Ryanodine receptor modulation by caffeine challenge modifies Na+ current properties in intact murine skeletal muscle fibres

We investigated effects of the ryanodine receptor (RyR) modulator caffeine on Na+ current (INa) activation and inactivation in intact loose-patch clamped murine skeletal muscle fibres subject to a double pulse procedure. INa activation was examined using 10-ms depolarising, V1, steps to varying voltages 0–80 mV positive to resting membrane potential. The dependence of the subsequent, INa inactivation on V1 was examined by superimposed, V2, steps to a fixed depolarising voltage. Current-voltage activation and inactivation curves indicated that adding 0.5 and 2 mM caffeine prior to establishing the patch seal respectively produced decreased (within 1 min) and increased (after ~2 min) peak INa followed by its recovery to pretreatment levels (after ~40 and ~30 min respectively). These changes accompanied negative shifts in the voltage dependence of INa inactivation (within 10 min) and subsequent superimposed positive activation and inactivation shifts, following 0.5 mM caffeine challenge. In contrast, 2 mM caffeine elicited delayed negative shifts in both activation and inactivation. These effects were abrogated if caffeine was added after establishing the patch seal or with RyR block by 10 μM dantrolene. These effects precisely paralleled previous reports of persistently (~10 min) increased cytosolic [Ca2+] with 0.5 mM, and an early peak rapidly succeeded by persistently reduced [Ca2+] likely reflecting gradual RyR inactivation with ≥1.0 mM caffeine. The latter findings suggested inhibitory effects of even resting cytosolic [Ca2+] on INa. They suggest potentially physiologically significant negative feedback regulation of RyR activity on Nav1.4 properties through increased or decreased local cytosolic [Ca2+], Ca2+-calmodulin and FKBP12.


Results
Currents obtained using the double pulse protocol. The experiments employed a loose-patch clamp preparation and recording layout fully described in the Materials and Methods section (Fig. 1). Voltagedependences of Na + current, I Na , including its activation and inactivation, were examined using a double test pulse protocol (Fig. 1D). In each sweep of the full protocol, the membrane was first held at resting membrane potential (RMP) for 5 ms. A hyperpolarising pre-pulse, duration 50 ms, imposed at times 5-55 ms from the beginning of the sweep to voltage V 0 = (RMP-40) mV, first removed any residual I Na inactivation at the RMP. This standardized the proportion of activatable Na + channels prior to application of the subsequent depolarising V 1 pulse, itself of duration 10 ms, at time 55-65 ms.
At early intervals immediately following solution change, a simplified protocol examined I Na in individual patches. This averaged results from 5 sweeps of a pulse protocol imposing a single voltage step to a single voltage V 1 at (RMP + 80) mV between 55 and 80 ms. I Na features at later time intervals following solution change were studied using the full procedure. This examined effects of a family of V 1 amplitudes varied between sweeps in +5 mV increments through test voltages between RMP and (RMP + 80) mV. This elicited currents reflecting the voltage dependence of I Na activation. A further V 2 pulse was then made to a constant, depolarised voltage of (RMP + 80) mV. This would activate all the Na + channels within the patch not inactivated by the preceding V 1 pulse. This made it possible to explore the voltage dependence of Na + current inactivation. For each value of V 1 , currents from 10 sweeps were averaged before examining the next value of V 1 . Each complete protocol applied to a single patch therefore involved collection of currents from 170 sweeps. Figure 1E exemplifies currents obtained by a single sweep at a single voltage, V 1 = (RMP + 35) mV following P/4 correction under control conditions. Downward, negative deflections denote inward while upward, positive deflections represent outward membrane currents. The V 0 pulse typically yielded no currents. The V 1 and associated V 2 steps each elicited transient inward currents representing the opening and inactivation of Na + channels. The magnitudes of their peaks, i 1 and i 2 , were measured relative to the currents measured just before the respective V 1 and V 2 steps.
