Somatic Integration of Single Ion Channel Responses of α7 Nicotinic Acetylcholine Receptors Enhanced by PNU-120596

Positive allosteric modulators of highly Ca2+-permeable α7 nicotinic acetylcholine receptors, such as PNU-120596, may become useful therapeutic tools supporting neuronal survival and function. However, despite promising results, the initial optimism has been tempered by the concerns for cytotoxicity. The same concentration of a given nicotinic agent can be neuroprotective, ineffective or neurotoxic due to differences in the expression of α7 receptors and susceptibility to Ca2+ influx among various subtypes of neurons. Resolution of these concerns may require an ability to reliably detect, evaluate and optimize the extent of α7 somatic ionic influx, a key determinant of the likelihood of neuronal survival and function. In the presence of PNU-120596 and physiological choline (∼10 µM), the activity of individual α7 channels can be detected in whole-cell recordings as step-like current/voltage deviations. However, the extent of α7 somatic influx remains elusive because the activity of individual α7 channels may not be integrated across the entire soma, instead affecting only specific subdomains located in the channel vicinity. Such a compartmentalization may obstruct detection and integration of α7 currents, causing an underestimation of α7 activity. By contrast, if step-like α7 currents are integrated across the soma, then a reliable quantification of α7 influx in whole-cell recordings is possible and could provide a rational basis for optimization of conditions that support survival of α7-expressing neurons. This approach can be used to directly correlate α7 single-channel activity to neuronal function. In this study, somatic dual-patch recordings were conducted using large hypothalamic and hippocampal neurons in acute coronal rat brain slices. The results demonstrate that the membrane electrotonic properties do not impede somatic signaling, allowing reliable estimates of somatic ionic and Ca2+ influx through α7 channels, while the somatic space-clamp error is minimal (∼0.01 mV/µm). These research efforts could benefit optimization of potential α7-PAM-based therapies.

A moderate, persistent activation of a7 nAChRs can be neuroprotective [3,6,7,8,9] and modeled ex vivo using low concentrations of nicotinic agonists (e.g., physiological choline; ,5-10 mM) enhanced by PNU-120596 [4,5] and possibly, other Type-II a7-PAMs [2,3]. Under these experimental conditions, detection of individual a7 channel openings in whole-cell experiments becomes possible in both voltage-and current-clamp configurations [4,5]. By contrast, in the absence of PNU-120596, openings of individual a7 channels cannot be distinguished from noise in whole-cell recordings and thus, the extent of a7-mediated ionic and Ca 2+ influx cannot be reliably detected and quantified. Therefore, PNU-120596 and possibly other a7-PAMs may be effective for enhancing and optimizing the potency of nicotinic agonists to a degree that permits low concentrations of nicotinic agonists such as physiological concentrations of choline to produce moderate, persistent activation of a7 nAChRs -effects that may support neuroprotection and cognitive performance in vivo. However, despite promising results, the initial optimism for potential therapeutic applications of PNU-120596 has been tempered by the concerns for cytotoxicity [9,10]. In general, the same concentration of a given nicotinic agent can be neuropro-tective, ineffective or neurotoxic due to differences in the expression density of functional a7 receptors and susceptibility to Ca 2+ influx among various subtypes of neurons. The resolution of these concerns and ambiguities may require an ability to reliably detect, evaluate, manipulate and optimize the extent of a7mediated somatic ionic and Ca 2+ influx, a key determinant of the likelihood of neuronal survival and function.
However, while PNU-120596 allows detection of individual a7 channel activity in whole-cell patch-clamp experiments, it remains unclear whether activation of each individual a7 channel impacts the entire soma or only specific subdomains located in the channel/electrode vicinity, reflecting somatic electrotonic properties. Such a somatic electrical compartmentalization may reduce neuronal excitability and obstruct integration and detection of ion channel activity across the neuronal soma, causing an underestimation of whole-cell a7-mediated persistent ionic influx. A reliable quantification of individual a7 channel activity in whole-cell requires the neuronal soma to act as an effective integrator, allowing an individual a7 channel to depolarize the entire soma and thus, be reliably detected across the entire soma as well as by a patch-clamp electrode regardless of the electrode location on the somatic membrane. The degree of compliance to this requirement in central neurons is currently unknown. This study tests the hypothesis that the electrotonic properties of large hypothalamic tuberomammillary (TM) neurons (.20 mm) and hippocampal CA1 interneurons (.15 mm) do not impede somatic signaling allowing reliable estimates of the net charge generated by individual a7 ion channels in whole-cell recordings in the presence of PNU-120596. These results may benefit the search for an optimum in a7-mediated Ca 2+ influx, neuronal excitability and survivability.

