Mechanics of Relaxation of the Human Heart

Denis Chemla, Catherine Coirault, Jean-Louis Hbert, and Yves Lecarpentier

Rapid and complete relaxation is a prerequisite for cardiac output adaptation to changes in loading conditions, inotropic stimulation, and heart rate. In the healthy human heart, the rate and extent of relaxation depend mainly on actomyosin cross bridge dissociation and on left ventricular end-systolic volume, rather than on the afterload level.

elaxation is the process by which heart muscle actively returns, after contraction, to its initial conditions of load and length. Over the past 25 years, physiologists, clinicians, and pharmacologists have shown increasing interest in trying to understand the regulation of myocardial relaxation. From a physiological point of view, rapid and complete relaxation is a prerequisite for cardiac output adaptation to changes in loading conditions, inotropic stimulation, and heart rate. From a clinical point of view, the relaxation phase could be impaired earlier and more profoundly than the contraction phase in numerous cardiac diseases. Thus clinicians have speculated that the diagnosis of isolated relaxation abnormalities may help with the early identication of a subgroup of patients who will subsequently develop systolic abnormalities. Finally, therapeutic interventions aimed at improving myocardial relaxation may be useful in the management of cardiac patients. It is widely accepted that myocardial relaxation depends essentially on the inactivating processes within the myocyte and on loading conditions (1). Indeed, relaxation reects the imbalance between the number of strongly bound actomyosin cross bridges and the total load imposed on the ventricle (afterload + preload). Afterload is determined by extracardiac and cardiac loading factors (Fig. 1), which potentially contribute to the so-called load dependence of relaxation (1). Recent studies have demonstrated that left ventricular (LV) end-systolic volume (ESV) also plays a key role in regulating cardiac relaxation. The present paper briey discusses the inuences of inactivation, load, and ESV on heart relaxation.

Physiological context
Between aortic valve closure and mitral valve opening, a major drop in LV pressure occurs while LV volume is minimal (ESV). The rapid decrease in LV stress boosts coronary diastolic lling, especially that of the left coronary artery. Thus LV relaxation appears to contribute to adequate coronary perfusion. After mitral valve opening, rapid myocardial lengthening and early diastolic LV lling occur at low left atrial pressure, such that pulmonary circulation is protected. The LV lls under the action of the instantaneous left atrial-to-LV pressure gradient. In the normal heart, in which left atrial pressure is
D. Chemla, C. Coirault, J.-L. Hbert, and Y. Lecarpentier are in the Service de Physiologie Cardio-Respiratoire, CHU de Bictre, Assistance PubliqueHpitaux de Paris, 94 275 Le Kremlin-Bictre, and Unit Inserm U451-LoaEnsta-Ecole Polytechnique, 91 125 Palaiseau, France.
78 News Physiol. Sci. Volume 15 April 2000

low, the rate of change and magnitude of this gradient are inuenced mainly by the rate and extent of LV pressure fall. The ability of the LV to relax both rapidly and completely is of paramount importance for optimal LV lling during the early diastolic period (6, 14). During exercise, LV stroke volume increases as a result of both increased contractility (via sympathetic stimulation) and increased end-diastolic LV volume/preload (via Frank-Starling mechanism). This must be paralleled by an enhanced LV lling volume. During exercise, both the enhanced LV lling volume and the shortened diastolic interval resulting from the increase in heart rate contribute to the marked increase in LV lling rate in early diastole. Given that there are only moderate changes in left atrial pressure during exercise, appropriate left atrial-to-LV pressure gradient is essentially dependent on the ability of the LV to enhance the speed of relaxation and create low or even negative minimum diastolic LV pressure during exercise (6, 14). In striated muscles, there is a difference in relaxation kinetics between skeletal and cardiac muscles. In most skeletal muscle, relaxation occurs according to an isotonic-isometric sequence, in which muscle lengthening precedes tension fall. Relaxation of the heart is auxotonic, i.e., it involves simultaneous changes in ventricular pressure and length/volume. However, it can be said that the heart relaxes according to a reverse isometric-isotonic sequence, in which isovolumic pressure drop occurs first, followed by isotonic relaxation (i.e., myocardial lengthening). If isotonic relaxation occurred rst, as observed in skeletal muscles, LV lling pressure would have to be of similar magnitude to systemic pressure. Therefore, under normal conditions, the specic sequence of relaxation of the heart appears to be benecial because it minimizes lling pressures, thus protecting pulmonary circulation. Because the initial pressure drop occurs at xed LV volume, essentially no work is performed by the isometrically relaxing LV. Furthermore, given that myocardial lengthening occurs at low pressure, the work performed during the early lling phase is also minimized. Overall, the specic sequence of relaxation of the heart could be especially benecial from a thermoenergetic point of view.

