Lubeluzole, (+)-(S)-1-[4-[N-(benzothiazol-2-yl)-N-methylamino]piperidin-1-yl]-3-(3,4-difluorophenoxy)-2-propanol, is a representative of a novel class of neuroprotective drugs currently being evaluated for usefulness in the management of acute ischemic stroke. 1–5 In a rat photochemical stroke model, curative i.v. lubeluzole restores neurologic function within a time window of 6 h. 6 The mechanism of action of lubeluzole has not been demonstrated clearly as yet, but the cellular pathways modulated by this agent are not believed to be dependent upon NMDA receptor blockade. 6 It has been suggested that lubeluzole interacts at a more focused point in the ischemic cascade that is below glutamate release, 7,8 such as at the level of NO after the generation of this free radical. 7,9 The neuroprotective effects of lubeluzole have also been reported to include blockade of voltage-gated Ca2+ channels 10,11 and in particular, blockade of voltage-gated Na+ channels. 12–16 Therefore, studies utilizing Na+ channel blockers as neuroprotectants have revealed that loss of the transmembrane sodium gradient is implicated in cellular damage in focal ischemia following middle cerebral artery occlusion, 17,18 in complete brain ischemia 19 as well as in myocardial ischemia. 20 Although blockade of neuronal Na+ channels has been perceived as the chief mechanism of neuroprotective action of lubeluzole, 12,14,15 possible interactions of the drug with cardiac Na+ channels have not been reported to date. 21 The aim of the present study was therefore to investigate the potential interaction of lubeluzole with cardiac Na+ channels. This was assessed in guinea-pig freshly isolated cardiac myocytes using the whole-cell patch-clamp technique.
Male guinea-pigs (SPF, Hartley, Charles River, France), weighing 200 to 300 g, were conditioned for 1 week at 21 ± 1°C with constant humidity of 55 ± 5%, a 12:12 hr light:dark cycle and free access to food and tap water, in accordance with French Law and the local ethics committee guidelines for animal research.
Whole-Cell Recording Technique
Single guinea-pig ventricular myocytes were isolated by the technique of Mitra and Morad. 22 Briefly, guinea-pigs were anaesthetized with sodium pentobarbitone (Sanofi Santé, France). Hearts were then quickly removed, cannulated, mounted at the base of a 60-cm high Langendorff column and perfused retrogradely with an oxygenated modified Tyrode's solution containing (in mM): NaCl 120.8, KCl 4.96, MgCl2 0.98, HEPES 19.97, Na-pyruvate 4.54, glucose 22.2. The pH was adjusted to 7.3 with NaOH. After perfusion for 7 minutes with a 0.45-mM calcium-Tyrode solution and for 6 minutes with a nominally Ca2+-free Tyrode solution, the enzymatic dissociation was begun in Tyrode solution complemented with protease (0.3 mg/ml, Sigma Chemicals, St. Louis, MO) and collagenase (1 mg/ml, Type IA, Sigma Chemicals, St. Louis, MO). The heart was then washed with a 0.05-mM calcium-Tyrode solution for 5 minutes and was cut into pieces in fresh solution and gently agitated. The cell suspension was placed in a Petri dish and incubated for 15 minutes at 38°C.
Ion currents were recorded by the patch-clamp technique in the whole-cell configuration. Pipettes with resistances of 1 to 3 MΩ were pulled from Pyrex capillaries (model 7740, Corning Glass Inc., NY) not fire-polished before use. A flow of solution (50–100 μL/min) from one of a series of 5 outlets continuously superfused the cell from which a recording was being made. A patch clamp amplifier (Axopatch 200A, Axon Instruments, Foster City, CA) was used. The resistance in series with the cell membrane was compensated by the dynamic series resistance control to provide the fastest possible capacity without ringing of the amplifier. Neither capacitive or leakage currents were compensated. Cell currents were digitized at 6 kHz and analyzed by computer (Desk-Pro 486/33 MHz, Compaq, Houston, TX) with interactive software (ACQUIS 1, G. Sadoc, Paris, France).
