DANTROLENE sodium is a unique skeletal muscle relaxant acting distally to the myoneural junction to block calcium release from inside the muscle cell. [1] In human volunteers, the intravenous administration of 2.4 mg/kg dantrolene produces a blood concentration of about 10 micro Meter and a 75% decrease in muscle twitch force. [2] The exact site(s) for dantrolene's action has not been determined, although indirect and conflicting evidence has been reported. [3–5] Studies using radiolabeled dantrolene have indicated that at least two binding sites exist in skeletal muscle sarcoplasmic reticulum membranes, [6–8] a high-(nM) and a low-(micro Meter) affinity site. The high-affinity binding site was not evident in membranes from cardiac and smooth muscle. [7] Knowledge about dantrolene's specific site of action is important because it is the only drug efficacious for treating malignant hyperthermia (MH), and although it can block skeletal muscle contraction only by 75%, no similar effects on cardiac or smooth muscle contractility have been noted. The ryanodine receptor protein (Ry¹) is the major calcium release channel in skeletal muscle, and its mutational forms are linked to MH susceptibility in pigs [9] and in some human families. [10–13] The Ry¹receptor remains a probable molecular target for dantrolene's action, even though previous studies produced conflicting results. [3–5] .

An analysis of the Adverse Metabolic Response to Anesthesia database in North American MH Registry* showed that 32 of 232 (13.8%) MH cases recrudesced and that 30 of these 32 patients received dantrolene therapy. While testing a hypothesis that DS acts on the Ry¹receptor protein, we discovered two effects of micromolar, therapeutic DS concentration on the channel's gating properties. The first observed effect was activation of the channel, which was followed in time by the second effect of inactivation. The activating effect of DS was shown to occur at concentrations 0.001 times the therapeutic, inactivating concentration. Subsequently, we showed that these low concentrations of DS can produce contracture in isolated normal and MHS skeletal muscle, providing one possible explanation for the recrudescence of MH among dantrolene-treated MH patients.

A heavy fraction of sarcoplasmic reticulum membranes (SR) was isolated from porcine longissimus dorsi according to methods previously described. [14] Muscle, biopsied from animals anesthetized with thiopental with their lungs mechanically ventilated with 100% Oxygen², was homogenized then centrifuged at varying speeds [14] to obtain a crude heavy SR membrane fraction. The membranes were finally suspended in 20 mM histidine (pH 6.8), 150 mM KCl, and 0.3 M trehalose and protease inhibitors and maintained at -75 degrees C. The animals from which the muscle was obtained were phenotyped to MH by in vitro gracilis muscle contracture testing and by a halothane-succinylcholine MH challenge protocol. [15] Single-channel reconstitution experiments were as previously reported. [16] The heavy SR membrane fraction concentrated with Ry¹receptor protein was incorporated into a phospholipid bilayer (palmitoyl - oleoyl - phosphatidylethanolamine:palmitoyl-oleoyl-phosphatidylcholine, 7:3) painted across a 250-micro meter diameter aperture separating two chambers. CsCH³SO³-equilibrated-agar Ag/AgCl electrodes were used to measure the Caesium sup + current flowing from one chamber (cis) to the other (trans). The cis chamber represents cytoplasmic and the trans chamber intraluminal reference to the SR membrane. The SR membrane fusion into the bilayer occurred in the cis chamber, and each chamber maintained at 25 degrees C contained 250 mM CsCH³SO³, 20 mM HEPES (pH 7.4), and 20 micro Meter Calcium²+. A pulsed-voltage protocol was used for sampling single-channel data. From a holding membrane potential of 0 mV, a 50-mV (cis) polarization was applied for 200 ms at a frequency of 1.4 Hz. Recordings of 250 episodes, each 2 s in duration, filtered at 2,500 Hz, and sampled at a rate of 10 kHz, were obtained for control predantrolene values. After the control recordings, dantrolene was added to the cis chamber and mixed for 1 min. A 0.5-mM stock solution of dantrolene sodium in water was prepared immediately before adding to the cis chamber. After adding and mixing dantrolene in the cis chamber, channel records were obtained as described above for the controls except that recordings continued for times ranging from 27 min up to 1 h among the channels. Data acquisition software and hardware (pClamp, TL-1 Interface, Axon, Burlingame, CA) were computer-interfaced. Analysis software by TRANSIT 2.0 (Department of Molecular Physiology and Biophysics, Baylor College of Medicine) was used to obtain measurements of open-state probability (Po), current amplitude (picoamperes, pA), and the open and closed time constants. The time constants were derived by fitting logarithmically binned histograms of the open and closed dwell times using an automated maximum likelihood method. The software program TRANSIT also produced current integral values by measuring the area (current amplitude x time) under each opening. Summation of the Caesium sup + current integral values during channel lifetimes provided data from which different activity states were identified. In the absence of dantrolene, the slope of the cumulative current integral versus time was relatively constant and identified as a control state. After adding dantrolene, activated (cumulative current integral slopes > control) or reduced activity (slope < control) states were identified. Consequently, for this study, three channel activity states were identified as control (+/-dantrolene), activated, and reduced activity. Among the channels recorded, five were tested in the presence of 1 nM and four with 5 micro Meter dantrolene. Nine variables were determined for each channel recording period represented by control, activated, and attenuated channel activity levels. Experimental design was repeated-measures with unbalanced subsampling. Contrasts were made to compare dantrolene concentration effect on the three different channel activity levels. Proc Mixed in SAS (SAS, Cary, NC) was used for a mixed model, repeated-measures analysis of variance to determine statistical significance (alpha = 0.05) of dantrolene effects on the channel variables. Corrections for multiple comparisons, when necessary, were made with Bonferroni's method. Transformations were required for the cumulative current integral values (ranked within channel with respect to size) and for the dwell-time constants (log transformation to normalize tau value distributions).