Inactivation
Inactivation is dened as the intracellular processes leading to dissociation of actomyosin cross bridges and to the lowering of intracellular Ca2+ concentration from 105 M to
0886-1714/99 5.00 2000 Int. Union Physiol. Sci./Am.Physiol. Soc.

FIGURE 1. Factors that determine left ventricular (LV) afterload at any time. LV afterload is determined by factors intrinsic to LV chamber (anatomical factors, intrinsic LV load) and by factors extrinsic to LV chamber (arterial load, other factors). All these factors may potentially be involved in load dependence of relaxation. RV, right ventricle.

107 M. The term lusitropy is often used in place of inactivation. The rate of relaxation is determined mainly by active Ca2+ pumping by the sarcoplasmic reticulum Ca2+ ATPase. Phosphorylation of phospholamban, a membrane-bound protein, removes its inhibitory effect on sarcoplasmic Ca2+ ATPase, thereby accelerating Ca2+ uptake and relaxation rate, especially under isotonic conditions. The rate of relaxation is also limited by 1) the afnity of troponin C (TnC) for Ca2+, especially under isometric conditions; 2) Ca2+ extrusion outside the cell, mainly via Na+/Ca2+ exchange; and 3) the number and kinetics of working cross bridges (Table 1). It is likely that the majority of working cross bridges detach during isovolumic relaxation. The detachment of cross bridges depends on ADP dissociation from the cross bridge and on ATP binding. After the release of inorganic phosphate and ADP, the actomyosin complex has a high afnity for ATP; ATP binding to myosin decreases the afnity of myosin for actin, thus leading to cross bridge detachment. Finally, myosin ATPase activity determines the cross bridge cycling rate and thus inuences relaxation. Troponin-tropomyosin interactions, cross bridge kinetics, and amplication of activation (cooperativity) can be modied by 1) mechanical changes in sarcomere length and, to a minor degree, in the strain put on cross bridges; 2) changes in neurohormonal state (e.g., cAMP, angiotensin II); and 3) cardiac endothelium-derived factors.

Inuence of load on relaxation

The physiologist has no direct estimate of all the inactivation processes within myocytes. Furthermore, an integrated approach should improve our understanding of how myocardial relaxation adapts to acute or chronic changes in venous return and arterial impedance. Thus clinically rele-

vant insights into the regulatory processes of relaxation have generally concerned the link between relaxation mechanics and loading conditions. The sensitivity of the timing of relaxation to the load imposed before the onset of contraction (systolic load) is a manifestation of the shortening-induced deactivation phenomenon: a twitch contracting against light or medium load ends earlier than the fully isometric twitch (1, 2, 8, 9). Thus the more the muscle is allowed to shorten the shorter the overall duration of the contraction-relaxation cycle. A clinical illustration of this load dependence is that increased systolic load tends to delay relaxation (1). Animal studies have shown that the rate of tension/pressure fall is hardly affected by preload changes. The isovolumic relaxation rate mainly depends on LV end-systolic pressure (ESP) and/or ESV (5). Consistent results have been reported in animals, with increased afterload having a slowing effect on the rate of pressure fall (i.e., there is an increase in the time constant tau of the monoexponential pressure fall) (1, 5, 6, 10). This load dependence of the isovolumic relaxation rate has been shown to be modulated by the inotropic state and the homogeneity of contraction, as well as the timing of imposition of peak systolic pressure (1, 5, 6, 8, 10). Recently, more complex effects of load on relaxation rate (accelerating or slowing effects) have been reported in dogs (7). The results obtained in the human heart are at variance with those obtained in animal studies. In humans, the rate of LV isovolumic relaxation appears to be essentially independent of loading conditions in healthy subjects, provided that changes in afterload are moderate. Conversely, the relaxation rate becomes gradually more load dependent as systolic dysfunction progresses (4). The unexpected nding regarding the load independence of the isovolumic relaxation rate in healthy humans has yet to be explained (4, 7, 11).
News Physiol. Sci. Volume 15 April 2000 79

Table 1.

