Functional profiling of neurons through cellular neuropharmacology

Abstract

We describe a functional profiling strategy to identify and characterize subtypes of neurons present in a peripheral ganglion, which should be extendable to neurons in the CNS. In this study, dissociated dorsal-root ganglion neurons from mice were exposed to various pharmacological agents (challenge compounds), while at the same time the individual responses of >100 neurons were simultaneously monitored by calcium imaging. Each challenge compound elicited responses in only a subset of dorsal-root ganglion neurons. Two general types of challenge compounds were used: agonists of receptors (ionotropic and metabotropic) that alter cytoplasmic calcium concentration (receptor–agonist challenges) and compounds that affect voltage-gated ion channels (membrane–potential challenges). Notably, among the latter are K-channel antagonists, which elicited unexpectedly diverse types of calcium responses in different cells (i.e., phenotypes). We used various challenge compounds to identify several putative neuronal subtypes on the basis of their shared and/or divergent functional, phenotypic profiles. Our results indicate that multiple receptor–agonist and membrane–potential challenges may be applied to a neuronal population to identify, characterize, and discriminate among neuronal subtypes. This experimental approach can uncover constellations of plasma membrane macromolecules that are functionally coupled to confer a specific phenotypic profile on each neuronal subtype. This experimental platform has the potential to bridge a gap between systems and molecular neuroscience with a cellular-focused neuropharmacology, ultimately leading to the identification and functional characterization of all neuronal subtypes at a given locus in the nervous system.

For any mechanistic investigation of nervous system function, it is essential to identify the neurons involved. However, this identification can be extremely challenging, because at any given locus in the mammalian nervous system, many functionally diverse neurons can be present that are difficult to discriminate from each other. This lack of cellular differentiation creates a barrier between systems and molecular neuroscience. Systems neuroscientists characterize properties of circuits, whereas molecular neuroscientists identify the signaling macromolecules that give neurons specific functional properties. Within invertebrate nervous systems, there is a precedent for identifying different types of neurons morphologically, anatomically, and physiologically, known as the “identified neuron approach” (1). Progress in understanding the mammalian nervous system has been impeded by the lack of a similar paradigm. The gap between systems and molecular neuroscience in the mammalian nervous system could be narrowed if there were a straightforward methodology to identify different neuronal subtypes.

We define neuronal subtype as a neuronal cell with a specific physiological function in contrast to a neuronal subpopulation or neuronal subclass, which may encompass multiple neuronal subtypes and be identified by the use of a single structural or functional marker. For example, dorsal-root ganglion (DRG) neurons may be subdivided into neuronal subclasses defined by staining with fluorescently labeled isolectin B₄ (IB4), which binds to the extracellular matrix proteoglycan, versican, present on the plasma membranes of a subset of relatively small neurons. However, neither the IB4-positive nor -negative subclass is homogeneous (e.g., both subclasses include capsaicin-sensitive and -resistant neurons) (2). A single marker is rarely unique to a specific neuronal subtype (3–5), particularly at a complex anatomical locus, where potentially hundreds of neuronal subtypes with different physiological roles may be present.

Although the mammalian nervous system has been studied intensively at the cellular level for decades, there is no anatomical locus where all of the neuronal subtypes have been identified. From a long tradition of anatomical studies pioneered by Santiago Ramón y Cajal (6), neurons that have been most intensively investigated, such as Purkinje and pyramidal cells, are those cells easily recognized by their striking cell shapes. However, even morphologically similar pyramidal cells have been classified into different subclasses based on different firing properties (7). In most regions of the nervous system, the vast majority of neurons are not easily differentiated morphologically, further complicating the task of parsing out a specific neuronal subtype from the surrounding anatomically similar neurons.

Most efforts to differentiate neuronal subtypes broadly use markers of mRNA or protein expression (3–5, 8–11). Only a few markers can be used simultaneously, and the expression of mRNA, or even protein, may not correlate with functional expression; however, functional expression is the critical parameter for any mechanistic or physiological study. The principle method for assaying neuronal function has been patch-clamp electrophysiology. However, it is severely limited by throughput; experiments are usually conducted on one neuron at a time.

In this report, we identify different neuronal subtypes using an experimental strategy that overcomes many of the limitations of other methods. For this study, we applied pharmacological agents (challenge compounds) to dissociated mouse lumbar DRG neurons, while monitoring the responses of >100 individual neuronal cells simultaneously by calcium imaging. Within DRG, >25 subtypes of neurons are believed to be present based on different sensory modalities. The divergent responses of individual cells to each challenge compound served as the primary criteria for distinguishing between neuronal subtypes. The rationale is that different neuronal subtypes express different receptors and ion channels in their plasma membranes, which create functional divergence. The cell bodies of DRG neurons generally are good surrogates for functional protein expression in axons and nerve endings based on consistency of responses to selective pharmacological agents obtained in cell bodies, nerve fibers, nerve endings, and behavioral studies in vivo with WT and KO mice (12).

