extensive axonal tree within the slice to warrant a detailed reconstruction were included in this study. To differentiate morphological groups of interneurons, we modified the original classification of interneurons in the monkey DLPFC based on Golgi impregnations (Lund and Lewis 1993). In the classification scheme presented here (Fig. 1), we used 2 readily observed morphological criteria: 1) the distribution of axonal arbors across layers, which reflects the predominant laminar targets of a neuron; and 2) a combination of axonal arborization patterns and terminal branch properties, which may reflect the targeting of different subdomains of cortical neurons.

Figure 1.

Flow chart for morphological identification of layer 2–3 DLPFC monkey interneurons. All interneurons were 3-dimensional computer reconstructed using the Neurolucida tracing system. Interneuron somata and dendrites are drawn in red and the axons in blue. Scale bar = 100 μm. Layers are represented with Arabic figures.

Figure 1.

Flow chart for morphological identification of layer 2–3 DLPFC monkey interneurons. All interneurons were 3-dimensional computer reconstructed using the Neurolucida tracing system. Interneuron somata and dendrites are drawn in red and the axons in blue. Scale bar = 100 μm. Layers are represented with Arabic figures.

Using the first criterion, we distinguished three groups of interneurons. The first group of cells projected their axons toward layer 1, forming relatively wide axonal plexi in that layer, and thus provided inhibitory inputs to the tufts of pyramidal cells. Cells with a similar axonal projection in rodents were recognized as Martinotti cells (MCs) (Kawaguchi and Kubota 1997; Wang et al. 2004; Ma et al. 2006; Silberberg and Markram 2007).

The second group of cells formed relatively narrow, vertically oriented projections to the deep layers (Fig. 2). Although all these interlaminar–intracolumnar cells connect upper layers with deep layers, they have different morphological appearances. The axon of one type formed basket-like structures and, most likely, specialized in innervating the perisomatic regions of other neurons. We classified these as “vertically oriented cells with baskets” (VOBCs). The remaining vertically oriented cells may target mostly distal parts of pyramidal cells dendrites as they resemble double bouquet cells (DBCs), described in multiple species (Somogyi and Cowey 1981; Kawaguchi 1995).

Figure 2.

Morphological varieties of interneurons projecting to deep layers. (A) Confocal reconstruction of DBC. Confocal (B1) and Neurolucida (B2) reconstructions of the same VOBC. At the Neurolucida drawing soma and dendrites of VOBC are designated in red and the axons in blue. Potential postsynaptic cells for VOBC are shown in orange. Scale bars for reconstructed cells = 50 μm. (C) Axon terminals of VOBC formed appositions on unstained somata of potential postsynaptic cells, which were observed with differential interference contrast. Scale bar = 20 μm.

Figure 2.

Morphological varieties of interneurons projecting to deep layers. (A) Confocal reconstruction of DBC. Confocal (B1) and Neurolucida (B2) reconstructions of the same VOBC. At the Neurolucida drawing soma and dendrites of VOBC are designated in red and the axons in blue. Potential postsynaptic cells for VOBC are shown in orange. Scale bars for reconstructed cells = 50 μm. (C) Axon terminals of VOBC formed appositions on unstained somata of potential postsynaptic cells, which were observed with differential interference contrast. Scale bar = 20 μm.

The third group of interneurons appeared to be the most numerous and morphologically heterogeneous. Axons of cells in this group were mostly distributed in layers 2–3, although a few axonal branches projected to deeper layers or to layer 1, but without clustering. According to their pattern of arborization and terminal branch properties, we distinguished 5 different morphological types within this group (Figs 3 and 4). Two of these morphological types, namely, neurogliaform cells (NGFCs) (Kawaguchi and Kubota 1997; Povysheva et al. 2007) and chandelier cells (ChCs) (DeFelipe 1999), are widely recognized in different species. The other 3 morphological types resemble different types of basket cells described in many species; however, only one type of these cells actually formed basket or “claw”-like structures around potential postsynaptic cells, whereas the other 2 types did not.

Figure 3.

Confocal reconstructions of cells with specialized axonal structures. (A) LPBC. Note that terminal parts of axon of LPBC formed appositions on unstained somata of potential postsynaptic cells, which were observed with differential interference contrast (inserts). (B) ChC. Axon terminals formed short vertical arrangements of boutons—axon cartridges. Scale bars for confocal images = 20 μm; scale bar for inserts = 10 μm.

Figure 3.

Confocal reconstructions of cells with specialized axonal structures. (A) LPBC. Note that terminal parts of axon of LPBC formed appositions on unstained somata of potential postsynaptic cells, which were observed with differential interference contrast (inserts). (B) ChC. Axon terminals formed short vertical arrangements of boutons—axon cartridges. Scale bars for confocal images = 20 μm; scale bar for inserts = 10 μm.

Figure 4.

Details of “curvy” (A1-A3) and “straight” (B1-B2) patterns of axonal arborization. A1 and A2: Confocal reconstruction of parts of axonal arbors from 2 CACs. A3: Example of curvy pattern of arborization from NGFC. B1 and B2: Straight pattern of arborization of 2 LACs. Scale bar = 10 μm.

Figure 4.

Details of “curvy” (A1-A3) and “straight” (B1-B2) patterns of axonal arborization. A1 and A2: Confocal reconstruction of parts of axonal arbors from 2 CACs. A3: Example of curvy pattern of arborization from NGFC. B1 and B2: Straight pattern of arborization of 2 LACs. Scale bar = 10 μm.

I. Interlaminar Cells Projecting to Layer 1

The MCs

All cells recognized as MCs (n = 14) had a multipolar somata that gave rise to 3–7 primary dendrites, each of which frequently branched forming an elaborate, generally vertically oriented, dendritic tree. The majority of MCs possessed spines on their nonbeaded dendrites. MCs formed the widest axonal arbor (up to 800 μm) in layer 1 (Fig. 5, Supplementary Fig. 1) and less prominent axonal clusters around their somata. Axonal braches were coarse, bearing large numerous beads, especially in layer 1. Some of the MCs sent a few axonal collaterals down to deeper layers.

Figure 5.

Results of a post hoc analysis (Fisher's LSD tests) of quantitative morphological properties. Cell types that are not statistically different are connected with the same color lines. Blue lines connecting cell types with the smallest values, green and yellow with medium, and red ones with the largest values. (A) Soma size. (B) Number of primary dendrites. Note that NGFCs are significantly different from all other morphological types by number of primary dendrites. (C) Horizontal axonal spread. (D) Vertical axonal spread.

Figure 5.

Results of a post hoc analysis (Fisher's LSD tests) of quantitative morphological properties. Cell types that are not statistically different are connected with the same color lines. Blue lines connecting cell types with the smallest values, green and yellow with medium, and red ones with the largest values. (A) Soma size. (B) Number of primary dendrites. Note that NGFCs are significantly different from all other morphological types by number of primary dendrites. (C) Horizontal axonal spread. (D) Vertical axonal spread.

II. Interlaminar Cells Projecting to Deep Layers

The VOBCs

These neurons (n = 13) were easily distinguished by their characteristic thick and smooth descending axonal trunks, which often increased in diameter during descent and could be followed to layer 6 or even the white matter. Each cell had 1–2 (rarely 3) main axonal trunks. Remarkably, the main trunks gave off short, curving, and robustly beaded collaterals, which formed claw-like or basket-like configurations around cells of different sizes and shapes. In layers 2–3, the neurons enclosed in the “baskets” usually had small round bodies and presumably were interneurons. In the deeper layers, both pyramidal and nonpyramidal somata were surrounded by the claws (Fig. 2).

VOBCs had oval or in a few cases multipolar somata. Thick primary dendrites arose vertically from each pole of the soma and bifurcated close to their origin. Shortly after the bifurcation, each branch divided again, finally producing 2 tufts of dendrites extending from the pia to layer 4 in a narrow column.

The morphological and physiological properties of these cells have not been described before in prefrontal cortex, although they resembled previously reported interneurons from monkey striate cortex “4A-Base 3B: Variety 8 local circuit neurons” (Lund and Yoshioka 1991) and “the first subtype” of CR-containing cells (Meskenaite 1997). VOBCs probably correspond to descending basket cells described in rats (Karube et al. 2004).

The DBCs

This group of cells (n = 21) included interneurons with fine descending (and ascending in some cells) axonal collaterals. Usually these interneurons did not form prominent main trunks. Slightly beaded axon collaterals of these cells tended to travel in isolation, branching rarely and did not form claw-like configurations around neurons. However, some cells included in this group had a web of fine recurrent collaterals of variable density.

III. Intralaminar Cells

These interneurons had very heterogeneous branching within layers 2–3 and seemed to specialize in targeting different domains of other neurons. Due to their distinct specialization, terminal branches in some types of these cells had a very specific morphological appearance and higher density of boutons. In the other cells, terminal parts of axons were indistinguishable from more proximal parts.

III. 1. Intralaminar Cells with Specialized Terminal Branches

The ChCs (n = 13).

These relatively small cells (axonal width 280 ± 80 μm) were recognized based on their characteristic axonal terminals, which formed short vertical arrangements of boutons, termed axon cartridges (Fig. 3, Supplementary Fig. 1), known to target the axon initial segment of pyramidal cells (DeFelipe 1999). The axonal arbor of ChCs was formed by extensive branching at shallow angles. Somata of these cells had a multipolar or bipolar shape. Few primary dendrites arose from the cell body and rarely branched into vertically oriented thick processes, which were slightly beaded.

Local arbor pericellular basket cells (n = 14).

These tiny cells (Fig. 3, Supplementary Fig. 1) were recognized by their dense, “curvy,” complex and very compact axonal arbor on both radial (220 ± 70 μm) and tangential dimensions (180 ± 40 μm). However, as an exception, some of these cells had a few relatively long descending axonal branches with numerous boutons. The pattern of arborization resembled that of ChCs, but prominently beaded terminal portions of axons did not form vertical arrays (Lund and Lewis 1993). Instead, they often formed claw-like pericellular structures, similar to VOBCs. The cells enclosed in these structures had small round somata and presumably were nonpyramidal neurons.

Somata of local arbor pericellular basket cells (LPBCs) had round or slightly elongated shape. Dendrites were aspiny and slightly beaded, some spread farther than the axonal arbor. Similar cells were described in cat visual cortex as clutch cells (Kisvarday et al. 1985) and as small basket cells in rodent neocortex (Kawaguchi and Kubota 1996; Karube et al. 2004; Markram et al. 2004).

III. 2. Intralaminar Cells without Specialized Terminal Branches

These cells were characterized by a relatively random distribution of boutons along the axon length without a visible increase in bouton density at the terminal parts of axons. Two main patterns of axonal arborization were frequently observed (Fig. 4). The first pattern had a curvy arborization with bended axonal segments between branching points that formed an axonal mesh of variable density. The second pattern, consisted of “straight” arborization, characterized by relatively long and straight segments that bifurcated at right or oblique angles and, in general, headed away from the soma.

The NGFCs.