The activating effects of low DS concentrations on channels from MHS (n = 5) and MHN (n = 6) pig muscle were compared by exposing each channel to DS concentrations increasing from 0.1 to 4.0 nM. Single-channel data acquisition and analysis was similar to that described above except that each record was 200 episodes, each 2 s duration, and only one 200-episodes control record was acquired before initiating the DS additions.

The effects of low DS concentrations on tension development in isolated human and dog skeletal muscle were tested as follows. Human vastus lateralis and dog gracilis muscle was obtained under general, MH-nontrigger anesthesia for diagnostic contracture testing by the North American MH Registry Protocol procedures. [17] Biopsied muscle fascicles were exposed to dantrolene in 37 degrees C Krebs-Ringer solution within 6 h of surgery. Small, 2–3-mm fascicles of human muscle were tied to a wooden applicator stick and chemically skinned [18] in a relaxing solution containing (mM): 172 K-propionate, 2.5 Mg-acetate, 5 K²ethyleneglycol-bis (beta aminoethylether) N,N,N',N'-tetraacetate (EGTA), 10 imidazole propionate (pH = 7.0), and 2.5 Na²K²ATP. From this bundle of skinned muscle, a single fiber was dissected, transferred to a chamber filled with 650 micro liter of relaxing solution maintained at 25 degrees C, and attached to a force transducer (Grass FT 03) for tension measurement. The force transducer signal was digitized and conditioned by a Digidata 1200 and a Cyberamp, respectively (Axon) and stored in a computer for subsequent analysis with pClamp6 software (Axon). Wash solution was identical to relaxing solution except the K-propionate was 185 mM and no K²EGTA was present. Caffeine and DS solutions were prepared in the washing solution immediately before use. The pCa levels were based on appropriate mixture of CaEGTA:K²EGTA calculated by the computer program described by Fabiato [19,20] using the constants provided by Orentlicher et al. [21] The protocol for tension measurements was as follows:(1) 1-min calcium (pCa = 6.6) loading into the sarcoplasmic reticulum;(2) two 20-s washes in washing solution;(3) 5-min exposure of fiber to 5 nM DS; and (4) exposure of fiber to a caffeine concentration. After step 3, steps 1 and 2 were repeated, and the next concentration of caffeine was added. Each fiber was tested with 20 mM caffeine at the beginning and end of each protocol. If the 20-mM caffeine response at the end of the protocol was < 80% of the initial response, the fiber data were discarded. The human and animal tissues used in these experiments were obtained by protocols approved by our institutional review committee.

