Schmelz, M., R. Schmidt, A. Bickel, H. E. Torebjörk, and H. O. Handwerker. Innervation territories of single sympathetic C fibers in human skin. J. Neurophysiol. 79: 1653–1660, 1998. Microneurography techniques were used to record action potentials from unmyelinated nerve fibers (C fibers) in the cutaneous fascicles of the peroneal nerve in healthy volunteers. C units were identified by their long latency responses to electrical stimulation of their terminals in the skin. Their responsiveness to mechanical or heat stimuli applied to the skin or to sympathetic reflex provocation tests was determined by transient slowing of conduction velocity following activation (marking technique). In a sample of 381 C units, 59 were unresponsive to mechanical and thermal stimulation of their endings, but responded to sympathetic reflex provocation tests, e.g., arousal or deep inspiration. They were classified as sympathetic efferent units. On average, conduction velocities of sympathetic units were lower (0.78 ± 0.12 m/s, mean ± SD) than those of mechano-heat (CMH) or mechanoresponsive (CM) afferent C units (0.91 ± 0.14 m/s). Endings of most of the sympathetic units were located in the skin of toes or in the foot dorsum. Innervation territories of 16 sympathetic units were mapped by means of conditioning transcutaneous electrical stimuli. Twelve units had one continuous skin territory, whereas two units had two and two other units had three and five separate territories, respectively. The mean innervated area was 128 mm2 (range: 24–350 mm2). Innervation territories of sympathetic units were of approximately the same size in different skin regions on the lower leg, foot, or toes. Based on responses to whole body cooling and warming, two units were tentatively classified as vasoconstrictor and sudomotor units, respectively. Eleven units were tested for responsiveness to iontophoresis of acetylcholine in their innervation territories. In five of them, activity was induced that was not due to central reflex activity but instead due to antidromic activation from the peripheral terminals. Iontophoresis of saline or histamine was ineffective. These findings confirm the existence of excitatory cholinergic receptors in the terminal membrane of some sympathetic units, possibly sudomotors.
In 1967 Hagbarth and Vallbo first described recordings of spike activity from single myelinated nerve fibers in humans (Hagbarth and Vallbo 1967). Further refinement of the technique has allowed recordings of integrated multiunit activity from C fibers to study sympathetic outflow to muscle or skin (Hagbarth et al. 1972; Normell and Wallin 1974; Wallin et al. 1974). This technique has provided insights into autonomic regulation as well as pathophysiologic mechanisms of diseases like hypertension (Wallin 1989) and autonomic dysfunction (Wallin and Elam 1993).
However, detailed microneurographic analysis of activity in single sympathetic C fibers has been impeded by their small spike amplitudes and their synchronized discharges. Therefore human sympathetic activity has been usually recorded as multiunit discharges, and there are very few previous (Hallin and Torebjörk 1970, 1974a,b) or more recent (Macefield and Wallin 1996; Macefield et al. 1994) studies of single-unit recordings.
An additional drawback of single sympathetic fiber recordings is the changing spike waveform during long-term recording, making it difficult to confidently attribute discharges to one and the same individual unit. To some extent, these problems have been overcome by the “marking technique,” which identifies single C units by their long response latencies to repeated electrical test stimuli. Reflex activation of sympathetic C fibers, or discharges in afferent C fibers evoked by exciting their peripheral endings in the skin, causes transient slowing of the conduction velocity of the activated unit that allows identification of afferent versus sympathetic C units (Hallin and Torebjörk 1974b). Recently, a computer-assisted form of the marking technique was introduced, which increased the efficiency and power of resolution of this method (Schmelz et al. 1995; Schmidt et al. 1995).
This paper reports novel findings on the innervation territories of single cutaneous sympathetic C-efferent units in humans obtained with the marking technique while also allowing a tentative classification into vasoconstrictor or sudomotor categories. In addition, we present evidence that some sympathetic axons, possibly sudomotor units, can be antidromically activated by stimulation of their terminals in the skin with acetylcholine (ACh).
Standard microneurography techniques were employed for recording from C fibers in cutaneous fascicles of the peroneal nerve dorsolateral to the fibular head in healthy subjects as described previously (Torebjörk 1974). Recordings of activity in sympathetic units from 18 subjects (age range 21–29 yr) are reported in this paper. All subjects gave their informed consent, and the study was approved by the local ethics committees.