NGFCs (n = 19) were readily identified based on their distinctive morphological features (Kawaguchi and Kubota 1997; Tamas et al. 2003; Povysheva et al. 2007). NGFCs had a small and almost perfectly round somata that gave rise to numerous radially distributed dendrites that formed a highly symmetrical, spherical structure (Supplementary Fig. 1). The shafts of the dendrites were fine, aspiny, and slightly beaded. NGFCs exhibited curvy axonal arborization, their axons arose either from the soma or from a primary dendrite, and almost immediately started to branch forming a dense, intertwined axonal mesh. The axonal collaterals were very thin and sparsely studded with fine beads. The zone occupied by the axons had a volume several times larger than that of the dendrites, but was still mainly confined within layers 2–3.

Curved arbor cells.

This group of cells (n = 41) was very heterogeneous because they consisted of the remaining, yet unclassified, cells with curving axon trajectory (Supplementary Fig. 1). The density of arborization within this cell group varied: smaller cells usually had denser and more curving arborization, whereas cells with larger axon spread had somewhat looser and more irregular axon mesh. There seemed to be a continuum between these cell subtypes. Axons of these cells could extend into layer 1, but in contrast with MCs, did not form a significant arbor in this layer. Some cells also had a few axonal processes descending to deeper layers. The dendrites spread predominantly in the vertical dimension, but some of them traveled horizontally, thus forming a dendritic tree with no particular pattern. The dendrites usually were sparsely spiny, but in some cells were smooth. These cells resembled neurons with local beaded axons and simple beaded axons, described previously in monkey DLPFC (Lund and Lewis 1993) or nest basket cells described in rat neocortex (Wang et al. 2002).

Linear arbor cells.

These cells represented a large population in our sample (n = 59). Linear arbor cells (LACs) were recognized by their linear course of axonal branches (Supplementary Fig. 1). The primary axon gave rise to a few long and stout main lateral branches. The latter gave rise to linear collaterals at right or oblique angles. The collaterals spread across several layers, but almost never reached layer 1. This cell type included wide, medium, and local arbor morphological varieties, according to their horizontal axonal span (Lund and Lewis 1993; Krimer et al. 2005).

The somato-dendritic component was often multipolar, but could be bitufted. The bulk of dendrites ascended parallel to each other in layers 1–3, forming a narrow columnar structure. The descending dendrites crossed layers 3–5 and could spread more horizontally than vertically. Dendrites were usually slightly beaded and aspiny. These cells morphologically resembled “large basket cells” described in other cortical regions and species (Jones and Hendry 1984; Markram et al. 2004); however, in our sample, the axons of these cells did not form well-defined pericellular baskets.

Quantitative Morphological Characteristics of Interneuron Subtypes

To assess the validity of the morphological classification of interneurons, we investigated whether the morphological groups identified above differed in certain quantitative morphological parameters. We measured somal size, number of primary dendrites, and horizontal and vertical axonal spreads for 95 neurons (Figs 5 and 6). According to a one-way ANOVA and post hoc tests, all 4 parameters differed among the 8 morphological groups in a statistically significant manner (P < 0.01), supporting our classification scheme. By somal size, interneurons were divided into 3 partially overlapping subgroups: 1) small—ChCs (74 ± 8 μm2) and LPBCs (81 ± 10 μm2); 2) medium—MCs (93 ± 14 μm2), LACs (104 ± 24 μm2), NGFCs (105 ± 30 μm2), DBCs (106 ± 26 μm2), and 3) large cells—VOBCs (112 ± 21 μm2) and curved arbor cells (CACs) (117 ± 27 μm2) (Fig. 5A). The shape and orientation of somata varied greatly across different types of interneurons. In the majority of interneurons, the main axis of somata was perpendicular (radial) to the pial surface and, respectively, the ratio between radial and tangential axes of the somata (Drd/Dtng ratio) was >1 (Fig. 6A). This ratio varied from 1.05 ± 0.16 in NGFCs that have almost round somata to 1.34 ± 0.24 in VOBSs with fusiform somata.

Figure 6.

Quantitative morphological parameters of different types of interneurons. (A) The ratio between radial and tangential extends of the somata. (B) Number of primary dendrites. (C) Vertical and horizontal axonal spans of different types of interneurons (n = 85 interneurons). Circles represent mean values and bars are standard deviations. Note that LPBCs, ChCs, DBCs, and VOBCs usually do not exceed one cortical column, whereas NGFCs, MCs, CACs, and LACs innervate adjacent columns as well.

Figure 6.

Quantitative morphological parameters of different types of interneurons. (A) The ratio between radial and tangential extends of the somata. (B) Number of primary dendrites. (C) Vertical and horizontal axonal spans of different types of interneurons (n = 85 interneurons). Circles represent mean values and bars are standard deviations. Note that LPBCs, ChCs, DBCs, and VOBCs usually do not exceed one cortical column, whereas NGFCs, MCs, CACs, and LACs innervate adjacent columns as well.

The number of primary dendrites ranged from 2 to 16 and NGFCs had the largest number of primary dendrites (10.9 ± 2.8), consistent with this characteristic feature of the NGFC type (Figs 5B and 6B). Both types of vertically oriented cells had the smallest number of primary dendrites (2.8–3.8), whereas the average number of dendrites varied from 4.2 to 6.0 for all other morphological subtypes.

Monkey interneurons had different horizontal (tangential) and vertical (radial) spreads of axons (Fig. 6C). The smallest vertical axonal extent (201 ± 69 μm) was detected for LPBCs; in these tiny cells, the axon arbors were mostly restricted to the layer of origin. Vertical spread of other intralaminar interneurons (ChCs, NGFCs, CACs, and LACs) was larger and varied from 450 to 550 μm. Ascending interlaminar MCs had only slightly larger vertical span (603 ± 230 μm) than intralaminar cells, whereas vertical axonal length was 2–3 times larger in descending VOBCs (1543 ± 835 μm) and DBCs (1125 ± 181 μm). By horizontal axonal spread, interneuronal morphological types formed 2 groups. The first group consisted of cells with narrow axonal spread (190–277 μm) and included LPBCs (191 ± 56 μm), VOBCs (209 ± 67 μm), DBCs (261 ± 77 μm), and ChCs (277 ± 75 μm). Thus, these cells may be considered intracolumnar cells. The other 4 morphological types had a broader axonal spread (NGFCs [439 ± 128 μm], CACs [545 ± 257 μm], MCs [581 ± 89 μm], LACs [661 ± 340 μm]) sufficient to innervate adjacent columns.

Next, we investigated whether the combination of these 4 types of measures (soma size, number of primary dendrites, vertical and horizontal axonal spreads) would statistically distinguish interneuron morphological types. To answer this question, we applied a multivariate design of ANOVA and found that all 4 commonly used multivariate tests Wilks’ Lambda (=0.059), Pillai's trace (=1.76), Hotelling–Lawley trace (=5.42), and Roy's largest root (=3.64) indicate a significant difference (P<0.001) among morphological groups. These results strongly support the suggested morphological classification.

Molecular Markers of Different Morphological Interneuron Subtypes

CaBPs and several neuropeptides tend to be expressed in different subpopulations of interneurons (Conde et al. 1994; Cauli et al. 1997; Kawaguchi and Kondo 2002). Although molecular markers do not map perfectly to different interneuron subtypes, their expression is an important general correlate of anatomical and electrophysiological attributes (Markram et al. 2004; Zaitsev et al. 2005). Therefore, to compare our suggested morphological classification with the previous schemes based on molecular markers, we tested a large set of interneurons for expression of three CaBPs (PV, CR, and CB) and 2 neuropeptides (SST and NPY). Each interneuron was usually tested for 2 markers (Fig. 7, Supplementary Fig. 2).

Figure 7.

Molecular characterization of monkey DLPFC interneurons. Double and triple immunostaining for Biocytin (blue) and CaBPs and neuropeptides (red and green alternatively) in VOBC, MC, LAC, and NGFC. Scale bars = 20 μm. Utilization of confocal microscopy with high-resolution objective (×40/1.3 N.A.) precluded bleaching of labeled cell from other depth of fields.

Figure 7.

Molecular characterization of monkey DLPFC interneurons. Double and triple immunostaining for Biocytin (blue) and CaBPs and neuropeptides (red and green alternatively) in VOBC, MC, LAC, and NGFC. Scale bars = 20 μm. Utilization of confocal microscopy with high-resolution objective (×40/1.3 N.A.) precluded bleaching of labeled cell from other depth of fields.

Our results established a correlation between the morphological types and the biochemical markers (Table 2), which is consistent with previous reports (Conde et al. 1994; Cauli et al. 1997; Kawaguchi and Kubota 1997). It is worth mentioning that LACs and CACs were perfectly segregated by this criterion: more than half of tested LACs expressed PV, which is considered to be a marker for soma-targeting fast-spiking (FS) basket cells (Kawaguchi and Kondo 2002; Freund and Katona 2007), whereas none of the CACs expressed PV. Instead, CACs contained CB, with the exception of a small population that expressed either CR or SST. Both varieties of vertically oriented cells usually expressed CR. LPBCs contained either CR or CB.

Table 2

Expression of CaBPs and neuropeptides in different types of interneurons

Cell type PV CR CB SS NPY 
MCs 0/1 n/a 3/4 4/5 n/a 
ChCs 3/4 0/3 n/a n/a n/a 
LPBCs 0/2 2/3 1/6 0/4 n/a 
NGFCs 0/3 0/2 8/11 0/4 4/10 
VOBCs 0/6 5/8 0/2 0/1 n/a 
DBCs 0/12 10/12 1/6 0/5 n/a 
LACs 14/24 0/9 1/8 0/8 0/2 
CACs 0/13 1/15 4/12 1/13 0/2 
Cell type PV CR CB SS NPY 
MCs 0/1 n/a 3/4 4/5 n/a 
ChCs 3/4 0/3 n/a n/a n/a 
LPBCs 0/2 2/3 1/6 0/4 n/a 
NGFCs 0/3 0/2 8/11 0/4 4/10 
VOBCs 0/6 5/8 0/2 0/1 n/a 
DBCs 0/12 10/12 1/6 0/5 n/a 
LACs 14/24 0/9 1/8 0/8 0/2 
CACs 0/13 1/15 4/12 1/13 0/2 

Note: n/a, not applicable.

As in our previous study (Zaitsev et al. 2005), we did not observe more than one CaBPs in any of the tested interneurons. However, colocalization of a CB with a neuropeptide was frequently detected in MCs (CB with SST) and in NGFCs (CB with NPY).

Electrophysiological Diversity of Monkey DLPFC Interneurons

Our previous study was primarily based on electrophysiological membrane properties of interneurons and has demonstrated that they are not a continuum, but consist of 3 unique physiological clusters (class) of neurons (Krimer et al. 2005). However, when morphology was included in the classification, the 3 basic electrophysiological clusters each included several morphological cell types. Therefore, the physiological types of interneurons obtained via cluster analysis combined functionally heterogeneous populations of interneurons. Here, we used morphological criteria as a starting point for the classification of interneurons. Eight morphological types of layer 2–3 interneurons have been identified, including 4 types that were not previously identified and thus had not been physiologically characterized. We believe that this approach allowed us to distinguish functionally more uniform groups of interneurons in the circuitry. An important question was if these different morphological cell types were distinct in their electrophysiological properties or not. To address this issue in detail, we measured a total of 21 different electrophysiological parameters from each neuron, including subthreshold and suprathreshold voltage responses to current injection (see Materials and Methods). All the cells included in the analysis had stable resting potentials that were more negative than −60 mV and overshooting APs.

