The mitochondrial Ca2+-activated K+ channel activator, NS 1619 inhibits L-type Ca2+ channels in rat ventricular myocytes

We examined the effects of the mitochondrial Ca(2+)-activated K(+) (mitoBK(Ca)) channel activator NS 1619 on L-type Ca(2+) channels in rat ventricular myocytes. NS 1619 inhibited the Ca(2+) current in a dose-dependent manner. NS 1619 shifted the activation curve to more positive potentials, but did not have a significant effect on the inactivation curve. Pretreatment with inhibitors of membrane BK(Ca) channel, mitoBK(Ca) channel, protein kinase C, protein kinase A, and protein kinase G had little effect on the Ca(2+) current and did not alter the inhibitory effect of NS 1619 significantly. The application of additional NS 1619 in the presence of isoproterenol, a selective beta-adrenoreceptor agonist, reduced the Ca(2+) current to approximately the same level as a single application of NS 1619. In conclusion, our results suggest that NS 1619 inhibits the Ca(2+) current independent of the mitoBK(Ca) channel and protein kinases. Since NS 1619 is widely used to study mitoBK(Ca) channel function, it is essential to verify these unexpected effects of NS 1619 before experimental data can be interpreted accurately.     

Voltage and Ca2+ activation of single large-conductance Ca2+-activated K+ channels described by a two-tiered allosteric gating mechanism

The voltage- and Ca2+-dependent gating mechanism of large-conductance Ca2+-activated K+ (BK) channels from cultured rat skeletal muscle was studied using single-channel analysis. Channel open probability (Po) increased with depolarization, as determined by limiting slope measurements (11 mV per e-fold change in Po; effective gating charge, q(eff), of 2.3 +/- 0.6 e(o)). Estimates of q(eff) were little changed for intracellular Ca2+ (Ca2+(i)) ranging from 0.0003 to 1,024 microM. Increasing Ca2+(i) from 0.03 to 1,024 microM shifted the voltage for half maximal activation (V(1/2)) 175 mV in the hyperpolarizing direction. V(1/2) was independent of Ca2+(i) for Ca2+(i) < or = 0.03 microM, indicating that the channel can be activated in the absence of Ca2+(i). Open and closed dwell-time distributions for data obtained at different Ca2+(i) and voltage, but at the same Po, were different, indicating that the major action of voltage is not through concentrating Ca2+ at the binding sites. The voltage dependence of Po arose from a decrease in the mean closing rate with depolarization (q(eff) = -0.5 e(o)) and an increase in the mean opening rate (q(eff) = 1.8 e(o)), consistent with voltage-dependent steps in both the activation and deactivation pathways. A 50-state two-tiered model with separate voltage- and Ca2+-dependent steps was consistent with the major features of the voltage and Ca2+ dependence of the single-channel kinetics over wide ranges of Ca2+(i) (approximately 0 through 1,024 microM), voltage (+80 to -80 mV), and Po (10(-4) to 0.96). In the model, the voltage dependence of the gating arises mainly from voltage-dependent transitions between closed (C-C) and open (O-O) states, with less voltage dependence for transitions between open and closed states (C-O), and with no voltage dependence for Ca2+-binding and unbinding. The two-tiered model can serve as a working hypothesis for the Ca2+- and voltage-dependent gating of the BK channel.     

Polyamines block Ca2+-activated K+ channels in pituitary tumor cells (GH3)

The effects of the natural polyamines, putrescine, spermidine and spermine on single calcium-activated potassium channels from clonal rat pituitary tumor cells (GH3) were studied. Applied to inside-out patches, polyamines were found to reduce the current amplitude and open probability of the channels in a dose- and voltage-dependent manner, indicating that polyamines act as fast blockers which sense a fraction of the electrical field in the channel pore. The K _d for spermine was 11.2 mm for the reduction of unitary current amplitude and 0.7 mm for the reduction of the open probability. The order of effectiveness was spermine > spermidine > putrescine. From fitting -functions to current amplitude histograms, blocking and unblocking rates were determined as 11.4 × 10⁴ sec^–1 and 21.9 × 10⁴ sec^–1, respectively. The reduction of the channel open probability was relieved by an increase of the Ca²⁺ concentration of the internal solution, indicating that polyamines compete with Ca²⁺ at the Ca²⁺ sensor of the channel. Putrescine antagonized the effect of spermine on the channel current amplitude. The results suggest that polyamines at intracellular millimolar concentrations suppress ion channel activity and therefore may effect electrical discharge behavior of excitable cells.This work was supported in part by the Austrian Fonds zur Förderung der wissenschaftlichen Forschung, P8587.     

