NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Zhu MX, editor. TRP Channels. Boca Raton (FL): CRC Press/Taylor & Francis; 2011.

Cover of TRP Channels

TRP Channels.

Show details

Chapter 14Time-Resolved Activation of Thermal TRP Channels by Fast Temperature Jumps

.

14.1. INTRODUCTION

Humans recognize thermal stimuli with distinct sensations of being noxious cold, cold, warm, and noxious hot. Recordings from peripheral nerve fibers have demonstrated the existence of thermally active neurons known as thermal receptors. When skin temperature is raised above 30°C, thermal receptors specialized for detection of warmth start to fire action potentials,1 and when the temperature is further increased to above about 45°C, the so-called nociceptors become active causing perception of pain. The activity of warm receptors reaches maximum discharge at or below 45°C and then decreases abruptly as the skin is warmed further.1 Thus, warm receptors cannot generate signals for differentiating noxious from innocuous hot stimuli. Cold receptors, on the other hand, show peak responses at about 25°–27°C,1,2 while cold nociceptors have an activation threshold below about 20°C. The response of cold receptors covers a relatively wide range of temperatures, which overlap with the response of warm receptors at near 37°C.1 Cold receptors become inactive upon warming. They are poor indicators of absolute temperature but are extremely sensitive to localized temperature changes, and their responses are slowly adapting.3

Despite extensive demonstration of thermally sensitive nerve fibers in vivo, the molecular entities for thermal transduction have remained enigmatic until the recent discovery of TRP channels (see, e.g., Ref. 4). Several TRP channels from different subfamilies show temperature-dependent activation. These include TRPV1-4 from the vanilloid subfamily, TRPM8 from the melastatin subfamily, and TRPA1 from the TRPA subfamily. TRPV1 was first shown to be a long-sought ion channel in the nociceptive sensory neurons that responded to capsaicin, the hot pungent ingredient of chili peppers.5 In addition to its vanilloid sensitivity, the cloned TRPV1 receptors, when heterologously expressed, are activated at temperatures above 42°C, consistent with the known temperature threshold of nociception.6 TRPV2-4 were identified as homologues of TRPV1.711 TRPV2 has an activation temperature threshold of above 50°C and is expressed in medium- to large-diameter myelinated neurons of the dorsal root ganglia (DRG). Both its temperature threshold and expression patterns support it as a candidate for the high-threshold heat response of Aδ fibers in vivo. TRPV3 and TRPV4 are activated by innocuous heat with reported temperature threshold ranges of 30°–40°C and ~25°–34°C, respectively. TRPV3 is also sensitive to compounds that induce warm feeling such as oregano, savory, and thyme, consistent with its role for warmth detection.12 Common to both TRPV3 and TRPV4 is their very low expression levels in the sensory neurons; instead, they are prominently found in skin keratinocytes.811 TRPV4 is also strongly expressed in the kidney, where it has been implicated in regulating the body fluid level as an osmosensor owing to its mechanical sensitivity to hypotonicity.13.14

Parallel to heat sensation, cold sensation can be chemically mimicked by plantderived cooling compounds such as menthol from mint oil. TRPM8 is a molecular target of menthol in sensory neurons15,16 and is activated by innocuous cool temperatures below 25°C, although its responsiveness continues into the noxious cold range. TRPA1 has been suggested to mediate noxious cold transduction,17 but its cold sensitivity remains a highly debated issue (see, e.g., Ref. 18). Nevertheless, its functions in nociception are supported by its co-expression with TRPV1 in nociceptors and its chemosensitivity to compounds such as mustard oil.19 Studies of the knockout of TRPV1, TRPV, and TRPM8 genes in mice have corroborated their contributions to thermal sensation (TRPV3 and TRPM8) or heat-related hyperalgesia (TRPV1),2025 while reports on the TRPA1 knockouts have invigorated the controversy regarding its involvement in cold sensation.26,27

The discoveries of the thermally active ion channels have now made it possible to study thermal sensation at the molecular level, just like the studies of other senses quite a long time ago. Importantly, these channels also raise biophysical questions about the mechanisms of thermal gating. Historically, ion channels have been mostly studied for gating by other variables such as voltage and chemical agonists. In contrast, the temperature is fundamentally different from these stimuli because thermal energy interacts with proteins globally. It is understood that voltage may be sensed by localized charges in a low dielectric environment (e.g., membranes), while agonists can be detected through stereochemically specific binding sites. However, it is not obvious whether such a “key-and-lock” mechanism also works for detection of thermal energy. The activation of thermal TRP channels by temperature has been observed in excised membrane patches (e.g., Ref. 28), indicating that their thermal sensitivity is membrane delimited. The temperature coefficients of the responses are high (for review, see Ref. 29), and single-channel analysis shows that the temperature has a localized effect primarily on the long closures between opening bursts, in spite of complex gating kinetics comprising multiple closed and open states.28

Both heat activation of TRPV1 and cold activation of TRPM8 exhibit large but compensatory enthalpy and entropy changes.28,30 As thermally gated TRP channels are also sensitive to voltage, Voets et al.31 suggested that temperature opens the channel through voltage sensors by shifting the voltage-dependent gating curve toward more physiological membrane potentials. Alternatively, a change of membrane potential can also alter thermal or agonist activation. Brauchi et al.30 and Matta and Ahern32 showed that such mutual dependence between stimuli can be better explained by a Monod–Wyman–Changeux (MWC)-type allosteric model. For cold receptors TRPM8, Latorre et al.33 also reported a temperature-independent Cole–Moore shift of voltage activation, suggesting that voltage sensors move separately from thermal sensors. Common to these models is that the energies from different stimuli contribute in an additive manner to stabilize the open conformations of the channel. Lately, Yao et al.34 showed instead that the thermal sensitivity interacts with voltage or agonist sensitivity in a nonadditive manner based on direct measurements of temperature responses of TRPV1. Their data support that the activation of thermal sensors is coupled to agonist binding or charge movement of voltage sensors.

