Fentanyl Decreases Ca²⁺Currents in a Population of Capsaicin-responsive Sensory Neurons

Sprague-Dawley weanling rats (40–100 g) of either sex were sacrificed with pentobarbital (0.2 mg/g, intraperitoneal) according to a protocol approved by the Animal Care Use Committee at the University of Wisconsin (Madison, Wisconsin). Dorsal root ganglia were removed from the cervical, thoracic, and lumbar regions of the spinal column using a dissecting microscope and placed in 0.25% trypsin solution (Sigma, St. Louis, MO) containing 2 mg/ml collagenase (Sigma) at 35°C for 45 min. The ganglia were then gently triturated using flame-polished Pasteur pipettes, and the cells were plated on glass coverslips coated with poly-L-lysine (Sigma). Cells were maintained in an incubator at 35°C in a solution of Dulbecco's Modified Eagle Medium (Gibco, Gaithersburg, MD) containing 10% fetal bovine serum (Sigma), 50 U/ml penicillin, and 50 μg/ml streptomycin (Sigma). Experiments were performed between 2 and 36 h after isolation. Responses were not noticeably different at various times after isolation, though after 2 days in culture, neurons developed extensive processes, making it difficult to voltage clamp the cells. Neurons tested ranged in size from approximately 18 to 50 μm in diameter (based on measurement of cell membrane capacitance and assuming spherical cells with specific membrane capacitance of 1 μF/cm²).

Patch pipettes were made from borosilicate glass using a two-stage pipette puller (Narishige USA, East Meadow, NY) and were heat-polished on a microforge (ALA Scientific Instruments, Westbury, NY) to a resistance of 1–3 MΩ. Pipettes were coated with Sylgard (World Precision Instruments, Sarasota, FL) to within 100 μm of the tip. Coverslips with cells attached were transferred to the recording chamber and perfused with external solutions. Whole cell currents were recorded in voltage clamp mode using a patch clamp amplifier (Axopatch 200B; Axon Instruments, Foster City, CA) using standard techniques. 17Data acquisition and analysis were performed using a Digidata 1200 A/D converter and pClamp version 6.0.4 (Axon) running on a microcomputer. Immediately after obtaining a tight seal and rupturing the cell membrane to achieve the whole-cell recording configuration, series resistance (R^s) and membrane capacitance (C^m) were estimated from the settings on the amplifier obtained after eliminating the capacitative current transient recorded during a small voltage step. The values for R^sand C^mobtained in this way closely matched the values calculated by integrating the capacitative transient to measure C^mand fitting the transient to an exponential with a time constant of R^sC^m.

Both I^Caand capsaicin-induced currents were recorded from individual cells at room temperature from a membrane holding potential of −80 mV. I^Cawas generally recorded first, but in some experiments, capsaicin was applied first, with no apparent differences in the results. I^Cawas elicited by square wave voltage steps of 90 ms duration to potentials ranging from −60 to +30 mV, or by repetitive voltage steps of 90 ms or less in duration to −10 mV. Currents were digitized at 5–10 kHz and filtered at 1–2 kHz. Leak and capacitative currents were subtracted from current records online by adding the currents obtained from four hyperpolarizing pulses each equal to one fourth the magnitude of the depolarizing test pulse (−P/4 protocol). Series resistance compensation was usually used to prevent voltage errors of greater than 5 mV.

The pipette solution contained the following: 100–120 mm CsCl, 2.5 mm MgCl², 10 mm EGTA, 10–40 mm HEPES, 2 mm MgATP, and 0.3 mm LiGTP. The pH was adjusted to 7.3 with tetraethylammonium hydroxide or CsOH, and aliquots were stored at −80°C until use. The external solution used to perfuse the cells between experiments and to record capsaicin-induced currents contained the following: 130 mm NaCl, 5 mm KCl, 1 mm MgCl², 2 mm CaCl², 10 mm glucose, and 10 mm HEPES. The pH was adjusted to 7.4 with NaOH. For recording I^Ca, the solution was switched to one containing the following: 135 mm tetraethylammonium chloride, 1–2 mm CaCl², 0–1 mm MgCl², 10 mm glucose, and 10 mm HEPES. The pH was adjusted to 7.4 with tetraethylammonium hydroxide or CsOH. In later experiments, sucrose was also added to adjust the osmolality of the external solutions to 310 mOsmol/kg and the osmolality of the pipette solutions to 290 mOsmol/kg. No differences were noted after osmolality correction was instituted.