Each patch was only subjected to a single application of the complete set of pulse protocols. This made differences between results arising from prolonged changes in the patch such as membrane blebbing unlikely 22 . Furthermore, following every set of sweeps a further bracketing protocol was then imposed using a single V 1 step to RMP. The resulting average i 2 value from the succeeding V 2 step was then compared with the i 2 value obtained by the first 10 sweeps of the preceding test protocol, for which V 1 was also RMP. Patch stability was confirmed, permitting further analysis, if the bracketing i 2 value was within 10% of the test i 2 value. Thus, only patches with a stable seal, I Na generation, and activation and inactivation characteristics were selected for further analysis. Pipette mounted at 45° to allow 90° contact of the bent tip with the fibre surface. The pipette is connected to a suction syringe 13 . (C) Equivalent circuit of loose patch clamp electrode on muscle membrane. Solution in the pipette clamped at V pip . The voltage error arising from the flow of currents through the patch resistance (R patch ) and the cell resistance (R cell ), along with the pipette resistance (R pip ) corrected mainly using a bridge circuit in the loose patch clamp amplifier 13 . (D) Complete set of pulse protocols carried out on a patch being investigated. This double test pulse protocol first superimposed a −40 mV hyperpolarising pre-pulse, lasting 50 ms, on the resting potential, RMP. This was followed by a depolarising V 1 pulse varying in amplitude to test voltages between RMP and (RMP + 80) mV in +5 mV increments. This was followed by a V 2 pulse to a strongly depolarised potential of (RMP + 80) mV. (E) The corresponding current trace during a single sweep in the protocol under control conditions where V 1 = (RMP + 35) mV, following P/4 protocol correction, giving peak currents I 1 and I 2 .
www.nature.com/scientificreports www.nature.com/scientificreports/ Separate experiments using 2.0 mM caffeine challenge gave concordant, complementary, results. Relative to the initial recordings in the absence of caffeine (Fig. 2C(i)), 2.0 mM caffeine initially produced a reduced I Na , but this was rapidly succeeded by increased I Na if the latter test recordings were made after lifting then restoring the patch seal ( Fig. 2C(ii)). These effects were attenuated when the 2.0 mM caffeine was added then test recordings made whilst contrastingly maintaining an intact patch seal ( Fig. 2D(i,ii)).
These findings suggested that both 0.5 and 2 mM caffeine reduced I Na within the first minute of exposure. This reduction persisted over 1, 2 and 3 min with 0.5 mM caffeine, but was succeeded by a marked increase in I Na with 2 mM caffeine. However, such effects required a direct access of the caffeine to the area of membrane beneath the patch pipette in which I Na was recorded. The subsequent explorations of caffeine effects on I Na using the full pulse procedure at the later times accordingly challenged with caffeine prior to formation of the patch seal. Results of simple pulse protocols to examine I Na at early times following caffeine challenge using two experimental procedures. Following obtaining control records prior to caffeine challenge (A-D) panel (i), in (A), the patch electrode was withdrawn prior to introduction of 0.5 mM caffeine, and then restored prior to imposition of test steps (ii) in the following 1-3 min. In (B) the patch seal was maintained before, during and following introduction of 0.5 mM caffeine through imposition of the test steps (ii). The latter procedure was repeated following pipette withdrawal and restoration of the patch seal (iii). (C,D) adopted identical respective procedures as (A,B) using a 2 mM caffeine concentration.
www.nature.com/scientificreports www.nature.com/scientificreports/ Effects on I na magnitudes of 0.5 mM caffeine challenge. The I Na recordings showed patterns that paralleled the previously reported contrasting [Ca 2+ ] i changes produced by 0.5 and 2.0 mM caffeine concentrations. Figure 3 illustrates current records from families of voltage steps at differing V 1 between RMP and (RMP + 80) mV under control conditions (A) and at different times following introduction of 0.5 mM caffeine (B). Initial rates of rise of current increased non-linearly with degree of depolarization elicited by the V 1 steps. So did the peak I Na which then reached a maximum at V 1 ≈ (RMP + 75) mV before decreasing with further depolarisation. The subsequent V 2 steps yielded currents whose peak I Na decreased with increasingly positive V 1 consistent with prior voltage-dependent channel inactivation reducing the proportion of remaining activatable Na + channels. The pulse protocols were repeated on different patches over successive time intervals following the introduction of caffeine. Figure 3B((i-iv)) illustrates typical families of traces obtained from 8, 20, 37 and 45 min following introduction of caffeine. Comparison of the successive sets of records suggested that peak I Na initially showed marked, ~80%, reductions within the 10-minute interval following challenge (Fig. 3B(i-ii)). This was followed by a recovery which then returned I Na to pre-treatment levels over a >40 min time-course ( Fig. 3B(iii-iv)).