Structurally, thermal TRP channels have a membrane topology similar to that of voltage-gated channels, consisting of six transmembrane segments (S1–S6) and a reentrant pore loop between S5 and S6. However, despite their apparent activation by membrane depolarization, they lack a highly charged S4 segment, which is common to voltage-gated channels and acts as a voltage sensor. They also have relatively large N- and C-termini, which harbor a variety of regulatory sites and protein–protein interaction domains, such as the ankyrin-like repeats found in TRPV and TRPA channels. The sequence homology of TRP channels is generally limited across subfamilies. The structural basis of the temperature sensitivity has been revealed in several mutagenesis studies. First, exchanges of the C-terminal domains between TRPV1 and TRPM8 reversed their hot and cold sensitivity.35 Second, mutations in the inner pore of TRPV1 also influenced its heat responses,36 while single residue mutations in the outer pore region impacted heat activation of TRPV3.37 Thus, it appears that the thermal sensitivity of these channels may spread over multiple regions of the proteins. In addition, both chemical and thermal activators induce similar structural changes in the S6 gate region,38 suggesting that a common gate is shared by different stimuli.

Studies of thermal TRP channels require controlled temperature perturbation. This chapter presents a review of thermal control methods for patch-clamp experiments and their applications for measuring the temperature dependence of thermal TRP channels. The technique of laser diode irradiation is emphasized. This approach offers a time resolution far superior to conventional thermoelectric heating while allowing for rapid modulation of laser output to clamp temperature. The instrumentation design of a submillisecond temperature clamp apparatus is described. Applicability of this system is demonstrated with the heat-gated TRPV1 channel by measurements of its activation rates and energy landscape of temperature gating. The results show that the time-resolved fast temperature jump experiments can provide new insights into the temperature gating mechanisms of thermal TRP channels.

14.2. THERMODYNAMIC PRINCIPLES OF TEMPERATURE GATING

Regardless of the structural basis of thermal gating, it is imperative to understand its thermodynamic basis. For a channel existing between two states, its open probability is dictated by a Boltzmann relationship

where ΔG represents the free energy change of the system from closed to open. At rest, the free energy has a profile with ΔG > 0 so that the closed state is favored. Application of a stimulus alters the profile to favor the open state instead. For voltage or agonist gating, the external electrical or chemical energy is added to reduce the free energy difference ΔG. For temperature gating, the perturbation results merely from thermal energy. The gating by temperature therefore relies on the entropy of the system to offset the free energy barriers of activation.

In terms of enthalpy and entropy, the Boltzmann relationship becomes

where ΔH and ΔS are, respectively, the enthalpy and entropy changes between the closed and open states, and R and T have their standard definitions. A few observations can be inferred from this relationship. First, the temperature dependence of opening is strictly determined by the enthalpy change (ΔH). In an analogy to voltage gating, the inverse of temperature (1/T) is equivalent to voltage, while the enthalpy change plays a similar role to the gating charges. The enthalpy change thus determines the slope of the Boltzmann curve of temperature responses. Second, the entropy change (ΔS) has no effect on the temperature dependence of opening, but it affects the midpoint of the Boltzmann curve of opening (T1/2 = ΔHS). Third, the sign of the enthalpy change determines the polarity of temperature sensitivity as to whether the channel is gated by heat or cold.

In theory, Equations 14.1 and 14.2 imply that temperature gating may occur to any ion channel, although the activation temperature range may be too high or too low, beyond the melting point of the channel protein. In this regard, the thermal TRP channels are unique in that they have evolved with an accessible range of temperatures for activation (overlapping with physiological temperatures) and temperature dependence that is adequately high so that appreciable activity occurs when the ambient temperature becomes only slightly above (or below) the threshold.

The temperature coefficient, or Q10, is commonly used to characterize the temperature dependence of the kinetics of a process. It measures the change of a rate when temperature is increased by 10°C and is related to the activation enthalpy (ΔH) by

where the approximation holds since T ≫ 10. Conversely, for a given Q10 value, the activation enthalpy is determined by

Image ch14_eq04

Figure 14.1 plots the relationship of ΔH versus Q10 on both linear and logarithmic scales. Enzymatic reactions typically have Q10 values between 2 and 3, which corresponds to an enthalpy of <20 kcal/mol. In this range, the enthalpy is sharply dependent on Q10. Thus, the term Q10 provides a sensitive measure of the temperature dependence of the reaction. For large Q10 values (e.g., >20), however, the rate of enthalpy increase becomes slow. Thermal TRP channels can have temperature dependence as high as ΔH ~ 100 kcal/mol,34 and in this range, the enthalpy becomes nearly insensitive to changes in Q10. In this regard, the term Q10 is not a sensitive descriptor for the temperature dependence of thermal TRP channels; the enthalpy itself seems to be more appropriate. By definition, Q10 is only pertinent to the rate of a process. In the studies of TRP channel gating, the temperature dependence is often evaluated from (pseudo) steady-state responses (the kinetics are hard to measure). The temperature coefficients calculated in this way reflect the equilibrium enthalpy change between the closed and open states (Equation 14.2) rather than the activation enthalpy between the closed and transition states (Equations 14.2 and 14.3).

FIGURE 14.1. Q10 versus activation enthalpy (ΔH) plotted on both linear (black; left-bottom axes) and logarithmic (dark gray; right-top axes) scales.

FIGURE 14.1

Q10 versus activation enthalpy (ΔH) plotted on both linear (black; left-bottom axes) and logarithmic (dark gray; right-top axes) scales. Typical ion channel gating has Q10 of 2–3.

14.3. MEASUREMENT OF TEMPERATURE DEPENDENCE

Because gating by temperature is governed by the thermodynamic properties of the channel, it is important to have precise estimates of the energetics for understanding the underlying mechanisms of gating. Below, we briefly summarize the experimental assessments of the temperature dependence of thermal TRP channels as they have been reported in the literature. The estimates of the temperature dependence sometimes exhibit significant variations with different types of measurements, which presumably could have resulted from the complexity of the experiments as well as the susceptibility of these channels to modulation by uncontrolled variables.

14.3.1. Pseudo-Equilibrium Analysis

By far, temperature ramps have appeared to be the most common protocol for measuring temperature dependence. In a typical experiment, whole-cell currents are continuously recorded while the temperature of the bath media is slowly changed. The ambient temperature of the bath solution may be changed through direct heating/cooling of the chamber or by superfusion with solutions that pass through an inline solution heater/cooler. Between the two, the inline solution heating/cooling appears to be more commonly used. Using a small-diameter tube coated with a platinum heating element, Dittert et al.39 demonstrated a remarkable heating rate (100°C/s). Reid et al.40 achieved a bipolar temperature change rate up to 4°C/s by restricting superfusion areas. More recently, Dittert et al.41 further significantly improved the speed up to –40° to 60°C/s through a combination of resistive heating and thermoelectric heating and cooling at the very end of the perfusion outlet. With commercially available inline solution heaters/coolers (e.g., SC-20/CL-100, Warner Instruments), a rate of ~1°C/s appears to be more common. The rise of temperature is generally nonlinear, typically with a time course being relatively linear during the onset and becoming slower (sublinear) when approaching the set point.