Capsaicin (N-vanillylnonanamide) stock solution was made in ethanol and refrigerated until use. For 10 μm capsaicin, the concentration of ethanol in the working solution was 0.1% (v/v). Concentrations of ethanol up to 1% (v/v) had either no effect or induced a small outward current. Concentrated stock solutions of fentanyl citrate, ω-conotoxin GVIA, ω-Agatoxin TK, ω-conotoxin MVIIC, and naloxone were made in distilled water and frozen. The nifedipine stock solution was made with DMSO and refrigerated. All reagents were from Sigma.

Agents were applied to cells by several methods. Cells were pretreated with some agents (ω-conotoxin MVIIC, ω-Agatoxin TK) by placing a coverslip of isolated neurons in a small dish filled with a solution containing one of the toxins for at least 20 min before removing the coverslip and placing it in the recording chamber, which did not contain toxin. For rapid applications of capsaicin, fentanyl, and some of the Ca²⁺channel blockers, cells were gravity perfused with solutions flowing through one of three large capillary tubes glued together in a row. The tubes were rapidly moved with a computer-driven motor (Warner, Hamden, CT), allowing cells to be perfused with different solutions of known concentrations with exchange rates of less than 1 s. Finally, in some experiments, solution changes were accomplished using a gravity-fed system that allowed complete exchange of the solution in the recording chamber in 10–20 s (chamber volume was approximately 700 μl).

The peak capsaicin-induced current was measured during a 5-s application of capsaicin at a membrane potential of −80 mV. I^Camagnitude was measured 20 ms after the voltage step, when the current was near maximal. Current magnitudes were normalized to cell size by dividing by membrane capacitance. Fentanyl-induced changes in I^Camagnitude were expressed as percent changes from the average of currents recorded before fentanyl application (control) and after washout of fentanyl (recovery). The inactivation time constant of I^Cawas estimated by fitting the decaying phase of the current to a single exponential using Clampfit (Axon). Averaged concentration–response relations were fit using nonlinear regression (GraphPad Software, San Diego, CA) to a logistic equation of the form:

where R is the response at concentration C, R^maxis the maximum response obtained, EC⁵⁰is the concentration at which the effect is half-maximal, and n^His the Hill coefficient.

Data are expressed as mean ± SEM. The Fisher exact test was used to determine the statistical significance of differences in the proportions of cells among groups. Other values were compared using one-way analysis of variance followed by the Tukey post hoc  test where appropriate (GraphPad Software). Statistical significance was assumed if P  was less than 0.05.

Capsaicin induced an inward current in 158 (83%) of 191 sensory neurons tested. Because repeated applications of capsaicin to the same neuron produced currents of progressively smaller magnitude, only one concentration of capsaicin was applied to each cell during the determination of the capsaicin concentration–response curve (fig. 1). The reversal potential of the capsaicin-induced current was approximately 0 mV. At 10 μm, capsaicin produced a near-maximal response. This concentration was used in subsequent experiments to determine the capsaicin response of individual neurons in order to decrease the risk of incorrectly categorizing a capsaicin-responsive cell as being unresponsive. The limits for categorizing cells as being capsaicin responsive or unresponsive were chosen after examining the distribution of capsaicin current responses. Neurons were considered to be responsive to capsaicin if the inward current magnitude was at least 100 pA or 5 pA/pF. Capsaicin-unresponsive cells generated less than 20 pA or 0 pA/pF of inward current. Six cells were excluded from analysis due to intermediate capsaicin responses (inward currents ranging from 30 to 88 pA or 1 to 6 pA/pF).