We found that all parameters except for RMP and DAP differed between groups in a statistically significant manner, according to a one-way ANOVA (Table 3), indicating that each morphological type had a specific combination of intrinsic electrical properties. Thus, we performed additional analyses in which the RMP and DAP were excluded. To determine how the electrophysiological properties differed between these eight morphological types, we used a post hoc Fisher’s LSD test (Table 4). This paired comparison analysis revealed that there were no morphological groups with an absolutely similar set of intrinsic properties, although the number of significantly different parameters between different groups varied. Some morphological varieties displayed many similarities in electrophysiological properties. For instance, CACs and MCs differed only in one out of 19 parameters, whereas CACs and LACs differed in 18 parameters. Below we briefly describe the most different electrophysiological properties between the morphological groups.

Table 3

Electrophysiological properties of different morphological types of monkey interneurons

 LAC, n = 59 ChC, n = 13 CAC, n = 41 DBC, n = 21 MC, n = 14 VOBC, n = 13 LPBC, n = 14 NGFC, n = 19 
RMP, mV −70 ± 7 −65 ± 8 −67 ± 9 −70 ± 7 −68 ± 9 −71 ± 9 −68 ± 8 −69 ± 7 
Rin, MΩ 215 ± 91 330 ± 144 437 ± 178 665 ± 430 430 ± 207 483 ± 189 458 ± 220 363 ± 150 
τ, ms 10.9 ± 3.9 10.1 ± 2.3 18.4 ± 5.3 16.5 ± 6.8 15.8 ± 4.9 15.7 ± 5.1 15 ± 4.5 12.5 ± 4.4 
Rb, pA 59 ± 34 40 ± 27 23 ± 15 23 ± 19 24 ± 9 25 ± 18 18 ± 13 36 ± 30 
Sag 0.3 ± 0.5 0.7 ± 0.6 0.6 ± 0.7 0.8 ± 0.7 1 ± 0.7 0.8 ± 0.7 1.1 ± 0.9 0.1 ± 0.4 
Hump 0.6 ± 0.6 0.4 ± 0.8 0.7 ± 0.7 0.9 ± 0.8 1.1 ± 0.9 0.9 ± 0.9 1.2 ± 0.7 0.2 ± 0.4 
RD 0.2 ± 0.5 0.5 ± 0.5 0.7 ± 0.8 0.8 ± 0.7 1 ± 0.7 0.5 ± 0.7 1.5 ± 0.8 0 ± 0 
Sum 1.1 ± 1.1 1.5 ± 1.5 2 ± 2 2.4 ± 1.8 3.1 ± 1.7 2.3 ± 1.8 3.8 ± 1.9 0.3 ± 0.5 
DAP, ms 118 ± 114 145 ± 174 100 ± 98 68 ± 77 53 ± 27 77 ± 57 64 ± 68 69 ± 39 
APT, mV −39 ± 5 −43 ± 5 −43 ± 5 −46 ± 4 −44 ± 6 −46 ± 5 −47 ± 3 −40 ± 5 
APA, mV 52 ± 12 52 ± 12 63 ± 13 59 ± 12 58 ± 11 68 ± 17 66 ± 9 55 ± 10 
APD, ms 0.38 ± 0.11 0.32 ± 0.06 0.68 ± 0.12 0.74 ± 0.19 0.68 ± 0.17 0.53 ± 0.12 0.52 ± 0.09 0.62 ± 0.13 
AHPAF, mV 21.8 ± 4.4 17.9 ± 4.2 14.4 ± 6.3 13.1 ± 5.7 14 ± 4.4 18.1 ± 3.4 13.7 ± 5 12 ± 3.6 
AHPAM, mV 1.3 ± 1.6 1.1 ± 1.8 2.6 ± 2.8 2.2 ± 1.7 1.9 ± 1.4 1.5 ± 1.5 3.1 ± 2.9 7.3 ± 3.7 
tAHP, ms 2.9 ± 2.8 2.9 ± 3.5 7.8 ± 8.4 6 ± 4.4 9.8 ± 9.7 4.7 ± 5 6.8 ± 5.9 13.5 ± 4.7 
Fr, Hz 52 ± 18 86 ± 29 30 ± 13 43 ± 13 34 ± 11 44 ± 11 39 ± 14 22 ± 6 
kISIs 4 ± 2.2 4.6 ± 2.1 8.3 ± 6.2 13.8 ± 21 6 ± 2 7.1 ± 3.7 6.7 ± 2.3 5.3 ± 2 
FrAI, % 1.9 ± 9.9 5.6 ± 14 11.5 ± 12.8 7.2 ± 11.3 4.8 ± 21.2 −0.9 ± 24.1 −3.5 ± 18.2 28 ± 11.9 
FrAL1, % 17 ± 20 11 ± 20 34 ± 19 25 ± 22 36 ± 19 18 ± 28 25 ± 19 48 ± 14 
FrAL2, % 16 ± 15 5.7 ± 15 25 ± 24 21 ± 20 35 ± 18 19 ± 22 26 ± 20 32 ± 17 
ΔAPA, % −0.5 ± 3.8 −2.5 ± 6.2 −7. 3 ± 8.8 −9.8 ± 5.1 −11.4 ± 8.9 −9.1 ± 5.5 −11 ± 7.4 1.2 ± 2.4 
ΔAHPA, % 3.9 ± 13.2 5.6 ± 22.9 30.4 ± 42.5 27.6 ± 27.1 21.1 ± 40.3 8.7 ± 21.9 14.3 ± 22.2 62.9 ± 41.9 
 LAC, n = 59 ChC, n = 13 CAC, n = 41 DBC, n = 21 MC, n = 14 VOBC, n = 13 LPBC, n = 14 NGFC, n = 19 
RMP, mV −70 ± 7 −65 ± 8 −67 ± 9 −70 ± 7 −68 ± 9 −71 ± 9 −68 ± 8 −69 ± 7 
Rin, MΩ 215 ± 91 330 ± 144 437 ± 178 665 ± 430 430 ± 207 483 ± 189 458 ± 220 363 ± 150 
τ, ms 10.9 ± 3.9 10.1 ± 2.3 18.4 ± 5.3 16.5 ± 6.8 15.8 ± 4.9 15.7 ± 5.1 15 ± 4.5 12.5 ± 4.4 
Rb, pA 59 ± 34 40 ± 27 23 ± 15 23 ± 19 24 ± 9 25 ± 18 18 ± 13 36 ± 30 
Sag 0.3 ± 0.5 0.7 ± 0.6 0.6 ± 0.7 0.8 ± 0.7 1 ± 0.7 0.8 ± 0.7 1.1 ± 0.9 0.1 ± 0.4 
Hump 0.6 ± 0.6 0.4 ± 0.8 0.7 ± 0.7 0.9 ± 0.8 1.1 ± 0.9 0.9 ± 0.9 1.2 ± 0.7 0.2 ± 0.4 
RD 0.2 ± 0.5 0.5 ± 0.5 0.7 ± 0.8 0.8 ± 0.7 1 ± 0.7 0.5 ± 0.7 1.5 ± 0.8 0 ± 0 
Sum 1.1 ± 1.1 1.5 ± 1.5 2 ± 2 2.4 ± 1.8 3.1 ± 1.7 2.3 ± 1.8 3.8 ± 1.9 0.3 ± 0.5 
DAP, ms 118 ± 114 145 ± 174 100 ± 98 68 ± 77 53 ± 27 77 ± 57 64 ± 68 69 ± 39 
APT, mV −39 ± 5 −43 ± 5 −43 ± 5 −46 ± 4 −44 ± 6 −46 ± 5 −47 ± 3 −40 ± 5 
APA, mV 52 ± 12 52 ± 12 63 ± 13 59 ± 12 58 ± 11 68 ± 17 66 ± 9 55 ± 10 
APD, ms 0.38 ± 0.11 0.32 ± 0.06 0.68 ± 0.12 0.74 ± 0.19 0.68 ± 0.17 0.53 ± 0.12 0.52 ± 0.09 0.62 ± 0.13 
AHPAF, mV 21.8 ± 4.4 17.9 ± 4.2 14.4 ± 6.3 13.1 ± 5.7 14 ± 4.4 18.1 ± 3.4 13.7 ± 5 12 ± 3.6 
AHPAM, mV 1.3 ± 1.6 1.1 ± 1.8 2.6 ± 2.8 2.2 ± 1.7 1.9 ± 1.4 1.5 ± 1.5 3.1 ± 2.9 7.3 ± 3.7 
tAHP, ms 2.9 ± 2.8 2.9 ± 3.5 7.8 ± 8.4 6 ± 4.4 9.8 ± 9.7 4.7 ± 5 6.8 ± 5.9 13.5 ± 4.7 
Fr, Hz 52 ± 18 86 ± 29 30 ± 13 43 ± 13 34 ± 11 44 ± 11 39 ± 14 22 ± 6 
kISIs 4 ± 2.2 4.6 ± 2.1 8.3 ± 6.2 13.8 ± 21 6 ± 2 7.1 ± 3.7 6.7 ± 2.3 5.3 ± 2 
FrAI, % 1.9 ± 9.9 5.6 ± 14 11.5 ± 12.8 7.2 ± 11.3 4.8 ± 21.2 −0.9 ± 24.1 −3.5 ± 18.2 28 ± 11.9 
FrAL1, % 17 ± 20 11 ± 20 34 ± 19 25 ± 22 36 ± 19 18 ± 28 25 ± 19 48 ± 14 
FrAL2, % 16 ± 15 5.7 ± 15 25 ± 24 21 ± 20 35 ± 18 19 ± 22 26 ± 20 32 ± 17 
ΔAPA, % −0.5 ± 3.8 −2.5 ± 6.2 −7. 3 ± 8.8 −9.8 ± 5.1 −11.4 ± 8.9 −9.1 ± 5.5 −11 ± 7.4 1.2 ± 2.4 
ΔAHPA, % 3.9 ± 13.2 5.6 ± 22.9 30.4 ± 42.5 27.6 ± 27.1 21.1 ± 40.3 8.7 ± 21.9 14.3 ± 22.2 62.9 ± 41.9 
Table 4

Results of Fisher’s LSD test

 LAC ChC CAC MC DBC VOBC LPBC NGFC 
LAC         
ChC        
CAC 18 11       
MC 14      
DBC 15     
VOBC 10    
LPBC 13 11   
NGFC 11 10 13 17 14 14  
 LAC ChC CAC MC DBC VOBC LPBC NGFC 
LAC         
ChC        
CAC 18 11       
MC 14      
DBC 15     
VOBC 10    
LPBC 13 11   
NGFC 11 10 13 17 14 14  

Note: Number of physiological parameters (out of 19), which were statistically different between morphological groups.

Subthreshold Membrane Properties of Different Morphological Types of Interneurons

We found several significant differences in the responses of cells of different morphological types to hyperpolarizing and subthreshold depolarizing current steps. For example, Fisher’s LSD test revealed that according to Rin interneuron types could be subdivided into 3 different groups: cells with low Rin (LACs and ChCs), with high Rin (DBCs), and with intermediate Rin (the remaining 5 morphological types). Shorter τ were found in LACs, ChCs, and NGFCs, than in other morphological types (Table 3).