For INa measurements, calcium current was blocked by addition of external Co2+ and K+ currents were abolished by a nominally K+-free medium and by addition of CsCl. The internal solution (pipette) contained (in mM): CsCl 130, CaCl2 1, MgCl2 2, HEPES 10, EGTA 15, Mg-ATP 4, D(+)-glucose 10, pH 7.4 (CsOH). The external solution contained (in mM): NaCl 25, tetraethylammonium chloride 80, MgCl2 2.5, CoCl2 3, HEPES 10, 4-aminopyridine 5, CsCl 5, D(+)-glucose 10, pH 7.4 (CsOH). All experiments were performed at room temperature (19–22°C).
To obtain rapid and uniform control of the membrane potential and to minimize voltage errors, INa was recorded using an external solution containing 25 mM NaCl. Moreover, small cells were preferred for selection, and those with INa amplitude greater than 5 nA were rejected.
Depolarizing voltage pulses were delivered at 0.2Hz. Pulse duration was 350 milliseconds. The amplitude of INa was measured as peak inward current with reference to the zero current measured before the test pulse. The maximal amplitude of INa obtained in the presence or absence of drug or vehicle was normalized with respect to control (baseline) values to determine the percentage of variation.
Lubeluzole dihydrochloride was synthesized by the Division of Medicinal Chemistry II, Centre de Recherche Pierre Fabre and was dissolved in distilled water.
Data are expressed as means ± SEM values. Student t test for paired data was used to determine statistical significance of data within groups. Any P value lower than P < 0.05 was considered significant (StatView 4.1, Abacus Concepts, Berkeley, CA). Curve fitting was performed using the non-linear least-square gradient-expansion Marquardt algorithm (Origin 4.0, Microcal Software Inc, Northampton, MA).
Sodium conductance (gNa) was calculated using the following equation:MATHwhere INa is the value of sodium current at a given test potential, V; (V-Vrev) is the ionic driving force, where Vrev is the reversal potential. Normalized conductance curves were then calculated and fitted to a Boltzmann distribution. From this equation the midpoint of activation (V0.5) and slope (s) of the curve were calculated:MATHwhere gmax is the maximal gNa and V is the test potential.
Data for steady-state sodium current inactivation (h) were fitted using the Boltzmann equation:MATHwhere V0.5 is the midpoint of inactivation, k is the slope factor, and Vpp is the 2-second prepulse voltage (from −120 mV to −40 mV with incremental steps of 5 mV).
The curve fitting routines for the unrecovered fraction (Itest/Icond) used to assess the rate of recovery from sodium current inactivation were obtained using the following double exponential decay function:MATHwhere A is the fraction of the total current described by a fast time constant (τ1), and B is the fraction of the total current described by a slow time constant (τ2).
Na Current Block by Lubeluzole
Figure 1A shows typical superimposed INa recordings obtained under control conditions, during exposure to lubeluzole (10 μM; 3 minutes) and during washout (10 minutes). INa was elicited by a depolarizing pulse to −30 mV from a holding potential of −80 mV. Amplitude of peak INa was reduced and this was reversed within 5 minutes following washout of the drug. Lubeluzole did not appear to significantly modify inactivation kinetics (Fig. 1A). INa was fitted by a monoexponential function by means of a non-linear regression algorithm according to the following equation:MATHLubeluzole (10 μM) failed to significantly affect INa decay (τ = 3.59 ± 0.7 versus 3.83 ± 0.5 milliseconds in baseline; n = 5; P = NS), obtained with a depolarizing pulse to −30 mV.
Inhibition of INa by lubeluzole was dependent upon the voltage of the resting state; peak INa current was reduced by 50.8 ± 4.9% (n = 5, P < 0.001) at a holding potential of −80 mV, by 33.5 ± 6.1% (n = 6, P < 0.05) at a holding potential of −100 mV, and by 27.2 ± 8.1% (n = 6, P < 0.05, data not shown) at a holding potential of −120 mV. Figure 1B shows that lubeluzole reduced peak INa obtained at a holding potential of −80 mV in a concentration-dependent manner with an IC50 value (geometric mean) of 9.5 μM (95% confidence limits: 3.5–21.9) and a Hill coefficient of 1.1.
Effect of Lubeluzole on Na+ Channel Activation and Inactivation
Na+ conductance activation curves in the presence and absence of lubeluzole (10 μM) are shown in Figure 2A. (Na+ conductance was normalized to maximal conductance). Half maximal voltage for Na+ conductance activation was not significantly affected by the drug (control: −56.3 ± 0.8 mV and lubeluzole (10 μM): −57.5 ± 1.3 mV, n = 6, NS; mean variation −0.69 ± 1.3 mV). This suggests that lubeluzole, at the concentrations studied, had no effect on voltage-dependent Na+ channel activation.