Intracellular processes limiting the rate of relaxation Afnity of troponin C for Ca2+ Ca2+ uptake by the Ca2+ ATPase of the sarcoplasmic reticulum Na+ / Ca2+ exchange Sarcolemmal Ca2+ ATPase Ca2+ buffers

Rate of Ca2+ dissociation from troponin C

Troponin-tropomyosin interaction with actin Crossbridge properties Cross bridge number, and proportion of strongly bound cross bridges Activation dependence Length dependence Load dependence Cross bridge detachment [ADP]i and [ATP]i dependence Activation dependence (cycling rate, myosin ATPase activity) Elasticity Viscous-like friction

Resynthesis of ATP Ca2+ afnity and sensitivity of myolaments are modulated by several factors, including sarcomere length, troponin I, C protein, myosin LC2, Ca2+, pH, inorganic phosphate, Mg2+, temperature, ionic strength, and neurohormonal and endothelium-derived factors. Ca2+-pumping ATPase of the sarcoplasmic reticulum requires phosphorylation of membrane protein phospholamban. Such phosphorylation is acheived either by cAMP-dependent pathway (-adrenergic stimulation) or by Ca2+ -calmodulin pathway. Na+ / Ca2+ exchanger is energetically linked to Na+/K+ ATPase. Energetic changes for phosphocreatine hydrolysis have a key role in regulating intracellular ADP and ATP concentrations.

The effects of load on the relaxation rate may be intrinsically linked to the afterload level (via increased strain put on cross bridges and cooperativity, changes in homogeneity, coronary circulation, or neurohormonal stimulation) or afterload timing. The effects of load may also be explained by changes in ESV, thus modifying intrinsic load and lengthdependent inactivation.

Rate and extent of relaxation: the role of ESV

The ability to reduce ESV contributes to the active restoration of LV dimension in early diastole. Reduced ESV enhances the rate and extent of the isovolumic pressure fall (3, 13, 14) and enhances the rate of myocardial active lengthening (15). In both cases, this accelerating effect has been observed irrespective of the loading conditions (3, 1315). Two mechanisms explain why the rate and extent of relaxation are so tightly linked to ESV. First, the amount of potential energy stored during contraction and released during relaxation is negatively related to the end-systolic length/volume; second, the decay of mechanical activity could be accelerated at short end-systolic length (length-dependent inactivation) (Table 2). Restoring forces are generated when the LV contracts below its equilibrium volume (Veq). The magnitude of restoring forces is inversely related to ESV. This suction effect contributes to LV lling at low physiological lling pressure and when higher lling rates are required (e.g., exercise). There is also an elastic recoil of external forces surrounding the myocytes, mainly the extracellular elastic components
80 News Physiol. Sci. Volume 15 April 2000

deformed and compressed during ventricular contraction (transmural and three-dimensional deformations). The (sarcomere) length dependence of activation has been related to length-dependent changes in myofilament overlap and interfilament lattice spacing, Ca2+ release by the sarcoplasmic reticulum, and the affinity of TnC for Ca2+. If these length-induced changes still operate at end systole, hastening of relaxation would occur at small ESV. Furthermore, shortening may be associated with changes in myofilament-bound Ca2+ and with a decreased likelihood of new actomyosin interactions.

Lusitropic reserve
A physiological approach to cardiac relaxation should also help improve our understanding of how relaxation adapts to changes in inotropy and heart rate. When contractility is increased, ESV decreases, helping to accelerate relaxation. Furthermore, increased contractility increases restoring forces for a given ESV below Veq and also increases the peak lengthening rate at any end-systolic length (3, 1315). Thus relaxation can be intrinsically promoted (i.e., independently of any effect on systolic shortening) thanks to the amount of lusitropy in reserve. -Adrenergic stimulation and increased heart rate are two major ways by which lusitropic reserve can be mobilized. Increased cAMP leads to increased cross bridge cycling and activates protein kinase A-induced phosphorylation of both phospholamban and troponin I (TnI). End-systolic

length modulates the extent to which -adrenergic stimulation exerts its positive lusitropic effects (12). Tachycardia induces a positive lusitropic effect by speeding up the activation/inactivation processes. Enhanced inotropy (the positive staircase phenomenon), increased homogeneity, and/or the associated cAMP increases may also be involved. It is widely acknowledged that, in addition to inactivation (lusitropy) and load, myocardial relaxation also depends on a third regulatory process: the homogeneity of the contractionrelaxation cycle (1). Indeed, the lusitropic reserve may be mobilized by interventions increasing the homogeneity of the contraction-relaxation cycle via different mechanisms that increase the speed of conduction, the extracellular temperature, the heart rate, and the inotropic state. Finally, it has been suggested that internal load would be activation/inactivation dependent.