Some interneurons exhibited a time-dependent rectification in response to hyperpolarizing current steps (Fig. 8). This rectification was characterized by a delayed depolarizing drift or “sag” that shifted the membrane potential toward the RMP and could be observed when the cells were hyperpolarized more than 10–20 mV relative to RMP. Such sag most likely was produced by the Ih current (Pape 1996; Robinson 2003). In the majority of interneurons that exhibited profound depolarizing sag, subsequent return to RMP caused a transient RD, which sometimes triggered a few spikes. Although the ionic mechanisms of such RD usually are associated with a low-threshold calcium current, Ih current can contribute to the RD as well (Pape 1996; Aizenman and Linden 1999). In agreement with this, we found a significant correlation (0.76) between sag and RD amplitudes.

Figure 8.

Representative examples of subthreshold responses in monkey interneurons. (A) Example of ramp depolarization observed in ChC. (B) Subthreshold responses of NGFC. (C and D) Subthreshold responses in CAC (C) demonstrated smaller amplitude of hump (arrow), sag (arrowhead), and RD (asterisk) than in LPBC (D). APs were truncated at (A and D). (E) Bar graph demonstrating averaged components of Sum values for different morphological types of interneurons; ANOVA revealed significant differences between morphological types by all 4 parameters (P < 0.01).

Figure 8.

Representative examples of subthreshold responses in monkey interneurons. (A) Example of ramp depolarization observed in ChC. (B) Subthreshold responses of NGFC. (C and D) Subthreshold responses in CAC (C) demonstrated smaller amplitude of hump (arrow), sag (arrowhead), and RD (asterisk) than in LPBC (D). APs were truncated at (A and D). (E) Bar graph demonstrating averaged components of Sum values for different morphological types of interneurons; ANOVA revealed significant differences between morphological types by all 4 parameters (P < 0.01).

In response to subthreshold depolarizing current steps, a few interneurons (n = 8), which were morphologically defined as ChCs (n = 4), CACs (n = 2), MC (n = 1), and DBC (n = 1), generated a depolarizing ramp (Fig. 8A). However, most ChCs (9 out of 13) did not exhibit any ramp depolarization. More often, monkey interneurons showed hump-like upward voltage deflection at the beginning of the responses to the current injections (Fig. 8). The magnitude of the hump was also correlated with the amplitude of sag (0.51) and of the RD (0.53). Because of these correlations, additional comparisons of different morphological cell types included a compound parameter that we termed “Sum,” which represented the sum of Sag, RD, and Hump amplitudes.

Responses to subthreshold depolarizing and hyperpolarizing current steps appeared to be cell-type specific. For example, majority of NGFCs (Fig. 8B) did not exhibit any or only a little sag or a hump; NGFCs did not exhibit RD as well. In contrast, practically all MCs, LPBCs, DBCs, VOBCs, and many of CACs exhibited more or less significant sag, RD, and hump. Moreover, in some of the MCs and LPBCs, rebound spikes were observed. LACs typically did not exhibit profound time-dependent rectification or RD, whereas they might display hump-like depolarization at the current level close to Rb.

Single Spike Properties

Among the parameters describing AP properties, spike duration was the most cell-type specific. Three different groups of cells were distinguished by this parameter. The shortest spike duration was observed in ChCs (0.32 ± 0.06 ms) and LACs (0.38 ± 0.11 ms); these values were typical for monkey FS cells (Krimer et al. 2005). Relatively brief spike duration was also registered for VOBCs (0.53 ± 0.12 ms) and LPBCs (0.52 ± 0.09 ms). The duration of spikes in the 4 other morphological types was longer; the average values varied from 0.62 to 0.74 ms.

Spike duration depends on many factors, among which is expression of different types of potassium channels. For example, narrow spike is associated with expression of Kv3-family channels (Rudy et al. 1999; Lau et al. 2000). Expression of different channels with distinct properties would also affect the shape of AP. Fast kinetics of AP precludes direct comparison and the analysis of AP shapes; however, there is a simple and informative way of examining the properties of APs by using phase plots (for details, see Materials and Methods). The phase plot for a membrane AP gives a direct read out of net ionic current as a function of voltage during the various phase of the AP (Bean 2007). In phase plots, AP is represented by a loop (Fig. 9). We observed 3 main types of the loop shapes, which were strongly associated with spike duration.

Figure 9.

Spike duration and its shape were different in distinct morphological types of interneurons. (A) Illustration of APs and their derivations from LAC and CAC. V is membrane voltage, and dV/dt is the time derivative of membrane voltage. Note the difference in spike duration between cells; faster and deeper repolarization rate in the narrow spike of LAC than in CAC. (B) On phase plots, APs were represented by loops, which had different shapes. In cells with the briefest spikes (LAC and ChC) they had a shape of rocking chair. Shape of the spikes of VOBCs and LPBCs in these plots resembled in appearance a snail. The third type of loops resembled in appearance a vertically oriented egg and was typical for interneurons with longer spike duration (CACs, MCs, DBCs, and NGFCs).

Figure 9.

Spike duration and its shape were different in distinct morphological types of interneurons. (A) Illustration of APs and their derivations from LAC and CAC. V is membrane voltage, and dV/dt is the time derivative of membrane voltage. Note the difference in spike duration between cells; faster and deeper repolarization rate in the narrow spike of LAC than in CAC. (B) On phase plots, APs were represented by loops, which had different shapes. In cells with the briefest spikes (LAC and ChC) they had a shape of rocking chair. Shape of the spikes of VOBCs and LPBCs in these plots resembled in appearance a snail. The third type of loops resembled in appearance a vertically oriented egg and was typical for interneurons with longer spike duration (CACs, MCs, DBCs, and NGFCs).

Spikes of LACs and ChC in a phase plot resembled a rocking chair. AP velocity (dV/dt) continuously increased from the threshold until membrane potential reached positive values. Then, near the peak of AP, velocity started to rapidly drop. This switch might indicate almost-synchronous activation of potassium channels at a membrane potential close to 0 mV; fast activation at voltages positive to −10 mV is typical only for Kv3 channels (Coetzee et al. 1999). During the repolarization phase, the velocity reached the most negative values at −10 to −20 mV and then steadily approached a value of 0 Vs−1. Voltage shift between the starting point and ending points of the loop indicates fast component of AHP, which was the largest in these cells.

Shape of the spikes of VOBCs and LPBCs in these plots resembled in appearance a snail. The depolarizing part of the loops was similar to the previous type, and velocity also reached maximum values at a positive membrane potential as before. However, after the turning point velocity decreased less rapidly than in LACs/ChCs. Change of velocity during repolarizing phase was also quite different from previous type. The velocity stayed relatively stable over a large range of membrane voltage, and thus, the outward current was relatively constant during repolarizing phase. The latter might indicate an involvement of several types of potassium channels with different kinetics.

The third type of loops resembled in appearance a vertically oriented egg and was typical for interneurons with longer spike duration (CACs, MCs, DBCs, and NGFCs). This was the most symmetrical loop because the maximum of net inward and outward current was reached approximately at half-height of AP, that is at negative membrane potential, and the voltage shift between starting and ending points of the loop was relatively small. Therefore, the repolarization was achieved via potassium channels that were activated at membrane potentials more negative than required for activation of Kv3 channels.

Spikes of LACs and ChCs were followed by large monophasic AHPs of the shortest latency (about 3 ms) (Fig. 10). These properties of AHP are typical for FS cells (Kawaguchi 1993) and may indicate a contribution of Kv3 channels, which are very fast deactivating channels (Rudy et al. 1999). VOBCs had AHP with a shape similar to LACs and ChCs, but with longer latency of the former. LPBCs, CACs, MCs, and DBCs exhibited two-component AHP, which indicates involvement of several types of potassium channels with different kinetics. Total amplitude of AHP in these cells was about 15–17 mV. Amplitudes of an early, fast- and of delayed, and medium-duration components of AHP had a relative ration of ∼6:1. Finally, distinct AHP pattern was observed in NGFCs. In these cells AHP also could be divided into 2 components, but the medium component had almost the same amplitude as the fast component and represented about 40% of the total AHP amplitude. This was in striking contrast with the other morphological types of interneurons. AHP in NGFCs also had the longest latency (13.5 ± 4.7 ms).

Figure 10.

AHP had different shape and latency in distinct morphological types of interneurons. (A) Representative examples of AHP shapes from different types of interneurons (see explanation in the text). (B) Bar graph representing the averaged amplitude of fast and medium components of AHP in different types of interneurons. Note that impact of medium component to the total AHPA was relatively small in LACs, ChCs, and VOBCs (5–8%), more profound in CACs, MCs, DBCs, and LPBCs (13–18%), and very large in NGFCs (40%). (C) Bar graph demonstrating AHP latency; cells with larger amplitude of medium component AHP exhibited longer AHP. Latency of AHP in NGF cells was almost 5 times longer than in LACs or ChC.

Figure 10.

AHP had different shape and latency in distinct morphological types of interneurons. (A) Representative examples of AHP shapes from different types of interneurons (see explanation in the text). (B) Bar graph representing the averaged amplitude of fast and medium components of AHP in different types of interneurons. Note that impact of medium component to the total AHPA was relatively small in LACs, ChCs, and VOBCs (5–8%), more profound in CACs, MCs, DBCs, and LPBCs (13–18%), and very large in NGFCs (40%). (C) Bar graph demonstrating AHP latency; cells with larger amplitude of medium component AHP exhibited longer AHP. Latency of AHP in NGF cells was almost 5 times longer than in LACs or ChC.

Firing Pattern Properties

To characterize the structure of firing pattern, we analysed the percentages of initial and late spike frequency adaptation. According to these parameters, we could distinguish 3 main types of temporal structure of firing (Fig. 11). Cells with nonadapting firing pattern exhibited very little, if any, adaptation (<25%). This pattern was typical for LACs and ChCs.

Figure 11.

Interneurons of different morphological types exhibited specific firing patterns. Firing patterns from ChC, VOBC, CAC, and NGFC represented the firing patterns that were typically observed in monkey interneurons. (A) Overlapped first (gray) and last (black) APs from the train. Note that amplitude of spikes and their duration did not change in ChC and NGFC during train, whereas in VOBC and CAC they changed. (B) Sweeps with the responses to the first suprathreshold current step and to the 2× threshold current step. Compare the shape of AHP after first spike and after other spikes in the train in different morphological types; first AHP was enlarged in VOBC, whereas in NGFC it was reduced. (C) Plots of instantaneous frequency against ISI number, calculated from subsequent sweeps with the responses to increasing suprathreshold current steps (step = 10 pA). Note the absence of significant spike frequency adaptation on ChC at any stimulation current steps, short initial facilitation and then moderate adaptation in VOBC, and different level of initial adaptation in CAC and NGFC.

Figure 11.