We have used an established technology, calcium imaging, to profile neuronal subtypes. Although this approach is not unprecedented, typically only a few pharmacological agents have been used to profile neuronal subtypes in a single experiment (13–15). We show the feasibility of applying many challenge compounds in a single experiment, and we have discovered that certain types of challenge compounds elicited a far greater spectrum of phenotypic responses than predicted. The successful application of many challenge compounds, coupled with the unexpected diversity of response phenotypes, establishes this experimental strategy as a powerful approach to distinguish between neuronal subtypes in a heterogeneous cell population. Using this experimental approach, we highlight a few unambiguous examples of neuronal subclasses in DRG cultures. Some of these neuronal subclasses probably conform to our narrow definition of a neuronal subtype (with a shared, specific physiological function), because each neuron within the subclass shares a common structural and functional profile (including both positive and negative markers) that clearly distinguishes it from other neuronal subclasses in the DRG cultures.

Results

Preparation and Calcium Imaging of Dissociated Mouse Lumbar DRG Neurons.

Fig. 1 shows bright-field and fluorescence images of cultured DRG neurons loaded with Fura-2-acetoxymethyl ester (Fura-2-AM) and imaged as described in Materials and Methods. The wide range of cell diameters observed in Fig. 1A is consistent with previous reports (16). Also shown are ratiometric images acquired before and during exposure to a high concentration of potassium (Fig. 1 C and D).

Fig. 1.

Images of dissociated mouse lumbar DRG neurons loaded with Fura-2-AM dye. A–D are images of the same field of view. Fluorescence images were acquired as described in Materials and Methods. (A) Bright-field image. (B) Fluorescence image acquired with 380-nm excitation and 510-nm emission filters. (C) Pseudocolored ratiometric calcium image obtained under control conditions (i.e., before depolarization). The color scale indicates that the resting cytoplasmic calcium concentration is relatively low (magenta and blue). (D) Pseudocolored ratiometric image obtained on depolarization of the neurons by 100 mM KCl. The color scale, same as the color scale in C, indicates that the cytoplasmic calcium concentration is relatively high in many cells (green, yellow, and red). (Scale bar: 30 μm.) Glial cells did not respond to 100 mM KCl with elevated cytoplasmic calcium, presumably because they lack voltage-gated calcium channels.

Cultured DRG neurons, similar to those neurons shown in Fig. 1, were challenged using two types of experimental protocols. The first type of protocol used agonists of receptors (ionotropic and metabotropic) previously reported to produce an increase in cytoplasmic calcium concentration, [Ca²⁺]_i, of DRG neurons. We refer to this experimental strategy as a receptor–agonist challenge (RA challenge). The second type of protocol used pharmacological agents targeted to voltage-gated ion channels, which perturb the membrane potential. A membrane depolarization that activates voltage-gated Ca channels will produce an increase in [Ca²⁺]_i. Thus, compounds that act on voltage-gated ion channels may attenuate or enhance the increase in [Ca²⁺]_i elicited by a membrane depolarization. We refer to this experimental strategy as a membrane–potential challenge (MP challenge).

RA Challenge Protocol.

With ionotropic receptors (ligand-gated ion channels), RA challenge compounds directly induce the influx of Ca²⁺ through Ca²⁺-permeable channels [e.g., transient receptor potential channel (TRP) V1 receptor] (17). With metabotropic receptors (G protein-coupled receptors), RA challenge compounds indirectly elevate cytoplasmic Ca²⁺ by downstream signaling pathways that ultimately trigger the release of Ca²⁺ from the endoplasmic reticulum (e.g., histamine receptors) (18–20). Six different RA challenge compounds were sequentially applied to DRG cultures, and responses from selected neurons are shown in Fig. 2 A–C.

Fig. 2.

Example calcium-imaging traces from RA challenges. Each trace is a single neuron's response to the challenge compounds indicated at the bottom of each panel. The x axis is the same for all traces in a given panel. The y axis (same for Figs. 4–6) is a relative measure of [Ca²⁺]_i determined by the 340/380 nm excitation ratio described in Materials and Methods. Challenge compounds were ACh, 1 mM acetylcholine; ATP, 10 μM adenosine 5′-triphosphate; AITC, 200 μM allyl isothiocyanate; C, 300 nM capsaicin; H, 50 μM histamine; M, 200 μM menthol; K, 100 mM KCl (A–C) or 25 mM KCl (D); and TEA, 10 mM tetraethylammonium chloride. Arrows indicate when each compound was applied to the bath. (A–C) Typical sequence of RA challenges. Challenge compounds were applied at 5-min intervals for ∼15 s. ACh was applied two times to show reproducibility of responses. KCl (100 mM) was applied at the end of the series, and therefore, nonresponsive cells could be excluded from additional analysis. The maximum response to KCl was cropped for some traces, and therefore, heights of lesser peaks would be more evident. (C and D) ATP, menthol, and AITC were used to differentiate between neurons only sensitive to menthol and neurons sensitive to all three challenge compounds, which are highlighted in blue. (D) Responses from one neuron of each type are shown. Neurons were also depolarized by KCl (25 mM) pulses at 7-min intervals before and after application of 10 mM TEA (indicated by horizontal bar), a blocker of voltage-gated K channels, to compare TEA with KCl-elicited responses. The red color is for emphasis only. It highlights the different types of responses to TEA.