Animals
Young adult male Sprague-Dawley rats (P22-30) were used in accordance with the Guide for the Care and Use of Laboratory Animals (NIH 865-23, Bethesda, MD), and experimental protocols were approved by the Animal Care and Use Committee of Southern Illinois University School of Medicine.

Dual-patch recordings from a single cell in brain slices
The recording chamber was perfused with oxygenated ACSF at a rate of 1 ml/min using a perfusion pump (Model73160-20, Cole-Parmer, Vernon Hills, IL). TM neurons of the posterior hypothalamus and interneurons of the CA1 region of the hippocampus in brain slices were visually selected for electrophys-iological recordings using an Olympus BX-51WI microscope (Olympus Inc, Center Valley, PA). Dual patch-clamp electrophysiological recordings were conducted using a MultiClamp-700B amplifier equipped with Digidata-1440A A/D converter (Molecular Devices, Sunnyvale, CA). Voltage and current traces were recorded in 20-100 s sweeps. No delays were allowed between sweeps. Data were filtered at 2-4 kHz, sampled at 10 kHz and stored on a personal computer. During analysis, data were filtered at 0.4-1 kHz for illustrations and comparison of traces. Patch pipettes were pulled using a Sutter P-97 horizontal puller (Sutter Instruments, Novato, CA). The pipette resistance was 4-6 MV when filled with the internal solution. In each experiment, a single neuronal soma was clamped with two patch electrodes: electrodes were advanced towards the opposite sides of the recorded soma under visual control and a cell-attached patch-clamp configuration with a gigaseal resistance was established for each electrode. The patch configuration of electrodes was later changed to whole-cell voltage-or current-clamp, as needed. The access resistances (,30 MV) were not compensated. Choline (5-10 mM) and PNU-120596 (1 mM) were present in ACSF during, and at least 40 min before, each experiment. The extracellular recording solution was identical to the ACSF used for the tissue preparation. The patch electrode solution contained (in mM): K-gluconate 140, NaCl 1, MgCl 2 2, Mg-ATP 2, Na-GTP 0.3, HEPES 10, KOH 0.42 (pH 7.4). Membrane voltages were not corrected for the liquid junction potential: V LJ (K-gluconate) = 16.2 mV.

The lack of effects of R f
In four experiments with hypothalamic TM neurons, the feedback resistance (R f ) of one or both of the recording amplifiers were switched from its whole-cell default value of 500 MV to 50 GV and step-like a7-mediated currents were recorded from the same cell by both electrodes in voltage-clamp. There were no visible differences between 500 MV and 50 GV traces (not shown) and the patterns of a7-mediated single channel openings recorded by the two electrodes remained identical: 100% of events detected by 50 GV amplifier were also detected by 500 MV amplifier and vice versa (not shown). Therefore, in whole-cell mode, the default feedback resistance (500 MV) provided a sufficient sensitivity for detecting as many a7 single channel events as does 50 GV feedback resistance and thus, reliable recordings of individual a7 channel activity in whole-cell did not require increasing R f .

Drugs
In this study, 1 mM PNU-120596 was used. This concentration lies near the EC50 for potentiating effects of PNU-120596 in heterologous systems (EC50,1. Choline-containing solutions were freshly made prior to each experiment from a 1 M stock which was kept at 220uC.