Although channel activity was measured from current produced by Caesium sup + conductance, the channels are identified as Ry¹receptor calcium channels by a characteristic unitary conductance of cesium (450 pS), inhibition by ruthenium red, activation by ATP, and transformation to a prolonged, suboptimal conducting state by ryanodine. [22] In a previous, preliminary study, [23] we observed activation followed by activity below control level when dantrolene was applied to channels at concentrations ranging from 5 to 25 micro Meter. Speculating that a low-affinity binding site may represent the dantrolene effect that reduces activity of the Ry¹receptor channel, we compared the effects of 1 and 5 micro Meter dantrolene on a single Ry¹receptor channel (Figure 1). Addition of 1 nM dantrolene was followed by a period of increased channel open-state probability that was never observed during control recordings (Figure 1). Overall, the open-state probability was increased from 0.02 to 0.07. Next, dantrolene concentration was increased from 1 to 5 micro Meter, and after a brief period of increased activity, the channel became inactivated by this higher concentration of dantrolene (Figure 1) and was never again observed in an activated state. Among four channels exposed to 5 micro Meter dantrolene, each was first activated, and this was followed by periods during which the channels had a marked reduction in activity. During the total recording time of 78.8 min for which these four channels were exposed to 5 micro Meter dantrolene, 37.7+/-4% of this time represented control level activity, 30.6+/-3% of time the channel was in active state, and 31.7+/-1.2% of time the channels were attenuated. The average values for the cumulative current integral, Po, open- and closed-state dwell times and time constants, and current amplitude were calculated for each of these states before and after 5 micro Meter dantrolene (Table 1). The five channels treated with 1 nM dantrolene were first recorded over a 10-min period that was analyzed for control values. After activation by the voltage pulse protocol described above (methods), these channels had an average open-state probability (Po) of 0.028+/- 0.004 and an amplitude of 12.5+/-0.64 pA and dwelt in the open state for an average of 0.37+/-0.031 ms. Two open time constants of 0.33+/-0.02 (tau^O1) and 4.9+/-0.82 (tau^O2) ms were obtained, and these represented 95.4% and 4.6% of the total dwell time histogram, respectively. The same procedure applied to histograms of the closed times provided two time constants of 8.1+/-1.05 (tau sub C1) and 20.7+/-2.04 (tau^C2) ms. A histogram of the current integral provides a history of the channel's conductance over the measured period (Figure 2). During the control period of measurement, this value averaged 90.1+/-12.9 pA *symbol* ms over the total recording time of 10 min.

Dantrolene (1 nM) produced an activating effect on the Ry¹receptor channel by increasing the probability of opening almost threefold from 0.028+/-0.004 to 0.078+/-0.01 (P < 0.01). During this dantrolene-activated state, the open-state time constant tau sub O1 value was increased (Table 1) and was statistically significantly different from the control values. The net result of dantrolene-induced increase in open-state probability is an increase in the total amount of current conducted by the channel over time. As represented in the histogram of the current integral for one of these channels (Figure 2), dantrolene increases the current integral above the control level, and these changes appear as bursts of increased activity occurring between periods of activity at the control, predantrolene level. A few activity states were identified as having activity below control levels. Because the current integral histograms for these channels did not display periods of high and low activity before dantrolene, this change in activity may represent on and off times for dantrolene binding to apparent high- and/or low-affinity sites of a single protein molecule. During the 73.6 min the channels were exposed to 1 nM dantrolene, the activated-state dwell times totaled 56.9+/-9% of the total time. During 30.4+/-7% of the total time, the channels had control level activity, and 12.7 +/-7% of the time was identified as reduced activity ranging from less than control levels to a completely closed channel. Some of the averaged channel variables in these different states are statistically described for the channels treated with 1 nM dantrolene (Table 1). To rule out the possibility that the crude SR membranes used to incorporate Ry¹receptor channels did not contain other proteins to which DS could bind and produce these effects, the Ry¹receptor was purified, reconstituted into liposomes, [24] and incorporated into the bilayer, where we found that DS had similar effects on the channel's activity (Figure 3(A)). To gain information regarding pharmacologic specificity, the more water-soluble analog of DS, azumolene, was exposed to an Ry¹receptor channel. As we observed for DS, azumolene at nanomolar concentration activated the channel, whereas micromolar concentrations markedly reduced channel activity (Figure 3(B)).*.