Electrical search procedure
The search procedure for identifying single C units has been described before in detail (Schmidt et al. 1995). Electrical stimuli (0.3 Hz, 0.3 ms, 40–100 V; delivered from a Grass S88 stimulator) were applied transcutaneously within the innervation territory of the impaled fascicle (lower leg or dorsum of foot) via a pointed steel electrode that was moved around on the skin, until a C-fiber response was obtained. At that site, two steel needles were inserted (5 mm apart) for intracutaneous electrical stimulation. Stimuli eliciting stable C-fiber responses were delivered at 4-s intervals (0.2-ms pulses of 10–40 V). These stimuli were well tolerated by the subjects over extended periods. They were described as slightly unpleasant, but not really painful. It is important to note that this technique works without recourse to mechanical stimulation of receptive fields that would limit the search to mechanosensitive afferent units.
Responsiveness of single C units to different stimuli was ascertained by the “marking” phenomenon described in detail elsewhere (Hallin and Torebjörk 1974b; Schmelz et al. 1995; Schmidt et al. 1995; Torebjörk 1974; Torebjörk and Hallin 1974). The response latency to regular 0.25-Hz electrical stimulation of the nerve terminals is transiently increased if additional action potentials are induced in the respective axon. The membrane mechanism for the conduction slowing is not entirely clear. A relatively long-lasting hyperpolarization of the axon due to activity-dependent ion pumps has been suggested to account for the slowing of C fibers in rat (Thalhammer et al. 1994). Our group has demonstrated that increased response latencies are reliably detected even after interpolation of only one single action potential in afferent C fibers (Schmelz et al. 1995). Correspondingly, ongoing activity of sympathetic units is detectable in the continuous records of evoked activity (Fig. 1), although collisions between antidromic and orthodromic impulses may conceal the latency changes if the electrical stimulation frequency is high (Hallin and Torebjörk 1974b). As seen in Fig. 1, each transient latency shift may be caused by a variable number of impulses, implying that the marking technique can yield at best semiquantitative data on unit activation, but no exact measure of discharge frequency.
C-unit responses to intracutaneous electrical stimulation were recorded on-line by a PC computer via an interface card (DAP, Microstar) with the SPIKE/SPIDI software package (Forster and Handwerker 1990). A suitable time segment after the electrical test pulse was selected and displayed on the computer screen in succeeding traces from top to bottom (Figs. 3, 5, and 6) allowing on-line assessment of latency changes.
Stimuli known to increase the skin sympathetic, sudomotor, and vasoconstrictor outflow in conscious humans were used to identify sympathetic efferents (Hagbarth et al. 1972; Hallin and Torebjörk 1970), i.e., unexpected loud noises, mental stress (by making rapid mathematical calculations), and inciting the subject to laugh or to inspire deeply. The efficacy of these maneuvers was verified by listening to the sympathetic multifiber background discharges from a loudspeaker. Mechanical and heating stimuli known to be suprathreshold for nociceptors were applied to the innervation territories of all efferent sympathetic units to test for possible excitability by natural stimulation. For mechanical stimulation a set of calibrated von Frey nylon filaments was used (Stoelting, Chicago, IL). Probing inside the innervation territory of sympathetic efferents with a stiff filament (750 mN) did not evoke discharges in any sympathetic unit. Heat stimuli were delivered from a halogen lamp controlled by feedback from a thermocouple attached to the skin (Beck et al. 1974). Skin temperature was increased by 0.25°C/s, from an adapting temperature of 32°C to a tolerance level of ∼50°C at which point subjects terminated the stimulus. None of the sympathetic units showed a graded response to this kind of stimulus that typically activates polymodal afferent C fibers. Few of the sympathetic fibers showed irregular discharges usually starting 1–2 s after peak temperature and outlasting the stimulus for up to 10 s (not shown). These discharges probably were due to sympathetic reflex activity following painful stimulation. During this poststimulus period, heat-sensitive C afferents are always silent.