The effect of lubeluzole on the voltage dependency of steady-state inactivation is summarized in Figure 2B. INa availability was studied using a standard preconditioning 2-second pre-pulse protocol. This 2-pulse sequence was repeated every 10 seconds at a holding potential of −120 mV, which allowed full recovery of any use-dependent block. Under control conditions (Fig. 2B), the solid line representing a single Boltzmann distribution had a half maximal inactivation (V0.5) of −67.2 ± 0.7 mV (n = 4). During exposure to lubeluzole (10 μM), V0.5 was shifted to −76.9 ± 4.3 mV (n = 4). The slope factor that reflects the steepness of the voltage dependence of inactivation was not significantly affected by lubeluzole (control: 5.1 ± 0.6 mV and lubeluzole (10 μM): 5.3 ± 0.2 mV, n = 4, NS).
Tonic Block Induced by Lubeluzole
The inhibitory effects of lubeluzole on cardiac INa can be divided into a tonic block and a use-dependent block (UDB) similar to the action of local anesthetic agents. Depolarization to −40 mV after a 90-second rest period at a holding potential of −120 mV, resulted in a marked reduction in INa amplitude compared with control after exposure to 10 μM lubeluzole (Fig. 3A). The tonic component of block was 27.4 ± 5.8% in presence of lubeluzole (10 μM) and 2.7 ± 1.4% in the presence of vehicle (Fig. 3B; n = 6, P < 0.01).
Use-Dependent Block Induced by Lubeluzole
In addition to tonic block, lubeluzole produced use-dependent block (UDB) of INa. Figure 4A shows the time course of UDB (n = 4) at 1 Hz for pulses of 4 different durations in the presence of 10 μM lubeluzole. Peak INa was elicited by applying the first depolarizing pulse at 1 Hz from a holding potential of −100 mV to −30 mV, and by the 30 following pulses in the presence of 10 μM lubeluzole, for which the pulse durations were varied from 10 to 200 milliseconds. The 200-millisecond train pulse (Fig. 4A) produced a greater degree of UDB than the 10-, 50-, or 100-millisecond pulses.
Rate of Recovery from Block Induced by Lubeluzole
The rate at which INa recovers from block induced by a test pulse represents one of the principal determinants of UDB. Figure 5 shows the effects of 10 μM lubeluzole on the rate of recovery from block. A standard 2-pulse protocol was used to determine the rate of recovery from block. The holding potential was maintained at −100 mV (Fig. 5A) or at −120 mV (Fig. 5B) and the depolarizing steps were elicited to −40 mV. Between each step, the potential recovered to its initial value. At a holding potential of −100 mV (Fig. 5A), INa recovered rapidly under control conditions, the recovery being described by the sum of 2 exponentials, one with a fast time constant (τ1 = 4.4 ± 6.2 milliseconds) and the other with a slow time constant (τ2 = 22.7 ± 1.5 milliseconds). During exposure to 10 μM lubeluzole, the time course of recovery was slowed markedly, with τ1 and τ2 values of 311 ± 144 and 672 ± 23 milliseconds, respectively (n = 5, P < 0.001).
At a holding potential of −120 mV (Fig. 5B), INa recovered rapidly under control conditions with τ1 and τ2 values of 5.8 ± 5.7 and 14.5 ± 2.8 milliseconds (n = 6), respectively. During exposure to 10 μM lubeluzole, the time course of recovery was slowed markedly, with τ1 and τ2 values of 331.8 ± 158 and 745 ± 218 milliseconds, respectively (n = 6, P < 0.05).This effect was partially reversible after 5 minutes perfusion with vehicle with τ1 and τ2 values of 31.8 ± 6.6 and 481 ± 32 milliseconds (n = 3), respectively (data not shown).