We used the protocol shown in Fig. 2 A–C and compiled data for the cellular responses of 2,026 mouse lumbar DRG neurons (Table 1). The fraction of cells that responded varied considerably from one challenge compound to the next challenge compound. A large proportion of the cells responded to capsaicin (39%), mustard oil [allyl isothiocyanate (AITC; 32%)], or ATP (76%), but only a minor fraction (<10%) responded to menthol, histamine, or acetylcholine (ACh) under these experimental conditions. We refer to experiments using the class of compounds that activate only a minor fraction of the DRG cells as diagnostic RA challenges.

Table 1.

Percentage of mouse lumbar DRG neurons that responded to RA challenge compounds

Challenge compound	Number of responsive cells	Cells (%)
ACh (1 mM)	158	7.8
ATP (10 μM)	1,536	75.8
Histamine (50 μM)	131	6.4
Menthol (200 μM)	178	8.8
AITC (mustard oil; 200 μM)	651	32.1
Capsaicin (300 nM)	791	39.0
KCl (100 mM)	2,026	100.0

Only data from viable neurons, defined by their ability to respond to 100 mM KCl at the end of the trial, are presented in this table and all subsequent tables.

Definition of DRG Neuron Subclasses Using Diagnostic RA Challenges: Histamine, ACh, and Menthol.

We used diagnostic RA challenges as a first step to identify a neuronal subtype. DRG neuron subclasses that responded to diagnostic RA challenge compounds could be further subdivided by their responses to other challenge compounds. We provide a few examples below. One example is the small percentage of cells that responded to histamine. Approximately one-half of the histamine-responsive neurons had the same phenotypic profile: these cells did not respond to the other diagnostic RA challenge compounds (ACh and menthol) or AITC but did respond to capsaicin and ATP (Fig. 2B and Table 2). The average area of the cell soma (cell area) for this class of neurons, 240 μm², was relatively small.

Table 2.

Selected subclasses of mouse lumbar DRG neurons defined by RA challenges

Diagnostic RA challenge compound			Profiling compounds
ACh	Menthol	Histamine	Cap	AITC	ATP	Distinctive functional phenotype*	Average cell size (μm2)	Number of cells†
+	−	−	−	−	−		670	28
+	−	−	+	−	+		270	30
−	+	−	−	−	−	+	150	33
−	+	−	−	+	+		270	24
−	−	+	+	−	+		240	65

*This phenotype is characterized by highly variable resting [Ca²⁺]_i.

^†Each neuron subclass represents a very small proportion (1.2–3.2%) of the total of 2,026 neurons examined.

One putative subtype of ACh-responsive cells did not respond to any of the other five challenge compounds (Fig. 2A and Table 2). Furthermore, none of those cells stained with Alexa-fluor-568–labeled IB4 (Fig. S1 and Table 3). A different subclass of ACh-responsive cells responded to capsaicin and ATP but not to menthol, histamine, or AITC (Fig. S1 and Table 2), and the majority of these cells stained with IB4 (Fig. S1 and Table 3). The ACh-sensitive cells that did not respond to any other challenge compound were predominantly large cells (average cell area = 670 μm²); in contrast, the subset of ACh-responsive cells that were also capsaicin- and ATP-sensitive was substantially smaller (average cell area = 270 μm²) (Fig. 3 and Table 2).

Table 3.

Seven selected subclasses of mouse lumbar DRG neurons defined by RA and/or MP challenges

DRG neuron subclass	Examples	Average cell size (μm2)	Positive functional markers (implicated protein expression)	Negative functional markers	IB4 stain	Cultured DRG neurons (%)
Large
1	Fig. 6	680	Direct κM-RIIIJ+* (KV1.2)	ACh−, pl14a−	IB4−	2.3
			Direct Dtx-K+ (KV1.1)
2	Figs. 2A, 3A, and 6A	670	ACh+ (AChRs)	Men−, Hist−, Cap−,  AITC−, ATP−	IB4−	1.4
Medium
3	Fig. S2	480	Indirect κM-RIIIJ+† (KV1.2)		IB4+/−	4.1‡
Small
4	Fig. 3B and S1	270	ACh+ (AChRs)	Men−, Hist−, AITC−	IB4+/−	1.5
			Cap+ (TRPV1)
			ATP+ (P2X/Y)
5	Fig. 2B	240	Hist+ (HistRs)	ACh−, Men−, AITC−	IB4+/−	3.2
			Cap+ (TRPV1)
			ATP+ (P2X/Y)
6	Fig. 2 C and D	270	Men+ (TRPM8)	ACh−, Hist−, Cap−	IB4+	1.2
			AITC+ (TRPA1)
			ATP+ (P2X/Y)
			TEA sustained response
7	Figs. 2 C and D and 3C	150	Men+ (TRPM8)	ACh−, Hist−, Cap−,  AITC−, ATP−	IB4−	1.6
			TEA spiky response