For analysis, data (currents) are often plotted versus temperature (1/T) on a semilog scale. The plot may be considered as a variant of the Arrhenius plot for rate constants, but unlike the Arrhenius plot, which is linear against 1/T, the current plot is generally nonlinear. It is typically biphasic, beginning with a slow increase at low temperature followed by a rapid increase above a certain threshold. The slow component arises from leak currents, while the fast component results from channel opening. The temperature coefficient of the channel activity is determined from the slope of the linear fit of the second component. According to the Boltzmann equation, when the open probability (Po) is low, the current–temperature relationship is approximately

and on the log scale, it becomes

Image ch14_eq06

The slope of the fit thus gives rise to ΔH/R. The asymptotic slope of the Boltzmann relationship consequently determines the enthalpy change between closed and open states. Once the enthalpy change is known, Q10 can be evaluated from Equation 14.3.

Although the approximation in Equation 14.5 is applicable only at low Po, the linearity of the Boltzmann relationship can extend to Po as high as ~0.2, as illustrated in Figure 14.2. Nevertheless, in practice, the choice of such a linear region may be less certain. The inevitable leak current will cause the linear asymptote to be actually curvilinear at low Po. The lack of explicit knowledge of Po also leaves the upper bound of the region largely empiric to determine. A small error in the choice of the region can lead to a large deviation in the fitting results because the activity of the channel is sharply dependent on temperature. Also of concern is that the analysis assumes that the gating of the channel can be considered at equilibrium as temperature is changed. The assumption is appropriate at high temperatures where the gating kinetics are relatively fast, but around the threshold temperature, the activation of the channel is slow relative to the rate of temperature change. Because the fitting is made within this range, the assumption that the channel is at equilibrium may fail if the rate of temperature change is not carefully controlled.

FIGURE 14.2. Boltzmann function plotted on both linear (black, left axes) and logarithmic (dark gray, right axes) scales.

FIGURE 14.2

Boltzmann function plotted on both linear (black, left axes) and logarithmic (dark gray, right axes) scales. On the log scale, the function is asymptotically linear at small values of Po (~<0.2).

14.3.2. Single-Channel Analysis

Without fast temperature jumps, single-channel analysis provides an alternative to resolve the kinetics of temperature gating. Because data are recorded at equilibrium, the control of temperature is relatively easy. With two miniature thermistors placed at the front and the back of a pipette tip, the temperature of the patch can be precisely monitored and clamped.28 Although recordings are made at equilibrium, their combinations across temperatures can lead to determination of the kinetic effects of temperature on gating. Single-channel analysis of TRPV1 shows that the apparent thermal sensitivity of whole-cell responses results predominantly from gating.28 The unitary conductance of the channel has temperature dependence similar to that of aqueous diffusion of electrolytes. The gating by temperature involves multiple states, at least three closed and three open states, similar to the gating by agonists (e.g., capsaicin and low pH).42,43 The channel in its open conformation is relatively independent of temperature. Instead, temperature mainly drives the long closures between opening bursts. As temperature is elevated, these closures become shortened so that the overall open probability increases. Statistical analysis of these long closures gives a temperature coefficient of Q10 > 9 (at +60 mV). By analogy to heat-induced protein unfolding, these events involve an enthalpy change similar to the denaturation of a globular protein of ~100 amino acids, suggesting that the gating by temperature may be accompanied with large structural rearrangements.

Single-channel experiments are technically challenging. For reliable analysis, good data quality is essential. For example, the patches have to be ensured to contain single channels. They have to be also extremely stable so that they can sustain a relatively long exposure to high temperatures (e.g., >50°C for TRPV1). Furthermore, because the response of thermal TRP channels such as TRPV1 often exhibits considerable variability across patches,34 it is important to have recordings for multiple experimental conditions (e.g., different temperatures) from the same patches so as to minimize the patch-related variations. In practice, the patches that can meet all these criterions are scarce, making the success rate of the experiments quite low. In addition, while single-channel data reveal gating events at high resolutions, their analysis can be difficult owing to the complex gating behavior of the channels.

14.3.3. Analysis of Thermal Sensitivity of Voltage Responses

Because thermal TRP channels are also activated by strong membrane depolarization, the thermal sensitivity of voltage responses provides another means for assessing the temperature dependence of the channels.3032 In this approach, the voltage-driven responses of the channel are measured at several ambient temperatures, which are typically below the threshold of thermal activation. The change of temperature affects the voltage responses by shifting the midpoint of channel opening versus voltage. The slope of the curve, which is interpreted as the effective gating charges, is independent of temperature. Because the voltage-dependent opening of the channels is observed even at room temperature, the determination of its thermal sensitivity requires temperature to be perturbed only at a moderate range (e.g., <40°C for TRPV1). The approach thus has the advantage of alleviating high-temperature exposures, which are experimentally challenging. On the other hand, it is also of concern that the measurements obtained at the moderate temperatures may contain a limited amount of information on temperature gating because the temperature sensors are only minimally activated at temperatures below the activation threshold.

The voltage-driven measurements from these experiments combine contributions from both voltage- and temperature-dependent gating. As a result, the interpretation of the data requires explicit modeling to separate them. At a minimum, it is necessary to know how the two activation pathways are interrelated. In one model, Voets et al.31 assume that all stimuli are directly coupled to the “gate” of the channel in an energetically additive manner. This is equivalent to a two-state model in which the opening or closing rate is driven simultaneously by all stimuli. A simple Boltzmann equation can describe the interplay between temperature and voltage gating,

where qFV accounts for the electrical energy of voltage sensors (q is the gating charge, V is the membrane potential, and F is Faraday’s constant), and other terms have the same definitions as in Equation 14.1.

