Online Supplementary Figures S1-S6. The mitochondrial Na + -Ca 2+ -exchanger, NCLX, regulates automaticity of HL-1 cardiomyocytes

Transcription

1 Online Supplementary Figures S-S6 for The mitochondrial Na + -Ca 2+ -exchanger, NCLX, regulates automaticity of HL- cardiomyocytes by Ayako Takeuchi *, Bongju Kim 2, and Satoshi Matsuoka 2 Department of Physiology and Biophysics, Graduate School of Medicine, Kyoto University, Yoshida-konoe, Sakyo-ku, Kyoto , Japan 2 Center for Innovation in Immunoregulative Technology and Therapeutics, Graduate School of Medicine, Kyoto University, Yoshidakonoe, Sakyo-ku, Kyoto , Japan Present address A.T. and S.M.: Department of Integrative Physiology, Faculty of Medical Sciences, University of Fukui, 23-3, Matsuokashimoaizuki, Eiheiji-cho, Yoshida-gun, Fukui, 90-93, Japan Correspondence and requests for materials should be addressed to A.T. ( atakeuti@u-fukui.ac.jp).

2 Figure S Figure S. Effect of NCLX knockdown on the time course of Ca 2+ mit increase in permeabilized HL- cells. Plasma membrane of the cells loaded with 5 µm Rhod-2, AM was permeabilized by incubating the cells with CLM mit containing 30 µm β-escin for 60 sec. Then cells were transferred to CLM mit without Ca 2+,incubated for 3 min, and the record was started. At 0 min, mitochondria were loaded with Ca 2+ in the CLM mit containing 0 µm Ca 2+. Note that the time course of Ca 2+ mit increase was almost identical between control sirna- and mnclx sirna-transfected cells, indicating that mitochondrial buffering capacity was not altered by NCLX knockdown.

3 Figure S2 Figure S2. Ca 2+ i and Ca 2+ mit in beating HL- cells. Plasmid harbouring Case2-mito, an indicator of mitochondrial Ca 2+, was transfected into HL- cells. Representative recordings of mitochondrial (Case2-mito; upper) and cytosolic (Rhod-3; lower) Ca 2+ in an identical cell are shown. Note that Ca 2+ mit oscillation was masked by noises while there was a clear cytosolic Ca 2+ oscillation.

4 Figure S3 Figure S3. Effect of a NCX mit inhibitor, CGP-3757, on spontaneous Ca 2+ i transients. Spontaneous Ca 2+ i transients were recorded from Indo- loaded cells. (a) Representative recordings from cells treated with 0-20 µm CGP-3757 for 5 min. (b) Bar graphs summarize the resting Ca 2+ i, peak Ca 2+ i and amplitude of Ca 2+ i transients expressed as Indo- fluorescence ratio (F405/F480), and cycle length of the spontaneous Ca 2+ i transients (N = 4). * P < 0.05, ** P < 0.0 vs 0 µm CGP-3757.

5 Figure S4 Figure S4. Effect of a I CaL inhibitor, nifedipine, on spontaneous Ca 2+ i transients. Spontaneous Ca 2+ i transients were recorded from Indo- loaded cells. (a) Representative recordings from cells treated with 0-20 µm nifedipine for 5 min. (b) Bar graphs summarize the resting Ca 2+ i, peak Ca 2+ i and amplitude of Ca 2+ i transients expressed as Indo- fluorescence ratio (F405/F480), and cycle length of the spontaneous Ca 2+ i transients (N = 4). * P < 0.05, ** P < 0.0 vs 0 µm nifedipine.

6 Figure S5 Figure S5. Effect of a mitochondrial Ca 2+ uniporter inhibitor, Ru360, on spontaneous Ca 2+ i transients. Spontaneous Ca 2+ i transients were recorded from Indo- loaded cells. (a) Representative recordings from cells before or after the treatment with 5 µm Ru360 for 30 min. (b) Graphs summarize the resting Ca 2+ i, peak Ca 2+ i and amplitude of Ca 2+ i transients expressed as Indo- fluorescence ratio (F405/F480), and cycle length of the spontaneous Ca 2+ i transients from cells before or after the treatment with 5 µm Ru360 for 5 min and 30 min (N=4). N.S., not significant vs before.