Sensitivity to both fentanyl and capsaicin was determined in 121 sensory neurons (table 1). Examples of these responses are illustrated in figure 2. Overall, 106 (88%) of 121 neurons tested responded to capsaicin, while 15 (12%) did not. Of the capsaicin-responsive cells, 62 (58%;fig. 2A) expressed Ca²⁺channels that were inhibited by fentanyl. In the other 44 capsaicin-responsive cells (42%;fig. 2B), fentanyl had no effect on I^Ca. Thirteen (87%) of the 15 capsaicin-unresponsive neurons expressed fentanyl-insensitive Ca²⁺channels (fig. 2C). In capsaicin-responsive, fentanyl-sensitive cells, the inhibition of I^Cawas concentration dependent, with an EC⁵⁰of 19 nm when fit to the logistic equation with a Hill coefficient of 1 (fig. 3). In fentanyl-insensitive neurons, fentanyl concentrations of up to 1 μm produced less than 10% inhibition of I^Ca(figs. 2 and 3). In six fentanyl-sensitive cells, naloxone (10 μm) reduced the fentanyl-induced inhibition of I^Cafrom 44 ± 10% to 2 ± 1% (P < 0.01).

A commonly held theory suggests that the cell bodies of primary nociceptive neurons tend to be relatively small, a reflection of the small axons of unmyelinated C fibers or thinly myelinated Aδ fibers that are thought to transmit noxious sensations. In the present study, assuming that cell capacitance reflects cell size, capsaicin-responsive cells were smaller, on average, than capsaicin-unresponsive cells (25 ± 1 pF, n = 106, vs.  38 ± 2 pF, n = 15, respectively;P < 0.05). Among the capsaicin-responsive cells, the fentanyl-insensitive cells (21 ± 1 pF, n = 44) were smaller than the fentanyl-sensitive cells (28 ± 2 pF, n = 62;P < 0.05). As can be seen from the histogram in figure 4, this is partly due to the presence of a number of very large capsaicin-responsive, fentanyl-sensitive neurons. The capsaicin-unresponsive, fentanyl-insensitive cells were the largest on average (39 ± 2 pF, n = 13;P < 0.05).

The average amplitude of capsaicin-induced currents recorded from each group of cells is shown in figure 5A. There was no difference in the current amplitude between the two groups of capsaicin-responsive neurons. The average current in the capsaicin-unresponsive group was an outward current of 1 ± 0 pA/pF.

Capsaicin-unresponsive cells had a significantly greater low-voltage–activated (LVA) Ca²⁺current density than either of the capsaicin-sensitive groups of cells (figs. 2C and 5B). Not only was the magnitude of LVA current larger, it was expressed in all 5 cells examined, compared to only 12 (55%) of 22 capsaicin-responsive, fentanyl-sensitive cells and 16 (67%) of 24 capsaicin-responsive, fentanyl-insensitive cells. The magnitude of LVA Ca²⁺current was similar in the two groups of capsaicin-responsive cells. Fentanyl had no effect on LVA Ca²⁺current.

There were no significant differences in the maximum amplitude of the high-voltage–activated (HVA) Ca²⁺current density among the three groups (fig. 5C). However, the inactivation rate of the HVA Ca²⁺current at a test potential of −10 mV was significantly faster in the fentanyl-sensitive group compared to the fentanyl-insensitive group of capsaicin-responsive cells (figs. 2 and 5D). HVA current inactivation was also rapid in the capsaicin-unresponsive neurons (fig. 5D). This may have been due to the presence of residual LVA Ca²⁺current at this test potential. There was no difference in the rate of HVA Ca²⁺current activation as assessed by comparing the time from the beginning of the voltage step to the time of peak inward current.