Lubeluzole is a novel neuroprotective agent for which blockade of neuronal voltage-gated sodium channels constitutes one of the major molecular targets. 12–15 Since no data are available on the interaction of lubeluzole with cardiac sodium channels, a study was undertaken to investigate this issue. The data show that lubeluzole produces a concentration-dependent, readily reversible block of rapid sodium current in cardiac myocytes. The drug shifts the sodium current inactivation curve to more hyperpolarized potentials, although no shift in the activation curve was apparent. This profile of action is similar to that seen with typical class I antiarrhythmic agents, lidocaine 23 and quinidine. 24 Lubeluzole therefore appears to block both neuronal and cardiac sodium channels within a similar range of concentrations.
Blockade of Cardiac Na+ Channels by Lubeluzole
Lubeluzole inhibits TTX-sensitive Na+ current in isolated hippocampal cells with an IC50 value of 3.1 μM 12 and protects against veratridine-induced Na+ overload and neurotoxicity in a hippocampal slice preparation (IC50 0.54 μM; 14). In the present study, the results demonstrate that lubeluzole also inhibits Na+ current in isolated ventricular cardiomyocytes with an IC50 value of 9.5 μM suggesting that lubeluzole has little or no selectivity for neuronal over cardiac Na+ channels. Therefore, concentrations of lubeluzole that exert neuroprotection by blockade of neuronal Na+ channels 13 are likely to affect cardiac sodium channel function.
Although the concentrations at which lubeluzole blocked cardiac Na+ channels in these voltage-clamp experiments were rather high compared with the effective plasma drug concentrations in a rat stroke model (0.23 μM, of which 1% was unbound 25), lubeluzole could have a more potent effect on cardiomyocyte sodium channels than might be expected from the IC50 value for inhibition of INa. Indeed, the IC50 value was determined at a holding potential of −80 mV and a low frequency (0.5 Hz). Lubeluzole inhibited INa more potently at more depolarized potentials, which is relevant to pathologic situations in which the cells are more depolarized such as cardiac ischemia. One limitation of the present experimental conditions could be the use of Co2+ in the external medium. However, Co2+ at a 3-mM concentration only slightly reduced the amplitude of the peak current suggesting that this non-significant effect of Co2+ on the amplitude of INa does not significantly interfere with the interaction of lubeluzole with Na+ channels.
Role of Channel State in Block by Lubeluzole
Lubeluzole produced both tonic block and UDB, findings which have also been reported with class I antiarrhythmic agents in a variety of animal models, 23 including man. 26 Lubeluzole-induced UDB was dependent on drug concentration, pulse duration, and interpulse interval. Each of these factors has been associated with altered affinity of the Na+ channel for drugs, depending on channel state as formalized in the modulated receptor theory. 27–30 According to the Hodgkin and Huxley model 31 of Na+ channel gating, Na+ channels exist in 3 primary states (rested, activated, and inactivated). It has subsequently been shown 28 that each state has its own specific binding affinity for class I antiarrhythmic agents.
Tonic Block Induced by Lubeluzole
It is established that tonic block is composed of rested and inactivated state block. Lubeluzole (10 μM) induced 27.4 ± 5.8% of tonic block at a holding potential of −120 mV. At the same potential, the steady-state inactivation curve (Fig. 2B) shows that the channels are in a rested state suggesting that under steady-state conditions, lubeluzole had an affinity for the rested state resulting in a decrease in the number of available sodium channels.
Use-Dependent Block Induced by Lubeluzole
Use-dependent block is produced when the drug-channel interaction is too slow to reach equilibrium within a single cycle of Na+ channel activation and inactivation. This use-dependent blocking action is explained by 2 different theories, the modulated receptor 27 and the guarded receptor hypotheses. 23 In the latter theory, the affinity of the drug binding site is constant, but access to the binding site is guarded by activation and/or inactivation gates, such that the forward binding rate is faster when the channel is open or inactivated than under resting conditions. The present experiments show that lubeluzole exerts pronounced UDB, which was dependent on the longer duration of the depolarizing pulse and a more depolarized level of holding potential. At a holding potential of −100 mV, a train of short pulses (10 milliseconds) produced a large degree of UDB and the extent of block was enhanced by a train of longer pulses (200 milliseconds), because at this holding potential, Na+ channels are partially in the inactivated state. These results suggest that lubeluzole has little effect on Na+ channels in their activated state, but binds to channels in the inactivated state (ie, lubeluzole produces inactivated gate (state)-dependent block, resulting in block of INa due to a decrease in the number of available channels). Since lubeluzole blocks inactivated channels with an affinity that is presumably greater than that for rested channels, at depolarized holding potentials (where tonic block is marked and most channels are distributed between drug-bound and unbound inactivated states), rapid train pulses can increase the fraction of channels in the inactivated state.