An alternative model is the allosteric MWC scheme as mentioned above. In this model, the channel is assumed to possess distinct structural domains as sensors for different stimuli, and these domains are independent of each other but all allosterically coupled to a common “gate.” A stimulus by itself does not evoke conformational changes for opening; instead, it serves to only increase the equilibrium constants of the intrinsic opening. Assuming a single sensor for each stimulus, the model gives rise to an open probability that depends on voltage and temperature by

Image ch14_eq08

where L, KV, and KT are, respectively, the equilibrium constants of intrinsic opening and activation of voltage and temperature sensors, and c and d are the corresponding allosteric coupling factors. The temperature dependence of Po arises from that of KT.

Voets et al.31 resolved the voltage and temperature dependence in Equation 14.7 by fitting voltage activation time courses. Brauchi et al.30 and Matta and Ahern32 determined the MWC model from the steady-state measurements (i.e., the conductance–voltage curves at multiple temperatures). Despite different mechanistic assumptions, the two models give rise to very similar estimates for both temperature dependence and voltage dependence. They both predict that changing the voltage shifts the midpoint or threshold of thermal activation and, conversely, changing the temperature shifts the midpoint of the conductance-voltage curves. The slopes of these gating curves are generally not affected.

14.4. FAST TEMPERATURE JUMPS

One of the important unknowns for temperature gating has been the rate of activation. This information is essential to understanding the gating mechanisms of the channels. Single-channel data show that temperature-independent events have lifetimes on the order of milliseconds.28 This implies that the activation by temperature can be as fast as in a few milliseconds. Thus, to resolve the time course of thermal activation would require “instant” temperature changes at a submillisecond resolution. As reviewed below, several efforts have been undertaken toward development of fast temperature jumps.

14.4.1. T-Jumps by Solution Superfusion

One strategy to improve the speed of temperature perturbation is to exploit fast solution exchangers. Solutions at control and test temperatures pass through separate perfusate delivery tubes. A fast switching mechanism is then used to select which perfusion tube is directed at the sample. Hayes et al.44 demonstrated the technique for thermal activation of TRPV1 using a step motor-driven exchanger (VC-6, Warner Instruments). The exchanger has a time resolution of ~30 ms for a maximal ~700-mm step. Temperature control is achieved by preheating test solutions through an inline heater (TC-324B, Warner Instruments). With this approach, they obtained a temperature rise from room temperature to ~50°C in ~1 s (equivalent to a rate of 25°C/s). The approach is potentially applicable for bipolar temperature perturbations, but its time resolution remains inadequate relative to the activation rate of the channels.

Another attempt to speed up temperature changes by solution superfusion is to minimize the volume of the solution to be heated. Microfluidic technology allows for fabrication of submillimeter compartments for precise control and manipulation of fluids. Pennell et al.45 have developed such a microfluidic chip with a channel of 250 μm wide and 25 μm deep. A platinum heater and a thin-film resistive sensor are placed along the channel on the outlet end for inline heating and temperature monitoring. The inlet end of the channel is connected to an open solution reservoir. With a flow rate of ~10 μL/min, the chip is capable of increasing the solution temperature at the outlet end from bath temperature (20°C) to 80°C at an optimum heating rate of 0.5°C/ms. The device therefore still has an inadequate time resolution.

14.4.2. T-Jumps by Laser Irradiation

Laser irradiation has long been used for thermal perturbation of biological processes including gating of ion channels (e.g., Ref. 46) and in vivo pain sensation (e.g., Ref. 47). For resolving the rate of biomolecular reactions, ultrafast temperature jumps on the order of nanoseconds have been demonstrated using high-energy pulsed lasers, such as an Nd-YAG laser with a pulse energy of a few hundreds of milliwatts and a pulse duration of ~10 ns.48 Ultrafast laser T-jumps generated in this way have proven useful for studying protein folding events in the microsecond regime, where they have provided critical data on the time scales of such elementary processes as formation of secondary protein structures.49,50

However, the high-energy pulsed lasers are expensive and offer a time resolution unnecessarily high for studying a process such as ion channel gating that occurs in milliseconds. In addition, the nanosecond temperature jumps produced by a pulsed YAG laser are not steady on a time scale of milliseconds or longer, although the decay of temperature may be negligible on a microsecond time scale (where the ultrafast folding of small proteins or secondary structures takes place). The flash lamp pumped lasers have an intrinsically low repetition rate, typically a few tens of hertz. As a result, feedback modulation of these lasers for producing constant temperature steps is impossible. Applicability of this approach to electrophysiological experiments involving live cells has also not been demonstrated.

For moderately fast temperature jumps in a micro- to minisecond regime, the requirement on laser power becomes considerably less demanding. Semiconductor laser diodes may therefore provide a more cost-effective substitution to the high-energy pulsed YAG lasers. Solid-state infrared laser diodes have steadily grown in power in recent years and become increasingly popular for their industrial applications. In the field of nociception, they have been exploited as a heat irradiation tool to stimulate peripheral nerves in vivo and cultured nociceptive neurons in vitro. For example, Baumann and Maternson51 reported the use of a laser diode with 500 mW at 970 nm for heating a blackened tip of an optical fiber placed near the preparation. They were able to record heat-evoked action potentials in cultured trigeminal ganglion neurons, which provide the first direct evidence for a thermal transduction mechanism in such sensory neurons. Miura and Kawatani52 and Jimbo et al.53 also explored the effects of laser irradiation on cultured nodose ganglion and DRG neurons, although the power of their lasers was too low (16–150 mW) to elicit significant thermal effects. To obtain a high output power, Greffrath et al.54 combined six 980-nm laser diode outputs into a single fiber output, so that they could obtain a maximum irradiation power of 15 W. With this device, they demonstrated heat-evoked currents in isolated DRG neurons. The laser irradiation results in an approximately sublinear temperature ramp, which reaches a temperature jump of ~30°C in 400 ms (~75°C/s). The whole-cell currents evoked by such heat stimuli exhibited a half activation time of t1/2≈ 28 ms.

14.5. A SUBMILLISECOND TEMPERATURE JUMP SYSTEM

The previous attempts with either solution superfusion or laser diode heating have not been able to obtain a time resolution necessary to time resolve the activation of thermal TRP channels. The failure raises questions about the fundamental feasibility of these approaches. In the following, a theoretical evaluation of the laser diode irradiation approach is first presented, followed by a description of a laser diode heating system we have recently developed. The system can be considered as a temperature clamp apparatus—it achieves for the first time a submillisecond time resolution and also the capability of holding temperature steady after the initial rise.