ACh, 1 mM acetylcholine; AChRs, acetylcholine receptors; AITC, 200 μM allyl isothiocyanate; ATP, 10 μM adenosine 5′-triphosphate; Cap, 300 nM capsaicin; Dtx-K, 100 nM Dendrotoxin-K; Hist, 50 μM histamine; HistRs, histamine receptors; IB4, isolectin B4; IB4^+/−, mix of IB4+ and IB4−; κM-RIIIJ, 1 μM κM-conopeptide RIIIJ; K_V1, K_V1 potassium channels; Men, 200 μM menthol; P2X/Y, P2X or P2Y receptors; pl14a, 16 μM conopeptide pl14a; TEA, 10 mM tetraethylammonium chloride; TRP (A1, M8, and V1), transient receptor potential channels; +, responsiveness to the compound; −, lack of responsiveness to the compound.

*κM-RIIIJ directly elicited a sustained calcium signal.

^†κM-RIIIJ indirectly amplified the high [K⁺]_o-elicited response. Indirect effects by κM-RIIIJ were scored if there was an increase in the [K⁺]_o-elicited peak height (calcium signal) >3 SDs over the peak height of the average [K⁺]_o-elicited peak height (control calcium signal) before application of κM-RIIIJ.

^‡Only medium-sized neurons with cell areas between 300 and 600 μm² were included in this percentage. However, the calculation for the average cell area of 480 μm² for subclass 3 included neurons of all diameters that responded indirectly to κM-RIIIJ.

Fig. 3.

Size distributions of neurons responsive to select RA challenge compounds. Data are shown for only a few neuronal subclasses. The x axis is the same for all panels. Cell area is the area of a cross-section of the cell soma. (A) ACh only refers to neurons that responded to ACh but not to any other RA challenge compounds. (B) ACh + ATP + Cap refers to neurons that responded to ACh, ATP, and capsaicin but no other RA challenge compounds. (C) Menthol only refers to neurons that responded to menthol but no other RA challenge compound.

Menthol-sensitive DRG neurons could also be subdivided on the basis of their responsiveness to other compounds. For example, a subset of menthol-sensitive DRG neurons was also sensitive to capsaicin, which other investigators have observed (13, 15, 21–23). A different putative DRG subtype responded exclusively to menthol (Fig. 2C and Table 2) and did not stain with IB4 (Table 3). On average, they were exceptionally small cells (average cell area = 150 μm²) (Table 2) that displayed unusually noisy and variable resting [Ca²⁺]_i (Fig. 2C). Part of the variability was characterized by a transient dip in the [Ca²⁺]_i baseline each time the static bath solution was replaced with a bath solution containing an RA challenge compound, with the exception of menthol or KCl, which elicited increases in [Ca²⁺]_i (Fig. 2C). The size distribution of this class of menthol-sensitive cells is shown in Fig. 3. Another putative DRG subtype responded to menthol, AITC, and ATP (Fig. 2C and Table 2) and stained with IB4 (Table 3). On average, they were larger cells than those cells that responded only to menthol (average cell area = 270 μm²) (Table 2).

MP Challenge Protocol.

Each neuron has molecular components with activation that promotes depolarization (e.g., voltage-gated Na and Ca channels) or hyperpolarization (e.g., voltage-gated K channels). On application of a depolarizing stimulus, the increase in [Ca²⁺]_i is determined by a balance between the actions of these two sets of components. Thus, the application of an MP challenge compound can be used to assess whether particular voltage-sensitive ion channels are functionally expressed in the plasma membrane of a neuron by monitoring a decrease or increase in the magnitude and/or kinetics of the calcium signal elicited on membrane depolarization (e.g., by elevating extracellular K⁺, [K⁺]_o).

Calcium signals were elicited by depolarizing neurons with 25-mM KCl pulses before and after application of an MP challenge compound, which is described in Materials and Methods and shown in Fig. 4. To assess how different DRG cells respond to various MP challenge compounds, we first applied two classical pharmacological agents: tetrodotoxin (TTX; it inhibits a broad spectrum of voltage-gated Na channels) and tetraethylammonium (TEA; a standard wide-spectrum, voltage-gated K-channel blocker). As expected, they produced opposite effects: TTX decreased the calcium signal in response to a standard KCl pulse, whereas TEA increased it, as described below.

Fig. 4.

MP challenge protocol illustrated with calcium-imaging traces. Each trace represents a single neuron's response. Each peak in a trace is the response to a 25-mM KCl pulse applied at 7-min intervals. The duration of each KCl pulse (∼15 s) is indicated by the vertical bars (1–8). The first four KCl pulses (1–4) served as internal controls to monitor the variability in the elicited calcium signals. The last three KCl pulses (6–8) allowed us to monitor the reversibility of a compound's effects. The compound of interest was applied at minute 23 and remained in the bath for 6 min, which is indicated by the horizontal bar in Upper. The fifth KCl pulse, at minute 29 (peak 5), shows that the compound caused an amplification of the high [K⁺]_o-elicited calcium signal, which was readily reversible (indicated by peaks 6–8). (Lower) Vehicle trials, in which only observation solution was applied to the bath at minute 23, served as external controls. The red color is for emphasis only. It highlights the difference between a response to a compound (Upper; compound trial) and a control without a compound (Lower; vehicle trial). At the end of the experiment, at minute 57, 300 nM capsaicin was applied for 1 min (black circles).

