^{1}

^{2}

^{1}

^{1}

^{1}

^{2}

^{1}

^{2}

^{*}

Conceived and designed the experiments: JB MB HT IBG AK. Performed the experiments: JB MB HT AK. Analyzed the data: JB MB HT IBG AK. Wrote the paper: JB MB IBG AK. Attempted to control entropy: AK.

The authors have declared that no competing interests exist.

Extracellular recordings in primates have identified two types of neurons in the external segment of the globus pallidus (GPe): high frequency pausers (HFP) and low frequency bursters (LFB). The aim of the current study was to test whether the properties of HFP and LFB neurons recorded extracellularly in the primate GPe are linked to cellular mechanisms underlying the generation of action potential (AP) firing. Thus, we recorded from primate and rat globus pallidus neurons. Extracellular recordings in primates revealed that in addition to differences in firing patterns the APs of neurons in these two groups have different widths (AP_{ex}). To quantitatively investigate this difference and to explore the heterogeneity of pallidal neurons we carried out cell-attached and whole-cell recordings from acute slices of the rat globus pallidus (GP, the rodent homolog of the primate GPe), examining both spontaneous and evoked activity. Several parameters related to the extracellular activity were extracted in order to subdivide the population of recorded GP neurons into groups. Statistical analysis showed that the GP neurons in the rodents may be differentiated along six cellular parameters into three subgroups. Combining two of these groups allowed a better separation of the population along nine parameters. Four of these parameters (F_{max}, AP_{amp}, AP_{hw}, and AHP_{s} amplitude) form a subset, suggesting that one group of neurons may generate APs at significantly higher frequencies than the other group. This may suggest that the differences between the HFP and LFB neurons in the primate are related to fundamental underlying differences in their cellular properties.

The external segment of the globus pallidus (GPe) is an intrinsic nucleus in the indirect pathway of the basal ganglia and is crucial to controlling their output

Surprisingly, the GPe of primates and humans are homologues of the globus pallidus (GP) in rodents _{h}), rebound firing, spike accommodation, spike frequency adaptation and spike afterhyperpolarization

The current study aimed to test whether properties of cells recorded extracellularly in the primate GPe can be linked to cellular mechanisms underlying the generation of AP firing. We compared extracellular recordings in the behaving primate with

LFB and HFP neurons differ considerably in their firing patterns

The recordings from primates suggested that the neurons in the GPe could be separated into two groups by the width of their AP. Furthermore, they suggested that these two populations differ in cellular properties that underlie the ability to produce high frequency firing. Since intracellular recordings from primates are not practical, it was decided to test this suggestion in an

To enable comparison between the intracellular and extracellular recordings we opted for spike-triggered averaging of the extracellular spikes using the intracellular APs as time markers. The shape of the average extracellular spike closely resembled the first derivative of the intracellular AP (

Given the similarity between the transmembrane current and the extracellular potential, we carried out all further experiments using the cell-attached configuration rather than an extracellular electrode. This substantially simplified the experiment. Recordings from each cell were made first in the cell-attached mode and then in the whole-cell mode. As previously reported, the spontaneous firing of GP neurons appeared to be regular