Changes in the voltage dependence of I na resulting from 0.5 mM caffeine challenge. Findings of this kind yielded I Na current-voltage activation (I-V) and inactivation curves plotting dependences of peak www.nature.com/scientificreports www.nature.com/scientificreports/ I Na on V 1 in response to the respective V 1 and V 2 voltage steps. These are plotted for families of records obtained at different intervals following onset of caffeine challenge (Fig. 4). These curves demonstrated that compared to pre-treatment controls (Fig. 4A(i)), I Na was reduced at 1-10, 11-25 and 26-40 min (Fig. 4A(ii-iv) respectively). However I Na had recovered by the final (41-55 min) time interval (Fig. 4A(iv)). The maximum values of peak I Na could be determined within the voltage range examined in the I-V curves obtained in the controls at 1-10 and 41-55, but not 11-25 and 26-40 min. Accordingly, the same determinations were made under control conditions and at 11-25, and 26-40 minutes with a modified pulse protocol (Fig. 5) that explored an increased V 1 range extending to (RMP + 120) mV and using a V 2 step to (RMP + 100) mV. This encompassed the V 1 at which maximum peak I Na was observed, both under the control conditions ( Fig. 5(i)) and at the selected times following addition of caffeine ( Fig. 5(ii,iii)). Together the I-V plots suggested that the greatest peak I Na occurred with depolarisations to (RMP + 105) mV at 11-25 min and at (RMP + 95) mV at 26-40 min respectively. Figure 6A plots values of these parameters in successive patches with time following application of caffeine. Tables 1A and 2A list these parameters as means ± SEM from explorations grouped in successive time intervals as adopted in Fig. 3. This made it possible to test for statistically significant differences from the control findings obtained before addition of 0.5 mM caffeine. These demonstrated that within 1-10 min after adding caffeine, there was a sharp reduction in I max . In addition, there was a negative shift in the V* value describing I Na inactivation. In contrast, V* values for I Na activation and k values for both activation and inactivation then remained unchanged. There then followed marked positive shifts in the V* of both I Na activation and I Na inactivation. The final I Na recovery to pretreatment values, was then accompanied by more gradual negative shifts in the voltage dependence of both I Na activation and I Na inactivation, with control I max , and V* values regained at 26-40 and 41-55 min respectively. These changes in I max and V* contrasted with the constant k values through these manoeuvres. Furthermore, they were abolished, as described fully below, by dantrolene-mediated RyR block.
Changes in magnitude and voltage dependence of I na resulting from 2 mM caffeine challenge. Caffeine challenge at a 2 mM concentration gave contrasting I Na changes. Following the initially decreased peak I Na within 2 min described above (Fig. 2C), the pattern assumed a transient increase in I Na followed by its recovery to control levels. Figure 7 shows families of I Na records obtained using the full pulse protocol as in Fig. 3 at the different times subsequent to 2.0 mM caffeine challenge. The current waveforms increased to a peak I Na followed by an inactivating decline (Fig. 7A). However, the families of currents demonstrated markedly greater I Na at 2-15 min following exposure to 2 mM caffeine, in contrast to the reduction in I Na produced by 0.5 mM caffeine. I Na then progressively declined to pre-treatment levels over the succeeding time intervals www.nature.com/scientificreports www.nature.com/scientificreports/  Figure 8A plots the resulting I-V and inactivation curves obtained at different time intervals, from which it was possible to obtain I max , V* and k values for the I-V and inactivation curves. Figure 9 plots these values for successive patches with time following application of caffeine and Tables 3A and 4A summarise the statistical analysis of their means ± SEM from explorations in the successive time intervals, comparing these against the control pretreatment results.
The I Na amplitudes, I max , increased threefold over the 2-15 min following addition of caffeine when compared to its control values (Compare Fig. 8A(i,ii)). I max then recovered in subsequent intervals to regain pre-treatment values at 26-40 minutes ( Fig. 8A(iii,iv)). There were delayed negative shifts in the V* of both I Na activation and I Na inactivation at 16-25 min followed by smaller positive shifts at 26-40 min (( Fig. 9A(ii)); Tables 3A and 4A). Again, in common with findings using 0.5 mM caffeine, values of k showed relatively little change through these protocols.