Additional experiments were performed to explore the possibility that the difference in the inactivation rates of I^Cabetween the two groups of capsaicin-responsive cells was due to a difference in the types of Ca²⁺channels expressed in these two populations of cells. Figure 6summarizes the data for all three groups of sensory neurons. ω-Conotoxin GVIA (GVIA; 1 μm) was used to irreversibly block N-type I^Ca. Subsequent perfusion with nifedipine (10 μm) blocked L-type current. The current that remained (P-/Q- and/or R-type) was considered GVIA and nifedipine resistant. There were no significant differences in the percentage of L-type, N-type, or GVIA/nifedipine-resistant I^Cabetween the two groups of capsaicin-responsive cells. On the other hand, the capsaicin-unresponsive neurons expressed significantly less N-type current and more GVIA/nifedipine-resistant current than either of the capsaicin-responsive groups. Capsaicin-unresponsive cells also had significantly less L-type current than the capsaicin-responsive, fentanyl-sensitive cells.

The approximate contributions of P-/Q- and R-type Ca²⁺channels to the GVIA/nifedipine-resistant current was estimated using another Ca²⁺channel toxin. ω-Conotoxin MVIIC (MVIIC; 3 μm) blocks P-/Q-type Ca²⁺channels slowly and irreversibly and blocks N-type channels rapidly and reversibly. 18Pretreatment of cells with MVIIC, followed by perfusion with MVIIC and nifedipine (10 μm), would thus block all but R-type Ca²⁺channels. Under these conditions, the average I^Cawas −64 ± 13 pA, or a current density of just −3 ± 1 pA/pF (n = 6). This compares to an average I^Cain the absence of any channel blockers of −3,458 ± 264 pA, or a current density of −141 ± 9 pA/pF (n = 49) for all groups. Thus, R-type current probably accounts for only about 2% of the whole cell current in these cells. In other words, it is likely that the majority of the GVIA/nifedipine-resistant current is attributable to P-/Q-type Ca²⁺channels.

Figure 7illustrates the effect of fentanyl on I^Cain the presence of various Ca²⁺channel blockers. In the presence of nifedipine and after MVIIC pretreatment, fentanyl (1 μm) reduced N-type I^Caby 62 ± 9% (n = 3) in sensitive cells (fig. 7A). Subsequent perfusion with MVIIC to block N-type channels eliminated the fentanyl response (fig. 7A), suggesting that R-type channels are not sensitive to fentanyl. In other experiments using pretreatment with ω-Agatoxin TK (500 nm) instead of MVIIC to irreversibly block P- and Q-type Ca²⁺channels, and in the presence of nifedipine, fentanyl (1 μm) reduced N-type I^Caby a similar amount (49 ± 7%, n = 4). In addition, after application of GVIA and during nifedipine perfusion, fentanyl (1 μm) reduced the P-/Q-type I^Caby 50 ± 8% (n = 4;fig. 7B). Assuming L- and R-type channels are insensitive to fentanyl, a 50% reduction by fentanyl (1 μm) in both N- and P-/Q-type I^Ca, which together comprise about 80% of whole cell current in sensitive cells (fig. 6), would produce the 40% reduction in whole cell current seen in earlier experiments (fig. 3).

Previous studies have shown that, in sensory neurons, the inhibition of I^Caby μ-opioid receptor agonists is quite variable. In random populations of acutely isolated sensory neurons, between 44 and 90% of cells studied are opioid sensitive (usually defined as ≥ 10% inhibition in I^Ca). 6–11Some of this variability in opioid sensitivity among cells might be explained by differences in the sensory modality transmitted by individual neurons. One might expect nociceptive neurons to be sensitive to the effects of opioids, whereas other neurons that transmit innocuous mechanical and thermal sensations should be relatively insensitive to opioids. 12,13The challenge, then, is to identify which neurons in an isolated, in vitro  preparation are nociceptors.