ESV, afterload, and lusitropy

We have discussed the respective inuences of LV pressure and volume on myocardial relaxation. It is important to remember when studying the hemodynamic correlates of relaxation rate that LVESP and ESV are not interchangeable variables. Thus the question arises as to whether relaxation is inuenced rather by LVESP (load dependence) or by ESV (length dependence). Although it is likely that both ESV and systolic load play a part in regulating the rate and extent of isovolumic relaxation, we suggest that ESV has a prominent role. Our proposition is based on the following considerations: 1) the intrinsic properties of the myolaments and the systolic storage of potential energy are highly dependent on myocardial length and only to a minor degree on load; 2) the effects of end-systolic length/vol-

ume on relaxation rate and extent have been shown to be consistent in animals and humans (3, 1315), whereas the effects of load are almost negligible in healthy humans (4) and still need to be claried in animal studies (1, 5, 7, 8); 3) load dependence of relaxation can be considered a manifestation of the deactivating effects induced by shortening (1); and 4) pressure fall and onset of lengthening both occur at end-systolic length. A simplied framework of afterload-independent/lengthdependent behavior may help improve our understanding of the regulation of the contraction-relaxation cycle in the healthy human heart (Fig. 2). Contraction and relaxation performances may be determined by LV volume at the onset of each phase and by inotropy/lusitropy, thus explaining the afterload independence of the contraction-relaxation cycle. This approach would integrate the regulatory role of myocardial length, the contractile properties of the myocyte, and the elastic properties of the LV chamber (intrinsic load).

Integrated function
Modulation of cardiac function is mediated mainly through changes in preload (Frank-Starling mechanism), neurohormonal stimulation, and heart rate, such that cardiac output can meet organ demand. Integrated function involves coordinated changes in contraction and relaxation, and further studies are needed to describe the different aspects of contraction-relaxation coupling (biochemical, genetic, mechanical, thermoenergetic, and electrical aspects). Right ventricular (RV) lling is facilitated by elastic recoil of the LV and by elastic recoil of the RV myocardium, leading to a piston-like motion of the tricuspid annulus. Elastic recoil

Table 2.

Main factors related to left ventricular end-systolic volume and factors that modulate rate and extent of relaxation Elasting recoil Ventricular restoring forces (ESV < Veq) Internal restoring forces (when present) Twisting/untwisting LVESP Stretching of great vessels and connective tissue Coronary compression Right ventricular load Myolament overlap Lattice spacing Instantaneous length and sliding velocity of the myolaments Troponin-tropomyosin interactions Amplication of activation (cooperativity) Afnity of troponin C for Ca2+ Modulation of sarcoplasmic reticulum function

Stored potential energy (intrinsic load)

Extrinsic load

Sarcomere length at end systole

Ventricular restoring forces are generated when LV contracts below its equilibrium volume (Veq), with Veq = volume when transmural pressure is 0 in fully relaxed state. At cellular level, internal restoring forces, when present, involve resistance to bending and double overlapping of thin laments at sarcomere length below slack length. LV end-systolic pressure (LVESP) is a linear function of end-systolic volume (ESV), slope of which is thought to reect cardiac inotropic state.
News Physiol. Sci. Volume 15 April 2000 81

FIGURE 2. A simplied, afterload-independent/length-dependent framework for regulation of contraction-relaxation cycle in healthy human heart. For a given heart rate, rate and extent of relaxation depend mainly on end-systolic volume (ESV) and inactivation (lusitropy). End-diastolic volume (EDV) and ESV are related through 1) regulatory processes whereby a constant stroke volume (SV) is maintained (ESV = EDV SV) and 2) inuence of venous return (EDV = ESV + venous return). When afterload increases, systolic function is preserved by increased preload and/or increased inotropy. Relaxation rate and extent may be maintained by preserved ESV and/or increased lusitropy. Preserved ESV could be explained by natural steepness of LV end-systolic pressure-ESV relationship and/or by homeometric regulation (increased inotropy thus preserving stroke volume). This latter mechanism may be combined with increased lusitropy, thus counteracting potential slowing effects of increased load. -Adrenergic stimulation also increases lusitropy when afterload is chronically increased. On the basis of this framework, length-dependent regulatory processes predominate over load-dependent processes in the contraction-relaxation cycle of healthy myocardium.