Interneurons of different morphological types exhibited specific firing patterns. Firing patterns from ChC, VOBC, CAC, and NGFC represented the firing patterns that were typically observed in monkey interneurons. (A) Overlapped first (gray) and last (black) APs from the train. Note that amplitude of spikes and their duration did not change in ChC and NGFC during train, whereas in VOBC and CAC they changed. (B) Sweeps with the responses to the first suprathreshold current step and to the 2× threshold current step. Compare the shape of AHP after first spike and after other spikes in the train in different morphological types; first AHP was enlarged in VOBC, whereas in NGFC it was reduced. (C) Plots of instantaneous frequency against ISI number, calculated from subsequent sweeps with the responses to increasing suprathreshold current steps (step = 10 pA). Note the absence of significant spike frequency adaptation on ChC at any stimulation current steps, short initial facilitation and then moderate adaptation in VOBC, and different level of initial adaptation in CAC and NGFC.

The firing frequency in CACs, MCs, LPBCs, VOBCs, and DBCs decreased gradually along the trains. The level of spike frequency adaptation did not change significantly within the range of stimulation currents used. This pattern was consistent with that of previously described as classic accommodating (Markram et al. 2004)/adapting (Gonzalez-Burgos et al. 2004) or regular spiking nonpyramidal cells (Kawaguchi and Kubota 1997). In some LPBCs and VOBCs, we observed a variation of this pattern (Krimer et al. 2005). In these cells, the first spike in the train had an enlarged AHP and the first ISI was much longer than the second one. The latter was followed by a typical adapting pattern.

Finally, NGFCs exhibited a unique firing pattern, with a temporal structure strongly dependent on stimulation current intensity. With near threshold depolarizing current steps, NGFCs displayed nonadapting properties of firing. However, an increase in stimulation current intensity strongly enhanced spike frequency adaptation. The changes in the level of adaptation were accompanied by a significant reduction in amplitude of the medium component AHP after the first spikes in trains (Fig. 12). In contrast to gradual firing adaptation in the previous group, in NGFCs spike frequency adapted extremely fast (within the first 3–5 spikes) and after that remained stable until the end of the current pulse. Thus, monkey NGFCs resemble the initial-adapting (i-Ad) cells described in mouse neocortex (Miyoshi et al. 2007).

Figure 12.

Shape of AHP of first spike in the trains was dependent on stimulation current in NGFC. Medium component of AHP was significantly reduced in amplitude with increase of depolarizing current. Note the appearance of an additional depolarizing component (arrow) with the increasing stimulation currents that leaded to significant reduction of total AHPA and to the reduction of the first ISI. APs are truncated. Light gray line shows the level of AHP, gray line the fast component of AHP, and dark gray line the most negative membrane potential after first spike.

Figure 12.

Shape of AHP of first spike in the trains was dependent on stimulation current in NGFC. Medium component of AHP was significantly reduced in amplitude with increase of depolarizing current. Note the appearance of an additional depolarizing component (arrow) with the increasing stimulation currents that leaded to significant reduction of total AHPA and to the reduction of the first ISI. APs are truncated. Light gray line shows the level of AHP, gray line the fast component of AHP, and dark gray line the most negative membrane potential after first spike.

Steady-state frequency of firing was measured at the end of the train, when adaptation was usually completed, and appeared to be cell-type specific (Fig. 11C). ChCs and LACs displayed the highest steady-state frequency (86 ± 29 and 52 ± 18 Hz, respectively), whereas NGFCs displayed the lowest frequency (22 ± 6 Hz). Other cell types demonstrated intermediate firing frequency in a range of 30–44 Hz.

To mathematically describe the irregularity of firing, we calculated coefficient of variance of ISIs (kISIs) during the steady-state phase. In majority of cells (166 out of 189) kISIs was less than 10%, indicating a high level of regularity in the firing pattern. Only 4 cells in our sample had kISIs >20%, which would classify them as irregular spiking cells. Morphologically, these cells included 2 DBCs, 1 MC, and 1 CAC. LACs and ChCs exhibited the lowest kISIs (4.0±2.2 and 4.6±2.1%, respectively), and the VO cells had the largest firing variance (13.8 ± 20.5%). This is in agreement with data from rat neocortex (Cauli et al. 1997).

Not only the AP frequency changed during trains but also the shape of the spikes also underwent substantial changes. Therefore, we measured the percent of change in APA and AHPA from the first to the last spike in the train (for details, see Materials and Methods). We did not detect any changes in APA for LACs, ChCs, and NGFCs, whereas in the other morphological groups decline of amplitude was about 6–12%. Total AHPA almost did not change in ChC or LACs, whereas it increased on 10–30% in adapting cells and even more (65%) in intial-adapting NGFCs.

To detect if other properties of spike shape were changing during the sweeps, we plotted the spike trains on the phase diagrams (Fig. 13). AP trains in ChC and LACs formed a set of identical superimposed loops, and the first AP in the train was indistinguishable from the rest. This might indicate that voltage gated ionic channels, involved in generation of APs, were completely recovering from spike to spike and that no additional types of channels were added to spike generation during the trains.

Figure 13.

Spike shape was preserved during trains in LAC and ChC, whereas in other morphological types it was changed. (A) Illustration of APs trains and their derivations from ChC and DBC. Note the difference in spike shapes of DBC during train was more obvious at the voltage derivation graph. V is membrane voltage, and dV/dt is the time derivative of membrane voltage. (B) Phase diagram of subsequent APs in the trains. First AP is shown in red. Note that shape of the first spike was undistinguishable from subsequent spikes in the train in ChC and LAC, slightly shifted to the left in NGFC. In other cell types, subsequent spikes had smaller area.

Figure 13.

Spike shape was preserved during trains in LAC and ChC, whereas in other morphological types it was changed. (A) Illustration of APs trains and their derivations from ChC and DBC. Note the difference in spike shapes of DBC during train was more obvious at the voltage derivation graph. V is membrane voltage, and dV/dt is the time derivative of membrane voltage. (B) Phase diagram of subsequent APs in the trains. First AP is shown in red. Note that shape of the first spike was undistinguishable from subsequent spikes in the train in ChC and LAC, slightly shifted to the left in NGFC. In other cell types, subsequent spikes had smaller area.

APs within the trains from NGFCs were also identically shaped, but in contrast to LACs and ChCs, the first spike was always slightly shifted to the left on the phase diagram. In the 5 other cell types, we usually observed a trend for a decrease in the loops’ area during trains, which translated to decrease in maximal amplitude of the underlying inward and outward currents. Such decrease leads to an increase in APD during trains (Fig. 13), which may significantly affect properties of synaptic transmission during trains of APs (Geiger and Jonas 2000).

The Morphological Types of Interneurons in the Monkey DLPFC Are Assembled into 3 Groups That Share Physiologically Relevant Electrical Properties

We identified 8 morphological types of monkey DLPFC interneurons. However, whereas electrophysiological differences were robust between some morphological types, the similarities between other morphological types in intrinsic membrane properties outweighed the differences. Electrophysiological diversity of interneurons results from the combined activity of different ion channels and from the morphology of the neurons (Mainen and Sejnowski 1996; Markram et al. 2004). Although many types of ion channels underlie electrophysiological properties of interneurons, recent studies indicate that the channels are frequently coexpressed together in only a few specific combinations (Toledo-Rodriguez et al. 2004, 2005). Thus, these specific combinations of channels can give rise to a finite number of distinct electrical classes of interneurons with particular combinations of membrane properties.

Statistically, specific combinations of variables can be revealed using factor analysis. For this purpose, we performed principal factor analysis of the electrophysiological data (see for details, Materials and Methods). A total of 162 interneurons were included in this analysis. In factor model, we retained 3 factors with eigenvalues greater than 1. These 3 factors accounted for 45% of the total variance. Interpretation of the factors was done according to their factor loadings (Table 5).

Table 5

Factor loading coefficients after varimax rotation for each of the 3 factors

Physiological parameters Factor 1 Factor 2 Factor 3 
Fr −0.62   
Rin 0.46 −0.45  
APD 0.58   
τ 0.63 −0.40  
AHPAM 0.63   
tAHP 0.65   
Sum  −0.52  
APA  −0.52  
Rb  0.55  
APT  0.65  
ΔAPA  0.65  
AHPAF −0.37  −0.57 
FrAL   0.64 
FrAI   0.69 
ΔAHPA   0.73 
kISIs    
Physiological parameters Factor 1 Factor 2 Factor 3 
Fr −0.62   
Rin 0.46 −0.45  
APD 0.58   
τ 0.63 −0.40  
AHPAM 0.63   
tAHP 0.65   
Sum  −0.52  
APA  −0.52  
Rb  0.55  
APT  0.65  
ΔAPA  0.65  
AHPAF −0.37  −0.57 
FrAL   0.64 
FrAI   0.69 
ΔAHPA   0.73 
kISIs    

The first factor was marked by high loadings on the parameters that associated with firing frequency. Cells with negative scores for this factor exhibited high frequency firing pattern with narrow spikes, followed by a deep monophasic AHP. Compared with other cells, these neurons had smaller Rin and time constant. In contrast, cells with a positive score for the first factor exhibited low-frequency firing and broader spikes followed by prolonged AHP with profound medium component. The second factor was marked by parameters describing mostly cells’ subthreshold properties. Interneurons with positive scores for this factor were more excitable, as they demonstrated low level of firing threshold and Rb, high Rin, profound sag, RD, and hump. These properties also correlated with larger AP amplitude, which had a tendency to decrease during the trains. The third factor was marked by parameters describing firing frequency adaptation. Cells with positive scores exhibited strong firing frequency adaptation and small amplitude of fast component AHP, whereas their total amplitude of AHP significantly increased during the trains.

Thus, 3 associations of electrophysiological parameters in monkey interneurons were discovered, and they were represented by 3 independent factors, which we named based on their key physiological properties as: factor of firing frequency, factor of interneuron excitability, and factor of firing frequency adaptation. According to a one-way ANOVA, all 3 factors were significantly different between distinct morphological types (F7,154 were 15.2, 34.0, 8.0 for firing frequency, interneuron excitability, and firing frequency adaptation, respectively, and for each of them P < 0.01). However, a post hoc Fisher’s LSD test revealed that some morphological types could not be distinguished by their factor scores (P < 0.01). We combined such morphological types together and found 3 clusters of morphological types, which shared the most similar combinations of physiological properties. These associations of morphological types can be easily observed in the 3-dimensional plot (Fig. 14). Our interpretation of these results is that monkey interneurons form 3 different physiological classes.

Figure 14.

The graph of the averaged factor scores of different morphological types. The first factor described properties, associated with frequency of firing, the second cell excitability, and the third adaptation. Morphological types formed 3 clusters, which shared the most similar combinations of physiological properties or formed 3 different physiological classes. First class consisted of ChCs and LACs, and according to their factor scores (high-frequency nonadapting firing pattern, low level of excitability), they could be recognized as FS cells. Second physiological class consisted of only NGFCs, which were i-Ad cells with high threshold of firing and the lowest firing frequency. Third class of c-Ad cells was formed by CACs, MCs, DBCs, VOBCs, and LPBCs, which all had positive scores for factor of excitability, distinguished them from all other morphological types. Cells of that class exhibited the moderate level of firing frequency and its adaptation.

Figure 14.