Effect of dantrolene on the I na responses to 0.5 and 2.0 mM caffeine. The final sequence of experiments examined the effects of the specific RyR inhibitor dantrolene both by itself and on these divergent actions of 0.5 and 2.0 mM caffeine on I Na using otherwise identical pulse protocols. Each experiment first obtained families of I Na traces in the absence of all test agents in two successive patches studied within a first 10 min interval. The bath solution was then substituted for one containing 10 µM dantrolene and 2 further patches investigated within the next 10 minutes. The final I Na records were then obtained after further replacing the bath solution with one containing both 0.5 or 2.0 mM caffeine and 10 µM dantrolene-containing solution. Patches were investigated using the same full pulse protocols as shown in Fig. 1.
Including dantrolene alone slightly increased peak I Na (Figs. 3C and 7C). Analysis of the I-V relationships (Figs. 4B and 8B) accordingly demonstrated small but significant increases in activation and inactivation I max compared to controls, but no effect on activation and inactivation V* and k values (Tables 5 and 6).
Simultaneous addition of dantrolene and 0.5 mM (Fig. 3C,D) or 2.0 mM caffeine (Fig. 7C,D) (Tables 1-4) resulted in currents whose magnitudes were similar to those obtained in the presence of dantrolene alone, at all the time intervals following the addition of caffeine. Dantrolene thus abrogated both the effects of 0.5 mM caffeine in producing transient decreases ( Fig. 3D(i-iv)), and the effects of 2 mM caffeine in producing transient increases in I Na (Fig. 7D(i-iii)). In parallel with the previous test experiments, I-V and inactivation relationships were  9B) in successive individual patches exposed to dantrolene. None of these parameters showed the contrasting pattern of time-dependent changes in activation and inactivation parameters shown with 0.5 and 2 mM caffeine challenge in the absence of dantrolene. Tables 1-4 summarise the statistical analysis of their means ± SEM from explorations in the successive time intervals comparing these (B) against the control pretreatment results (A). In both analyses the activation and inactivation I max , V * and k values were respectively statistically indistinguishable from those obtained under dantrolene challenge alone when caffeine was absent. Again values of k remained constant through the experiments.
Neither challenge by dantrolene, nor that by caffeine, whether by itself or in combination with dantrolene, noticeably affected I Na kinetics. When quantified by times to peak current at voltages yielding maximum I Na in the I-V curves, control experimental groups subsequently used to test effects of dantrolene alone, or 0.5 or 2 mM caffeine in the absence of or following further dantrolene challenge, gave times of 0.49 ± 0.010, 0.48 ± 0.0093 and 0.50 ± 0.012 ms respectively. Dunnett multiple comparison tests demonstrated that none of the subsequent interventions produced significantly alterations for all the time intervals assessed in Tables 1-4 (all P ≫ 0.05).
These control experiments thus demonstrate that (a) the RyR inhibitor dantrolene abrogated the contrasting actions of either 0.5 mM or 2.0 mM caffeine in transiently decreasing or increasing I Na with accompanying changes in V* to previously established altered RyR-mediated changes in [Ca 2+ ]. (b) Their demonstration of an increased I Na in the presence of dantrolene alone complements the demonstrations of increased I Na produced by 2 mM caffeine in suggesting some background inhibition of I Na at even background resting [Ca 2+ ] i .