7 Figure S6 Figure S6. Western blot analysis of HCN4 and Ca 2+ handling proteins (Cav.2, Cav3., NCX, and SERCA2) in control- and mnclx sirna-transfected cells. Membrane proteins were extracted (BioVision, USA) from four different batches of HL- cells transfected with control or mnclx sirna, and were resolved by SDS-PAGE. The gels were transferred to PVDF membrane and used for Western blot analysis. Antibodies were purchased from Santa Cruz for Cav.2, Cav3., and SERCA2, Abcam for HCN4, NCX and β-actin, respectively. (a) Representative data. β-actin was used as reference. (b) Bar graph represents the expression levels of proteins in NCLX knockdown cells compared to those in control cells (N=7).

8 Figure S7 Figure S7. Effect of a SERCA inhibitor, thapsigargin, on spontaneous Ca 2+ i transients. Spontaneous Ca 2+ i transients were recorded from Indo- loaded cells. (a) Representative recordings from cells treated with 5 µm thapsigargin at the interval of 5 min for 20 min. (b) Graphs summarize the resting Ca 2+ i, peak Ca 2+ i and amplitude of Ca 2+ i transients expressed as Indo- fluorescence ratio (F405/F480), and cycle length of the spontaneous Ca 2+ i transients (N = 7-8). * P < 0.05, ** P < 0.0 vs 0 min.

9 Figure S8 Figure S8. Simulation of SERCA amplitude reduction. (a) The amplitude factor of SERCA was reduced to 50%, and simulated for 30 sec. Black and red lines represent simulated data from 00% and 50% SERCA, respectively. (b) Effects of varying amplitude factor of SERCA on the cycle length and resting Ca 2+ i.

10 Figure S9 Figure S9. Effect of NCLX knockdown on the reactive oxygen species and ATP level. (a) Reactive oxygen species. Cells were loaded with5 µm MitoSOX (Life Technologies), a mitochondrial superoxide indicator. Images were obtained with 488 nm excitation and emission (LSM70), using a x40 oil objective lens. Mean fluorescence intensities after subtracting background were presented (N=5). (b) Cellular ATP level measured by using CellTiter-Glo Luminescent Cell viability Assay (Promega) (N=8). con; control sirna-transfected cells. NCLX KD; NCLX sirna-transfected cells. a.u., arbitrary unit. N.S., not significant vs control sirna.

11 Online Supplementary Data for The mitochondrial Na + -Ca 2+ exchanger, NCLX, regulates automaticity of HL- cardiomyocytes by Ayako Takeuchi *, Bongju Kim 2, and Satoshi Matsuoka 2 Department of Physiology and Biophysics, Graduate School of Medicine, Kyoto University, Yoshida-konoe, Sakyo-ku, Kyoto , Japan 2 Center for Innovation in Immunoregulative Technology and Therapeutics, Graduate School of Medicine, Kyoto University, Yoshida-konoe, Sakyo-ku, Kyoto , Japan Present address A.T. and S.M.: Department of Integrative Physiology, Faculty of Medical Sciences, University of Fukui, 23-3, Matsuokashimoaizuki, Eiheiji-cho, Yoshida-gun, Fukui, 90-93, Japan Correspondence and requests for materials should be addressed to A.T. ( atakeuti@u-fukui.ac.jp).