Rate of Recovery from Lubeluzole Block
Interpulse interval represents the second major influence on UDB. UDB develops whenever the interval between pulses is too short to allow complete recovery from the block that developed during the previous pulse. Therefore, the time course for the development of UDB will depend on the amount of recovery from block that occurs during the diastolic interval. The finding that lubeluzole slows recovery from inactivation block is consistent with previous studies with typical class I antiarrhythmic agents. 23,24,26 The present results indicate the presence of a biexponential recovery process during exposure to lubeluzole, the process being best described by a fast time constant representing the recovery of drug-free Na+ channels and a slow process representing the recovery of drug-associated Na+ channels. Based on the guarded receptor hypothesis, at a holding potential of −100 mV, the access of the drug to its binding site would be prevented by the inactivation gate. Under these conditions, the drug would hardly access its binding site, but would also have difficulty leaving the site, which might reflect slow kinetics of drug-receptor interaction, resulting in an enhanced degree of UDB because of the longer time the drug was present on its binding site. Thus the delayed recovery process observed with lubeluzole can be explained parsimoniously by a slow transition between multiple drug-bound inactivated states. Further studies are warranted to clarify the effects of lubeluzole on multiple inactivated states.
Collectively, the results indicate that lubeluzole produces both tonic and use-dependent block of cardiac sodium channels due to interaction of the drug with channels in the inactivated state. In addition, the data suggest that lubeluzole interacts with both neuronal and cardiac sodium channels at similar concentrations.
We are grateful to Anne Leparq for excellent technical assistance.
1. Aronowski J, Strong R, Grotta JC. Combined neuroprotection and reperfusion therapy for stroke
: effect of lubeluzole and diaspirin cross-linked hemoglobin in experimental focal ischemia. Stroke
. 1996; 27:1571–1577.
2. Diener HC, Hacke W, Hennerici M, et al. Lubeluzole in acute ischemic stroke
: a double-bind, placebo-controlled phase II trial. Stroke
. 1996; 27:76–81.
3. Grotta J. Lubeluzole treatment of acute ischemic. Stroke
. 1997; 28:2338–2346.
4. Diener HC, Cortens M, Ford G, et al. Lubeluzole in acute ischemic stroke
treatment: A double-blind study with an 8-hour inclusion window comparing a 10-mg daily dose of lubeluzole with placebo. Stroke
. 2000; 31:2543–2551.
5. Kaste M. Thrombolysis in ischaemic stroke
—present and future: role of combined therapy. Cerebrovasc Dis. 2001: 11(S1):55–59.
6. De Ryck M, De Keyser J, Lesage A, et al. Lubeluzole reduces mortality and improves functional outcome in the treatment of acute ischemic stroke
: early preclinical and clinical findings. In: Grotta J, Miller LP, Buchan AM, eds. Ischemic Stroke. Recent Advances in Understanding Therapy.
International Business Communications; 1995:271–275.
7. Lesage AS, Preters L, Leysen JE. Lubeluzole, a novel long-term neuroprotectant, inhibits the glutamate-activated nitric oxide synthase pathways. J Pharmacol Exp Ther. 1996; 279:759–766.
8. Koinig H, Vornik V, Rueda C, et al. Lubeluzole inhibits accumulation of extracellular glutamate in the hippocampus during transient global cerebral ischemia. Brain Res. 2001; 898:297–302.
9. Maiese K, Tenbroeke M, Kue I. Neuroprotection of lubeluzole is mediated through the signal transduction pathways of nitric oxide. J Neurochem. 1997; 68:710–714.
10. Marrannes R, De Prins E, Clincke G. Influence of lubeluzole on voltage-sensitive Ca2+
channels in isolated rat neurons. J Pharmacol Exp Ther. 1998; 286:201–214.
11. Marrannes R, De Prins E. Site of action of lubeluzole on voltage-sensitive Ca2+
channels in isolated dorsal root ganglion cells of the rat: influence of pH. J Pharmacol Exp Ther. 2000; 295:531–545.