Block by TTX.

The reduction by TTX of the high [K⁺]_o-induced calcium signal varied from cell to cell, which is shown in Fig. 5 A–D. We expected the application of KCl (25 mM) to activate voltage-gated Na channels by moderately depolarizing the cells. Our hypothesis was that the opening of voltage-gated Na channels would be required to further depolarize the membrane sufficiently to activate high-voltage–activated Ca channels. However, TTX significantly reduced (by >3 SDs below the mean internal control value) the height of each peak in only a small subset of cells (<20%), with the high [K⁺]_o-induced calcium signals of most cells largely nonresponsive to TTX (Fig. 5E).

Fig. 5.

Selected calcium-imaging traces from MP challenges. Each trace represents a single neuron's response. These data were obtained by using the MP challenge protocol exemplified by Fig. 4; x axis is the same for all traces, and the red color is for emphasis only. It highlights the different types of responses to each challenge compound. (A–E) Effects of 1 μM TTX, which was applied for 6 min starting at minute 23 (indicated by horizontal bars). (F–J) Effects of 25 mM TEA, which was applied for 6 min starting at minute 23 (bars). At the end of the experiment, 300 nM capsaicin was applied for 1 min at minute 57 (black circles).

Diverse Responses to TEA.

TEA, at 25 mM, has been shown to block most of the TEA-sensitive (sustained) K currents in small-diameter rat DRG neurons without blocking A-type currents (24). In a separate study, 5 mM TEA applied to current-clamped, small-diameter rat DRG neurons depolarized the membrane, reduced the threshold for action potential generation, and extended the duration of an action potential (25). In view of these studies, we used TEA at both 25 and 5 mM concentrations.

A surprising variety of responses were induced by TEA. The same types of effects were elicited with either 25 or 5 mM TEA. Some DRG neurons responded directly to TEA addition with sustained, elevated [Ca²⁺]_i and/or repetitive [Ca²⁺]_i spikes of variable intermittency (Fig. 5 F and G, respectively). In addition to these direct effects, many DRG neurons responded to TEA only with an enhanced response to the subsequent KCl pulse (Fig. 5 H and I). The enhancement took the form of increased peak height, width, or both. In all cells, the effects of TEA were readily reversible. A subset of cells was unresponsive to TEA; their high [K⁺]_o-induced calcium signals were unaffected by TEA (Fig. 5J).

Definition of DRG Neuron Subclasses Using Diagnostic MP Challenges: κM-Conopeptide RIIIJ, Dendrotoxin-K, and Conopeptide pl14a.

As noted above for TEA, K-channel blockers may elicit different types of phenotypic responses in different cells. Some neurons responded directly to the challenge compound. This response typically occurred only in a small proportion of neurons and could be used to define subclasses in the same way as the diagnostic RA challenge compounds described above.

The addition of 1 μM κM-conopeptide RIIIJ (κM-RIIIJ), a conopeptide previously reported to be highly selective for K_V1.2-containing channels (26), caused a direct response with sustained, elevated [Ca²⁺]_i in a fraction of large-diameter cells (Fig. 6). Large-diameter neurons with definitive direct responses to κM-RIIIJ did not respond to ACh (Fig. 6A). Thus, they delineate a different subclass of cells from the ACh-responsive, large-diameter cells described above.

Fig. 6.

Selected traces from large-diameter neurons (cell area > 600 μm²) illustrating the effects of voltage-gated K-channel blockers. Each trace represents a single neuron's response. The x axis is the same for all traces in a given panel. Challenge compounds were ACh, 1 mM acetylcholine; C, 300 nM capsaicin; K, 25 mM KCl; pl14a, 16 μM conopeptide pl14a; RIIIJ, 1 μM κM-conopeptide RIIIJ; and Dtx, 100 nM Dendrotoxin-K. Arrows and bars indicate when each challenge compound or KCl was applied; the latter was applied at 7-min intervals except as noted. The red color is for emphasis only. It highlights differences in responsiveness to the challenge compounds used. (A) ACh was applied (at minute 1) to identify responders (blue is for emphasis only), after which time the KCl pulses were given before and after application of κM-RIIIJ (bar). These neurons were resistant to capsaicin, which is typical for large-diameter DRG neurons. (B) KCl pulses presented before and after application of κM-RIIIJ (first bar) and conopeptide pl14a (second bar). These neurons were resistant to capsaicin. (C) KCl pulses applied before and after application of κM-RIIIJ (first bar) and Dtx-K (second bar).