The graph of the averaged factor scores of different morphological types. The first factor described properties, associated with frequency of firing, the second cell excitability, and the third adaptation. Morphological types formed 3 clusters, which shared the most similar combinations of physiological properties or formed 3 different physiological classes. First class consisted of ChCs and LACs, and according to their factor scores (high-frequency nonadapting firing pattern, low level of excitability), they could be recognized as FS cells. Second physiological class consisted of only NGFCs, which were i-Ad cells with high threshold of firing and the lowest firing frequency. Third class of c-Ad cells was formed by CACs, MCs, DBCs, VOBCs, and LPBCs, which all had positive scores for factor of excitability, distinguished them from all other morphological types. Cells of that class exhibited the moderate level of firing frequency and its adaptation.

The first class consisted of ChCs and LACs and, according to their factor scores (high frequency non-adapting firing pattern, low level of excitability), these cells can be recognized as nonadapting FS cells, described in different species (Foehring et al. 1991; Kawaguchi and Kubota 1997; Gonzalez-Burgos et al. 2005).

The second physiological class consists of only NGFCs, which are i-Ad cells with a high-threshold of firing and the lowest firing frequency. Although NGFCs in rodent neocortex also constitute a separate electrophysiological class (Kawaguchi and Kubota 1997), their membrane properties are quite different from those in monkey, and the commonly used term “late-spiking interneurons” does not fit the physiological properties of monkey NGFCs (Povysheva et al. 2007).

The third physiological class was formed by CACs, MCs, DBCs, VOBCs, and LPBCs, each of which had positive scores for the factor of excitability and thus had the lowest firing threshold and Rb and the largest Rin. Cells of this class exhibited intermediate values of firing frequency and level of adaptation. In contrast to NGFCs that adapted extremely fast, the firing frequency of these interneurons decreased gradually along the train and, thus, they might be best described as continuous-adapting (c-Ad) cells. Within this physiological class, LPBCs and VOBCs had lower scores on the factor of firing frequency adaptation as compared with CACs, MCs, DBCs (difference is significant for P < 0.1). Because LPBCs and VOBCs also exhibited significantly shorter spike duration and had different shape of AP at “phase plot” diagram than other c-Ad cells, we delineated monkey c-Ad interneurons into 2 physiological subclasses, c-Ad1 and c-Ad2. The c-Ad1 subclass includes interneurons with typical c-Ad firing pattern and is represented morphologically by CACs, MCs, and DBCs, whereas the c-Ad2 subclass includes c-Ad interneurons with less adaptation and shorter spike duration. Morphologically they belong to LPBCs and VOBCs.

Discussion

In this study, we delineated 8 morphological types of monkey interneurons based on 1) the distribution of axonal arbors between layers and 2) the overall pattern of axonal arborization and terminal branching. We characterized the physiological and molecular properties of these 8 morphological types, 4 of which had not been previously characterized electrophysiologically in primates. For this physiological classification, we employed principal factor analysis and defined 3 factors, generally describing interneuron excitability, firing frequency, and level of adaptation of the latter. Monkey interneurons exhibited 3 basic combinations of factor scores, which were recognized as three distinct electrophysiological classes. The main results of this study are summarized in Table 6.

Table 6

Summary of interneuron diversity in layers 2–3 of monkey DLPFC

Morphological type Unique morphological features Molecular markers Shape of AP at phase plot diagram Physiological class 
LAC “Straight” pattern of axonal arborization PV Rocking chair FS 
ChC Axon cartridges PV Rocking chair FS 
NGFC “Curvy” arborization; numerous thin radially distributed dendrites CB (and NPY) Vertically oriented egg i-Ad 
MC Dense axonal cluster in layer 1 SS (and CB) Vertically oriented egg c-Ad1 
CAC Curvy arborization; few predominantly vertically distributed dendrites CB or CR in some cells Vertically oriented egg c-Ad1 
DBC Thin vertically oriented axon collaterals with minimal branching CR (CB in some cells) Vertically oriented egg c-Ad1 
VOBC Thick smooth vertically oriented axon trunks with short beaded curving horizontal collaterals CR Snail c-Ad2 
LPBC Tiny cells; axon terminals form “claw”-like structures CR (CB in some cells) Snail c-Ad2 
Morphological type Unique morphological features Molecular markers Shape of AP at phase plot diagram Physiological class 
LAC “Straight” pattern of axonal arborization PV Rocking chair FS 
ChC Axon cartridges PV Rocking chair FS 
NGFC “Curvy” arborization; numerous thin radially distributed dendrites CB (and NPY) Vertically oriented egg i-Ad 
MC Dense axonal cluster in layer 1 SS (and CB) Vertically oriented egg c-Ad1 
CAC Curvy arborization; few predominantly vertically distributed dendrites CB or CR in some cells Vertically oriented egg c-Ad1 
DBC Thin vertically oriented axon collaterals with minimal branching CR (CB in some cells) Vertically oriented egg c-Ad1 
VOBC Thick smooth vertically oriented axon trunks with short beaded curving horizontal collaterals CR Snail c-Ad2 
LPBC Tiny cells; axon terminals form “claw”-like structures CR (CB in some cells) Snail c-Ad2 

Morphology-Based Classification of Interneurons in Monkey DLPFC

Inhibitory interneurons vary greatly in their morphology; however, the axonal arborization can reveal the anatomical identity of an interneuron because interneurons seem to be particularly specialized to target different domains of the postsynaptic cell membrane, different layers of a column, and different columns (Markram et al. 2004). Four morphological types (NGFCs, ChCs, MCs, and DBCs) described here are commonly recognized in different species and have very distinctive morphological characteristics. We found these types to express specific molecular markers in agreement with previously published data (Conde et al. 1994; DeFelipe 1997; Ma et al. 2006). Four other delineated morphological types are not conventionally accepted. Although we tried to find the closest morphological analogues to them from other cortical regions of monkey and other species, some differences with previously described types were presented.

We have described for the first time in monkey cortex CR-positive vertically oriented interneurons forming basket-like structures around postsynaptic cells. Although we do not have evidence that these cells actually form synaptic contacts on the cell bodies of other neurons, in some studies (Markram et al. 1997; Karube et al. 2004), a high probability of finding a synapse at the electron microscopy level for similar structures was shown. Neocortical interneurons that form basket-like structures do not usually form a vertically oriented axonal arbor and have been previously shown to express PV or cholecystokinin (CCK)/cannabinoid (CB1) receptor (Thomson and Bannister 2003). We did not test VOBCs for expression of CCK or CB1; however, the highest density of CB1 immunoreactive axons was previously observed in layer 4 of monkey DLPFC, whereas CB1 mRNA signal is more abundant in superficial layers (Eggan and Lewis 2007). These findings suggest that CB1-immunoreactive cells might have a vertically oriented descending axonal arbor. Therefore, whether CCK and CB1 receptors are expressed in VOBCs from monkey DLPFC is an important question for further investigation.

LPBCs, CACs, and LACs of primates are presumably correlates of the three types of basket cell from rodents: small, nest, and large basket cells respectively (Markram et al. 2004). However, with the exception of LPBCs, cells of the 2 other types in our study did not form basket-like structures, and, thus, they are not referred to as “basket” cells. CACs and LACs represented the most numerous subpopulations in our sample, accounting for more than half of all interneurons studied here. We distinguished these cell types according to their patterns of axonal arborization.

LAC type includes morphological varieties with different axonal horizontal span; however, we did not find any significant correlation between axonal spread and the intrinsic membrane properties or with molecular markers (we found PV almost in 60% of the tested LACs). Therefore, in this study we did not further subdivide them into large, medium, or local arbor cells, as was done previously (Krimer et al. 2005).

CACs were not as homogeneous as LACs in respect to the expression of CaBPs and neuropeptides. We found them to express CB, CR, or SST, with majority of the tested cells not expressing any marker. Some of CACs may express CCK, as it was frequently reported to be expressed in cells with similar morphology in rodents (Kawaguchi and Kubota 1998; Wang et al. 2002; Freund and Katona 2007).

Electrophysiological Classes of Monkey DLPFC Interneurons

Currently, several different classification schemes to distinguish physiological groups of interneurons have recognized 2–15 different classes of interneurons (Kawaguchi and Kubota 1997; Gibson et al. 1999; Beierlein et al. 2000; Gupta et al. 2000; Bacci et al. 2003). These classification schemes are convenient for electrophysiological studies in the same species because they operate with only a few electrophysiological parameters; however, they do not work well across species. For example, NGFCs from monkey DLPFC do not exhibit late-spiking properties (Povysheva et al. 2007), which is a distinct feature of rat NGFCs (Kawaguchi and Kubota 1997). Therefore, a more complete combination of electrophysiological parameters is required for accurate cross-species correlations of interneurons.

In this study, we employed principal factor analysis, which allowed us to define 3 complex variables (factors), generally describing interneuron excitability, firing frequency, and a level of firing frequency adaptation. Monkey interneurons exhibited 3 basic combinations of factor scores, which were recognized as 3 distinct electrophysiological classes: nonadapting fast spiking, c-Ad, and i-Ad cells. Although these suggested terms reflect mostly difference between interneuron classes in firing frequency adaptation, other electrophysiological properties were dissimilar between these classes as well.

Some evidence suggests that the described electrophysiological factors are not just a convenient classification tool, but that they may reflect the level of expression of specific sets of ion channels in different types of interneurons (Toledo-Rodriguez et al. 2004; Sugino et al. 2006). For example, expression of a “PV cluster” of genes, contained HCN2, Kv3.1, Kv1.2, Kv1.6, Kv1.1, PV, Kv 3.2, HCN1, Kvβ1, and Caα1A was correlated with high frequency of firing, narrow spike, large AHPAF, and low Ri. (Toledo-Rodriguez et al. 2004). All these physiological parameters load factor 1 and would then explain why PV-positive LACs and ChCs were significantly different by this factor from all the other non-PV types of interneurons.

Of note, we observed important differences in the electrophysiological features of interneurons from those reported in rodent studies. We did not observe stuttering or bursting cells and only a few irregular-spiking cells in monkey layers 2–3 DLPFC, although these cells were consistently recognized in rodents (Kawaguchi 1995; Cauli et al. 1997; Wang et al. 2002). In addition NGFCs exhibited firing properties different from those described in rodents. An important question is whether these differences reflect differences between species, cortical regions, or developmental stage. Such question is particularly relevant in the case of the primate DLPFC, an area apparently absent in the neocortex of rodents (Preuss 1995). Whether the appearance of specialized areas during evolution of the primate neocortex (Krubitzer 2007) is associated with acquisition of GABA neuron classes that allow specialized forms of information processing in cortical microcircuits needs further investigation.

Although the accurate matching of different mammalian species by a certain phase of their brain development is a challenging task (Clancy et al. 2001), we do not think that differences in developmental stage account for the observed species differences in membrane properties for the following reasons. First, membrane properties of rat neocortical neurons have been shown to undergo changes through the time span of postnatal development and seem to achieve a mature state by the third to fourth postnatal week in different cortical regions (Maravall et al. 2004; Zhang 2004; Povysheva et al. 2007; Oswald and Reyes 2008) and in many classification studies 3- to 4-week-old rats were used. Second, recently we directly compared the membrane properties of NGFCs from young (P19–P28) and adult rats (P56–P135) with NGFCs from young adult monkeys (Povysheva et al. 2007) and found significant interspecies differences, whereas almost all sub- and suprathreshold properties of rat NGFCs in the young rats were similar to those found in adults.