12. Osikowska-Evers BA, Wilhelm D, Nebel U, et al. The effects of the novel neuroprotective compound lubeluzole on sodium current and veratridine-induced sodium load in rat brain neurons and synaptosomes. J Cereb Blood Flow Metab. 1995; 15:S380.
13. Urenjak J, Obrenovitch TP. Pharmacological modulation of voltage-gated Na+
channels: a rational and effective strategy against ischemic brain damage. Pharmacol Rev. 1996; 8:21–67.
14. Ashton D, Willems R, Wynants J, et al. Altered Na+
-channel function as an in vitro model of the ischemic penumbra action of lubeluzole and other neuroprotective drugs. Brain Res. 1997; 745:210–221.
15. Culmsee C, Junker V, Wolz P, et al. Lubeluzole protects hippocampal neurons from excitotoxicity in vitro and reduces brain damage caused by ischemia. Eur J Pharmacol. 1998; 342:193–201.
16. Blackburn-Munro G, Ibsen N, Erichsen HK. A comparison of the anti-nociceptive effects of voltage-activated Na+
channel blockers in the formalin test. Eur J Pharmacol. 2002; 445:231–238.
17. Graham SH, Chen J, Lan J, et al. Neuroprotective effects of a use-dependent blocker of voltage-dependent sodium channels, BW619C89, in rat middle cerebral artery occlusion. J Pharmacol Exp Ther. 1994; 269:854–859.
18. Rataud J, Debarnot F, Mary V, et al. Comparative study of voltage-sensitive sodium channels blockers in focal ischemia and electric convulsions in rodents. Neurosci Lett. 1994; 172:19–23.
19. Xie Y, Dengler K, Zacharias E. Effects of sodium channel blocker tetrodotoxin (TTX) on cellular ion-homeostasis in rat brain subjected to complete ischemia. Brain Res. 1994; 652:216–224.
20. Ver Donck L, Borgers M, Verdonck F. Inhibition of sodium and calcium overload pathology in the myocardium: a new cytoprotective principle. Cardiovasc Res. 1993; 27:349–357.
21. Le Grand B, Dordain-Maffre M, John GW. Lubeluzole-induced prolongation of cardiac action potential in rabbit Purkinje fibres. Fundam Clin Pharmacol. 2000; 14:159–162.
22. Mitra R, Morad M. A uniform enzymatic method for dissociation of myocytes from hearts and stomachs of vertebrates. Am J Physiol. 1985; 249:H1056–H1060.
23. Starmer CF, Grant AO, Strauss HC. Mechanisms of use-dependent block of sodium channels in excitable membranes by local anesthetics. Biophys J. 1984; 46:15–27.
24. Snyders DJ, Hondeghem LM. Effects of quinidine on the sodium current of guinea-pig ventricular myocytes. Circ Res. 1990; 66:565–579.
25. De Ryck M, Keersmaekers R, Duytschaever H, et al. Lubeluzole protects sensorimotor function and reduces infarct size in a photochemical stroke
model in rats. J Pharmacol Exp Ther. 1996; 279:748–758.
26. Jia H, Furukawa T, Singer DH, et al. Characteristics of lidocaine block of sodium channels in single human atrial cells. J Pharmacol Exp Ther. 1993; 284:1275–1284.
27. Hille B. Local anesthetics: hydrophylic and hydrophobic pathways for the drug-receptor reaction. J Gen Physiol. 1977; 69:497–515.
28. Hondeghem LM, Katzung BG. Time- and voltage-dependent interactions of antiarrhythmic drugs with cardiac sodium channels. Biochim Biophys Acta. 1977; 472:373–398.
29. Hondeghem LM, Katzung BG. Antiarrhythmic agents. The modulated receptor mechanism of action of sodium and calcium channel-blocking drugs. Annu Rev Pharmacol Toxicol. 1984; 24:387–423.
30. Snyders DJ, Bennett PB, Hondeghem LM. Mechanisms of drug-channel interaction. In: Fozzard HA, Haber E, Jennings RB, et al, eds. The Heart and Cardiovascular System.
Raven Press; 1992:2165–2193.
31. Hodgkin AL, Huxley AF. A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol. 1952; 117:500–544.
Keywords:© 2003 Lippincott Williams & Wilkins, Inc.
cardiomyocyte; lubeluzole; sodium channels; state-dependent block; stroke