Analysis
Only current-clamp data were used for automated analysis of ion channel activity (hypothalamic TM neurons, n = 16 and hippocamal CA1 interneurons, n = 8) using Clampfit-10 software, however all available data were visually analyzed in both current-and voltage-clamp modes. Both visual and automated analyses resulted in identical conclusions for both types of neurons tested. For automated analysis, two approaches were used. The first approach evaluated the standard deviation (S.D.) of the original and subtracted voltage traces: voltage traces recorded by electrode #2 were subtracted from traces recorded by electrode #1 using ''Arithmetic'' command (Clampfit-10) and then the values of S.D. of the original and subtracted traces were obtained using ''Statistics'' command (Clampfit-10). The second approach evaluated cross-correlations of the original voltage traces, estimated the correlation coefficient (R) and the probability that the estimated values of R were obtained by chance (i.e., p-value). The cross-correlation of traces was evaluated using ''Crosscorrelation'' command with a 610 ms lag. The value of R corresponding to 0 ms lag was chosen for statistical analysis and the t-value was determined from R as previously described [11]. The p-value was then calculated using its definition [11]

Hypothalamic TM neurons
Large histaminergic TM neurons were identified in coronal hypothalamic slices on the basis of their size and morphology, location within the slice, and the expression of high densities of functional a7 nAChRs [12] as well as strong I h and I A currents [13]. To activate a7 nAChRs, hypothalamic slices were perfused with ACSF containing 5 mM choline and 1 mM PNU-120596. In two experiments, 10 mM choline was used. To detect a7 singlechannel events in whole-cell current-clamp experiments, spontaneous firing was inhibited by keeping the membrane voltage near 270 mV by injecting continuous hyperpolarizing currents (50-120 pA). The inter-patch distances were made as large as possible and images of the recorded TM neurons were taken during each experiment. The inter-patch distances were then measured off-line ( Figures 1A and 2A, open circles). The average inter-patch distance was 21.062.9 mm (n = 16).
Reliability of physiological responses. The reliability of data obtained from the two recording electrodes was tested in each experiment: one of the electrodes was transitioned into currentclamp ( Figure 1B, top trace) or voltage-clamp ( Figure 1C, top trace), while the other electrode remained cell-attached ( Figure 1B, C, bottom traces). Action potentials recorded by the electrode held in current-clamp ( Figure 1B, top trace) were detected as synchronous membrane capacitance transients by the electrode held cell-attached ( Figure 1B, bottom trace). By contrast, current deviations corresponding to a7 single channel openings ( Figure 1C, close arrow) and spontaneous synaptic activity ( Figure 1C, open arrows) recorded by the electrode held in voltage-clamp were undetected by the electrode held cell-attached ( Figure 1C, bottom trace). These data demonstrate that the two recording electrodes did not cross-talk and voltage/current events recorded by one electrode were not relayed to another electrode by any means except directly from the recorded neuron.  Step-like deviations recorded in hypothalamic TM neurons were completely and reversibly blocked by 20 nM methyllycaconitine (MLA), a selective antagonist of a7 nAChRs (n = 4, Figure 2D, E, F).
Analysis of standard deviations. Subtraction of traces recorded by electrode #2 (Figures 2B 1 and 2C 1 , bottom traces) from the corresponding traces recorded by electrode #1 ( Figures 2B 1 and 2C 1 Figures 2B 1 and 2B 2 ). These results were consistent with the results of visual analysis indicating a lack of mismatch in the patterns of single a7 ion channel openings observed in dual-patch current-clamp (n = 14, Figure 2B 1 ) and voltage-clamp (n = 16, Figure 2C 1 ) experiments. To evaluate differences in the values of SD 1 , SD 2 and SD 3 , a one-way ANOVA test and the Bonferrani post-test were used. The values of SDs were found to be significantly different (F(2,39) = 31.16, p,0.0001). The Bonferonni post-test determined that differences between SD 1 and SD 2 were statistically insignificant (p.0.05), while differences between SD 1 and SD 1-2 as well as SD 2 and SD 1-2 were statistically significant (p,0.05).
Analysis of cross-correlations. Episodic (100 s long) voltage traces recorded from the same TM neuron by two patch electrodes were analyzed to determine the degree of identity (i.e., cross-correlation) among individual a7 ion channel events. As a result, the correlation coefficient was determined for each pair of voltage traces. Figures 3A, B illustrate typical examples of wholecell voltage traces recorded from a TM neuron by two patch electrodes ( Figure 3A) and the values of correlation coefficient as a function of time lag (in ms) obtained in the same experiment ( Figure 3B). The correlation coefficient corresponding to a 0 ms lag was then used for estimation of the t-and p-values (see Methods). In the example shown in Figure 3B, the correlation coefficient R = 0.9993 corresponding to lag = 0 ms is marked by a small dot at the intersection of dashed lines. The mean correlation coefficient evaluated over all current-clamp experiments was R = 0.99660.005 (n = 14) indicating an extremely strong correlation between pairs of voltage traces ( Figure 3A). This correlation was found to be highly significant (p,0.0001, n = 14, see Methods).