Glucocorticoids stimulate the activity of large-conductance Ca2+ -activated K+ channels in pituitary GH3 and AtT-20 cells via a non-genomic mechanism

The effects of glucocorticoids on ion currents were investigated in pituitary GH3 and AtT-20 cells. In whole-cell configuration, dexamethasone, a synthetic glucocorticoid, reversibly increased the density of Ca2+ -activated K+ current (IK(Ca)) with an EC50 value of 21 +/- 5 microM. Dexamethasone-induced increase in IK(Ca) density was suppressed by paxilline (1 microM), yet not by glibenclamide (10 microM), pandinotoxin-Kalpha (1 microM) or mifepristone (10 microM). Paxilline is a blocker of large-conductance Ca2+ -activated K+ (BKCa) channels, while glibenclamide and pandinotoxin-Kalpha are blockers of ATP-sensitive and A-type K+ channels, respectively. Mifepristone can block cytosolic glucocorticoid receptors. In inside-out configuration, the application of dexamethasone (30 microM) into the intracellular surface caused no change in single-channel conductance; however, it did increase BKCa -channel activity. Its effect was associated with a negative shift of the activation curve. However, no Ca2+ -sensitiviy of these channels was altered by dexamethasone. Dexamethasone-stimulated channel activity involves an increase in mean open time and a decrease in mean closed time. Under current-clamp configuration, dexamethasone decreased the firing frequency of action potentials. In pituitary AtT-20 cells, dexamethasone (30 microM) also increased BKCa -channel activity. Dexamethasone-mediated stimulation of IK(Ca) presented here that is likely pharmacological, seems to be not linked to a genomic mechanism. The non-genomic, channel-stimulating properties of dexamethasone may partly contribute to the underlying mechanisms by which glucocorticoids affect neuroendocrine function.     

Rabbit distal colon epithelium: III. Ca2+-activated K+ channels in basolateral plasma membrane vesicles of surface and crypt cells

Summary In the mammalian distal colon, the surface epithelium is responsible for electrolyte absorption, while the crypts are the site of secretion. This study examines the properties of electrical potential-driven⁸⁶Rb⁺ fluxes through K⁺ channels in basolateral membrane vesicles of surface and crypt cells of the rabbit distal colon epithelium. We show that Ba²⁺-sensitive, Ca²⁺-activated K⁺ channels are present in both surface and crypt cell derived vesicles with half-maximal activation at 5×10^–7 m free Ca²⁺. This suggests an important role of cytoplasmic Ca²⁺ in the regulation of the bidirectional ion fluxes in the colon epithelium.The properties of K⁺ channels in the surface cell membrane fraction differ from those of the channels in the crypt cell derived membranes. The peptide toxin apamin inhibits Ca²⁺-activated K⁺ channels exclusively in surface cell vesicles, while charybdotoxin inhibits predominantely in the crypt cell membrane fraction. Titrations with H⁺ and tetraethylammonium show that both high-and low-sensitive⁸⁶Rb⁺ flux components are present in surface cell vesicles, while the high-sensitive component is absent in the crypt cell membrane fraction. The Ba²⁺-sensitive, Ca²⁺-activated K⁺ channels can be solubilized in CHAPS and reconstituted into phospholipid vesicles. This is an essential step for further characterization of channel properties and for identification of the channel proteins in purification procedures.     

Properties of the apamin-sensitive component of Ca2+-dependent K+ current in smooth muscle cells of the guinea pigtaenia coli

With the help of a standard voltage-clamp technique, we investigated transmembrane ion currents in isolated smooth muscle cells of the guinea pigtaenia coli. In Ca²⁺-dependent K⁺ current, we identified and studied the properties of an apamin-sensitive voltage-independent component carried through the channels of low conductance (in many publications called small conductance,I _SK(Ca)). This component did not show the temporal inactivation;I _SK(Ca) was insensitive to the action of 4 mM tetraethylammonium, but was completely blocked by 500 nM of apamin. It was shown thatI _SK(Ca) is very sensitive to changes in the intracellular Ca²⁺ concentration ([Ca²⁺] _i ): a decrease in [Ca²⁺] _i up to 50 nM resulted in the almost complete blockade of the current. The entry of Ca ions into a cell from the external solution through the voltage-operated Ca²⁺ channels of L-type was not an obligatory condition for activation ofI _SK(Ca). The current-voltage relationship forI _SK(Ca) had a maximum within the voltage range of +20 to +50 mV. Neirofiziologiya/Neurophysiology, Vol. 32, No. 2, pp. 87–94, March–April, 2000.     