14.5.1. Feasibility of Laser Diode Heating

The heating performance with laser irradiation may be assessed by numerical simulations.55 The analysis also provides insights on the choice of laser diodes in terms of wavelength and necessary optical power. Briefly, the temperature rise of the solution resulting from illumination by a laser beam can be modeled by the standard heat conduction equation, i.e.,

Image ch14_eq09

where cp is the specific heat capacity of water, κ is the thermal conductivity, ε is the optical absorption coefficient, and u is the spatial distribution of laser power. For a first-order approximation, one may assume a laser source producing a collimated emission beam, so that the spatial distribution of power has a simple cylindrical geometry, i.e.,

Image ch14_eq10

where R is the beam radius, and P is the total output power. Other assumptions that may be invoked to further simplify the equation include a negligible loss of power of the laser beam due to water absorption and a constant specific heat capacity of water independent of temperature. Under these conditions, the heat conduction equation may be readily solved by a partial differential equation solver.

High-power laser diodes that are applicable for heating typically have wavelengths around either 980 or 1460 nm. Today, a single emitter diode of 980 nm can output an optical power of >10 W. The longer wavelength diode generally has much less power (e.g., <3 W). The 980-nm wavelength overlaps with a minor absorption peak of water, while the 1460-nm wavelength is close to a major peak in the near-infrared region. The simulation suggests that with a diode of 3 W at 1460 nm, a temperature jump from room temperature to 60°C (saturating temperature for thermal TRP channels) could be achieved in <150 μs. The rise of temperature was almost linear with a rate of ~280°C/ms. On the other hand, with a 10-W 980-nm laser diode, the same temperature jump requires a considerably longer irradiation time (~4 ms), whereas the rise of temperature becomes sublinear. Thus, a diode of 1460 nm, albeit with a lower power, is much more efficient. Its superior performance occurs because of a higher absorption coefficient of water at longer wavelengths (0.46/cm at 980 nm and 30/cm at 1460 nm). The simulation supports the conclusion that a submillisecond temperature jump should be possible using a single emitter diode with a wavelength at ~1500 nm or an array of twenty 980-nm diodes.

The irradiation area is also an important factor for consideration of heating performance. A larger area is experimentally preferred but limits the brightness of the laser beam. The use of optical fibers for illumination typically limits the size to be around 100 μm in diameter. This size is large enough to cover most mammalian cells, while giving a light intensity strong enough for a submillisecond temperature rise. With the 1460-nm diode, the temperature within the laser beam (±50 μm) is nearly uniform, and outside the beam, it falls sharply. Thus, the spatial profile of the temperature changes resulting from laser diode irradiation is useable for single-cell experiments.

14.5.2. Laser Diode Instrumentation

The theoretical simulations suggest that it is possible to produce submillisecond temperature jumps using a single emitter laser diode with an appropriate wavelength. One implementation of such a system is diagramed in Figure 14.3, which employs a 1060-nm diode with a maximum 4-W output and is capable of a temperature jump from room temperature to ~60°C in 0.75 ms (Figure 14.4).55 The laser beam emitted from the diode is first collimated and then launched into a multimode fiber, which has a diameter of 100 μm and a numeric aperture NA = 0.22. Simple singleelement lenses, such as an aspheric lens, are adequate for both collimation and fiber coupling. A good launch efficiency (>75%) is readily achievable using a common XY translation lens mount (0.25 mm/revolution) and a flexcture Z-axis translator (1 μm/ revolution).

FIGURE 14.3. Schematic drawing of a submillisecond temperature jump system using a single emitter laser diode.

FIGURE 14.3

Schematic drawing of a submillisecond temperature jump system using a single emitter laser diode. The laser beam was launched into a multimode fiber with a core diameter of 100 μm, and the other end of the fiber was placed close to the samples (more...)

FIGURE 14.4. Activation of TRPV1 by fast temperature jumps.

FIGURE 14.4

Activation of TRPV1 by fast temperature jumps. Top: Macroscopic current responses of the channel from an outside-out patch (Vh = –100 mV). Bottom: A family of temperature jumps. (Adapted from Yao, J. et al., Biophysical Journal 96, 3611–3619, (more...)

The diode needs to be powered by a constant current source with a relatively high output current (>10 A) and a low compliance voltage (1–2 V). Such highpower laser diode drivers usually come in two configurations: continuous wave (CW) or pulsed. For ease of control and modulation, a quasi-pulsed controller is preferred, which allows for separate controls on current amplitude and output pulsing. A separate voltage power supply may be used for quick dial of the output current. The pulsing can be controlled through a computer. A computer-based instead of a microprocessor-based autonomous control for pulsing is preferred because of the need to store a relatively large amount of data as discussed below. To maintain a consistent optical output and also prevent overheating, the diode needs to operate at a constant temperature, typically room temperature by thermoelectrical cooling.

Feedback control on the diode output is implemented by modulating the pulse input of the power supply. A quasi-CW controller allows for a variable pulse width ranging from CW to a minimal duration as limited by the pulse rise times. Thus, both pulse frequency (duty cycle) and durations can be exploited for modulation using a common feedback control algorithm such as PID. For quick prototyping, a simple on-and-off modulation scheme proves adequate for maintaining a constant temperature. When the temperature is above its set point, the laser is pulsed off; otherwise, the laser is on. The control algorithms are readily implemented with any programmable digital processors, although a computer-based control through a National Instrument (NI) multifunctional acquisition card has been used in our implementation. NI provides both visual Labview and object-oriented library for low-level controls of its boards, thus making programming possible in high-level languages such as C/C++.

14.5.3. Temperature Calibration

The actual temperature at the beam spot may be calibrated using an open pipette with the tip placed at the beam center. The pipette is filled with a normal saline solution. A small voltage is applied to the pipette to induce a measurable current. Laser irradiation of the pipette tip causes an increase of the current. The temperature rise is related to the current change by

Image ch14_eq11

where To and T are the temperatures before and after laser irradiation, respectively, Io and I are the corresponding currents, and Q10 is the temperature coefficient of the electrolyte conductivity at To. The value of Q10 may be predetermined by placing an open electrode in a hot bath solution. As the solution is cooled, the current through the pipette and the ambient temperature are simultaneously recorded. The current is then plotted on a log scale against 1/T (Arrhenius plot). The plot is generally linear in the temperature range of interest. The slope of the plot then gives rise to the temperature dependence of the electrolyte conductance. With our normal saline buffer, we typically obtained a Q10 value of 1.2–1.4.