Functional Implications

In this study, we defined 8 morphologically distinct groups of monkey interneurons, which were validated by their different intrinsic physiological and molecular properties (Table 6) that possibly play distinct roles in the neocortical circuitry. Insight into these roles can be inferred from the different lateral extent of their axonal arbor. For example, DBCs, VOBCs, LPBCs, and ChCs have a narrow axonal arbor, suggesting they contribute to information processing within the elementary cortical functional column. In contrast, the other cell types have a wider axonal arbor spread, suggesting they mediate lateral interactions between the neighboring columns.

Functions of monkey ChCs and LACs (FS PV-positive cells) may be similar to those of rodents. For example, they may exert strong perisomatic inhibition on pyramidal cells and be involved in regulation of synchronous and oscillatory activity of large populations of pyramidal cells (Freund 2003; Buzsaki and Draguhn 2004; Freund and Katona 2007). Monkey FS interneurons like those of rodents may also provide feed-forward inhibition of pyramidal cells (Pouille and Scanziani 2001; Povysheva et al. 2007). Importantly, in striking contrast to rodents, these cells in primates are not the predominant interneuron subpopulation (Conde et al. 1994; Kawaguchi and Kubota 1997), which may reflect species differences in the organization of neocortex.

We found that NGFCs in monkey DLPFC constitute a separate physiological class of i-Ad cells with very low firing frequency. In rodent, these cells have been shown to have a unique function in neocortical circuitry because they provide long-lasting inhibition of pyramidal cells via synapses containing both GABAA and GABAB receptors. Thus, NGFCs appear to be specialized for sparse temporal operations tuned for long-lasting metabotropic effects, which, in turn, may result in sustained modulation of cortical excitability (Tamas et al. 2003).

Although c-Ad cells had many common physiological properties, they still can play a distinct role in neocortical circuitry because they target different domains of pyramidal cells. For example, MCs were shown to innervate distal tufts of pyramidal cells’ apical dendrites in layer 1 and to mediate disynaptic inhibition between neocortical pyramidal cells (Wang et al. 2004; Kapfer et al. 2007; Silberberg and Markram 2007). DBCs also target mostly distal compartments of pyramidal cells but in deep cortical layers (DeFelipe 1997). Specific targets of CACs may be similar to the targets of nest basket cells from rat neocortex, which were reported to innervate mostly perisomatic regions of pyramidal cells (Wang et al. 2002). Physiological properties of c-Ad cells, such as low level of firing threshold, high Rin, and facilitating synaptic inputs to and outputs from pyramidal cells (Thomson and Deuchars 1997; Reyes et al. 1998; Gonzalez-Burgos et al. 2004), result in a reliable excitation of these interneurons by weak repetitive excitatory inputs, which puts c-Ad cells in a position to provide an effective feedback inhibition to pyramidal cells, preventing their overactivation. The functional significance of VOBCs and LPBCs remains to be determined. These 2 cell types have unusually fast spikes for adapting cells described in rats. Morphological properties of their axons, forming claw-like structures around somata of postsynaptic cells, point to perisomatic inhibition of pyramidal cells and interneurons within a cortical column.

Supplementary Material

Supplementary material can be found at http://www.cercor.oxfordjournals.org/.

Funding

The National Institutes of Health (MH067963 and MH051234).

The authors thank Mrs Olga Krimer and Mr James Kosakowski for their excellent technical assistance. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health or the National Institutes of Health. Conflict of Interest: None declared.