The apparent high-affinity DS binding site was characterized for its DS concentration dependency, which was compared between MHS and MHN channels from pig skeletal muscle. The open-state probability of MHN channels increased biphasically, with DS concentrations increasing from 0.1 to 4.0 nM (Figure 4). The first phase appeared to be saturated at about 0.5 nM, and the second phase at 2.0 nM (Figure 4). The Po was increased twofold at 0.5 nM and sixfold at 2.0 nM DS (Figure 4). Channels from MHS muscle had 1.8+/-0.39-fold increase in Po at 0.5 nM, but from 1 to 4 nM DS, no further increase in Po was observed. The open-state dwell time histograms were logarithmically binned and fitted for time constants, for which two were found. The shorter time constant, tau^O1averaged 0.26+/-0.02 ms and 0.48+/-0.17 ms for MHN and MHS channels, respectively, and these dwell time constants represented 95.9+/-0.02 and 95.8+/-0.025% of the total histograms, respectively. The tau^O1values increased with increasing DS concentrations with a biphasic response (Figure 5) similar to that observed for DS effects on Po. However, the tau^O1value was increased by 1.5-fold compared to a sixfold increase for Po. Similar to the DS concentration dependency for Po, the biphasic tau^O1responses showed saturation at 0.5 and 1.0 nM DS (Figure 5(A)). As previously reported, [22] the tau^O1value for MHS channels is greater than for MHN channels. The DS concentrations (0.1–4 nM) had no statistically significant effect on tau^O1(Figure 5(A)). The longer open time constants, tau^O2, were not statistically significantly affected by DS application to MHN or MHS channels (Figure 5(B)). Two closed-state time constants were fitted for MHS and MHN channels and DS, 0.1–4 nM, had no effect on these values (Figure 6).*.

To determine whether these effects of nM DS could occur in the muscle cell, chemically skinned, single human fibers were loaded with calcium and exposed to 5 nM DS, and then calcium was released by adding caffeine. Increasing caffeine concentration from 2 to 20 mM produced an increase in amount of calcium released and in consequent tension force (Figure 7). When the fiber was exposed to 5 nM DS for 5 min after the calcium loading, the subsequent response to 2–20 mM caffeine-induced tensions was reduced by as much as 40% of the control response (Figure 7(A)). This indicates that, during the 5 min exposure to 5 nM DS, some of the loaded calcium was either lost from the SR storage or was made unavailable for release. The rate of tension development, dT/dt, was also decreased by 5 nM DS, but this was only evident at 8 and 20 nM caffeine concentrations (Figure 7(B)). Exposure of two vastus lateralis muscle fascicles from an MH-positive diagnosed patient to 20 nM dantrolene resulted in isometric contracture tension in each, the largest being 1.3 g (Figure 8(A)). In four other fascicles, two each from patients with MH-negative contracture test results, 1-h exposure to 20 nM dantrolene did not produce a contracture (data not shown). Gracilis fascicles from MH dogs treated in vitro with halothane produced the MH-phenotypic contracture response, and when DS is applied to these halothane-treated fascicles, an additional contracture response can occur (Figure 8(B)).

In vastus lateralis fascicles from another MHS patient, 20 nM DS potentiated the twitch tension in three of three fascicles tested without evoking isometric contracture tension (Figure 9). Increasing DS to 40 nM produced no further effect, but an increase to 10 micro Meter resulted in a 72% reduction in twitch tension and blockade of the contracture response to 3% halothane (Figure 9).

Although these in vitro studies show that low concentrations of dantrolene can cause a release of calcium from the sarcoplasmic reticulum stores, such findings do not prove in vivo dantrolene effects in normal or MHS patients. Data extracted from adverse metabolic response to anesthesia reports to the North American MH Registry* are interesting in this regard. Among 232 reported patients with clinical episodes suspicious for MH, 32 (13.8%) of these were reported to have recrudescence of the syndrome. Of these 32 patients, 30 had received an average initial DS dose of 120 mg. Among the 122 patients treated with DS, 24.6% were reported to have recrudesced.

Results obtained from these investigations provide evidence at the cellular and molecular levels for two effects of DS on the regulation of [Calcium²+] in skeletal muscle cells. In biopsied skeletal muscle fascicles from MHS humans, nanomolar concentrations of DS can increase twitch tension and produce isometric contracture in these genetically predisposed cells. Presumably, these tension increases are a consequence of increased myoplasmic [Calcium²+] that is produced by dantrolene. At micromolar concentration, which corresponds to the therapeutic blood concentration, DS produces a reduction in twitch tension and blocks the MH muscle's contracture response to halothane. Thus, the genetically predisposed MH muscle cell is capable of expressing both of the DS concentration-dependent effects on contractility. Using chemically skinned single-muscle fibers, the muscle cell function is simplified to calcium uptake and release by the sarcoplasmic reticulum and to the generation of tension by the contractile elements. In this experimental model, we were able to demonstrate that nanomolar concentrations of DS caused a release of calcium from the SR, whereas it had no effect on the amount of calcium loaded in the presence of MgATP. This finding suggests that the calcium release channel could be opened by nanomolar concentrations of DS, allowing calcium to flow down its concentration gradient into the myoplasm.