Mapping innervation territories
Innervation territories were mapped by conditioning electrical stimulation as described previously (Schmelz et al. 1994). The tip of a pointed steel electrode (1 mm2) was moistened with electrode gel and gently attached to the skin for focal electrical stimulation with the indifferent electrode (steel plate, 3 × 5 cm) placed on the lower thigh. Trains of three pulses were delivered from a constant current stimulator (Type 15E25 insulated stimulator output unit, Disa, Denmark) at a moderately unpleasant intensity (5–10 mA, duration 0.2 ms, interval 200 ms). To exclude activation of sympathetic efferents via a reflex induced by the mapping stimulus, the trains were time locked to the regular electrical pulses with an interval of 50 ms between the last of the three mapping pulses and the next regular pulse. The principle of this technique is shown in Fig. 2.
Test sites were spaced at 2-mm intervals, and sites at which transcutaneous electrical stimuli activated the respective C unit were marked on the skin. Testing was carefully performed this way for distances of at least 2 cm in all directions away from the border of the documented innervation territory, to detect remote territories. Depending on the size of the innervation territory, such an analysis often required hundreds of test stimuli and stable recordings for several hours. During this time the temperature of the laboratory was maintained at a constant thermoneutral level, and visual and acoustic were restricted to minimize reflex activation of the unit under study. Innervation territories were traced on acetate sheets (transparencies) at the end of the experiment, and areas were planimetrically assessed after digitizing.
Body warming and cooling
To classify the sympathetic efferents, the development of ongoing activity during whole body warming and cooling was studied in three experiments. Subjects were warmed by use of preheated blankets (40°C) and warm packs that were placed on thorax and abdomen for at least 30 min. Cooling was achieved by wetting the skin with ice water and increasing evaporation by directing an air stream from a fan to the wet skin. The leg from which the C fiber was recorded was excluded from these warming and cooling maneuvers. The effectiveness of these procedures was checked by measuring skin temperatures in the innervation territory, which decreased from a baseline level of ∼30°C to values <25°C during the cooling procedure (35 min) and reached >33°C after 45 min of warming.
Because it has been assumed that the terminal membrane of sudomotor units contains excitatory cholinergic receptors responsible for axon reflex sweating (Coon and Rothman 1939), acetylcholine (ACh, 10% wt/vol in distilled water) was iontophoretically applied to the innervation territory of 11 units. The applicator was identical to a device that has been used for histamine iontophoresis by our group (Handwerker et al. 1987; Magerl et al. 1990) and also for iontophoresis of ACh in psychophysical experiments (Vogelsang et al. 1995). It had a diameter of 6 mm. A positive current of 1 mA was applied for 60 s, resulting in a transferred charge of 60 mC. Iontophoretic applications of saline (9 units) and histamine (4 units) were used as controls.
Identification of sympathetic C units
In a sample of 381 C units, 59 sympathetic efferents were identified by their responsiveness to sympathetic provocation tests. A typical example shown in Fig. 3 demonstrates clear increases in latency in association with sudden noise, the Valsalva maneuver, and during mental stress. Under control conditions (20–22°C room temperature, absence of arousal stimuli) most sympathetic efferents were spontaneously active (asterisk in Fig. 3). None of the sympathetic units showed responses to mechanical stimulation in the innervation territory. Conversely, none of the afferent C fibers investigated with identical techniques responded to sympathetic provocation tests.
The numbers of sympathetic efferents located in the toes, in the dorsum of the foot, and in the lower leg were 13, 40, and 6, respectively. Compared with a corresponding sample of afferent mechano-heat (CMH) and mechanoresponsive (CM) units (12, 107 and 30 in toes, foot, and lower leg, respectively), a higher proportion of sympathetic efferents was found in the toe region (toe vs. foot, P = 0.005; toes vs. lower leg, P < 0.001; χ2 test). Because all afferent and efferent units were searched with transcutaneous electrical stimuli, there was no regional bias that could explain this result.
Conduction velocities of the sympathetic fibers ranged from 0.2 to 0.95 m/s with a mean of 0.78 ± 0.12 (SD)m/s, which was significantly lower (P < 0.001; unpaired t-test) compared with the conduction velocities of mechanosensitive C fibers (CMH and CM, n = 149), which were 0.91 ± 0.14 m/s (data pooled from Schmidt et al. 1995, 1997).
Evidence against conduction block in sympathetic fibers at the recording site
This study employing the marking phenomenon provides information related to a recent controversy on the interpretation of microneurographic recordings. It has been suggested that conduction in most of the recorded units is blocked at the recording site (Wall and McMahon 1985). However, our results clearly show that sympathetic provocation tests cause a slowing of conduction velocity distal to the recording site in all sympathetic units. Because reflex activity in efferent fibers travels distally, action potentials of the units under study were apparently conducted across the recording site to the periphery without being blocked.