The laser beam spot on the cover slip can be located with the help of fluorescent cells. The red (mRFP) or yellow (YFP) fluorescent proteins are preferable choices because the green lasers, which are readily available at low cost, can be exploited for excitation of these fluorescent proteins. The green laser line also needs to be coupled to the illumination fiber. Once the beam area is identified, the center of the beam can be estimated. A video camera can greatly help the identification and also allow for more precise controls. The resultant center of the beam is marked on a computer screen for subsequent experiments.

Presently, it is technically difficult to simultaneously perform patch-clamp recording and real-time monitoring of temperature at the patches during actual experimentation. Miniature thermal sensors are either too bulky in size or too slow in response time. Thermally sensitive dyes are also difficult to use and suffer from problems such as bleaching. A more practical alternative we have found is to use a two-step procedure in which the desired temperature jumps are first generated with an open electrode and then applied to patched cells or excised patches by playing back the diode control protocols that were stored during the first step. The temperature jumps obtained in the first step appear to be quite reproducible in the second step. In practice, it is found that the reproducibility does not have a very high sensitivity to the patch positions, presumably owing to a relatively uniform temperature distribution within the beam. Proper alignments are readily achievable by positioning the patches at the beam center as previously marked on the computer screen during the first step. The vertical positions of the patch can be determined from the focal distance of the open pipette tip.

14.5.4. Rapid Cooling

To generate a square temperature pulse, it is necessary to rapidly cool the sample following a temperature jump. Passive cooling by heat dissipation is slow, taking hundreds of milliseconds to return to the ambient temperature. To accelerate cooling, it is necessary to superfuse cells with solutions at the control temperature. One possible solution is to use a fast solution exchanger in conjunction with the laser. The two can be synchronized so that during heating, the solution stream is outside the laser beam, whereas when the laser is switched off, the exchanger is simultaneously activated to move the solution stream to the beam center. The solution exchanger can be attached to a third manipulator mounted on the back of the microscope stage to produce a solution flow perpendicular to the fiber and the patch pipette. An exchange time <1 ms can be obtained with a double-barrel capillary installed on a piezo actuator.34

14.6. APPLICATIONS OF FAST TEMPERATURE JUMPS

The application of fast temperature jumps to study thermal TRP channels is still at a fledgling stage. One immediate application is to resolve the time course of temperature gating (Figure 14.4). It turns out that although a large amount of energy (~100 kcal/mol) is involved, the activation by temperature is surprisingly fast. With TRPV1, it occurs on the order of a few milliseconds. It is significantly faster than activation by chemical agonists. Capsaicin (100 μM) involves a half-activation time over a hundred milliseconds. Low pH is faster but still takes tens of milliseconds. Thus, temperature is the fastest stimulus to activate the channel. In contrast to its fast kinetics, the temperature response shows a more limited steady-state dynamic range, which is only approximately half of the capsaicin response at the same temperature.34,56 Temperature alone does not seem to fully open the channel.

With fast temperature jumps, one can expose patches to elevated temperatures only for a short period. This is important because the reduced thermal stress makes experimentation at high temperatures more feasible, for example, for measuring temperature response at hyperpolarizing membrane potentials, which occurs at a temperature range of 40°–55°C. The experiments at these temperatures provide information on the gating of the channel mainly by temperature (i.e., in absence of charge movement).

At hyperpolarizing potentials, TRPV1 shows relatively simple activation kinetics. In response to a temperature jump, the activation of current follows a single exponential time course. The temperature dependence of both steady-state response and time course can be described by a simple two-state model where the opening rate is driven by temperature while the closing has nominal but negative temperature dependence (i.e., sensitive to cold instead of heat). By analysis of thermal sensitivity of voltage gating, Voet et al.31 also reached the conclusion that the opening of the channel is mainly temperature dependent, although the temperature dependence of the closing rate from these experiments was found to be the opposite.

The energetics of thermal gating obtained from temperature jump measurements are considerably larger than those inferred from voltage-driven responses at different temperatures. Between closed and open, the enthalpy of the channel increases by ~100 kcal/mol. This energy is equivalent to an electrical energy of moving ~71 unit charges across 60 mV (i.e., about 5 times the energy of voltage-gated channels) or melting of 10 phospholipids.57 The gating by temperature thus evolves from a low enthalpy and entropy closed state to a high enthalpy and strongly entropic (disordered) open state. The opening of the channel involves an activation enthalpy of ~85 kcal/mol, which accounts for most of the large equilibrium enthalpy change between closed and open. Thus, the open state of the channel is energetically similar to the transition state. The channel becomes open only at the very end of the large energetic change. Despite this large amount of energy involved, the activation by temperature (thermal energy) is made fast because the entropy change mostly cancels the enthalpy change leaving the free energy difference quite moderate.

14.7. SUMMARY

Thermal TRP channels are activated on a time course of milliseconds and over a broad temperature range from a few Celsius degrees to >50°C. Such fast time responsiveness in conjunction with the wide temperature dynamics has brought challenges to instrumentation of rapid temperature controls for time-resolved measurements of these channels in live cells. Thermoelectric or resistive approaches are slow in time, thereby limiting their uses mostly to steady-state temperature-dependent measurements. They also inevitably incur prolonged thermal stresses on patches and channels, which can give rise to unintended thermal effects such as leak currents and/or rundown of channels, thereby complicating analysis of data. Optical approaches such as laser diode irradiation are more efficient for rapid temperature perturbation. With a proper choice of wavelength, a single-emitter laser diode is capable of producing a temperature jump with a rise time in submilliseconds. Implementation of constant temperature jumps is also possible by exploiting the capability of laser diodes for rapid modulation. A practical, useable system by laser irradiation can be readily implemented using off-shelf optics and electronic drivers. With the availability of such systems, we are now in a position for time-resolved measurements of temperature responses.