12 This is a supplemental data that describes details of the computer model of spontaneously beating HL- cardiomyocytes. The basic frame of the model was adapted from the human atrial cell model by Grandi et al.. Major modifications from the Grandi model are ) an introduction of mitochondrial Ca 2+ handling, 2) an introduction of hyperpolarization-activated cation current (I ha ) and voltage-dependent T-type Ca 2+ current (I CaT ), both of which are characteristic ion channels in HL- cells 2-5, and 3) modifications of parameters for several ion channel equations to reproduce the experimental data in HL- cardiomyocytes,4,6-8. The model cell has six compartments; extracellular space, junctional space (JS), subsarcolemmal space (SL), sarcoplasmic reticulum (SR), myoplasm (cytoplasm), and mitochondria. The model scheme is shown in Figure 6a in the main text. Despite the variability of expression levels and/or current densities of particular carriers among the individual cells, the phenotypes of a cellular syncytium are essentially identical for multiple cellular recordings (also refer to 9 ). Therefore although this HL- model corresponds to a single cell, the concept of the model is an averaged phenotype of the cells from multicellular preparations. Units are mm for concentration and ms for time. Mathematical formulas are described in the following tables (Tables -8). The model was created with a Java-based simulation platform, simbio 0 and the simultaneous differential equations were integrated by Euler method with adaptive time step. Figures -4 demonstrate the characteristics of channel properties in the model compared with reported experimental recordings using HL- cardiomyocytes. Detailed protocols are described in the legends. In Figures 5 and 6, configuration of action potential, major ionic currents and Ca 2+ transients in each compartment are shown. The cycle length of the spontaneously generated action potentials is ms, which well corresponds to the experimental data, ± 7.52 ms (see Table in the main text). There is no experimental report on mitochondrial Ca 2+ (Ca 2+ mit) concentration in rapidly beating HL- cells (~3.5 Hz). However, several groups reported that Ca 2+ mit concentrations can go up to tens or hundreds of micromolar -3. Ca 2+ mit concentration in the model is within the range. In spite of the large cytosolic Ca 2+ transients, oscillation of Ca 2+ mit was hardly detected under our experimental condition (see Supplementary Fig. S2). Our model estimates small Ca 2+ mit oscillation with the amplitude of ~20 nm (Figure 6), which is comparable to the value of ~30 nm measured in rabbit ventricular myocytes 4. 2

13 Table. Abbreviations R F T Ca o Ca i Ca JS Ca SL Ca SR Ca mit Ca totalmit Na o Na i Na mit K o K i Mg i Gas constant, C mv/k/mmol Faraday s constant, C/mmol absolute temperature, 30 K Extracellular Ca 2+,.8 mm Cytoplasmic Ca 2+ concentration in mm Junctional space Ca 2+ concentration in mm Subsarcolemmal space Ca 2+ concentration in mm Sarcoplasmic reticulum Ca 2+ concentration in mm Mitochondrial free Ca 2+ concentration in mm Mitochondrial total Ca 2+ concentration in mm Extracellular Na + concentration, 40 mm Cytoplasmic Na + concentration, 5.0 mm Mitochondrial Na + concentration, 4.0 mm Extracellular K + concentration, 5.4 mm Cytoplasmic K + concentration, 20 mm Mitochondrial membrane potential, -50 mv Cytoplasmic Mg concentration,.0 mm Table 2. Cell property Reference Total cellular volume Vol i m 3 Myoplasm volume Vol myo 0.65 Vol i m 3 Junctional space Vol JS volume Vol i m 3 Subsarcolemmal space volume Vol SL 0.02 Vol i m 3 SR volume Vol SR Vol i m 3 Mitochondria volume Vol mit 0.2 Vol i m 3 5 Cell capacitance Cm 40 pf 3 Fractional current of junctional space F JS 0. Fractional current of subsarcolemmal space F SL - F JS 3

14 Fractional current through I CaL of junctional space Fractional current through I CaL of subsarcolemmal space F JSCa 0.9 F SLCa - F JSCa Table 3. Ion fluxes at junctional and subsarcolmmal membrane I Na ; Voltage-dependent Na + current conductance G Na 23 ns/pf exp exp exp exp exp, , 2.7 exp exp0.3485, , exp ,, 40 4

15 I NaK ; Na + /K + pump current exp0.0052, 40 exp exp0.057 exp0. 32, amplitude A NaK.26 pa/pf dissociation constant Km Nai.0 mm dissociation constant Km Ko.5 mm σ exp exp exp I Kr ; Rapidly activating K + current (adapted from 6 ) 5

16 I Ks ; Slowly activating K + current I Kp ; Plateau K + current conductance G Kr 0.73 ns/pf 5 exp exp exp conductance G Ks ns/pf 3.8 exp exp 4.2 ln pnak

17 conductance G Kp ns/pf exp I to ; Transient outward K + current (adapted from ) conductance G to ns/pf exp exp exp exp I Kur ; Transient outward K + current (adapted from ) conductance G Kur ns/pf 6 exp exp exp 0 7

18 I ClCa ; Ca 2+ activated Cl - current exp 0 I K ; Inward rectifier K + current (adapted from ) conductance G K 0.5 ns/pf.02 exp exp exp exp conductance G ClCa ns/pf 8