To successfully map innervation territories of sympathetic C units, a fairly good signal-to-noise ratio and low level of ongoing discharge is necessary to reliably detect the spots from which the unit was antidromically activated, and hence determine the exact boundaries of the innervation territory. The units successfully mapped in this sample had a signal-to-noise ratio of at least 2:1 and an average ongoing activity of less than two “markings” per minute throughout the mapping procedure. Thirty-two of 59 units had lower signal-to-noise ratios or irregular spontaneous activity of higher frequency, which made the mapping unreliable, and they were not considered for further analysis. Another obvious prerequisite for successful mapping was a stable recording for this time-consuming procedure. In nine units the mapping was incomplete because the units were lost.
In 18 units we were able to carry out the complete mapping protocol. Two sympathetic units, responsive to reflex activation by deep inspiration and sudden noises, could not be excited by transcutaneous electrical stimulation even at tolerance intensity.
The innervation territories of 16 successfully mapped sympathetic units are shown in Fig. 4. Twelve of the units had continuous innervation territories. However, the remaining 4 units had 2 (n = 2), 3, or 5 separate, discontinuous territories (Fig. 4). The shapes of the territories were highly irregular with extensions in various directions. No preference was observed for extensions along the presumed course of the parent axon. The average innervation area was 128 mm2 (range 24–350 mm2). The median diameter in the direction of maximal extension was 21 mm (range 9–32 mm). In contrast to afferent CMH and CM units, sympathetic units showed no trend toward smaller innervation territories at more distal sites on the leg. Details on area and diameter of the innervation territories in the different regions of the leg are given in Table 1.
Classification of sympathetic units by their responsiveness during thermoregulation
In three experiments it was possible to assess thermoregulatory responses (to whole body warming and cooling) in sympathetic fibers after their innervation territories had been mapped. As expected, the temperature changes in the skin led to gradual and prolonged latency shifts of the C-unit responses due to changes of conduction velocity (Fig. 5). These gradual shifts were readily distinguished from the marking phenomenon, which is characterized by a transient increase in latency due to reflex activation. One unit had a conduction velocity of 0.78 m/s and a continuous innervation territory of 223 mm2. This unit showed almost no activity when the subject was warmed, whereas cooling markedly increased activity (Fig. 5). According to its activation pattern, the unit was tentatively classified as a vasoconstrictor unit. It should be noted that this unit still could be activated by arousal stimuli during the warming period. A second unit with a conduction velocity of 0.68 m/s and an innervation territory of 74 mm2 showed the opposite activation pattern by being depressed during cooling and activated during warming. The activation pattern of this unit is therefore consistent with a sudomotor or vasodilatator unit.
A third unit with a territory of 25 mm2 did not respond at all to body cooling or warming and exhibited no spontaneous activity during the 3-h recording period. However, reflex discharges were reliably elicited by arousal stimuli independent of whole body heating or cooling. This unit was also unresponsive to ACh iontophoresis (not shown).
Responsiveness to iontophoresis of ACh
To test for excitatory cholinergic receptors in the terminal membranes, ACh was applied by iontophoresis (1 mA, 60 s; see Methods) in the innervation territories of 11 sympathetic units. Five of the 11 units tested were clearly activated after termination of iontophoresis, which excludes a direct action of the current. Figure 6 shows an example. Activation typically started with a delay of a few seconds after termination of the current application and lasted for 3–5 min.
This prolonged response was apparently due to depolarization of the terminals by ACh leading to antidromically conducted spikes. It was unlikely to have been due to reflex activation, particularly because iontophoresis of ACh usually did not induce any sensations that outlasted the period of current application. Sensations during ACh application were not reported to be painful. Furthermore, there were no signs of increased discharges in sympathetic fibers during control iontophoresis of saline, providing additional evidence for a direct pharmacological action of ACh on the sympathetic nerve terminals. None of the four sympathetic units additionally tested with iontophoresis of histamine (1 mA, 20 s) were excited, despite the fact that histamine induced a wheal at the application site and a flare reaction in the surrounding area (Magerl et al. 1990).