REFERENCES

1.
Spray D.C. Cutaneous temperature receptors. Annual Review of Physiology. 1986;48:625–638. [PubMed: 3085583]
2.
Dubner R, Sumino R, Wood W.I. A peripheral “cold” fiber population responsive to innocuous and noxious thermal stimuli applied to monkey's face. Journal of Neurophysiology. 1975;38:1373–1389. [PubMed: 815515]
3.
Hensel H. Thermoreceptors. Annual Review of Physiology. 1974;36:233–249. [PubMed: 19400663]
4.
Clapham D.E. TRP channels as cellular sensors. Nature. 2003;426:517–524. [PubMed: 14654832]
5.
Caterina M.J, Schumacher M.A, Tominaga M, Rosen T.A, Levine J.D, Julius D. The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature. 1997;389:816–824. [PubMed: 9349813]
6.
Cesare P, McNaughton P. A novel heat-activated current in nociceptive neurons and its sensitization by bradykinin. Proceedings of the National Academy of Sciences of the United States of America. 1996;93:15435–15439. [PMC free article: PMC26422] [PubMed: 8986829]
7.
Caterina M.J, Rosen T.A, Tominaga M, Brake A.J, Julius D. A capsaicin-receptor homologue with a high threshold for noxious heat. Nature. 1999;398:436–441. [PubMed: 10201375]
8.
Smith G.D, Gunthorpe J, Kelsell R.E, editors. et al. TRPV3 is a temperature-sensitive vanilloid receptor-like protein. Nature. 2002;418:186–190. [PubMed: 12077606]
9.
Xu H.X, Ramsey I.S, Kotecha S.A, editors. et al. TRPV3 is a calcium-permeable temperature-sensitive cation channel. Nature. 2002;418:181–186. [PubMed: 12077604]
10.
Peier A.M, Reeve A.J, Andersson D.A, editors. et al. A heat-sensitive TRP channel expressed in keratinocytes. Science. 2002;296:2046–2049. [PubMed: 12016205]
11.
Guler A.D, Lee H, Iida T, Shimizu I, Tominaga M, Caterina M. Heatevoked activation of the ion channel, TRPV4. Journal of Neuroscience. 2002;22:6408–6414. [PubMed: 12151520]
12.
Xu H, Delling M, Jun J.C, Clapham D.E. Oregano, thyme and clovederived flavors and skin sensitizers activate specific TRP channels. Nature Neuroscience. 2006;9:628–635. [PubMed: 16617338]
13.
Strotmann R, Harteneck C, Nunnenmacher K, Schultz G, Plant T.D. OTRPC4, a nonselective cation channel that confers sensitivity to extracellular osmolarity. Nature Cell Biology. 2000;2:695–702. [PubMed: 11025659]
14.
Liedtke W, Choe Y, Marti-Renom M.A, editors. et al. Vanilloid receptor-related osmotically activated channel (VR-OAC), a candidate vertebrate osmoreceptor. Cell. 2000;103:525–535. [PMC free article: PMC2211528] [PubMed: 11081638]
15.
McKemy D.D, Neuhausser W.M, Julius D. Identification of a cold receptor reveals a general role for TRP channels in thermosensation. Nature. 2002;416:52–58. [PubMed: 11882888]
16.
Peier A.M, Moqrich A, Hergarden A.C, editors. et al. A TRP channel that senses cold stimuli and menthol. Cell. 2002;108:705–715. [PubMed: 11893340]
17.
Story G.M, Peier A.M, Reeve A.J, editors. et al. ANKTM1, a TRP-like channel expressed in nociceptive neurons, is activated by cold temperatures. Cell. 2003;112:819–829. [PubMed: 12654248]
18.
Latorre R. Perspectives on TRP channel structure and the TRPA1 puzzle. Journal of General Physiology. 2009;133:227–229. [PMC free article: PMC2654079] [PubMed: 19237586]
19.
Jordt S.E, Bautista D.M, Chuang H.H, editors. et al. Mustard oils and cannabinoids excite sensory nerve fibres through the TRP channel ANKTM1. Nature. 2004;427:260–265. [PubMed: 14712238]
20.
Caterina M.J, Leffler A, Malmberg A.B, editors. et al. Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science. 2000;288:306–313. [PubMed: 10764638]
21.
Davis J.B, Gray J, Gunthorpe M.J, editors. et al. Vanilloid receptor-1 is essential for inflammatory thermal hyperalgesia. Nature. 2000;405:183–187. [PubMed: 10821274]
22.
Moqrich A, Hwang S.W, Earley T.J, editors. et al. Impaired thermosensation in mice lacking TRPV3, a heat and camphor sensor in the skin. Science. 2005;307:1468–1472. [PubMed: 15746429]
23.
Bautista D.M, Siemens J, Glazer J.M, editors. et al. The menthol receptor TRPM8 is the principal detector of environmental cold. Nature. 2007;448:204–208. [PubMed: 17538622]
24.
Colburn R.W, Lubin M.L, Stone D.J Jr., editors. et al. Attenuated cold sensitivity in TRPM8 null mice. Neuron. 2007;54:379–386. [PubMed: 17481392]
25.
Dhaka A, Murray A.N, Mathur J, Earley T.J, Petrus M.J, Patapoutian A. TRPM8 is required for cold sensation in mice. Neuron. 2007;54:371–378. [PubMed: 17481391]
26.
Bautista D.M, Jordt S.E, Nikai T, editors. et al. TRPA1 mediates the inflammatory actions of environmental irritants and proalgesic agents. Cell. 2006;124:1269–1282. [PubMed: 16564016]
27.
Kwan K.Y, Allchorne A.J, Vollrath M.A, editors. et al. TRPA1 contributes to cold, mechanical, and chemical nociception but is not essential for hair-cell transduction. Neuron. 2006;50:277–289. [PubMed: 16630838]
28.
Liu B, Hui K, Qin F. Thermodynamics of heat activation of single capsaicin ion channels VR1. Biophysical Journal. 2003;85:2988–3006. [PMC free article: PMC1303577] [PubMed: 14581201]
29.
Benham C.D, Gunthorpe M.J, Davis J.B. TRPV channels as temperature sensors. Cell Calcium. 2003;33:479–487. [PubMed: 12765693]
30.
Brauchi S, Orio P, Latorre R. Clues to understanding cold sensation: thermodynamics and electrophysiological analysis of the cold receptor TRPM8. Proceedings of the National Academy of Science of the United States of America. 2004;101:15494–15499. [PMC free article: PMC523461] [PubMed: 15492228]
31.
Voets T, Droogmans G, Wissenbach U, Janssens A, Flockerzi V, Nilius B. The principle of temperature-dependent gating in cold- and heat-sensitive TRP channels. Nature. 2004;430:748–754. [PubMed: 15306801]