Discussion
The present experiments explored for the existence of physiologically effective feedback loops between ryanodine receptor (RyR) activation by tubular depolarisation and Na v 1.4 channel function that would thereby modify surface membrane excitability in skeletal muscle in vivo. This mechanism would complement feedforward, excitation-contraction-coupling, processes by which Nav1.4-mediated surface membrane depolarisation triggers RyR-mediated release of intracellularly stored sarcoplasmic reticular (SR) Ca 2+ . Feedback alterations in Na v 1.4 availability and/or voltage sensitivity are potentially physiologically important particularly under conditions www.nature.com/scientificreports www.nature.com/scientificreports/ of intense and prolonged muscle activation including tetanic stimulation and metabolic exhaustion. The latter changes could compromise SR Ca 2+ -ATPase activity through ATP depletion, thereby elevating [Ca 2+ ] i . Any consequent effects of Ca 2+ on Nav1.4 function would complement its other actions promoting slow RyR inactivation 19 and increasing sarcolemmal Ca 2+ -activated large conductance K + channel (K Ca 1.1) conductance in downregulating excitation-contraction coupling 23 . Such feedback mechanisms would also bear on clinical conditions affecting skeletal muscle excitability. K + -aggravated myotonia is associated with slowed I Na kinetics and impaired I Na inactivation that accompany Nav1.4 mutations in its EF-hand-like domain 24 . Other genetic, K + and cold-aggravated myotonias are associated with reduced Ca 2+ -dependent Na + channel inhibition similarly attributed to Nav1.4 mutations 6 . Finally, the weakness observed in dystrophic muscle is associated with elevated The experiments here examined effects of direct pharmacological manipulation of RyR-mediated SR Ca 2+ release by the RyR agonist caffeine 14 on both Na v 1.4 channel availability and its voltage-dependent activation and inactivation properties in intact skeletal myocytes. They utilised previously established contrasting effects of different applied caffeine concentrations on spectrofluometrically measured resting [Ca 2+ ] i 15,16 . Low concentrations of caffeine (0.5 mM) then increased [Ca 2+ ] i persistently over the subsequent 3-10 min recording intervals in isolated rabbit skeletal SR or murine skeletal muscle fibre preparations 15 . In contrast, ≥1.0 mM caffeine induced early [Ca 2+ ] i peaks rapidly followed by [Ca 2+ ] i decreases below even resting-levels within 80-90 sec 16,17 .
Here, Na + currents, I Na were measured in intact mammalian gastrocnemius skeletal muscle using the loose patch clamp technique. This avoided membrane disruption and perturbations of [Ca 2+ ] i homeostasis inherent in other experimental measurement procedures involving conventional cell-attached patch clamp studies, which additionally often include the Ca 2+ -sequestrating EGTA in the pipette solutions. Furthermore, the latter can require study of isolated or cultured cells with enzymatically treated membranes as opposed to the intact in situ skeletal myocytes examined here. The loose-patch configuration also permitted sequential and multiple recordings with the same standardized pipette in successive patches before and after pharmacological challenge as required in the present experiments 26 . The features of the resulting I Na concurred with those obtained on earlier occasions 26 and with other voltage clamp methods applied to mammalian skeletal muscle 27 . The magnitude, voltage-dependence and quantified activation (current-voltage, I-V) and inactivation properties of I Na were compared at different times following introducing caffeine. Further experiments blocking these effects with further applications of the specific RyR inhibitor dantrolene attributed the observed changes specifically to altered RyR-mediated SR Ca 2+ release. www.nature.com/scientificreports www.nature.com/scientificreports/ The observed contrasting alterations in I Na closely paralleled the previously reported [Ca 2+ ] i changes following caffeine challenge. Thus, I Na was reduced early (1-2 min) following either 0.5 or 2 mM caffeine challenge. This effect was abrogated by reducing access of caffeine to tubular and therefore tubular-sarcoplasmic reticular triad regions in which RyRs are concentrated, in regions within the patch from which the recordings were obtained. The latter was accomplished by establishing patch seals before rather than after introduction of caffeine to the extracellular space. This would alter direct access of caffeine over a surface membrane area close to the 30 μm 2 cross sectional area of the electrode tip in contact with the membrane within which I Na was recorded. Subsequent recordings following the 0.5 and 2 mM caffeine challenge obtained by identical pulse protocols respectively demonstrated contrasting increases and decreases in peak I Na followed by their recovery to pretreatment levels. These findings corroborate and extend previous independent reports that I Na was reduced by www.nature.com/scientificreports www.nature.com/scientificreports/ increased RyR activity following 8-CPT induced Epac pathway activation 13 . The I Na changes were quantified by systematically exploring a consistent range of V 1 test voltages thereby deriving families of I Na records. These in turn yielded activation, I-V, and inactivation curves giving I max , V* and k values. These respectively quantified total available I Na and the position and steepness of their voltage dependences. Sustained reductions in I max induced by 0.5 mM caffeine paralleled the sustained [Ca 2+ ] i increases previously reported following caffeine mediated activation of RyR-mediated SR Ca 2+ release 15 . The increased I Na observed in the presence of 2 mM www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ caffeine similarly paralleled reported contrastingly decreased [Ca 2+ ] i attributed to slow RyR inactivation 16,17 . The latter findings further suggested significant inhibition of I Na by even normal background [Ca 2+ ] i .