Pharmacological gating modulation of small- and intermediate-conductance Ca2+-activated K+ channels (KCa2.x and KCa3.1)

This short review discusses pharmacological modulation of the opening/closing properties (gating) of small- and intermediate-conductance Ca²⁺-activated K⁺ channels (K_Ca2 and K_Ca3.1) with special focus on mechanisms-of-action, selectivity, binding sites, and therapeutic potentials. Despite K_Ca channel gating-modulation being a relatively novel field in drug discovery, efforts in this area have already revealed a surprising plethora of pharmacological sites-of-actions and channel subtype selectivity exerted by different chemical classes. The currently published positive modulators show that such molecules are potentially useful for the treatment of various neurodegenerative disorders such as ataxia, alcohol dependence, and epilepsy as well as hypertension. The negative K_Ca2 modulators are very effective agents for atrial fibrillation. The prediction is that further unraveling of the molecular details of gating pharmacology will allow for the design of even more potent and subtype selective K_Ca modulators entering into drug development for these indications.     

Ca2+-activated K+ permeability in human erythrocytes: Modulation of single-channel events

Elevated levels of intracellular Ca²⁺ activate a K⁺-selective permeability in the membrane of human erythrocytes. Currents through single channels were analysed in excised inside-out membrane patches. The effects of several ions that are known to inhibit K⁺ fluxes are described with respect to the single-channel events. The results suggest that the blocking ions can partly move into the channels (but cannot penetrate) and interact with other ions inside the pore. The reduction of single-channel conductance by Cs⁺, tetraethylammonium and Ba²⁺ and of single-channel activity by quinine and Ba²⁺ is referred to different rates of access to the channel. The concentration- and voltage-dependent inhibition by ions with measurable permeability (Na⁺ and Rb⁺) can be explained by their lower permeability, with single-file movement and ionic interactions inside the pore.     

The beta subunit increases the Ca2+ sensitivity of large conductance Ca2+-activated potassium channels by retaining the gating in the bursting states

Coexpression of the beta subunit (KV,Cabeta) with the alpha subunit of mammalian large conductance Ca2+- activated K+ (BK) channels greatly increases the apparent Ca2+ sensitivity of the channel. Using single-channel analysis to investigate the mechanism for this increase, we found that the beta subunit increased open probability (Po) by increasing burst duration 20-100-fold, while having little effect on the durations of the gaps (closed intervals) between bursts or on the numbers of detected open and closed states entered during gating. The effect of the beta subunit was not equivalent to raising intracellular Ca2+ in the absence of the beta subunit, suggesting that the beta subunit does not act by increasing all the Ca2+ binding rates proportionally. The beta subunit also inhibited transitions to subconductance levels. It is the retention of the BK channel in the bursting states by the beta subunit that increases the apparent Ca2+ sensitivity of the channel. In the presence of the beta subunit, each burst of openings is greatly amplified in duration through increases in both the numbers of openings per burst and in the mean open times. Native BK channels from cultured rat skeletal muscle were found to have bursting kinetics similar to channels expressed from alpha subunits alone.     