Equipotential somatic space in hypothalamic TM
neurons. To evaluate potential errors in clamping the membrane voltages (i.e., space-clamp) related to large (,20 mm) inter-patch distances ( Figures 1A and 2A), in five experiments, one of the two electrodes was always held in current-clamp, while the second electrode switched between voltage-and current-clamp modes. In these experiments, the average inter-patch distance was 21.7063.76 mm (n = 5). When the second electrode was held in voltage-clamp ( Figure 3C, bottom trace), the mean amplitude of step-like voltage deviations recorded by the first electrode held in current-clamp was 0.2260.10 mV (n = 5, Figure 3C, top trace) and thus, of the same magnitude as SD 1-2 . By contrast, when both electrodes were held in current-clamp, the mean amplitude of step-like voltage deviations was 2.8760.64 mV (n = 5, Figure 3A), i.e., a .10-fold greater value. These results demonstrate that in TM neurons, the somatic space-clamp error is small (,10%) and develops as a slow function of the somatic distance (,0.01 mV/ mm). These data present the TM somatic space as nearly equipotential.

Hippocampal CA1 interneurons
To determine whether the observed effects are pertinent to other a7 nAChR-expressing neurons, the identical dual-patch experimental protocol was applied to hippocampal CA1 interneurons expressing functional a7 nAChRs [5,14] with single-channel activity observable in whole-cell patch-clamp experiments [5]. The results obtained from hippocampal CA1 interneurons and hypothalamic TM neurons were similar. Interneurons were identified on the basis of their morphology and location within the CA1 Stratum Radiatum region of the hippocampus. To activate a7 nAChRs, hippocampal slices were perfused with ACSF containing 5 mM choline and 1 mM PNU-120596. In four experiments, 10 mM choline was used. To detect a7 singlechannel events in whole-cell current-clamp experiments, occasional spontaneous action potentials were prevented by keeping the membrane voltage near or below 270 mV by injecting continuous hyperpolarizing currents (50-150 pA). Similar to the case of hypothalamic TM neurons, the inter-patch distances were made as large as possible to introduce the largest electrical resistance between the two patches. However, hippocampal CA1 interneurons were generally smaller than TM neurons and therefore, the average inter-patch distance for interneurons was shorter. Images of the recorded hippocampal CA1 interneurons were taken during each experiment and the inter-patch distances were measured ( Figures 4A, open circles). The average inter-patch distance was 14.761.6 mm (n = 8).  p,0.01). The Bonferonni post-test determined that differences between SD 1 and SD 2 were statistically insignificant (p.0.05), while differences between SD 1 and SD 1-2 as well as SD 2 and SD 1-2 were statistically significant (p,0.05). As in the case of hypothalamic TM neurons, step-like deviations recorded in hippocampal CA1 interneurons were completely and reversibly blocked by 20 nM MLA (n = 4, Figure 4D, E, F). Figures 5A, B illustrate examples of typical voltage traces recorded from a hippocampal CA1 interneuron by two patch electrodes ( Figure 5A) and the value of correlation coefficient as a function of time lag (in ms) obtained in the same experiment ( Figure 5B). In this example, the correlation coefficient R = 0.9991 corresponding to lag = 0 ms is marked by a small dot at the intersection of dashed lines ( Figure 5B). Analysis of crosscorrelations among events recorded by the two patch electrode indicated an extremely strong correlation between pairs of voltage traces with the mean correlation coefficient, ,R. = 0.99260.009 (n = 8). This correlation was found to be highly significant (p,0.0001, n = 8).