In addition to κM-RIIIJ, two other selective K-channel antagonists are reported to target voltage-gated K channels in the same subfamily (shaker or K_V1). One of these antagonists, a conopeptide from Conus planorbis (pl14a), is selective for K_V1.6 (27), whereas a kunitz domain small protein from mamba venom, Dendrotoxin-K (Dtx-K), is reported to be highly selective for K_V1.1 (28). In previous reports, the targeting selectivities of κM-RIIIJ, pl14a, and Dtx-K were assessed by tests on heterologously expressed homomeric channels (26–28).

Here, we compared the direct responses to pl14a and Dtx-K with the direct responses to κM-RIIIJ to evaluate the overlap between the neurons responsive to pl14a and Dtx-K with the large-diameter neurons that directly respond to κM-RIIIJ. For the large-diameter DRG neurons, the response to κM-RIIIJ was inversely correlated with the response to pl14a (Fig. 6B). For neurons in which κM-RIIIJ produced robust direct effects, pl14a produced no direct effects and vice versa. Even more striking was the fact that those cells that did respond directly to either agent had different phenotypes. Thus, as illustrated in Fig. 6B, the smooth, sustained rise in [Ca²⁺]_i observed with κM-RIIIJ was strikingly different from the spiky response to pl14a. In contrast, there was considerable overlap between the cells that responded directly to Dtx-K and the cells that responded directly to κM-RIIIJ (Fig. 6C). The direct response of most cells to Dtx-K was similar in phenotype to the response observed for κM-RIIIJ (i.e., a smooth, sustained increase in [Ca²⁺]_i).

Similar to the variability in responses to TEA, some DRG neurons responded to κM-RIIIJ only indirectly on elevation of [K⁺]_o (Fig. S2B). Such cells were larger in diameter, on average (∼480 μm²), than the histamine- or menthol-sensitive cells (average cell sizes ≤ 270 μm²) (Table 2). Thus, the indirect responses constitute another functional phenotype that also subdivides DRG neurons.

A noteworthy feature of experiments with Dtx-K was the differential reversibility observed among the responsive neurons (Fig. S3). In some neurons, the effect of Dtx-K was rapidly reversed on washout (Fig. S3C), whereas in other neurons, the toxin's effect was either slowly reversible or almost irreversible (Fig. S3 D and E).

Although a moderate fraction of the total number of cells responded to selective K-channel inhibitors, those cells that respond with characteristic functional phenotypes, which are illustrated in Figs. 5 and 6 and Figs. S2 and S3, comprise only a small, coherent subset of neurons with other correlated phenotypic properties. Accordingly, the experiments described above show that a specific type of phenotypic response to selective K-channel antagonists can be used as a diagnostic MP challenge.

Additional Characterization of DRG Neuron Subtypes by Combining RA and MP Challenges.

The approach that we used to define neuronal subclasses also provides opportunities for their further functional characterization. The menthol-sensitive cells described above can be easily identified in the population of DRG neuronal subtypes. In the experiment shown in Fig. 2D, ATP, menthol, and AITC were applied to differentiate between two putative subtypes of menthol-sensitive neurons. The menthol-sensitive subtype that was insensitive to other RA challenges was responsive to TEA, an MP challenge; furthermore, in each instance, a characteristic spiky response to TEA was observed. In contrast, a spiky response to TEA was never observed in the cells that were sensitive to menthol, ATP, and AITC. Instead, those cells responded either only indirectly upon high [K⁺]_o-induced depolarization or directly to TEA with a smooth, sustained increase in [Ca²⁺]_i (Fig. 2D).

Dtx-K, κM-RIIIJ, and pl14a were also tested after application of menthol. The menthol-sensitive neurons that did not respond to other RA challenges were also resistant to all of these MP challenges, suggesting that inhibition of a different subset of voltage-gated K channels (non-K_V1 family sensitive to TEA) is responsible for initiating the spiky response to TEA observed in Fig. 2D. Thus, after a neuronal subtype has been identified by a set of markers, it may be further characterized by additional RA and MP challenge compounds.

In Table 3, we summarize the data obtained for seven selected subclasses of DRG neurons that were defined by diagnostic RA and/or MP challenges, cell size, and IB4 staining. Some of these subclasses probably conform to our narrow definition of DRG subtype because of their highly consistent cross-correlation with multiple functional and structural markers.

Discussion

Functional Profiling.

This report describes experiments performed on mouse DRG neurons in which we used a variety of functional markers to define several distinctive neuronal subclasses. This analysis has parallels to the early anatomical characterization of neurons; the classes first recognized by Ramón y Cajal were those classes with the most striking morphology. Using our approach, DRG neuronal subclasses that exhibit the most distinctive functional properties are the easiest to define. Thus, we have chosen to highlight a few neuronal subclasses with clearly distinctive functional properties. A future objective is to use our experimental approach, coupled with cluster analysis, to develop a taxonomy of all neuronal subtypes within the DRG.