A simple approach used by many investigators interested in nociceptor function is to focus their studies on small sensory neurons. This is because most nociceptor axons are unmyelinated C fibers or thinly myelinated Aδ fibers, and the cell somata associated with these axons are generally smaller. 19,20However, many small C- and Aδ-fiber-type sensory neurons do not transmit nociceptive information, and some nociceptors even conduct in the Aαβ-fiber range. 21,22Size restriction alone, therefore, does not guarantee a population of nociceptive neurons.

Other characteristics have been used to identify nociceptive neurons in isolated cultures of dorsal root ganglion neurons. These include differences in the shape of the somatic action potential, the presence of neuropeptides such as substance P and calcitonin gene–related peptide, and the presence of tetrodotoxin-resistant sodium channels, as well as other types of ion channels. 23–25Although these are generally thought of as “nociceptor properties,” they do not always cosegregate in individual neurons, and none of these properties have been shown to be specific or sensitive tests for nociceptors.

In the present study, nociceptive neurons were identified by their sensitivity to capsaicin. Capsaicin is a vanilloid compound isolated from hot peppers that produces burning pain when injected intradermally or arterially or applied to mucous membranes. It is an agonist for the vanilloid receptor VR1, a membrane protein involved in the transduction of noxious heat. 14–16VR1 is expressed primarily by small- to medium-sized neurons of sensory ganglia. 14,15Single-unit recordings from peripheral nerves innervating the skin demonstrate that capsaicin selectively activates heat-sensitive nociceptors. The majority of heat-sensitive nociceptive neurons are C-fiber polymodal nociceptors, which respond to noxious heat, noxious mechanical stimuli, and noxious chemical stimuli. Some C- and Aδ-fiber-type neurons respond to noxious heat and noxious mechanical stimuli (mechano-heat nociceptors), while purely mechanical nociceptors, which do not respond to noxious heat, are insensitive to capsaicin. Nerve fibers transmitting nonnoxious stimuli, such as light touch and hair sensors, are rarely found to be capsaicin responsive. 26–29Capsaicin responsiveness, therefore, is a sensitive and specific test for identifying heat-sensitive nociceptors.

Approximately 80% of all the cells in the present study responded to capsaicin. This is in close agreement with other studies that have reported the capsaicin sensitivity of small sensory neurons from young rats, 30,31though in older rats, the percentage of capsaicin-sensitive neurons is less, ranging from about 50 to 65%. 24,25,32,33It also correlates with the finding that approximately 75% of C-fiber single units recorded from a cutaneous nerve in the rat are polymodal nociceptors. 22 

The remaining 20% of the cells in the present study did not respond to capsaicin. Because of their size (calculated diameters of approximately 20–40 μm), these cells may have been mostly C-fiber-type low-threshold mechanoreceptors 19,20as these are among the most common nonnociceptive unmyelinated fibers found in rat nerves. 22On the other hand, the two capsaicin-unresponsive cells that were sensitive to fentanyl might have been high-threshold mechanical nociceptors. These neurons may have been underrepresented in this study due to the fact that they often conduct in the Aδ range 21and therefore may have been larger, on average, than most of the cells studied here.

The capsaicin-unresponsive cells expressed a unique distribution of voltage-activated Ca²⁺channels. Most of the HVA Ca²⁺current in these cells was P-/Q-type current, while in both capsaicin-responsive groups, N-type current was predominant. In addition, all of the capsaicin-insensitive cells expressed a large T-type Ca²⁺current, an association that has been demonstrated previously. 32–34While a T-type current was observed in many of the capsaicin-sensitive neurons as well, it was never as large as in the capsaicin-insensitive neurons. Scroggs and Fox 35also described a population of “medium-sized” (33–37 μm in diameter) sensory neurons with a large T-type Ca²⁺current component, though capsaicin sensitivity was not tested in that study.