Ca2+-activated K+ channels in the apical membrane ofNecturus choroid plexus

Summary The properties of Ca²⁺-activated K⁺ channels in the apical membrane of theNecturus choroid plexus were studied using single-channel recording techniques in the cell-attached and excised-patch configurations. Channels with large unitary conductances clustered around 150 and 220 pS were most commonly observed. These channels exhibited a high selectivity for K⁺ over Na⁺ and K⁺ over Cs⁺. They were blocked by high cytoplasmic Na⁺ concentrations (110mm). Channel activity increased with depolarizing membrane potentials, and with increasing cytoplasmic Ca²⁺ concentrations. Increasing Ca²⁺ from 5 to 500nm, increased open probability by an order of magnitude, without changing single-channel conductance. Open probability increased up to 10-fold with a 20-mV depolarization when Ca²⁺ was 500nm. Lowering intracellular pH one unit, decreased open probability by more than two orders of magnitude, but pH did not affect single-channel conductance. Cytoplasmic Ba²⁺ reduced both channel-open probability and conductance. The sites for the action of Ba²⁺ are located at a distance more than halfway through the applied electric field from the inside of the membrane. Values of 0.013 and 117mm were calculated as the apparent Ba²⁺ dissociation constants (K _d (0 mV) for the effects on probability and conductance, respectively. TEA⁺ (tetraethylammonium) reduced single-channel current. Applied to the cytoplasmic side, it acted on a site 20% of the distance through the membrane, with aK _d (0 mV)=5.6mm. A second site, with a higher affinity,K _d (0 mV)=0.23mm, may account for the near total block of chanel conductance by 2mm TEA⁺ applied to the outside of the membrane. It is concluded that the channels inNecturus choroid plexus exhibit many of the properties of maxi Ca²⁺-activated K⁺ channels found in other tissues.     

Intracellular alkalinization induces Ca2+ influx via non-voltage-operated Ca2+ channels in rat aortic smooth muscle cells

In smooth muscle, the cytosolic Ca2+ concentration ([Ca2+](i)) is the primary determinant of contraction, and the intracellular pH (pH(i)) modulates contractility. Using fura-2 and 2',7'-biscarboxyethyl-5(6) carboxyfluorescein (BCECF) fluorometry and rat aortic smooth muscle cells in primary culture, we investigated the effect of the increase in pH(i) on [Ca2+](i). The application of the NH(4)Cl induced concentration-dependent increases in both pH(i) and [Ca2+](i). The extent of [Ca2+](i) elevation induced by 20mM NH(4)Cl was approximately 50% of that obtained with 100mM K(+)-depolarization. The NH(4)Cl-induced elevation of [Ca2+](i) was completely abolished by the removal of extracellular Ca2+ or the addition of extracellular Ni2+. The 100mM K(+)-induced [Ca2+](i) elevation was markedly inhibited by a voltage-operated Ca2+ channel blocker, diltiazem, and partly inhibited by a non-voltage-operated Ca2+ channel blocker, SKF96365. On the other hand, the NH(4)Cl-induced [Ca2+](i) elevation was resistant to diltiazem, but was markedly inhibited by SKF96365. It is thus concluded that intracellular alkalinization activates the Ca2+ influx via non-voltage-operated Ca2+ channels and thereby increases [Ca2+](i) in the vascular smooth muscle cells. The alkalinization-induced Ca2+ influx may therefore contribute to the enhancement of contraction.     

Ca2+-activated K+ conductance of the human red cell membrane: Voltage-dependent Na+ block of outward-going currents

Summary Human red cells were prepared with various cellular Na⁺ and K⁺ concentrations at a constant sum of 156mm. At maximal activation of the K⁺ conductance,g _K(Ca), the net efflux of K⁺ was determined as a function of the cellular Na⁺ and K⁺ concentrations and the membrane potential,V _m , at a fixed [K⁺]_ex of 3.5mm.V _m was only varied from (V _m –E _K)25 mV and upwards, that is, outside the range of potentials with a steep inward rectifying voltage dependence (Stampe & Vestergaard-Bogind, 1988).g _K(Ca) as a function of cellular Na⁺ and K⁺ concentrations atV _m =–40, 0 and 40 mV indicated a competitive, voltage-dependent block of the outward current conductance by cellular Na⁺. Since the present Ca²⁺-activated K⁺ channels have been shown to be of the multi-ion type, the experimental data from each set of Na⁺ and K⁺ concentrations were fitted separately to a Boltzmann-type equation, assuming that the outward current conductance in the absence of cellular Na⁺ is independent of voltage. The equivalent valence determined in this way was a function of the cellular Na⁺ concentration increasing from 0.5 to 1.5 as this concentration increased from 11 to 101mm. Data from a previous study of voltage dependence as a function of the degree of Ca²⁺ activation of the channel could be accounted for in this way as well. It is therefore suggested that the voltage dependence ofg _K(Ca) for outward currents at (V _m –E _K)>25 25 mV reflects a voltage-dependent Na⁺ block of the Ca²⁺-activated K⁺ channels.     