Spontaneous firing rate | Hz | 4±3 | 17±7 | 11±7 |

STD of spontaneous firing rate | Hz | 1±0.3 | 2.3±1 | 1.6±0.8 |

CV of spontaneous firing rate | 0.6±0.4 | 0.2±0.1 | 0.3±0.2 | |

Fano factor of spontaneous firing rate | ms | 0.4±0.3 | 0.4±0.2 | 0.3±0.3 |

V_{m} |
mV | −54±4 | −53±4 | −54±4 |

R_{in} |
MΩ | 306±72.5 | 260±36.5 | 217±48.5 |

Sag | mV | 42±8 | 18±7 | 23±12 |

Sag ratio | 0.7±0.1 | 0.8±0.1 | 0.8±0.1 | |

F_{max} |
Hz | 108.5±53 | 266.6±170 | 121±48.5 |

Current inducing 63% of F_{max} |
pA | 288.5±246 | 521.5±513 | 319±256 |

Monophasic AHP | + | ― | + | |

Biphasic AHP | ― | + | ― | |

AHP_{s} amplitude |
mV | 18±2 | 14±3 | 16±3 |

AHP_{f} amplitude |
mV | N/A | 13±4 | N/A |

AP adaptation ratio | 0.5±0.1 | 0.65±0.1 | 0.75±0.1 | |

AP_{amp} |
mV | 80±7 | 69±8 | 77.5±9 |

AP_{hw} |
ms | 0.6±0.1 | 0.4±0.15 | 0.5±0.1 |

AP threshold | mV | −42±5 | −44±5 | −43±4.5 |

Rebound firing | ― | + | ― | |

Firing pattern | irregular | regular | regular | |

Spontaneous firing rate | Hz | 8±5 | 14±6 | 12±10 |

STD of spontaneous firing rate | Hz | 1.3±0.5 | 2.2±1 | 1.6±0.7 |

CV of spontaneous firing rate | 0.3±0.2 | 0.2±0.1 | 0.3±0.2 | |

Fano factor of spontaneous firing rate | ms | 0.3±0.2 | 0.4±0.3 | 0.3±0.2 |

AP_{ex} duration |
ms | 0.8±0.2 | 0.5±0.1 | 0.7±0.2 |

Although spontaneous firing frequency measured intracellularly and extracellularly appeared similar, their weak correlation allows only partial correlation of intracellular and extracellular firing. Since the recordings from primates suggested that the population of neurons could be divided using the width of the AP, we further characterized the relation between the width of the intracellular and extracellular potentials in the rat.

The half-width of the intracellular AP (AP_{hw}) was calculated by measuring the width of the AP at the midpoint between threshold and peak. The half-width of the intracellular AP could also be extracted from the first order numerical derivative of the intracellular AP by measuring the time delay between the minimum and maximum values of this derivative (^{2} = 0.82).

Can the AP half-width recorded in rats be used as a parameter to dissect the neuronal populations of the GP into groups as in primate? To investigate this we constructed histograms for the intracellular (

The visual separation allowed us to divide the population into three groups. One group of neurons, referred to as type A neurons, responded to depolarizing current injection with burst firing (

To further investigate the differences between the three proposed groups we quantified the sag of the membrane potential and calculated the R_{in} of each cell type from current-voltage (I–V) curves of the neurons (_{h} was still not activated. In fact, for types A and C the I–V curve was linear over the entire the current range applied in this study (_{h}. The parameters extracted from these curves are reported in

_{h} on membrane potential. _{max} and current required to reach 63% of maximal firing rate were extracted by exponential fit for type A neurons. Values and error bars are mean ± S.E. Bii, As in Aii but for type B neurons. Cii, As in Aii but for type C neurons.

Next, we generated current-frequency (F–I) curves to analyze the firing activity of each cell type in response to depolarizing current injections (_{max}) of type A, B, and C neurons was 108.5±53 Hz (n = 14), 266.6±170 Hz (n = 24), and 121±48.5 Hz (n = 38), respectively. Statistical analysis of these differences appears in