could be especially important for pump function, given the extremely low RV lling pressure. Investigation of the interplay between relaxation and myocardial compliance did not fall within the scope of the present paper. Relaxation could be impaired earlier and more profoundly than contraction in numerous cardiac diseases (6, 10). Slowed and incomplete relaxation has a negative effect on the lling function of the heart, especially in cases in which chamber compliance is decreased, diastolic duration is shortened, or the limit of preload reserve is reached.

This study was supported by grants from PHRC AOM96174, Assistance Publique-Hpitaux de Paris.

References
1. Brutsaert, D. L., and S. U. Sys. Relaxation and diastole of the heart. Physiol. Rev. 69: 12281315, 1989. 2. Chemla, D., Y. Lecarpentier, J. L. Martin, M. Clergue, A. Antonetti, and P. Y. Hatt. Relationship between inotropy and relaxation. Am. J. Physiol. Heart Circ. Physiol. 250: H1008H1016, 1986. 3. Courtois, M., C. J. Mechem, B. Barzilai, and P. A. Ludbrook. Factors related to end-systolic volume are important determinants of peak early diastolic transmitral ow velocity. Circulation 85: 11321138, 1992. 4. Eichorn, E. J., J. E. Willard, L. Alvarez, A. S. Kim, D. B. Glamann, R. C. Risser, and P. A. Grynburn. Are contraction and relaxation coupled in patients with and without congestive heart failure? Circulation 85: 21322139, 1992. 5. Gaasch, W. H., A. S. Blaustein, C. W. Andrias, R. P. Donahue, and B. Avitall. Myocardial relaxation. II. Hemodynamic determinants of rate of left ventricular isovolumic pressure decline. Am. J. Physiol. Heart Circ. Physiol. 239: H1H6, 1980. 6. Gaasch, W. H., and M. M. LeWinter. Left Ventricular Diastolic Dysfunction and Heart Failure. Philadelphia, PA: Lea and Febiger, 1994. 7. Gillebert, T. C., A. F. Leite-Moreira, and S. G. de Hert. Relaxation-systolic pressure relation. A load-independent assessment of left ventricular contractility. Circulation 95: 745752, 1997.

Conclusions
For a given heart rate, systolic function of the healthy human heart depends on end-diastolic volume (preload) and on activation (inotropy). It has long been accepted that myocardial relaxation depends essentially on afterload and inactivation (lusitropy). Recent studies support an alternative framework, in which relaxation depends mainly on ESV and inactivation. This approach could explain the load independence of the relaxation rate in the healthy human heart. This framework may help improve our understanding of the pathophysiological aspects of human heart relaxation.
82 News Physiol. Sci. Volume 15 April 2000

8. Gillebert, T. C., S. U. Sys, and D. L. Brutsaert. Inuence of loading patterns on peak length-tension relation and on relaxation in cardiac muscle. J. Am. Coll. Cardiol. 13: 483490, 1989. 9. Lecarpentier, Y., A. Waldenstrm, M. Clergue, D. Chemla, P. Oliviero, J. L. Martin, and B. Swynghedauw. Major alterations in relaxation during cardiac hypertrophy induced by aortic stenosis in guinea pig. Circ. Res. 61: 107116, 1987. 10. Lorell, B. H., and W. Grossman. Diastolic Relaxation of the Heart (2nd ed.). Boston, MA: Kluwer Academic, 1994. 11. Nitenberg, A., Y. Lecarpentier, I. Antony, and D. Chemla. Dipyridamole slows the rate of isovolumic pressure fall in patients with normal coronary arteries. Eur. Heart J. 16: 17211725, 1995.