Use of the marking technique for analysis of activity in single sympathetic efferent units
The marking technique was originally developed to differentiate between electrically evoked afferent and sympathetic efferent C-unit responses in human skin nerves (Hallin and Torebjörk 1974b) and was used presently to study activity in single sympathetic efferent axons. Although this technique allows only qualitative or semiquantitative assessment of C-unit activity (Schmelz et al. 1995), it has the advantage of allowing long-term recordings from single units. Over recording periods of several hours, spike activity can be definitively allocated to a single axon based on the characteristic latency of that particular unit after electrical stimulation of its terminals. Because of the long conduction distance between fiber terminals in the foot or lower leg and the recording site at the knee level, superposition of spikes from different axons is rare.
However, the marking technique has limitations. First, each marking in our recordings may have derived from one or more impulses; the number of discharges in the unit under study cannot be accurately estimated by marking. Furthermore, collision between electrically evoked antidromic and orthodromic action potentials might lead to an underestimate of ongoing activity. If this happens on the way from the recording electrode to the sympathetic ganglia, it would induce no change of the spike latency to stimulation of the terminals at the recording site, and hence the discharge rate may be underestimated. If it happens, however, on the way from the skin to the recording electrode, it would lead to a drop out of one electrically induced spike response. As discussed in detail by Hallin and Torebjörk (1974b), the probability of detecting antidromic activity in the centripetally conducted C response of sympathetic fibers will be reduced if the stimulation frequency is high, and if recordings are made at a long distance from the ganglion. We have tried to minimize these limitations by using low-frequency (0.25 Hz) stimulation and recording at knee level. However, measuring the response latency every 4 s reduces the temporal resolution of the method. Each marking indicates that one or more impulse occurred in the sympathetic fiber during the preceding 4 s, and the number of markings cannot exceed 15 per minute.
Bias in the present sample of mapped sympathetic units
Another obvious limitation of the marking technique is that units with a high level of spontaneous activity will be excluded from the analysis of innervation territories, because the ongoing barrage precludes stable latencies in response to intracutaneous electrical stimulation. The sample of 16 mapped sympathetic units reported here is thus biased in favor of units with low ongoing activity. Further evidence for this bias is that the mean ongoing activity was less than two markings per minute, which is lower than the mean discharge rates of single sympathetic fibers (9–36 imp/min) recorded in humans with conventional microneurography methods (Hallin and Torebjörk 1974a; Macefield and Wallin 1996). However, such a methodological bias may not necessarily be relevant to the entire sample of 58 sympathetic units. They were all searched by transcutaneous electrical stimuli in the same way as we searched for afferent C-fiber activity. Therefore we feel justified in comparing the innervation densities of afferent and efferent C fibers in different regions of the leg, and we thus suggest that the higher proportion of sympathetic efferents found in the toe region may be real and not a sampling artifact.
Innervation territories of sympathetic efferent units
The innervation territories of 16 sympathetic fibers were mapped by conditioning transcutaneous electrical stimuli. This mapping procedure was used in previous studies to analyze receptive-field organization of afferent C fibers that were responsive or unresponsive to mechanical stimuli (Schmelz et al. 1994, 1996). In case of the CM or CMH sensitive units, the borders of the innervation territories determined by electrical stimulation were essentially identical with the borders of receptive fields determined by von Frey hair stimulation within a tolerance limit of ∼2 mm. Therefore the transcutaneous electrical stimulation method probably yields reliable assessments of the terminal arborization of single cutaneous C units in relation to the skin surface. We found that extensions of the innervation territory did not follow the presumed trajectory of the parent axon in any of the sympathetic units, indicating that true innervation territories were mapped as projected onto the skin surface. However, we cannot exclude the possibility that deep branches of afferent as well as efferent C units were not detected by our transcutaneous stimulation procedure. In two units, quite strong transcutaneous electrical stimuli were ineffective, whereas intracutaneous stimulation was effective. Presumably these units innervated deeper structures. Interestingly, the innervation territories of sympathetic units that could be mapped were of the same order of magnitude as those of afferent CM and CMH units, although the tendency toward smaller innervation territories in more distal parts of the foot, which we observed for the afferent units (Schmidt et al. 1997), was not presently observed for efferent units.