P | P (AvsB) | P (AvsC) | P (BvsC) | P (ACvsB) | |

Spontaneous firing rate | 3e-4 |
8e-5 |
0.006 |
0.04 |
0.002 |

CV of firing rate | 0.01 |
0.002 |
0.01 |
0.7 | 0.1 |

Fano factor of firing rate | 0.5 | 0.8 | 0.3 | 0.3 | 0.5 |

V_{m} |
0.2 | 0.1 | 0.9 | 0.08 | 0.06 |

R_{in} |
1e-5 |
2e-6 |
2e-5 |
0.5 | 0.03 |

Sag ratio | 9e-4 |
5e-5 |
0.006 |
0.2 | 0.02 |

Sag | 1e-8 |
2e-9 |
4e-7 |
0.05 |
5e-4 |

F_{max} |
8e-8 |
5e-6 |
0.01 |
4e-6 |
2e-7 |

Current inducing 63% of F_{max} |
0.01 |
0.009 |
0.004 |
0.8 | 0.3 |

AP_{amp} |
1e-4 |
2e-4 |
0.4 | 3e-4 |
4e-5 |

AP_{hw} |
6e-6 |
5e-5 |
0.06 | 6e-5 |
4e-6 |

AHP_{s} amplitude |
9e-4 |
2e-4 |
0.02 |
0.05 |
0.004 |

AP threshold | 0.7 | 0.4 | 0.6 | 0.9 | 0.7 |

AP adaptation ratio | 2e-6 |
0.01 |
7e-7 |
0.006 |
0.2 |

Spontaneous firing rate | 0.03 |
0.009 |
0.5 | 0.06 | 0.2 |

CV of spontaneous firing rate | 0.9 | 0.8 | 0.8 | 0.99 | 1 |

AP_{ex} duration |
3e-6 |
0.02 |
0.03 |
3e-5 |
1e-4 |

Fano factor of spontaneous firing rate | 0.2 | 0.1 | 0.3 | 0.2 | 0.1 |

Visual dissection of the neuronal population may lead to selection bias. In addition, it was not clear whether the parameters used in the separation process (_{max}, AHP_{s} amplitude, AP adaptation ratio, and AP_{ex} duration) were significant for all three pairs of groups tested.

As noted above, there were clear differences between the I–V and F–I curves of group B and those of the two other groups (

_{ex}) calculated as in _{in}) calculated as in _{s}). AHP amplitude was measured from threshold. _{h}. Sag was calculated as the difference between the maximal deflection of the membrane potential following a hyperpolarizing step and the deflection at the end of the pulse.

In this study we recorded from rat brain slices in order to determine whether the differences between LFB and HFP neurons in the globus pallidus of primates are due to cellular or to network properties. To do this, we compared intracellular and extracellular characteristics of the AP recorded from neurons in the rat globus pallidus and classified the population of recorded neurons into functional groups. We then used the properties of these groups to predict the cellular properties of neurons in the primate GPe.

Initially, we found that the extracellular width of the AP (AP_{ex}) in behaving primate could differentiate the neurons in the two groups in the GPe. We then showed that the AP_{ex} in the rat could serve as a quantitative measure for the intracellular AP_{hw}. Analysis of the response of GP neurons to current injection in the whole-cell mode suggested that the neurons in the rat could be divided into three groups. Statistical analysis of these groups revealed that six of the parameters (intracellular spontaneous firing rate, Sag, F_{max}, AHP_{s} amplitude, AP adaptation ratio, and AP_{ex}) significantly divided the population into three groups. Combining two groups into one super-group increased the number of parameters dividing the population to 9. These data from the rat thus hint that the neuronal population in the primate GPe may be composed of two neuronal types with different cellular properties.

The division of neurons in the primate GPe into LFB and HFP groups was established in the early days of basal ganglia electrophysiology

Given the differences between the extracellular waveforms of primate LFB and HFP neurons (

Several methods have been used to classify populations of neurons into subgroups. In the GP these methods have mostly included visual inspection of the recorded data followed by extraction of visually salient parameters

The parameters in our data did not display a normal distribution, thus we used non-parametric methods to assess the statistical differences among the groups. The division into the different groups was based on a few parameters and was verified by the division for other parameters. However, the separation of the data into groups using the parameters did not enable automatic clustering into either two or three groups. The data reside in a high dimensionality space derived from the number of parameters. The sparseness of the data points leads to the problem known as the “curse of dimensionality”

Thus, our current study does not provide a clear dissection of the GP population into groups, even though statistical analysis suggests that the population may be divided into 2–3 groups, unlike the database modeling suggesting that the population is homogenous _{max}, AP_{amp}, AP_{hw}, and AHP_{s} amplitude). Furthermore, the width of the extracellular potential was significantly different between the HFPs and LFBs in the primate GPe. This hints that the observed firing modes of these two types of neurons in the primate GPe may be due to different cellular properties. Thus, there could be a basic mechanistic difference between them, possibly a differential expression of either voltage-gated sodium or potassium conductances. This possible mechanistic model may be important for research on the dynamic properties of the GPe. Furthermore, the parallels observed here between the properties of neurons in the rat GP and the primate GPe may help combine system level studies of the primate GPe with cellular studies in the rodent GP. It is important to note that, since recording intracellularly from neurons in primates is not possible, the differences in AP shapes may be partly induced by network activity.