These alterations in total available I Na were accompanied by time-dependent changes in properties of the I-V and inactivation curves. With 0.5 mM caffeine challenge, the initial decreases in I Na were accompanied by negative shifts in inactivation V* over 1-10 min. This was followed, over 11-25 min, by positive shifts in both I Na activation and I Na inactivation V*. The final I Na recovery to pretreatment values between 26-55 min was accompanied by opposite, negative, shifts in activation and inactivation V*. With 2 mM caffeine challenge, the contrasting,  Table 5. Mean I max , V * and k values of activation I-V curves before and after addition of 10 μM dantrolene. P values indicate significance of differences compared to control values. *Denotes a significant difference compared to controls where P < 0.05.  Table 6. Mean I max , V * and k values of inactivation I-V curves before and after addition of 10 μM dantrolene. P values indicate significance of differences compared to control values. *Denotes a significant difference compared to controls where P < 0.05. (2020) 10:2199 | https://doi.org/10.1038/s41598-020-59196-9

Condition
www.nature.com/scientificreports www.nature.com/scientificreports/ declining phases of the transient I Na increases accompanied delayed transient negative shifts in both activation and inactivation V*. In contrast, values of k of either activation or inactivation little changed throughout these protocols.
Finally, the effects of the specific RyR inhibitor dantrolene were investigated both by itself and in combination with identical caffeine concentrations and pulse protocols. Dantrolene specifically stabilises RyR closed states by selectively decreasing Ca 2+ affinities of RyR activation sites 21 . Previous reports had demonstrated that it inhibited caffeine-induced increments in mammalian SR Ca 2+ permeability 28 and caffeine-induced contractures in murine diaphragm muscle 29 .
In the present experiments dantrolene abrogated all the effects of either 0.5 or 2 mM caffeine, whether on I max , or activation or inactivation I Na or V*. In contrast, dantrolene by itself produced only small increases in I max without affecting activation and inactivation, V* and k values. Together with the increased I Na produced by 2 mM caffeine this latter finding suggests that [Ca 2+ ] i exerts significantly inhibits I Na even at background RyR activity. These explorations using dantrolene thus trace all the present actions of caffeine on I Na specifically to its effects on RyR-mediated SR Ca 2+ release.
These findings together demonstrate and characterise RyR-Nav1.4 feedback phenomena complementing their established feedforward interactions underlying excitation contraction coupling (Fig. 10A). The observed contrasting effects of caffeine challenge on I max and V* combined with previously reported [Ca 2+ ] i changes under similar experimental conditions 15-17 were incompatible with straightforward combined negative shifts in Na + channel activation and inactivation V* resulting from simple screening effects arising from an increased [Ca 2+ ] i 30,31 . However, they directly correlate with previous in vitro biochemical and biophysical reports on skeletal muscle Na v 1.4 and RyR. They establish for these a physiological context, possibly occurring within a common microdomain, in working muscle, as opposed to in vitro systems for the first time.