Action of ACh on sympathetic terminals
Five of 11 sympathetic units responded to iontophoretic application of ACh. This provides the first direct evidence that activation of cholinergic, probably nicotinic, receptors (MacMillan and Spalding 1969) in the terminal membrane of sympathetic units induces prolonged, antidromic neuronal discharges. They might then pass orthodromically into axon collaterals, a prerequesite for axon reflex sweating. It has been shown that an identical stimulus applied to the skin of the foot dorsum induces axon reflex sweating that covers an area with a median radius of 26 mm in the direction of its greatest extension (Riedl et al. 1998). This radius fits well with the mean diameter of innervation territories of sympathetic units encountered in this study (median maximal diameter: 21 mm). The fit between the mean extension of the sweating response and the innervation territory of the sympathetic units assessed in this study provides further evidence for the accuracy of our mapping technique. Iontophoresis of saline or histamine did not activate sympathetic units, in accordance with the finding that axon reflex sweating is induced by ACh, but not by histamine (Benarroch and Low 1991; Riedl et al. 1998).
Our data clearly show that only a subset of sympathetic terminals are endowed with excitatory cholinergic receptors that can induce spike responses, and one might assume that the activated fibers were at least in part sudomotors. However, other types of sympathetic units may be equipped with excitatory nicotinic receptors. Iontophoresis of ACh also induces axon reflex vasodilatation, and this is probably due to activation of afferent units that release vasodilatory neuropeptides (Blumberg and Wallin 1987). However, sympathetic vasodilator units have been found in animal (Bell et al. 1985; Gregor et al. 1976) and human (Lundberg et al. 1989) skin, and their possible contribution to axon reflex vasodilatation cannot be excluded. There is no indication that ACh also induces axon reflex vasoconstriction in the skin; however, this might be masked by the concurrent vasodilatation. ACh or nicotine evokes the release of noradrenaline from sympathetic terminals in the walls of arteries in animal preparations (Bultmann et al. 1991; Niedergaard and Schrold 1977). This indicates the existence of nicotinic receptors in at least some vasoconstrictor terminals associated with vasculature, although the mechanism does not necessarily involve antidromic spike activity. Therefore it is presently uncertain whether the ACh test will provide a tool to differentiate between sudomotor and vasoconstrictor efferents.
Classes of sympathetic units in the skin
Sympathetic efferents to the skin of the leg and foot are functionally heterogeneous. Two groups of neurons have been identified, vasoconstrictor and sudomotor units (Bini et al. 1980; Normell and Wallin 1974). As mentioned above, there is evidence that also vasodilators exist in animal (Bell et al. 1985; Gregor et al. 1976) and human (Lundberg et al. 1989) skin. However, the vasodilators found by Jänig and co-workers in cat hairy skin (Bell et al. 1985; Gregor et al. 1976) were not contributing to sympathetic arousal reflex activity, whereas all of the present sympathetic efferents did.
In the present study we have tried to differentiate between vasoconstrictor and sudomotor units on the basis of their different functions in thermoregulation. Of course, putative vasodilator units might behave similarly to sudomotors in thermoregulation. Differentiation between vasoconstrictor and sudomotor units on the basis of their behavior in thermoregulation is rather tedious, because sweating and vasoconstriction depend on both skin and body core temperature. It is difficult and time consuming to change the latter. Whereas a quantative differentiation of units is not practicable due to the requirement of stable long-term recordings, a pharmacological means of differentiation may prove feasible. However, it remains to be shown in future experiments how such pharmacological tests relate to the functional classification of sympathetic units.
In summary, the marking technique has been used to map, for the first time, the innervation territories of single sympathetic C fibers in human skin. The technique appears useful for semiquantitative analysis of single-unit activity in sympathetic units during long-lasting experiments. It will be of interest to explore in the future if the ACh activation test described here is specific for sudomotors, and whether the terminal arborizations of sudomotor and vasoconstrictor units are different.
We thank Prof. Earl Carstens for helpful comments and for improving the style of the manuscript.
This work was supported by the Bank of Sweden Tercentenary Foundation (to H. O. Handwerker), the Max Planck Award for International Cooperation (to H. E. Torebjörk), the Swedish Foundation for Brain Research (grant to R. Schmidt), the Swedish Medical Research Council, Project 5206, and the Deutsche Forschungsgemeinschaft (SFB 353).
Address for reprint requests: M. Schmelz, Institut für Physiologie und Experimentelle Pathophysiologie, Universitätsstr. 17, 91054 Erlangen, Germany.
- Copyright © 1998 the American Physiological Society