19 I Clb ; Background Cl - current (adapted from ) conductance G Clb ns/pf I CaL ; L-type Ca 2+ current (adapted from ) permeability P Ca cm/sec permeability P Na cm/sec permeability P K cm/sec exp exp exp 9 exp exp

20 exp exp exp exp exp 0.75 exp 0.75 exp 0.75 exp 0.75 exp 0.75 exp 0.75 exp 0.75 exp I CaT ; T-type Ca 2+ current (adapted from 7 ) conductance G CaT ns/pf 30 exp exp.068 exp

21 I Cab ; Background Ca 2+ current 48 exp exp 0.05 exp I pca ; Sarcolemmal Ca 2+ pump current amplitude A pca pa/pf dissociation constant Km pca mm conductance G Cab ns/pf I NCX ; Plasmalemmal Na + -Ca 2+ exchange current (adapted from ) amplitude A NCX 4.0 pa/pf dissociation constant Kd act mm dissociation constant Km Cai mm dissociation constant Km Cao.3 mm dissociation constant Km Nai 2.29 mm

22 dissociation constant Km Nao 87.5 mm exp 2 exp nu 0.27 k sat exp exp 2 3 exp I ha ; Hyperpolarization-activated cation current (adapted from 7 ) conductance G ha 0.2 ns/pf 70 exp exp exp 5 7 2

23 Buffers; Cytosolic Ca 2+ buffers association constant kon TnClow 32.7 /mm ms dissociation constant koff TnClow / ms total TnC low concentration Bmax TnClow 0.07 mm association constant kon TnChCa 2.37 /mm ms dissociation constant koff TnChCa / ms association constant kon TnChMg /mm ms dissociation constant koff TnChMg / ms total TnC high concentration Bmax TnChigh 0.4 mm association constant kon CaM 34.0 /mm ms dissociation constant koff CaM / ms total CaM concentration Bmax CaM mm association constant kon myocinca 3.8 /mm ms dissociation constant koff myocinca / ms association constant kon myocinmg /mm ms dissociation constant koff myocinmg / ms total Myocin concentration Bmax myocin 0.4 mm association constant kon SRB 00 /mm ms dissociation constant koff SRB 0.06 / ms total SRB concentration Bmax SRB 0.07 mm 3

24 Junctional space Ca 2+ buffers (adapted from ) association constant kon JSlow 00 /mm ms dissociation constant koff JSlow.3 / ms total JS low concentration Bmax JSlow association constant kon JShigh 00 /mm ms dissociation constant koff JShigh 0.03 / ms total JS high concentration Bmax JShigh.55 0 mm mm 4

25 Subsarcolemmal space Ca 2+ buffers (adapted from ) association constant kon SLlow 00 /mm ms dissociation constant koff SLlow.3 / ms total SL low concentration Bmax SLlow mm association constant kon SLhigh 00 /mm ms dissociation constant koff SLhigh 0.03 / ms total SL high concentration Bmax SLhigh mm Ca 2+ fluxes; J myosl ; Ca 2+ diffusion between myoplasm and subsarcolemmal space permeability P myosl 5000 m 3 /ms J JS_SL ; Ca 2+ diffusion between junctional space and subsalcolemmal space permeability P JS_SL m 3 /ms Table 4. SR Ca 2+ fluxes I SERCA ; SR Ca 2+ pump current (adapted from ) maximum velocity Vmax 0.05 mm/ms 5

26 dissociation constant Km f mm dissociation constant Km r.7 mm I RyR ; Ca 2+ current through RyR channel (adapted from ) 5. ec50 SR 0.25 mm ko Ca 0.0 mm 2 /ms ko m 0.06 /ms ki Ca 0.5 /mm ms ki m /ms ks 00 /ms 2 I leak ; Leak Ca 2+ current through SR (adapted from ) maximum velocity Vmax /ms 6

27 2 Calsequestrin; Ca 2+ buffer in SR (adapted from ) association constant kon csqn 00 /mm ms dissociation constant koff csqn 65.0 / ms total calsequestrin concentration Bmax csqn 0.3 mm Table 5. Mitochondrial Ca 2+ fluxes I CaUni ; Mitochondrial Ca 2+ uniporter current (adapted from 8 ) mit 0.2 i 0.34 amplitude A CaUni 23.2 L/ms A 2 2 exp2 exp 2 I NCXmit ; Mitochondrial Na + -Ca 2+ exchanger current (adapted from 8 ) dissociation constants KdNa mit 38.0 mm KdNa i 32.0 mm KdCa mit mm KdCa i mm amplitude A NCXmit mmol/ms 7