Thick sagittal slices of 300 µm were obtained from rat somatosensory cortex, striatum and GP using previously described techniques _{3}, 1.25 Na_{2}HPO_{4}, 1.5 CaCl_{2}, 1 MgCl_{2}, 25 glucose and 0.5 Na-ascorbate (pH 7.4 with 95% O_{2}/5% CO_{2}). Slices were cut using an HR2 Slicer (Sigman Electronic, Germany) and transferred to a submersion-type chamber, where they were maintained for the remainder of the day. Experiments reported here were carried out at 34°C. The GP nucleus and individual GP neurons were visualized using infrared differential interference contrast (IR-DIC) microscopy. The recording chamber was constantly perfused with oxygenated ACSF. The standard pipette solution contained (in mM): 130 K-gluconate, 10 KCl, 10 HEPES, 4 MgATP, 10 Na-phosphocreatin, 0.5 EGTA and 0.3 GTP (Sigma) (pH 7.2 with KOH). The reference electrode was an Ag-AgCl pellet placed in the bath. The 10 mV liquid junction potential measured under the ionic conditions reported here was not corrected for.

Cell-attached and whole-cell recordings were obtained from the soma of GP neurons with an Axopatch-200B amplifier (Axon Instruments). Voltage was filtered at 10 kHz and sampled at 20 kHz, unless stated otherwise, using patch pipettes (4–8 MΩ) pulled from thick-walled borosilicate glass capillaries (2.0 mm outer diameter, 0.5 mm wall thickness, Hilgenberg, Malsfeld, Germany).

All off-line analysis was carried out with Offline Sorter 2.8.6 (Plexon), NeuroExploler 4.007 (Nex Technologies), Matlab R2007b (Mathworks) and IgorPro 5.0 (WaveMetrics) on a personal computer. The frequency of spontaneous extracellular and intracellular firing was calculated from spikes extracted from 5 min of continuous recording.

The duration of the extracellular action potential (AP_{ex}) corresponding to the intracellular half-width (see

The action potential threshold was extracted by numerically calculating the time-dependent second derivative of the membrane potential. The threshold was defined as the point at which the second derivative exceeded 50% of its maximal value.

The amplitude of the intracellular action potential (AP_{amp}) was measured from threshold to the peak of the AP.

The half-width (AP_{hw}) of the intracellular AP was measured as the width at half height of the AP.

The amplitude of the action potential fast afterhyperpolarization (AHP_{f}) was measured from threshold to the peak of the membrane afterhyperpolarization.

The amplitude of the action potential slow afterhyperpolarization (AHP_{s}) was measured from threshold to the second peak of the membrane after-hyperpolarization.

Input resistance (R_{in}) was measured by injecting several hyperpolarizing current steps, recording the response of the membrane potential to these steps, subtracting the resting membrane potential, and calculating R_{in} from linear current-voltage (I–V) curve fitting.

Time- and voltage-gated anomalous rectification (Sag) reflecting the activation of nonspecific cationic current (I_{h}) was measured similarly to input resistance and calculated as the difference between minima of the current-voltage curves and the current-sag curves. Sag ratio was calculated as the ratio between these points.

Current-frequency (F–I) curves were generated by injecting several depolarizing current steps via the patch pipette and analyzed using exponential curve fitting. The maximal firing rate (F_{max}) and the current that induced 63% of F_{max} were extracted from each fit.

The action potential adaptation ratio was calculated as the ratio between the last and first action potential amplitudes in depolarizing current steps of 50 pA. The existence of rebound firing and the pattern of depolarizing firing in response to injected depolarizing currents were qualitatively evaluated.

Experimental results were consistently obtained from cells from at least 7 rats. All the results for a particular experiment were pooled and displayed as mean ± S.D., unless otherwise stated. A Kruskal-Wallis (KW) one- way analysis of variance by ranks tested equality of the measured cell population based on extracellular and intracellular parameters. A Mann-Whitney-Wilcoxon (MWW) test determined which of the measured extracellular or intracellular parameters divided the GP cell population into three groups.

We thank Dr. M. Dror, P. Malamud, K. McCairn and I. Miller for animal care and K. McCairn and I. Miller also for assistance with recording.