Based on cross-correlations of multiple markers, we have identified a few subclasses of DRG neurons, which are summarized in Table 3. These subclasses include one histamine-sensitive, two ACh-sensitive, and two menthol-sensitive as well as large-diameter neurons that responded directly to κM-RIIIJ and medium-diameter neurons that responded indirectly to κM-RIIIJ. These neurons comprise only a small minority (∼15% altogether) of all of the different cells in the culture. Some of the subclasses also showed highly consistent cross-correlations with several markers (e.g., consistent responses to functional markers and consistent IB4 staining). Thus, they probably conform to our narrow definition of neuronal subtypes, with specific common physiological functions. However, many more DRG subtypes remain to be identified, and all of the subtypes require additional characterization.

Menthol-Sensitive Neurons.

Two putative subtypes of menthol-sensitive neurons that we have described in this paper are distinguished from each other by strikingly divergent functional phenotypes. Only one subtype responded to AITC and ATP and stained with IB4, which we will call the MA+ subtype (i.e. positive responses to menthol, ATP and AITC). The other subtype was exceptionally small, IB4-negative, and did not respond to any other RA challenge compounds; these cells also displayed highly variable resting [Ca²⁺]_i (Fig. 2C). We will call this subtype the M+ subtype (i.e., exclusively menthol-positive).

The facile identification of the M+ and MA+ subtypes should allow us to begin to systematically characterize their cellular neurophamacology. For example, we can investigate the source of variability in [Ca²⁺]_i observed in the M+ subtype, which may be caused by activity of TRPM8 at room temperature. Traces in Fig. 2C show that there was a transient dip in the [Ca²⁺]_i baseline of the M+ cells each time the static bath solution was replaced (except menthol- and KCl-elicited increases in [Ca²⁺]_i). Each dip may be caused by a slight warming on addition of solution to a slightly cool background produced by evaporative cooling of the static bath. Presumably, the warming reduces TRPM8 activity (i.e., fewer channels open), causing the dip, followed by evaporative cooling, which increases TRPM8 activity and thus, restores the [Ca²⁺]_i baseline after each dip. Some of the baseline variability may also be related to action potential bursting observed for some cold-sensitive units (29) and/or activity of ryanodine receptors, etc.

By using additional challenge compounds and experimental protocols, it should be feasible to identify the molecular isoforms of various ion channels and receptors present in each menthol-sensitive subtype. For instance, what voltage-gated Ca-channel isoforms are present, and do they differ between the M+ and MA+ subtypes? We can even begin to address which K channels are present in each subtype and determine whether homomeric or heteromeric channels are responsible for the functional phenotypes observed. The constellation of such receptors and ion channels in different neuronal subtypes constitute the molecular basis of their divergent functional properties. Thus, the cellular neuropharmacology of the menthol-sensitive subtypes should provide a far more refined characterization of these cells.

The menthol receptor, TRPM8, is strongly implicated in cold sensing (21, 30), but the precise physiological roles of the different menthol-sensitive DRG subtypes are less clear. As each neuronal subtype is further characterized, the properties uncovered may provide a guide to their physiological roles in vivo. For example, it is apparent from the divergent phenotypes elicited by TEA (Fig. 2D) that the complement of voltage-gated K channels in the two subtypes of menthol-sensitive cells likely differ. Notably, other investigators have shown that sensory trigeminal neurons responsive to innocuous cool temperatures (low-threshold cold thermoreceptors) show high expression of TRPM8 and low expression of K_V1 channels, whereas neurons responsive to noxious cold temperatures (high-threshold cold thermoreceptors) show low expression of TRPM8 and high expression of K_V1 channels, with K_V1 acting as an excitability brake in high-threshold, cold-sensitive neurons (14).

In our experiments, the M+ subtype showed robust menthol responses (Fig. 2C) (suggesting high TRPM8 expression), insensitivity to K_V1 blockers (suggesting low K_V1 expression), and instability in [Ca²⁺]_i at room temperature, all consistent with the hypothesis that the M+ subtype is a low-threshold cold thermoreceptor. In contrast, the MA+ subtype typically responded relatively weakly to menthol (Fig. 2C) (suggesting relatively low TRPM8 expression), whereas the responses to ATP (31–33) and AITC are implicated in nociception; the AITC receptor, TRPA1, is implicated in noxious cold nociception (34–38). These data, coupled with the relatively stable [Ca²⁺]_i at room temperature, are all consistent with the hypothesis that the MA+ subtype is a high-threshold cold thermoreceptor. We have obtained preliminary temperature-sensitivity data for the M+ and MA+ neurons in support of these hypotheses, and a detailed characterization of these two DRG subtypes is presently being carried out.

K-Channel Antagonists as Challenge Compounds.

In the experiments described above, a variety of K-channel antagonists were used. The most extensively analyzed was κM-RIIIJ, reported to have high selectivity for K_V1.2 compared with other homomeric voltage-gated K channels. A distinctive subclass of large-diameter cells responded directly to this conopeptide, distinguishing this subset of large-diameter neurons from the subclass sensitive only to ACh (Fig. 6A).