Activation of Ca(2+)-dependent K(+) channels contributes to rhythmic firing of action potentials in mouse pancreatic beta cells

We have applied the perforated patch whole-cell technique to beta cells within intact pancreatic islets to identify the current underlying the glucose-induced rhythmic firing of action potentials. Trains of depolarizations (to simulate glucose-induced electrical activity) resulted in the gradual (time constant: 2.3 s) development of a small (<0.8 nS) K(+) conductance. The current was dependent on Ca(2+) influx but unaffected by apamin and charybdotoxin, two blockers of Ca(2+)-activated K(+) channels, and was insensitive to tolbutamide (a blocker of ATP-regulated K(+) channels) but partially (>60%) blocked by high (10-20 mM) concentrations of tetraethylammonium. Upon cessation of electrical stimulation, the current deactivated exponentially with a time constant of 6.5 s. This is similar to the interval between two successive bursts of action potentials. We propose that this Ca(2+)-activated K(+) current plays an important role in the generation of oscillatory electrical activity in the beta cell.     

Voltage dependence of the Ca2+-activated K+ conductance of human red cell membranes is strongly dependent on the extracellular K+ concentration

Summary The conductance of the Ca²⁺-activated K⁺ channel (g _K(Ca)) of the human red cell membrane was studied as a function of membrane potential (V _m ) and extracellular K⁺ concentration ([K⁺]_ex). ATP-depleted cells, with fixed values of cellular K⁺ (145mm) and pH (7.1), and preloaded with 27 m ionized Ca were transferred, with open K⁺ channels, to buffer-free salt solutions with given K⁺ concentrations. Outward-current conductances were calculated from initial net effluxes of K⁺, correspondingV _m , monitored by CCCP-mediated electrochemical equilibration of protons between a buffer-free extracellular and the heavily buffered cellular phases, and Nernst equilibrium potentials of K ions (E _K) determined at the peak of hyperpolarization. Zero-current conductances were calculated from unidirectional effluxes of⁴²K at (V _m –E _K)0, using a single-file flux ratio exponent of 2.7. Within a [K⁺]_ex range of 5.5 to 60mm and at (V _m –E _K) 20 mV a basic conductance, which was independent of [K⁺]_ex, was found. It had a small voltage dependence, varying linearly from 45 to 70 S/cm² between 0 and –100 mV. As (V _m –E _K) decreased from 20 towards zero mVg _K(Ca) increased hyperbolically from the basic value towards a zero-current value of 165 S/cm². The zero-current conductance was not significantly dependent on [K⁺]_ex (30 to 156mm) corresponding toV _m (–50 mV to 0). A further increase ing _K(Ca) symmetrically aroundE _K is suggested as (V _m –E _K) becomes positive. Increasing the extracellular K⁺ concentration from zero and up to 3mm resulted in an increase ing _K(Ca) from 50 to 70 S/cm². Since the driving force (V _m –E _K) was larger than 20 mV within this range of [K⁺]_ex this was probably a specific K⁺ activation ofg _K(Ca). In conclusion: The Ca²⁺-activated K⁺ channel of the human red cell membrane is an inward rectifier showing the characteristic voltage dependence of this type of channel.     

Ca2+-activated K+ conductance of human red cell membranes exhibits two different types of voltage dependence

Summary The voltage dependence for outward-going current of the Ca-activated K⁺ conductance (g _k (Ca)) of the human red cell membrane has been examined over a wide range of membrane potentials (V _m) at constant values of [K⁺]_ex, [K⁺]_c and pH_c, the intact cells being preloaded to different concentrations of ionized calcium. Outward-current conductances were calculated from initial net effluxes of K⁺ and the corresponding (V _m-E_k) values. The basic conductance, defined as the outward-current coductance at (V _m-E_k) 20 mV and [K⁺]_ex 3mM (B. Vestergaard-Bogind, P. Stampe and P. Christophersen,J. Membrane Biol. 95:121–130, 1987) was found to be a function of cellular ionized Ca. At all degrees of Ca activationg _K(Ca) was an apparently linear function of voltage (V _m range –40 to +70 mV), the absolute level as well as the slope decreasing with decreasing activation. In a simple two-state model the constant voltage dependence can, at the different degrees of Ca activation, be accounted for by a Boltzmann-type equilibrium function with an equivalent valence of 0.4, assuming chemical equilibrium atV _m=0 mV. Alternatively, the phenomenon might be explained by a voltage-dependent block of the outward current by an intracellular ion. Superimposed upon the basic conductance is the apparently independent inward-rectifying steep voltage function with an equivalent valence of 5 and chemical equilibrium at the givenE _K value.Abbreviations CCCP carbonyl cyanidem-chlorophenylhydrazone - DIDS 4,4-diisothiocyanostilbene-2,2-disul     