32.
Matta J.A, Ahern G.P. Voltage is a partial activator of rat thermosensitive TRP channels. Journal of Physiology. 2007;585:469–482. [PMC free article: PMC2375500] [PubMed: 17932142]
33.
Latorre R, Brauchi S, Orta G, Zaelzer C, Vargas G. ThermoTRP channels as modular proteins with allosteric gating. Cell Calcium. 2007;42:427–438. [PubMed: 17499848]
34.
Yao J, Liu B.L, Qin F. Kinetic and energetic analyses of thermally activated TRPV1 channels. Biophysical Journal. 2010;99:1743–1753. [PMC free article: PMC2941006] [PubMed: 20858418]
35.
Brauchi S, Orta G, Salazar M, Rosenmann E, Latorre R. A hot-sensing cold receptor: C-terminal domain determines thermosensation in transient receptor potential channels. Journal of Neuroscience. 2006;26:4835–4840. [PubMed: 16672657]
36.
Susankova K, Ettrich R, Vyklicky L, Teisinger J, Vlachova V. Contribution of the putative inner-pore region to the gating of the transient receptor potential vanilloid subtype 1 channel (TRPV1) Journal of Neuroscience. 2007;27:7578–7585. [PubMed: 17626219]
37.
Grandl J, Hu H, Bandell M, Bursulaya B, Schmidt M, Petrus M, Patapoutian A. Pore region of TRPV3 ion channel is specifically required for heat activation. Nature Neuroscience. 2008;11:1007–1013. [PMC free article: PMC2685190] [PubMed: 19160498]
38.
Salazar H, Jara-Oseguera A, Hernandez-Garcia E, editors. et al. Structural determinants of gating in the TRPV1 channel. Nature Structural and Molecular Biology. 2009;16:704–710. [PubMed: 19561608]
39.
Dittert I, Vlachova V, Knotkova H, Vitaskova Z, Vyklicky L, Kress M, Reeh P.W. A technique for fast application of heated solutions of different composition to cultured neurones. Journal of Neuroscience Methods. 1998;82:195–201. [PubMed: 9700692]
40.
Reid G, Amuzescu B, Zech E, Flonta M.L. A system for applying rapid warming or cooling stimuli to cells during patch clamp recording or ion imaging. Journal of Neuroscience Methods. 2001;111:1–8. [PubMed: 11574114]
41.
Dittert I, Benedikt J, Vyklicky L, Zimmermann K, Reeh P.W, Vlachova V. Improved superfusion technique for rapid cooling or heating of cultured cells under patch-clamp conditions. Journal of Neuroscience Methods. 2006;151:178–185. [PubMed: 16129494]
42.
Hui K.Y, Liu B.Y, Qin F. Capsaicin activation of the pain receptor, VR1: Multiple open states from both partial and full binding. Biophysical Journal. 2003;84:2957–2968. [PMC free article: PMC1302858] [PubMed: 12719227]
43.
Ryu S.J, Liu B.Y, Qin F. Low pH potentiates both capsaicin binding and channel gating of VR1 receptors. Journal of General Physiology. 2003;122:45–61. [PMC free article: PMC2234467] [PubMed: 12835470]
44.
Hayes P, Meadows H.J, Gunthorpe M.J, editors. et al. Cloning and functional expression of a human orthologue of rat vanilloid receptor-1. Pain. 2000;88:205–215. [PubMed: 11050376]
45.
Pennell T, Suchyna T, Wang J, Heo J, Felske J.D, Sachs F, Hua S.Z. Microfluidic chip to produce temperature jumps for electrophysiology. Analytical Chemistry. 2008;80:2447–2451. [PubMed: 18302344]
46.
Moore L.E, Holt J.P Jr., Lindley B.D. Laser temperature-jump technique for relaxation studies of the ionic conductances in myelinated nerve fibers. Biophysical Journal. 1972;12:157–174. [PMC free article: PMC1484058] [PubMed: 5057748]
47.
Bromm B, Neitzel H, Tecklenburg A, Treede R.D. Evoked cerebral potential correlates of C-fibre activity in man. Neuroscience Letters. 1983;43:109–114. [PubMed: 6669318]
48.
Kubelka J. Time-resolved methods in biophysics. 9. Laser temperature-jump methods for investigating biomolecular dynamics. Photochemistry and Photobiology Science. 2009;8:499–512. [PubMed: 19337664]
49.
Williams S, Causgrove T.P, Gilmanshin R, Fang K.S, Callender R.H, Woodruff W.H, Dyer R.B. Fast events in protein folding: helix melting and formation in a small peptide. Biochemistry. 1996;35:691–697. [PubMed: 8547249]
50.
Munoz V, Thompson P.A, Hofrichter J, Eaton W.A. Folding dynamics and mechanism of beta-hairpin formation. Nature. 1997;390:196–199. [PubMed: 9367160]
51.
Baumann T.K, Martenson M.E. Thermosensitivity of cultured trigeminal neurons. Society of Neuroscience Abstracts. 1994;20:1379.
52.
Miura A, Kawatani M. Effects of diode laser irradiation on sensory ganglion cells from the rat. Pain Research. 1996;11:175–183.
53.
Jimbo K, Noda K, Suzuki K, Yoda K. Suppressive effects of low-power laser irradiation on bradykinin evoked action potentials in cultured murine dorsal root ganglion cells. Neuroscience Letters. 1998;240:93–96. [PubMed: 9486480]
54.
Greffrath W, Nemenov M.I, Schwarz S, Baumgartner U, Vogel H, Arendt-Nielsen L, Treede R.D. Inward currents in primary nociceptive neurons of the rat and pain sensations in humans elicited by infrared diode laser pulses. Pain. 2002;99:145–155. [PubMed: 12237192]
55.
Yao J, Liu B, Qin F. Rapid temperature jump by infrared diode laser irradiation for patch-clamp studies. Biophysical Journal. 2009;96:3611–3619. [PMC free article: PMC2711624] [PubMed: 19413966]
56.
Tominaga M, Caterina M.J, Malmberg A.B, editors. et al. The cloned capsaicin receptor integrates multiple pain-producing stimuli. Neuron. 1998;21:531–543. [PubMed: 9768840]
57.
Heimburg T. Thermal Biophysics of Membranes. Weinheim, Germany: Wiley-VCH; 2007.
Copyright © 2011 by Taylor and Francis Group, LLC.
Bookshelf ID: NBK92811PMID: 22593957

Views

  • PubReader
  • Print View
  • Cite this Page

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Similar articles in PubMed

See reviews...See all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...