In addition to κM-RIIIJ, two other relatively selective K-channel antagonists were used, the pl14a conopeptide and Dtx-K, and they were reported to have high selectivity for K_V1.6 and K_V1.1, respectively. It is clear that the subset of neurons that responded directly to pl14a comprises a different neuronal subclass from the large neurons that responded directly to κM-RIIIJ. Although fewer cells were analyzed, the neurons that responded directly to pl14a were, on average, smaller in size. Thus, different subsets of neurons presumably have inversely proportional expression levels of K_V1.2 and K_V1.6. In addition, the phenotypes elicited by the two conopeptides were different: a subset of DRG neurons responded to κM-RIIIJ with a smooth, sustained increase in [Ca²⁺]_i, whereas a spiky response was observed in the neurons that responded directly to pl14a (Fig. 6B). In contrast, there was considerable overlap in DRG neurons that responded to κM-RIIIJ and Dtx-K when direct responders were scored (Fig. 6C). The results suggest that the K_V1.1 subunit overlaps significantly with K_V1.2 in large-diameter DRG neurons, consistent with previous expression studies (24).

The results using Dtx-K are particularly notable; in some neurons, the effects of Dtx-K were rapidly reversible but only slowly reversible in other neurons (Fig. S3), suggesting that Dtx-K may block different heteromers of K_V1.1/1.X with variable affinity in different neuronal subtypes.

Prospectives.

The experiments using ACh or K-channel antagonists as challenge compounds provide some insight into an important long-term direction of the cellular neuropharmacological approach described in this work. K channels and nicotinic acetylcholine receptors (nAChRs) are examples of ion channels that endow the nervous system with functional complexity at the level of individual macromolecules, because the functional receptor/ion channel is multimeric. Although there is a limited number of subunit genes, the subunits can combine into heteromers, making an enormous array of different combinations possible. For such ion channel families, it has been challenging to identify the functional roles for the individual molecular isoforms. The standard approach to identify function is gene KOs. The problem for multimeric complexes, such as the K-channel or nAChR families, is that KO of a single subunit does not just abolish one isoform but all potential heteromeric combinations containing the subunit encoded by the KO gene. This problem leads to a complex phenotype that does not reveal the function of any individual molecular isoform but rather, is the result of ablating all isoforms that contain that subunit.

The platform that we have developed makes it possible, in principle, to identify particular neuronal subtypes that functionally express specific nAChR- and K-channel isoforms. With the appropriate neuropharmacological tools (e.g., the α- or κ-conopeptides among other reagents), the molecular isoform(s) expressed in specific neuronal subtypes can be identified. Therefore, this identification opens the door to characterize the functional role of a specific receptor/ion channel isoform in the relevant neuronal circuitry. If one knows the molecular isoform(s) expressed in a specific neuron, inhibition of each molecular isoform should allow an assessment of the change in the properties of the circuit pertaining to that cell. In this way, the cellular neuropharmacology of individual neuronal subtypes provides a potential bridging technology between systems and molecular neuroscience.

Materials and Methods

Mouse DRG Dissection and Cell Culture.

Detailed procedures are provided in SI Materials and Methods. Briefly, WT C57BL/6 mice between the ages of 20–30 d postnatal were euthanized with CO₂ just before the dissection of DRGs. Dissection, cell dissociation, and cell culture methods were essentially as reported previously (39), except the minimal essential media contained 10 mM Hepes and glial-derived neurotrophic factor was used at a final concentration of 20 ng/mL.

Calcium Imaging.

Imaging was performed as detailed in SI Materials and Methods. Briefly, cells loaded with Fura-2 dye were excited intermittently with 340- and 380-nm light, whereas fluorescence emission was monitored at 510 nm. An image was captured at each excitation wavelength and the 340/380 nm ratio of fluorescence intensity was acquired (usually) one time per second to monitor the relative changes in [Ca²⁺]_i for each cell over time. Thus, we obtained the 340/380-nm ratiometric images shown in Fig. 1 and the 340/380-nm ratiometric data (traces) shown in Figs. 2–6. In a given experiment, >100 cells were individually imaged simultaneously. Increases in [Ca²⁺]_i were elicited by ∼15-s application of various RA challenge compounds or elevated [K⁺]_o in observation solution. MP challenge compounds in observation solution were applied to cells for time periods indicated in Figs. 1–6. For RA challenge compounds, we generally chose high concentrations to avoid issues of dose response. The concentration of ACh was chosen to be saturating or nearly saturating for all subtypes of ACh receptors. We applied ATP at the high end of its expected physiological concentration (in muscle interstitium) (39). The other RA challenge compound concentrations are consistent with common literature values that have been used to identify responsive sensory neurons previously (13–15, 19, 35). For MP challenge compounds, we chose either high concentrations to obtain broad-spectrum block of K or Na channels (i.e., TEA and TTX, respectively) or relatively low concentrations that were expected to provide subtype-selective block of particular K-channel subtypes [i.e., Dtx-K (K_V1.1) (28), κM-RIIIJ (K_V1.2) (26), and conopeptide Pl14a (K_V1.6) (27)]. After calcium imaging, cells were stained with Alex-fluor-568–labeled IB4 and imaged. Data for each experiment were screened manually for cells that did not respond to elevated [K⁺], washed out of the field of view, or produced irreversible high [Ca²⁺]_i during the experiment. Such cells were excluded from additional analysis.