Equipotential somatic space in hippocampal CA1
interneurons. Similar to the case of hypothalamic TM neurons, the space-clamp error in hippocampal CA1 interneurons was found to be small (,10%). When both electrodes were held in current-clamp ( Figure 5A), the mean amplitude of step-like voltage deviations in interneurons was 2.4661.05 mV (n = 5). By contrast, when one of the electrodes was held in current-clamp ( Figure 5C, top trace) and the second electrode was held in voltage-clamp ( Figure 5C, bottom trace), the mean amplitude of voltage deviations in interneurons was 0.1260.07 mV (n = 5) and thus, similar to the magnitude of SD 1-2 . In these experiments, the average inter-patch distance was 14.261.6 mm (n = 5). These results demonstrate that as in the case of hypothalamic TM neurons, in hippocampal CA1 interneurons the somatic space-clamp develops as a slow function of the somatic distance (,0.01 mV/mm). Therefore, the somatic space of hippocampal CA1 interneurons also appears to be nearly equipotential.

Discussion
A moderate persistent activity of highly Ca 2+ -permeable a7 nAChRs can be neuroprotective [3,6,7,8,9] and may be achieved by endogenous levels of choline enhanced by a7-PAMs (e.g., PNU-120596) [9]. These conditions can be modeled in whole-cell current-and voltage-clamp experiments in brain slices, while quantification of a7-mediated persistent ionic influx integrated by neuronal somata could provide a rational basis for optimization of the levels of a7 nAChR activation that promotes neuronal survival and function. Likewise, the results of this study may help to define conditions that reduce cytotoxicity, a key concern in the use of Type-II a7 PAMs, because these compounds robustly enhance a7 activation and Ca 2+ influx [10].
In the presence of physiological choline (,5-10 mM) and 1-2 mM PNU-120596 in ACSF, persistent openings of individual a7 ion channels can be detected in whole-cell patch-clamp experiments in hypothalamic and hippocampal slices [4,5]. Under these and similar conditions, the extent of a7-mediated persistent ionic influx may act as a key determinant of the likelihood of neuronal survival and function [10]. However, quantification of this influx remained elusive because activation of individual a7 channels may not be integrated across the neuronal soma and reliably detected in whole-cell patch-clamp experiments. A reliable quantification of individual a7 channel activity in whole-cell recordings requires neuronal soma to act as an effective integrator, allowing an individual a7 channel to depolarize the entire soma and thus, be detected by a patch-clamp electrode regardless of the electrode location on the somatic membrane. By contrast, somatic electric compartmentalization may obstruct somatic integration and detection of a7-mediated step-like voltage/current deviations and lead to errors in estimation of a7-mediated somatic ionic influx and net charge recorded in whole-cell experiments in the presence of nicotinic agonists and a7-PAMs. It would also make the patterns of a7 single-channel events depend on the location of recording electrode on the neuronal surface. Such a dependence was not observed in this study, as in all sixteen experiments with hypothalamic TM neurons and eight experiments with hippocampal CA1 interneurons, the patterns of a7 ion channel activity recorded from a single neuron by two electrodes were identical ( Figures 2B, C, 3A, B, C, 4B, C and 5A, B, C). Moreover, the results of simultaneous voltage-and current-clamp recordings from a single neuron ( Figure 3C and 5C) revealed that the somatic space-clamp error in both TM neurons and CA1 interneurons is minimal (,10%) and the accumulation of voltage error was a slow function of somatic distance (,0.01 mV/mm). Therefore, these results do not support the presence of somatic compartmentalization in hypothalamic TM neurons and hippocampal CA1 interneurons and confirm the hypothesis that, in the presence of PNU-120596 and possibly other a7-PAMs, activation of an individual a7 ion channel depolarizes the entire soma and is detected and integrated across the entire soma.