Experiments on single Ry¹receptor protein molecules incorporated into a lipid bilayer provided direct evidence that DS could alter the gating properties of this calcium release channel. At the nanomolar concentration range, DS produces a marked increases in the open-state probability and dwell time of the channel, a finding that could explain how low concentrations of DS alter contractility in the intact muscle cell. When the calcium release channel is exposed to DS at micromolar concentration, an opposite effect is observed, i.e., the channel open-state probability is markedly reduced. This finding is consistent with the well established pharmacologic effect of DS to reduce electromechanically coupled twitch tension by 70–80% and may be the mechanism by which DS prevents and treats MH in patients.

These antithetical effects of DS may relate more to the properties of the ryanodine receptor protein than to the ligand, because similar opposing effects of the alkaloid ryanodine on Ry¹receptor are well recognized. [25–27] One explanation for these dual effects of DS is that two (or more) binding sites with different affinities for dantrolene exist on the Ry¹receptor protein molecule. Previous studies reported two different binding sites, one at nanomolar and the other at micromolar dantrolene concentrations. [6–8] The low-affinity binding site was attributed to the distribution of DS into the lipid matrix of the SR membrane rather than binding to SR protein. [7] In our study, activation of Ry¹receptor channel gating was the dominant effect for nanomolar DS, because the channels exhibited increased gating 57% of the time and deactivation occurred only 13% of the time. In contrast, channels exposed to 5 micro Meter DS dwelled in a deactivated state 32% of the time; 2.5 times longer than channels exposed to 1 nM DS. There are several possible explanations for these findings. Because the more hydrophilic DS analog, azumolene, produced activation of Ry¹receptor at nanomolar concentrations, partitioning of these molecules into the lipid matrix is not likely to cause the observed effects. It may be that two different binding sites exist for DS on the Ry¹receptor molecule; one binding with high and the other with low affinity. Because the Ry¹receptor molecule is comprised of four identical subunits, it may be that each subunit has a DS binding site and that the affinity decreases as each site becomes occupied. The binding of calmodulin to each of the ryanodine receptor subunits [26] and the antithetical effects of calmodulin on the calcium release channel [27] suggest that certain ligands can have these effects. Further studies are required to address these complex issues.

The findings of this study raise a clinically important question regarding the possibility that DS could have adverse effects in the genetically predisposed MH patient. When, during the normal course of DS elimination, the therapeutic, micromolar DS concentration is reduced to the nanomolar concentration range, the putative high affinity DS binding site effect would dominate, and the Ry¹receptor channel would open to release Calcium²+ into the myoplasm. If such a DS-induced calcium leak occurs in metabolically compromised MH muscle cells, myoplasmic [Calcium²+] may increase to levels that reproduce the MH syndrome. In metabolically uncompromised cells, such a calcium leak may be controlled without serious consequences. The high incidence of MH syndrome recrudescence among dantrolene-treated MH patients suggests that our findings may have clinical relevance and represent a clinical paradox.

The genetic predisposition to MH in some human families is linked to one of several different mutations in Ry¹receptor, [10–13] presenting another confounding possibility that some Ry¹receptor mutations are adversely susceptible to the nanomolar DS effects and others are not. In this regard, it is interesting that nanomolar DS produced a greater increase in opening of the wild-type channel from pig muscle than it did from the MH mutant channels. Because only 5% humans demonstrate the pig defect, and the Ry¹receptor may be involved in only about 50% of the mutations that demonstrate phenotypic MH, the increased opening caused by dantrolene in normal Ry¹receptors may be of considerable significance. In muscles in which the ability to handle Calcium²+ appropriately is compromised by a variety of genetic defects, the effect of dantrolene may be sufficient to induce hypermetabolism, which may proceed to an MH episode. It may be that the Ry sub 1 receptor mutation in pigs results in an attenuation of the nanomolar DS effects. The mutation predisposing to MH in dogs is not the pig MH mutation.** In isolated MH dog muscle, addition of nanomolar DS to a halothane-induced contracture produced more contracture. This experiment suggests that inappropriately low dosage of DS during an MH clinical crisis could exacerbate the syndrome. Although our investigation unequivocally identifies an effect of low DS to increase myoplasmic [Calcium²+], it does not prove that this can cause recrudescence of the MH syndrome. Further studies are necessary to determine whether our findings are of clinical importance.

The authors thank Samantha Evans, for technical help; Robert James, M.S., for statistical analysis; and Wilson Somerville, Ph.D., and Adela Larimore, for editorial work.

*Allen G: Personal communication. 1995.

**Hogan K: Personal communication.