First, the early I Na reductions with 0.5 and 2 mM caffeine, persistent with 0.5 mM caffeine, were consistent with the reported effects of either direct Ca 2+ -binding to a Na v 1.4 EF hand site or indirect Ca 2+ binding to Na v 1.4 via its binding to CaM in reducing I Na (Fig. 10C) . Secondly, the accompanying early negative V* shifts 1-10 min following 0.5 mM caffeine challenge paralleled reported negative shifts in voltage-dependent inactivation produced by the indirect Ca 2+ binding process 2,6-8 . This was selective for Nav1.4 as opposed to Nav1.5, reflecting Nav1.4 lacking the CaM N-lobe binding site present in Nav1.5 1 . Thirdly, the subsequent (11-25 min) positive shifts in I Na activation and inactivation V* directly paralleled reported reductions in FKPB12 binding affinity for RyR following RyR activation. An increased free [FKBP12] i would then accompany the resulting SR Ca 2+ release elevating [Ca 2+ ] i . in intact fibres 9,10 ; FKBP12 is known to positively shift V* activation and inactivation in Na v 1.5 11 , thought closely homologous to Na v 1.4 4 . This action is probably indirect and delayed: co-immunoprecipitation assays appeared to exclude direct FKBP12-Na v 1.5 interactions 11 . Fourthly, negative shifts in both I Na activation and inactivation V* values eventually accompanied the reductions in I Na and recoveries from I Na elevation, 26-55 or 16-40 min following 0.5 and 2.0 mM caffeine challenge respectively. These were compatible with a subsequent, previously reported, slow RyR inactivation 19 . This would reduce RyR open channel probabilities and Ca 2+ release whilst permitting continued SR Ca 2+ re-uptake and possible surface membrane Ca 2+ expulsion by Na + /Ca 2+ exchange 32,33 . These would reverse the increases in [Ca 2+ ] i and [FKBP12] i 19 (Fig. 10D). Finally, the effects of the RyR inhibitor dantrolene (Fig. 10B) abrogating all these contrasting reductions and increases in I Na associated with different caffeine concentrations through the identical pulse protocols associated all these findings with feedback effects arising specifically from the RyR. Mice were killed by cervical dislocation by Home Office-licensed personnel in accordance with Schedule 1 of The Animals (Scientific Procedures) Act 1986. The gastrocnemius and soleus muscles were then immediately isolated, with the distal (Achilles) tendon secured by a knot to a ligature and the proximal tendon still attached to its bony origin. Overlying connective tissue was excised to allow access to the myocyte surface. The isolated muscles were then immediately placed in a Sylgard-bottomed experimental bath containing KH solution with the soleus muscle facing down maximizing electrode access to gastrocnemius fibres. Experiments were carried out at stable temperatures around 20 °C. The muscle was secured in a taut and fixed position by pinning down the bony origin and the ligature with A1 insect pins (Fig. 1A).

Materials and Methods
Loose patch clamp pipettes were pulled from borosilicate glass capillaries (GC150-10; Harvard Apparatus, Cambridge, UK) using a Brown-Flaming microelectrode puller (Model P-97; Sutter Instrument Co. Novato, CA). They were mounted under a microscope with ×250 magnification with a calibrated eyepiece graticule, and a small indentation made on one side of the pipette using a diamond knife. A transverse force was then applied distal to the indentation, giving a fracture perpendicular to the long axis of the pipette. Squarely-broken tips were then fire-polished under visual guidance at ×400 magnification using an electrically-heated nichrome filament to smooth the edges of the tip. The internal diameters of the pipettes were measured under ×1000 magnification. Pipettes of internal tip diameters between 28-38 µm were used in the experiments. Such loose patch recordings thus provide a complete representation of the properties of both surface and tubular membrane directly beneath these seals. Finally, the pipettes were bent at about 1 mm from the tip to make a 45° angle with the pipette's long axis. This allowed the tips to contact the muscle surface at 90° when mounted at an angle of 45° below the horizontal on the recording amplifier head stage (Fig. 1B). The pipettes were mounted onto a pipette holder with an Ag/AgCl electrode held within the inside of the pipette. The distal half of the pipette was filled with solution from the bath, using suction provided by a syringe via an air-filled connection to the pipette holder. The bath was actively grounded at reference potential, using an Ag/AgCl electrode. These Ag/AgCl electrodes maintained electrical connection between the organ bath and pipette with the electronic circuit.