These conclusions are supported by previous observations that the amplitudes of a7-mediated step-like currents in voltageclamp (i,4-5 pA) and voltages in current-clamp (V,8-10 mV) in TM neurons comply with the Ohm's law (i = V/r), where the values of r fall near the values of TM input resistance (,500 MV) [4]. Therefore, a7 single-channel currents elicit voltage deviations that recharge the majority of the somatic membrane and not a small segment of it. In the latter case, the value of r would have been proportionally smaller, indicating the presence of electrical compartmentalization in the vicinity of the recording electrode. Furthermore, in experiments with hippocampal CA1 pyramidal neurons in the presence of 2 mM PNU-120596 and 10 mM choline, all recorded action potentials were generated by clearly detected a7-mediated step-like depolarizations [5]. These observations argue against the presence of single a7 channel activity undetected in whole-cell recordings because such an undetected activity would have produced randomly occurring action potentials not associated with a7-mediated steplike depolarizations clearly observed in these current-clamp recordings. The present study directly demonstrates that in the presence of PNU-120596, the patterns of activity of individual a7 nAChRs as seen by the soma are independent of the somatic position of electrodes and therefore, despite their large sizes, somata of hypothalamic TM neurons (.20 mm) and hippocampal CA1 interneurons (.14 mm) act as nearly equipotential volumes and effective signal integrators. This conclusion justifies and allows evaluation of a7-mediated persistent ionic and Ca 2+ influx integrated by somata of hypothalamic TM neurons and hippocampal CA1 interneurons under various experimental conditions [4]. Similar conclusions may apply to other central neurons. Although it appears beneficial for neurons to express functional a7 nAChRs, as moderate activation of these highly Ca 2+permeable receptor-channels can be therapeutic [1,3,6,7,8], inadequate (i.e., deficient or extensive) activation of a7 nAChRs may cause neuronal dysfunction, damage and death. Specifically, a prolonged activation of a7 nAChRs by endogenous choline and/ or ACh in the presence of a7-PAMs may be cytotoxic and may disrupt neuronal circuitry. Therefore, the search for an optimum in activation of a7 nAChRs and Ca 2+ entry is critical in designing therapeutic approaches aimed at supporting neuronal survival and function [9]. The results of this study suggest that the rate of a7mediated persistent ionic influx and its Ca 2+ component integrated by the soma can be reliably estimated in conventional whole-cell recordings in the presence of PNU-120596 in hypothalamic TM neurons and hippocampal CA1 interneurons because electrotonic properties of these cells do not impede somatic signaling. These, results may help to reveal and quantify conditions that promote neuroprotection. Once the rate of a7 persistent ionic influx is determined, a correlation between this rate and the level of neuroprotection can be established and optimized for various subtypes of a7-expressing neurons.
Taken together, this and previous studies demonstrated that in the presence of physiological choline (i.e., 5-10 mM) and 1-2 mM PNU-120596, patch-clamp whole-cell experiments can provide reliable estimates of the number of simultaneously open a7 channels (N total P open ,0.27), a7-mediated persistent ionic influx (,1.4 pA) and Ca 2+ influx (,0.14 pA) integrated by the TM soma [4]. Similar estimates performed in hippocampal CA1 pyramidal neurons [5] indicated a .10-fold weaker expression of functional pyramidal a7 nAChRs compared to hippocampal CA1 interneurons and the TM. The ability to evaluate and manipulate the level of persistent activation of a7 nAChRs by physiological choline in the presence of a7-PAMs may reveal important relationships between the extent of a7-mediated somatic persistent ionic influx and neuronal survival and function in the search for a Ca 2+ optimum [9]. By contrast, in the absence of PNU-120596, openings of individual a7 channels cannot be distinguished from noise in whole-cell recordings and thus, cannot be reliably detected, quantified or optimized. Therefore, these observations suggest an intriguing possibility of establishing and optimizing a quantitative relationship between the level of neuroprotection by nicotinic agonists in the presence of Type-II a7-PAMs and the extent of a7-mediated somatic persistent ionic and Ca 2+ influx evaluated within the relevant ex vivo models.