References

Aizenman
CD
Linden
DJ
Regulation of the rebound depolarization and spontaneous firing patterns of deep nuclear neurons in slices of rat cerebellum
J Neurophysiol.
 , 
1999
, vol. 
82
 (pg. 
1697
-
1709
)
Avoli
M
Williamson
A
Functional and pharmacological properties of human neocortical neurons maintained in vitro
Prog Neurobiol.
 , 
1996
, vol. 
48
 (pg. 
519
-
554
)
Bacci
A
Rudolph
U
Huguenard
JR
Prince
DA
Major differences in inhibitory synaptic transmission onto two neocortical interneuron subclasses
J Neurosci.
 , 
2003
, vol. 
23
 (pg. 
9664
-
9674
)
Bean
BP
The action potential in mammalian central neurons
Nat Rev Neurosci.
 , 
2007
, vol. 
8
 (pg. 
451
-
465
)
Beierlein
M
Gibson
JR
Connors
BW
A network of electrically coupled interneurons drives synchronized inhibition in neocortex
Nat Neurosci.
 , 
2000
, vol. 
3
 (pg. 
904
-
910
)
Buzsaki
G
Draguhn
A
Neuronal oscillations in cortical networks
Science.
 , 
2004
, vol. 
304
 (pg. 
1926
-
1929
)
Cauli
B
Audinat
E
Lambolez
B
Angulo
MC
Ropert
N
Tsuzuki
K
Hestrin
S
Rossier
J
Molecular and physiological diversity of cortical nonpyramidal cells
J Neurosci.
 , 
1997
, vol. 
17
 (pg. 
3894
-
3906
)
Clancy
B
Darlington
RB
Finlay
BL
Translating developmental time across mammalian species
Neuroscience.
 , 
2001
, vol. 
105
 (pg. 
7
-
17
)
Coetzee
WA
Amarillo
Y
Chiu
J
Chow
A
Lau
D
McCormack
T
Moreno
H
Nadal
MS
Ozaita
A
Pountney
D
, et al.  . 
Molecular diversity of K+ channels
Ann N Y Acad Sci
 , 
1999
, vol. 
868
 (pg. 
233
-
285
)
Conde
F
Lund
JS
Jacobowitz
DM
Baimbridge
KG
Lewis
DA
Local circuit neurons immunoreactive for calretinin, calbindin D-28K or parvalbumin in monkey prefrontal cortex—distribution and morphology
J Comp Neurol.
 , 
1994
, vol. 
341
 (pg. 
95
-
116
)
DeFelipe
J
Types of neurons, synaptic connections and chemical characteristics of cells immunoreactive for calbindin-D28K, parvalbumin and calretinin in the neocortex
J Chem Neuroanat.
 , 
1997
, vol. 
14
 (pg. 
1
-
19
)
DeFelipe
J
Chandelier cells and epilepsy
Brain.
 , 
1999
, vol. 
122
 (pg. 
1807
-
1822
)
Eggan
SM
Lewis
DA
Immunocytochemical distribution of the cannabinoid CB1 receptor in the primate neocortex: a regional and laminar analysis
Cereb Cortex.
 , 
2007
, vol. 
17
 (pg. 
175
-
191
)
Foehring
RC
Lorenzon
NM
Herron
P
Wilson
CJ
Correlation of physiologically and morphologically identified neuronal types in human association cortex in vitro
J Neurophysiol.
 , 
1991
, vol. 
66
 (pg. 
1825
-
1837
)
Freund
TF
Interneuron diversity series: rhythm and mood in perisomatic inhibition
Trends Neurosci.
 , 
2003
, vol. 
26
 (pg. 
489
-
495
)
Freund
TF
Katona
I
Perisomatic inhibition
Neuron.
 , 
2007
, vol. 
56
 (pg. 
33
-
42
)
Gabbott
PLA
Bacon
SJ
Local circuit neurons in the medial prefrontal cortex (areas 24a,b,c, 25 and 32) in the monkey .2. Quantitative areal and laminar distributions
J Comp Neurol.
 , 
1996
, vol. 
364
 (pg. 
609
-
636
)
Gabbott
PLA
Dickie
BGM
Vaid
RR
Headlam
AJN
Bacon
SJ
Local-circuit neurones in the medial prefrontal cortex (areas 25, 32, and 24b) in the rat: morphology and quantitative distribution
J Comp Neurol.
 , 
1997
, vol. 
377
 (pg. 
465
-
499
)
Geiger
JRP
Jonas
P
Dynamic control of presynaptic ca2+inflow by fast-inactivating K+ channels in hippocampal mossy fiber boutons
Neuron.
 , 
2000
, vol. 
28
 (pg. 
927
-
939
)
Gibson
JR
Beierlein
M
Connors
BW
Two networks of electrically coupled inhibitory neurons in neocortex
Nature.
 , 
1999
, vol. 
402
 (pg. 
75
-
79
)
Gonchar
Y
Burkhalter
A
Three distinct families of GABAergic neurons in rat visual cortex
Cereb Cortex.
 , 
1997
, vol. 
7
 (pg. 
347
-
358
)
Gonzalez-Burgos
G
Krimer
LS
Povysheva
NV
Barrionuevo
G
Lewis
DA
Functional properties of fast spiking interneurons and their synaptic connections with pyramidal cells in primate dorsolateral prefrontal cortex
J Neurophysiol.
 , 
2005
, vol. 
93
 (pg. 
942
-
953
)
Gonzalez-Burgos
G
Krimer
LS
Urban
NN
Barrionuevo
G
Lewis
DA
Synaptic efficacy during repetitive activation of excitatory inputs in primate dorsolateral prefrontal cortex
Cereb Cortex.
 , 
2004
, vol. 
14
 (pg. 
530
-
542
)
Gupta
A
Wang
Y
Markram
H
Organizing principles for a diversity of GABAergic interneurons and synapses in the neocortex
Science.
 , 
2000
, vol. 
287
 (pg. 
273
-
278
)
Jones
EG
Hendry
SHC
Peter
A
EG
Jones
Basket cells
Cerebral cortex 1, cellular components of the cerebral cortex
 , 
1984
New York
Plenum
(pg. 
309
-
336
)
Kapfer
C
Glickfeld
LL
Atallah
BV
Scanziani
M
Supralinear increase of recurrent inhibition during sparse activity in the somatosensory cortex
Nat Neurosci.
 , 
2007
, vol. 
10
 (pg. 
743
-
753
)
Karube
F
Kubota
Y
Kawaguchi
Y
Axon branching and synaptic bouton phenotypes in GABAergic nonpyramidal cell subtypes
J Neurosci.
 , 
2004
, vol. 
24
 (pg. 
2853
-
2865
)
Kawaguchi
Y
Groupings of nonpyramidal and pyramidal cells with specific physiological and morphological characteristics in rat frontal cortex
J Neurophysiol.
 , 
1993
, vol. 
69
 (pg. 
416
-
431
)
Kawaguchi
Y
Physiological subgroups of nonpyramidal cells with specific morphological characteristics in layer II/III of rat frontal cortex
J Neurosci.
 , 
1995
, vol. 
15
 (pg. 
2638
-
2655
)
Kawaguchi
Y
Kondo
S
Parvalbumin, somatostatin and cholecystokinin as chemical markers for specific GABAergic interneuron types in the rat frontal cortex
J Neurocytol.
 , 
2002
, vol. 
31
 (pg. 
277
-
287
)
Kawaguchi
Y
Kubota
Y
Physiological and morphological identification of somatostatin- or vasoactive intestinal polypeptide-containing cells among GABAergic cell subtypes in rat frontal cortex
J Neurosci.
 , 
1996
, vol. 
16
 (pg. 
2701
-
2715
)
Kawaguchi
Y
Kubota
Y
GABAergic cell subtypes and their synaptic connections in rat frontal cortex
Cereb Cortex.
 , 
1997
, vol. 
7
 (pg. 
476
-
486
)
Kawaguchi
Y
Kubota
Y
Neurochemical features and synaptic connections of large physiologically-identified GABAergic cells in the rat frontal cortex
Neuroscience.
 , 
1998
, vol. 
85
 (pg. 
677
-
701
)
Kisvarday
ZF
Martin
KAC
Whitteridge
D
Somogyi
P
Synaptic connections of intracellularly filled clutch cells: a type of small basket cell in the visual cortex of the cat
J Comp Neurol.
 , 
1985
, vol. 
241
 (pg. 
111
-
137
)
Krimer
LS
Zaitsev
AV
Czanner
G
Kroner
S
Gonzalez-Burgos
G
Povysheva
NV
Iyengar
S
Barrionuevo
G
Lewis
DA
Cluster analysis-based physiological classification and morphological properties of inhibitory neurons in layers 2-3 of monkey dorsolateral prefrontal cortex
J Neurophysiol.
 , 
2005
, vol. 
94
 (pg. 
3009
-
3022
)
Krubitzer
L
The magnificent compromise: cortical field evolution in mammals
Neuron.
 , 
2007
, vol. 
56
 (pg. 
201
-
208
)
Lau
D
Vega-Saenz de Miera
E
Contreras
D
Ozaita
A
Harvey
M
Chow
A
Noebels
JL
Paylor
R
Morgan
JI
Leonard
CS
, et al.  . 
Impaired fast-spiking, suppressed cortical inhibition, and increased susceptibility to seizures in mice lacking Kv3.2 K+ channel proteins
J Neurosci.
 , 
2000
, vol. 
20
 (pg. 
9071
-
9085
)
Letinic
K
Zoncu
R
Rakic
P
Origin of GABAergic neurons in the human neocortex
Nature.
 , 
2002
, vol. 
417
 (pg. 
645
-
649
)
Lewis
DA
Hashimoto
T
Volk
DW
Cortical inhibitory neurons and schizophrenia
Nat Rev Neurosci.
 , 
2005
, vol. 
6
 (pg. 
312
-
324
)
Lund
JS
Lewis
DA
Local circuit neurons of developing and mature macaque prefrontal cortex: Golgi and immunocytochemical characteristics
J Comp Neurol.
 , 
1993
, vol. 
328
 (pg. 
282
-
312
)
Lund
JS
Yoshioka
T
Local circuit neurons of macaque monkey striate cortex: III. Neurons of laminae 4B, 4A, and 3B
J Comp Neurol.
 , 
1991
, vol. 
311
 (pg. 
234
-
258
)
Ma
YY
Hu
H
Berrebi
AS
Mathers
PH
Agmon
A
Distinct subtypes of somatostatin-containing neocortical interneurons revealed in transgenic mice
J Neurosci.
 , 
2006
, vol. 
26
 (pg. 
5069
-
5082
)
Mainen
ZF
Sejnowski
TJ
Influence of dendritic structure on firing pattern in model neocortical neurons
Nature.
 , 
1996
, vol. 
382
 (pg. 
363
-
366
)
Maravall
M
Stern
EA
Svoboda
K
Development of intrinsic properties and excitability of layer 2/3 pyramidal neurons during a critical period for sensory maps in rat barrel cortex
J Neurophysiol.
 , 
2004
, vol. 
92
 (pg. 
144
-
156
)
Markram
H
Lubke
J
Frotscher
M
Roth
A
Sakmann
B
Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex
J Physiol.
 , 
1997
, vol. 
500
 (pg. 
409
-
440
)
Markram
H
Toledo-Rodriguez
M
Wang
Y
Gupta
A
Silberberg
G
Wu
CZ
Interneurons of the neocortical inhibitory system
Nat Rev Neurosci.
 , 
2004
, vol. 
5
 (pg. 
793
-
807
)
McBain
CJ
Fisahn
A
Interneurons unbound
Nat Rev Neurosci.
 , 
2001
, vol. 
2
 (pg. 
11
-
23
)
Meskenaite
V
Calretinin-immunoreactive local circuit neurons in area 17 of the cynomolgus monkey, Macaca fascicularis
J Comp Neurol.
 , 
1997
, vol. 
379
 (pg. 
113
-
132
)
Miyoshi
G
Butt
SJB
Takebayashi
H
Fishell
G
Physiologically distinct temporal cohorts of cortical interneurons arise from telencephalic Olig2-expressing precursors
J Neurosci.
 , 
2007
, vol. 
27
 (pg. 
7786
-
7798
)
Molyneaux
BJ
Arlotta
P
Menezes
JRL
Macklis
JD
Neuronal subtype specification in the cerebral cortex
Nat Rev Neurosci.
 , 
2007
, vol. 
8
 (pg. 
427
-
437
)
Oswald
A-MM
Reyes
AD
Maturation of intrinsic and synaptic properties of layer 2/3 pyramidal neurons in mouse auditory cortex
J Neurophysiol.
 , 
2008
, vol. 
99
 (pg. 
2998
-
3008
)
Pape
HC
Queer current and pacemaker: the hyperpolarization-activated cation current in neurons
Annu Rev Physiol.
 , 
1996
, vol. 
58
 (pg. 
299
-
327
)
Pouille
F
Scanziani
M
Enforcement of temporal fidelity in pyramidal cells by somatic feed-forward inhibition
Science.
 , 
2001
, vol. 
293
 (pg. 
1159
-
1163
)
Povysheva
NV
Zaitsev
AV
Kroner
S
Krimer
OA
Rotaru
DC
Gonzalez-Burgos
G
Lewis
DA
Krimer
LS
Electrophysiological differences between neurogliaform cells from monkey and rat prefrontal cortex
J Neurophysiol.
 , 
2007
, vol. 
97
 (pg. 
1030
-
1039
)
Preuss
TM
Do rats have prefrontal cortex—the rose-woolsey-akert program reconsidered
J Cogn Neurosci.
 , 
1995
, vol. 
7
 (pg. 
1
-
24
)
Reyes
A
Lujan
R
Rozov
A
Burnashev
N
Somogyi
P
Sakmann
B
Target-cell-specific facilitation and depression in neocortical circuits
Nat Neurosci.
 , 
1998
, vol. 
1
 (pg. 
279
-
285
)
Robinson
RB
Hyperpolarization-activated cation currents: from molecules to physiological function
Annu Rev Physiol.
 , 
2003
, vol. 
65
 (pg. 
453
-
480
)
Rudy
B
Chow
A
Lau
D
Amarillo
Y
Ozaita
A
Saganich
M
Moreno
H
Nadal
MS
Hernandez-Pineda
R
Hernandez-Cruz
A
, et al.  . 
Contributions of Kv3 channels to neuronal excitability
Ann N Y Acad Sci
 , 
1999
, vol. 
868
 (pg. 
304
-
343
)
Silberberg
G
Markram
H
Disynaptic inhibition between neocortical pyramidal cells mediated by martinotti cells
Neuron.
 , 
2007
, vol. 
53
 (pg. 
735
-
746
)
Soltesz
I
Diversity in the neuronal machine: order and variability in interneuronal microcircuits
 , 
2006
Oxford
Oxford University Press
Somogyi
P
Cowey
A
Combined Golgi and electron microscopic study on the synapses formed by double bouquet cells in the visual cortex of the cat and monkey
J Comp Neurol.
 , 
1981
, vol. 
195
 (pg. 
547
-
566
)
Sugino
K
Hempel
CM
Miller
MN
Hattox
AM
Shapiro
P
Wu
CZ
Huang
ZJ
Nelson
SB
Molecular taxonomy of major neuronal classes in the adult mouse forebrain
Nat Neurosci.
 , 
2006
, vol. 
9
 (pg. 
99
-
107
)
Szabadics
J
Varga
C
Molnar
G
Olah
S
Barzo
P
Tamas
G
Excitatory effect of GABAergic axo-axonic cells in cortical microcircuits
Science.
 , 
2006
, vol. 
311
 (pg. 
233
-
235
)
Tamas
G
Lorincz
A
Simon
A
Szabadics
J
Identified sources and targets of slow inhibition in the neocortex
Science.
 , 
2003
, vol. 
299
 (pg. 
1902
-
1905
)
Thomson
AM
Bannister
AP
Interlaminar connections in the neocortex
Cereb Cortex.
 , 
2003
, vol. 
13
 (pg. 
5
-
14
)
Thomson
AM
Deuchars
J
Synaptic interactions in neocortical local circuits: dual intracellular recordings in vitro
Cereb Cortex.
 , 
1997
, vol. 
7
 (pg. 
510
-
522
)
Toledo-Rodriguez
M
Blumenfeld
B
Wu
CZ
Luo
JY
Attali
B
Goodman
P
Markram
H
Correlation maps allow neuronal electrical properties to be predicted from single-cell gene expression profiles in rat neocortex
Cereb Cortex.
 , 
2004
, vol. 
14
 (pg. 
1310
-
1327
)
Toledo-Rodriguez
M
Goodman
P
Illic
M
Wu
CZ
Markram
H
Neuropeptide and calcium-binding protein gene expression profiles predict neuronal anatomical type in the juvenile rat
J Physiol.
 , 
2005
, vol. 
567
 (pg. 
401
-
413
)
Wang
Y
Gupta
A
Toledo-Rodriguez
M
Wu
CZ
Markram
H
Anatomical, physiological, molecular and circuit properties of nest basket cells in the developing somatosensory cortex
Cereb Cortex.
 , 
2002
, vol. 
12
 (pg. 
395
-
410
)
Wang
Y
Toledo-Rodriguez
M
Gupta
A
Wu
CZ
Silberberg
G
Luo
JY
Markram
H
Anatomical, physiological and molecular properties of Martinotti cells in the somatosensory cortex of the juvenile rat
J Physiol.
 , 
2004
, vol. 
561
 (pg. 
65
-
90
)
Xu
XM
Roby
KD
Callaway
EM
Mouse cortical inhibitory neuron type that coexpresses somatostatin and calretinin
J Comp Neurol.
 , 
2006
, vol. 
499
 (pg. 
144
-
160
)
Yanez
IB
Munoz
A
Contreras
J
Gonzalez
J
Rodriguez-Veiga
E
DeFelipe
J
Double bouquet cell in the human cerebral cortex and a comparison with other mammals
J Comp Neurol.
 , 
2005
, vol. 
486
 (pg. 
344
-
360
)
Zaitsev
AV
Gonzalez-Burgos
G
Povysheva
NV
Kroner
S
Lewis
DA
Krimer
LS
Localization of calcium-binding proteins in physiologically and morphologically characterized interneurons of monkey dorsolateral prefrontal cortex
Cereb Cortex.
 , 
2005
, vol. 
15
 (pg. 
1178
-
1186
)
Zhang
Z-w
Maturation of layer V pyramidal neurons in the rat prefrontal cortex: intrinsic properties and synaptic function
J Neurophysiol.
 , 
2004
, vol. 
91
 (pg. 
1171
-
1182
)