28 Ca 2+ buffer in mitochondria 8.0 exp exp α 2 where = (k2+k4), = (k+k3) A 2 2 total CaBuf mit concentration B maxmit 0.6 mm dissociation constant Kd CaBufmit 0.00 mm Table 6. Nernst potentials ln ln ln ln 8

29 ln Table 7. Ion concentrations Na + concentrations K + concentrations Cl - concentrations 0 Ca 2+ concentrations _ 2 _ 9

30 Table 8. Membrane potential a I Kr (pa/pf) Li et al., 20 0 model Vm (mv) b I to (pa/pf) Lu et al., 20 model Vm (mv) Figure. Channel properties of I Kr and I to. a. Current-voltage relationships of I Kr. Red line is a simulation result by the model and black circles are experimental data of E403-sensitive peak current in HL- cardiomyocytes 7. The membrane potential was held at -50 mv and depolarized by second test pulses, then repolarized back to the holding potential. b. Current-voltage relationships of I to. Red line is a simulation result by the model and black circles are experimental data of I to current in HL- cardiomyocytes 8. The membrane potential was held at -40 mv and depolarized by 0.3 second test pulses, then repolarized back to the holding potential. Data are expressed as current density (pa/pf). 20

31 a Normalized current experiment (Rao et al., 2009) inactivation activation Vm (mv) inactivation activation b Normalized I CaL Vm (mv) model experiment (Rao et al., 2009) model Figure 2. Channel properties of I CaL. a. Voltage dependence of I CaL. Red and blue lines are simulation results of the model for I CaL inactivation and activation, respectively. Black and white circles are experimental data in HL- cardiomyocytes 6 of I CaL inactivation and activation, respectively. b. Current-voltage relationships of I CaL. Green line is a simulation result by the model and black circles are experimental data of I CaL measured as nicardipine-sensitive current in HL- cardiomyocytes 6. The membrane potential was held at -80 mv and depolarized by 0. second test pulses, then repolarized back to the holding potential. The data are normalized to the maximum current. The maximum current density of I CaL was reported to be between -22 and -35 pa/pf, and we set it at -35 pa/pf, according to 3. a.0 b 0 Normalized current I CaT (pa/pf) Vm (mv) experiment (Deng et al., 2009) inactivation activation inactivation activation Vm (mv) model experiment (Deng et al., 2009) model 2

32 Figure 3. Channel properties of I CaT. a. Voltage dependence of I CaT. Red and blue lines are simulation results of the model for I CaT inactivation and activation, respectively. Black and white circles are experimental data in HL- cardiomyocytes of I CaT inactivation and activation, respectively 4. b. Current-voltage relationships of I CaT. Green line is a simulation result by the model and black circles are experimental data of I CaT in HL- cardiomyocytes 4. The membrane potential was held at -80 mv and depolarized by 0.3 second test pulses, then repolarized back to the holding potential. Data are expressed as current density (pa/pf). a c Fractional activation b activation (s) pa 500 ms -40 mv 0 mv -70 mv Vm (mv) -20 mv Figure 4. Channel properties of I ha. a, b. Voltage dependence of I ha fractional activation (a) and activation time constants (b). Red lines are simulation results of the model and black circles are experimental data in HL- cardiomyocytes 2. c. Current-voltage relationships of fully activated I ha in HL- cell model. The membrane potential was hyperpolarized during.5 s to -20 mv from a -40 mv holding potential to fully activate I ha. Tail current was obtained by subsequent depolarization to potentials ranging from -70 to 0 mv (in 5 mv steps) as indicated in the voltage protocol in the lower part of c. The traces well reproduced the experimental data in HL- cells 2. 22

33 Current (pa/pf) Current (pa/pf) Current (pa/pf) Current (pa/pf) Current (pa/pf) Vm (mv) Vm I Na I CaL I CaT I pca I bca I ClCa I bcl Current (pa/pf) Current (pa/pf) Current (pa/pf) Current (pa/pf) Current (pa/pf) Current (pa/pf) I K I Kur I ha I NCX I NaK I Kr I Ks I to I pk 200 ms Figure 5. Configurations of action potential and major ionic currents. 23