12. Suard, I., N. Pry-Man, C. Coirault, J. C. Pourny, Y. Lecarpentier, and D. Chemla. Relaxant effects of isoproterenol in isolated cardiac muscle: inuence of loading patterns. Am. J. Physiol. Heart Circ. Physiol. 267: H1814H1823, 1994. 13. Udelson, J. E., S. L. Bacharach, R. O. Cannon III, and R. O. Bonow. Minimum left ventricular pressure during -adrenergic stimulation in human subjects. Evidence for elastic recoil and diastolic suction in the normal heart. Circulation 82: 11741182, 1990. 14. Yellin, E. L., S. Nikolic, and W. M. Frater. Left ventricular lling dynamics and diastolic function. Progr. Cardiovasc. Dis. 32: 247271, 1990. 15. Zile, M. R., and W. H. Gaasch. Mechanical loads and the isovolumic and lling indices of left ventricular relaxation. Progr. Cardiovasc. Dis. 32: 333346, 1990.

The Time Course of Signaling at Central Glutamatergic Synapses

Peter Jonas
Glutamate is the main excitatory transmitter in the mammalian CNS, mediating fast synaptic transmission primarily by activation of AMPA-type glutamate receptor channels. Both synaptic structure and a cell-specic molecular switch in the AMPA receptor subunit expression are involved in the regulation of the synaptic signaling time course.

hemical synaptic transmission involves the release of a transmitter into the synaptic cleft and the subsequent activation of postsynaptic receptors. The most extensively characterized synapse is the neuromuscular junction, formed between motor axons and skeletal muscle bers. The transmitter acetylcholine (ACh) is released from synaptic vesicles in multimolecular packets, or quanta, and activates nicotinic ACh receptors (AChRs) in the postsynaptic membrane. This generates postsynaptic conductance changes of fast time course and large amplitude, which induce reliable activation of the muscle ber. In the mammalian central nervous system (CNS), glutamate is the main excitatory transmitter. The principles of synaptic communication at glutamatergic synapses, however, appear to be more complex than at the neuromuscular junction. First, glutamatergic synapses differ in morphological properties, such as number of release sites and presence of dendritic spines. Second, synapses in different circuitries differ substantially in impact and time course of synaptic signaling. Finally, glutamate activates several different types of ionotropic and metabotropic receptors: -amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptors (AMPARs), N-methyl-D-aspartate receptors (NMDARs), kainate receptors (KARs), and metabotropic glutamate receptors coupled to either inositol 1,4,5-trisphosphate (IP3; class 1 mGluRs) or cAMP-signaling cascades (class 2 and 3 mGluRs) (10). On the basis of extensive work in the last ten years, we are beginning to understand how cellular and molecular factors shape functional differences between glutamatergic synapses.

The time course of the excitatory postsynaptic currents

The patch-clamp technique in brain slices developed by Bert Sakmann and co-workers allows us to record from neurons in the CNS with high resolution in an intact synaptic environment. Analysis of several synapses revealed that fast glutamate-mediated signaling is predominantly mediated by AMPARs (5, 9, 15). However, the time course of signaling differs substantially between synapses. This is illustrated in Fig. 1, which shows AMPAR-mediated evoked excitatory postsynaptic currents (EPSCs) at three well-characterized synapses: the mossy ber-CA3 pyramidal cell synapse in the hippocampus (9), the granule cell-basket cell synapse in the dentate gyrus (5), and the calyx synapse on neurons in the medial nucleus of the trapezoid body (MNTB) in the brain stem (1). Throughout the CNS, slow AMPAR-mediated evoked EPSCs are generated at excitatory synapses on hippocampal and neocortical principal neurons (decay time constant ~5 ms at 22C). In contrast, fast AMPAR-mediated evoked EPSCs are generated at glutamatergic synapses on hippocampal and neocortical inhibitory interneurons and on neurons in the auditory pathway (decay time constant ~12 ms).

Synchrony vs. asynchrony of quantal release

Which factors determine the time course of the AMPARmediated postsynaptic conductance change? First, asynchrony of quantal release needs to be considered (Fig. 2A; Refs. 3, 5, and 8). Evoked release of transmitter involves several steps: invasion of the action potential into the presynaptic terminal, activation of voltage-gated Ca2+ channels, Ca2+ entry into the
News Physiol. Sci. Volume 15 April 2000 83

P. Jonas is in the Physiologisches Institut der Universitt Freiburg, D-79104 Freiburg, Germany.
0886-1714/99 5.00 2000 Int. Union Physiol. Sci./Am.Physiol. Soc.