Demarcation of Ca2+ transport processes in guinea pig stomach smooth muscle

Summary Microsomal fractions were isolated from gastric antrum and fundus smooth muscle of guinea pigs. Ca²⁺ uptake into and Ca²⁺ release from the membrane vesicles were studied by a rapid filtration method, and Ca²⁺ transport properties of the different regions of the stomach were compared. ATP-dependent Ca²⁺ uptake was similar in microsomes isolated from both regions. This uptake was increased by oxalate and was not affected by NaN₃. Oxalate affected Ca²⁺ permeability of both antrum and fundus microsome vesicles similarly. Fundus microsome vesicles preincubated in 100mm NaCl and then diluted to 1/20 concentration with Na⁺-free medium had significantly higher ATP-independent Ca²⁺ uptake than vesicles preincubated in 100mm KCl and treated the same way. This was not true for antrum vesicles. Monensin abolished Na⁺-dependent Ca²⁺ uptake, and NaCl enhanced Ca²⁺ efflux from fundus microsome vesicles. The halflife values of Ca²⁺ loss from fundus vesicles in the presence of NaCl were significantly smaller than those in the presence of KCl. The release of Ca²⁺ from the vesicles within the first 3 min was accelerated by NaCl to three times that by KCl. However, NaCl had ro effect on Ca²⁺ release from antrum microsome vesicles.Results suggest two distinct mechanisms of stomach membrane Ca²⁺ transport: (1) ATP-dependent Ca²⁺ uptake and (2) Na⁺–Ca²⁺ exchange; the latter in the fundus only.     

Permeation of Ca2+ through K+ channels in the plasma membrane of Vicia faba guard cells

Summary The whole-cell patch-clamp method has been used to measure Ca²⁺ influx through otherwise K⁺-selective channels in the plasma membrane surrounding protoplasts from guard cells of Vicia faba. These channels are activated by membrane hyperpolarization. The resulting K⁺ influx contributes to the increase in guard cell turgor which causes stomatal opening during the regulation of leaf-air gas exchange. We find that after opening the K⁺ channels by hyperpolarization, depolarization of the membrane results in tail current at voltages where there is no electrochemical force to drive K⁺ inward through the channels. Tail current remains when the reversal potential for permeant ions other than Ca²⁺ is more negative than or equal to the K⁺ equilibrium potential (–47 mV), indicating that the current is due to Ca²⁺ influx through the K⁺ channels prior to their closure. Decreasing internal [Ca²⁺] (Ca _i ) from 200 to 2 nm or increasing the external [Ca²⁺] (Ca _o ) from 1 to 10 mm increases the amplitude of tail current and shifts the observed reversal potential to more positive values. Such increases in the electrochemical force driving Ca²⁺ influx also decrease the amplitude of time-activated current, indicating that Ca²⁺ permeation is slower than K⁺ permeation, and so causes a partial block. Increasing Ca _o also (i) causes a positive shift in the voltage dependence of current, presumably by decreasing the membrane surface potential, and (ii) results in a U-shaped current-voltage relationship with peak inward current ca. –160 mV, indicating that the Ca^2– block is voltage dependent and suggesting that the cation binding site is within the electric field of the membrane. K⁺ channels in Zea mays guard cells also appear to have a Ca _i -, and Ca _o -dependent ability to mediate Ca²⁺ influx. We suggest that the inwardly rectiying K⁺ channels are part of a regulatory mechanism for Ca _i . Changes in Ca _o and (associated) changes in Ca _i regulate a variety of intracellular processes and ion fluxes, including the K⁺ and anion fluxes associated with stomatal aperture change.This work was supported by grants to S.M.A. from NSF (DCB-8904041) and from the McKnight Foundation. K.F.-G. is a Charles Gilbert Heydon Travelling Fellow. The authors thank Dr. R. MacKinnon (Harvard Medical School) and two anonymous reviewers for helpful comments.