34 Ca 2+ i Ca 2+ SR Concentration (nm) Concentration (mm) Ca 2+ JS Ca 2+ mit Concentration (M) Concentration (M) (nm) (ms) Ca 2+ SL Concentration (M) ms Figure 6. Ca 2+ transients in each compartment. Inset in Ca 2+ mit shows very small Ca 2+ mit oscillation, amplitude of which is ~20 nm. Supplementary Data References. Grandi, E. et al. Human atrial action potential and Ca 2+ model: sinus rhythm and chronic atrial fibrillation. Circ. Res. 09, (20). 2. Sartiani, L., Bochet, P., Cerbai, E., Mugelli, A. & Fischmeister, R. Functional expression of the hyperpolarization-activated, non-selective cation current I f in immortalized HL- cardiomyocytes. J. Physiol. 545, 8-92 (2002). 3. Yang, Z., Shen, W., Rottman, J.N., Wikswo, J.P. & Murray, K.T. Rapid stimulation causes electrical remodeling in cultured atrial myocytes. J. Mol. Cell Cardiol. 38, (2005). 4. Deng, C. et al. Pharmacological effects of carvedilol on T-type calcium current in murine HL- 24

35 cells. Eur. J. Pharmacol. 62, 9-25 (2009). 5. Bahrudin, U. et al. Impairment of ubiquitin-proteasome system by E334K cmybpc modifies channel proteins, leading to electrophysiological dysfunction. J. Mol. Biol. 43, (20). 6. Rao, F. et al. Involvement of Src in L-type Ca 2+ channel depression induced by macrophage migration inhibitory factor in atrial myocytes. J. Mol. Cell Cardiol. 47, (2009). 7. Li, P. et al. Reciprocal control of herg stability by Hsp70 and Hsc70 with implication for restoration of LQT2 mutant stability. Circ. Res. 08, (200). 8. Lu, Y.Y. et al. Extracellular matrix of collagen modulates intracellular calcium handling and electrophysiological characteristics of HL- cardiomyocytes with activation of angiotensin II type receptor. J. Card. Fail. 7, (20). 9. Yang, Z. & Murray, K.T. Ionic mechanisms of pacemaker activity in spontaneously contracting atrial HL- cells. J. Cardiovasc. Pharmacol. 57, (20). 0. Sarai, N., Matsuoka, S. & Noma, A. simbio: a Java package for the development of detailed cell models. Prog. Biophys. Mol. Biol. 90, (2006).. Montero, M. et al. Chromaffin-cell stimulation triggers fast millimolar mitochondrial Ca 2+ transients that modulate secretion. Nat. Cell Biol. 2, 57-6(2000). 2. Filippin, L., Magalhães, P.J., Di Benedetto, G., Colella, M. & Pozzan, T. Stable interactions between mitochondria and endoplasmic reticulum allow rapid accumulation of calcium in a subpopulation of mitochondria. J. Biol. Chem. 278, (2003). 3. Pinton, P., Leo, S., Wieckowski, M.R., Di Benedetto, G. & Rizzuto R. Long-term modulation of mitochondrial Ca 2+ signals by protein kinase C isozymes. J. Cell Biol. 65, (2004). 4. Lu, X. et al. Measuring local gradients of intramitochondrial [Ca 2+ ] in cardiac myocytes during sarcoplasmic reticulum Ca 2+ release. Circ. Res. 2, (203). 5. Bossen, E.H., Sommer, J.R. & Waugh, R.A. Comparative stereology of mouse atria. Tissue Cell 3, 7-77 (98). 6. Nygren, A. et al. Mathematical model of an adult human atrial cell: the role of K + currents in repolarization. Circ. Res. 82, 63-8 (998). 7. Kurata, Y., Hisatome, I., Imanishi, S. & Shibamoto, T. Dynamical description of sinoatrial node pacemaking: improved mathematical model for primary pacemaker cell. Am. J. Physiol. Heart Circ. Physiol. 283, H2074-H20 (2002). 8. Kim, B., Takeuchi, A., Koga, O., Hikida, M. & Matsuoka, S. Pivotal role of mitochondrial Na+ Ca²+ exchange in antigen receptor mediated Ca 2+ signalling in DT40 and A20 B lymphocytes. J. Physiol. 590, (202). 25