Only about 60% of the capsaicin-responsive neurons were sensitive to fentanyl. The inhibition of I^Cain these cells was concentration dependent and reproducible with little variability from cell to cell. The I^Cain these cells inactivated more rapidly than the I^Cain the fentanyl-insensitive nociceptors. Others have reported that opioids primarily inhibit a transient I^Cacomponent, and this is generally presumed to be due to current flowing through N- and/or P-/Q-type channels. 6,10,36While the present results confirm these observations, they also show that not all N- and P-/Q-type channels are inhibited by opioids. I^Cawas not inhibited by fentanyl when recorded from capsaicin-responsive neurons with slowly inactivating I^Caor when recorded from capsaicin-unresponsive neurons. It is not obvious how the opioid sensitivity of Ca²⁺channels is related to their inactivation rate. Differences in G-protein expression or the phosphorylation state of the Ca²⁺channels might be involved as both can markedly alter I^Cakinetics in addition to their ability to affect I^Camagnitude. On the other hand, a recent study in young rats demonstrated that less than half of small sensory neurons in the L4–L5 dorsal root ganglion express μ-opioid receptors. 37This would explain the fentanyl insensitivity but not the difference in I^Cainactivation rates.

Few other studies have attempted to characterize the opioid sensitivity of putative nociceptive sensory neurons. Schroeder and McCleskey 7tested the effect of DAMGO, a μ-opioid receptor agonist, on a group of putative nociceptive neurons identified by using antibodies against a specific oligosaccharide present only on cells that project to laminae I and II in the dorsal horn of the spinal cord, the site of most nociceptive input. They found that while the DAMGO response was much less variable in the labeled neurons, the average magnitude of the inhibition of I^Cain the labeled neurons (25%) was actually less than that observed in random samples of neurons (38%). This result may be partly explained by the fact that while primary nociceptive neurons do terminate extensively in laminae I and II, they also project to other areas of the dorsal horn. 21,38It is also possible that nociceptors projecting to deeper laminae are more likely to be, for example, Aδ mechanical nociceptors, which may be more sensitive to opioids than other types of nociceptors. Another limitation of this labeling technique is that some nonnociceptive neurons also project to laminae I and II, which could explain their finding that not all of the labeled neurons were sensitive to DAMGO.

Using a different approach, Taddese et al.  39studied the opioid responsiveness of neurons from the trigeminal ganglion that were labeled when they took up a fluorescent indicator placed in the tooth pulp. They reasoned that neurons innervating tooth pulp are a pure population of nociceptors based on the assumption that the only sensory modality transmitted from the tooth is pain. They found that DAMGO decreased I^Caby an average of 21% in small labeled neurons compared to 13% in small randomly selected neurons from the trigeminal ganglion. However, their results also show that only 79% of small nociceptors (≤ 30 μm in diameter) were inhibited by DAMGO, and the response rate for nociceptors of all sizes was only 49%. In the present study, 58% of all nociceptors were opioid sensitive using a more diverse population of sensory neurons and, arguably, a simpler method of identifying nociceptors.

Taddese et al.  39found that smaller nociceptors were much more likely to be opioid sensitive than larger nociceptors. They postulated that this may be related to the finding that opioids better alleviate more chronic, aching types of pain that are transmitted by more slowly conducting C-fiber axons, compared to the sharp shooting pain that is likely transmitted by rapidly conducting A-fiber-type axons from larger nociceptive neurons. The present results do not support this theory, however, since the fentanyl-sensitive cells were larger, on average, than the fentanyl-insensitive cells. There were also a few extremely large cells that were opioid sensitive. Nevertheless, it would be beneficial to learn more about the opioid-insensitive nociceptors that have been identified in this and other studies. Such knowledge may make it possible to develop new analgesics that specifically target these nociceptors, which could then be used in combination with neuraxial opioids to provide improved control of acute and chronic pain.

The author thanks Craig Levenick, B.S. (Research Specialist, Department of Anesthesiology, University of Wisconsin, Madison, Wisconsin), for expert technical assistance and Hirochika Komai, Ph.D. (Associate Scientist, Emeritus, Department of Anesthesiology, University of Wisconsin), for numerous discussions and suggestions.