The pipette was lowered onto the gastrocnemius muscle membrane surface and gentle suction applied to allow seal formation around the resulting membrane patch. The central regions of the muscle (crosshatched region in scheme shown in Fig. 1A) were targeted for each experiment as this region was the most horizontal, permitting optimal (~90°) contact between pipette tip and muscle. The loose patch clamp configuration clamps the voltage on the extracellular side of the membrane within the pipette seal. Therefore, positive and negative potentials applied within the pipette respectively hyperpolarise and depolarise the membrane potential of the patch relative to the cell resting membrane potential (RMP). In this report, the signs of the voltage steps are inverted and membrane potentials expressed relative to RMP, a convention adopted by previous papers using this technique 12,13,20,26 . The Ag/AgCl electrode within the pipette both measured the intra-pipette potential relative to the ground reference potential of the bath and passed the clamp current. Figure 1C shows the equivalent circuit of a typical patch. The loose patch technique involves a relatively low seal resistance (typically < 2 MΩ). This ohmic leakage and series resistance error [(seal resistance + patch resistance) and pipette resistance], along with additional pipette capacitance currents were largely corrected by a bridge circuit within the loose patch clamp amplifier 26 . The remaining linear leakage and capacitive currents were corrected using a P/4 protocol in which the test steps were followed by four control voltage steps a quarter the amplitude and of opposite signs to the test step. These were of a sign and/or amplitude that did not activate the voltage www.nature.com/scientificreports www.nature.com/scientificreports/ conductances, hence only eliciting leak currents. These responses were measured, summed and subtracted from the current values obtained from the original test step within the loose patch clamp 26 .
An IBM-compatible computer connected to the loose patch clamp amplifier, which was connected to the headstage that held the recording pipette was used to deliver voltage clamp steps relative to the RMP. Study of each patch commenced with strongly depolarising +80 mV, 15 ms pulses in order to confirm the presence of ion channels. Only patches showing clear, resolvable inward currents with kinetics characteristic of voltage-gated Na + channels were investigated with the full set of pulse protocols. The patch clamp currents were sampled by the computer at a 50 kHz digital sampling rate and filtered with a DC-10 kHz bandwidth, using a 10 kHz Bessel low pass filter.
Experiments were initially performed with the muscle preparation immersed in control KH solution. Around 2-3 different patches were investigated using the complete set of pulses within 20 minutes after transferring the muscle preparation into the bath. Only muscles that gave stable control results were further studied in the presence of pharmacological challenge. For each pharmacological condition, data from patches collected from at least 3 muscle preparations were used. To achieve conditions comparable to previous studies on [Ca 2+ ] i , the bathing solution was then rapidly replaced with KH solution containing either 0.5 mM and 2.0 mM caffeine. The resulting effects on Na + current amplitude, and activation and inactivation characteristics were then examined at increasing times from a sequence of different patches following onset of caffeine challenge. Averaged currents from the 10 sweeps obtained for each pulse voltage within the experimental protocol were then normalised to the pipette tip cross-sectional area to give current densities, using the formula relating current density (in pA/µm 2 ), measured current (in nA) and pipette diameter (in µm): The resulting peak I 1 and I 2 current densities obtained from the V 1 and V 2 steps in the double pulse protocols generated activation and inactivation current-voltage (I-V) curves, respectively, for each patch. The rising phase of the activation I-V curve for each patch was fitted to a two-state Boltzmann function relating the peak current density for any V 1 excursion (I 1 ) to its maximum value (I max ), the voltage at half maximal peak current (V*), the excursion corresponding to the V 1 step (V) and the steepness factor (k): I 1 = I max /[1 + exp{(V* − V)/k}]). The inactivation I-V curves were fitted to a similar Boltzmann function relating the peak current density I 2 for the V 2 step following any given V 1 excursion to its maximum value (I max ), the voltage at half maximal current (V * ), the excursion of the V 1 step (V) and a steepness factor (k): Successive families of ionic currents revealed alterations in current characteristics with time following caffeine challenge. These are represented by displaying currents, current-voltage curves and computed values of I max , V * and k as obtained from each patch averaged to give means ± SEM for currents obtained within successive time intervals following introduction of pharmacological agents. Curve fitting procedures of the I-V curves used fitting algorithms in the open source programme SciDAVis (Version 1.25-mac). Derived parameters obtained under different conditions were tested for significant differences to a P < 0.05 significance level by ANOVA with a post-hoc Dunnett's test.

Ethical approval. This research has been regulated under the Animals (Scientific Procedures) Act 1986
Amendment Regulations 2012 following ethical review and approval by the University of Cambridge Animal Welfare and Ethical Review Body (AWERB). All procedures were completed by Home Office-licensed personnel and fell within the scope of Schedule 1 of the UK Animals (Scientific Procedures) Act (1986).

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.