| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
L78P) of the muscle nicotinic acetylcholine receptor with unusual single channel properties
1 Department of Pharmacology, University College London, London WC1E 6BT, UK
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
L78P) is known to cause a congenital slow channel myasthenic syndrome. We have investigated the changes in receptor function that result in the mutant receptor producing prolonged endplate currents, and consequent muscle damage. The rate constants for channel gating and for the binding and dissociation of acetylcholine were investigated by analysis of single ion channel recordings. A conventional mechanism with two non-equivalent binding sites, and variations upon this mechanism, were fitted to data using a maximum likelihood method that uses the Hawkes-Jalali-Colquhoun (HJC) treatment of missed brief events. The mutant receptor produced prolonged activations, bursts of openings that cause a slow decay of simulated synaptic currents. The main reason for the longer bursts of openings seen with mutant receptor was a decrease in the rate of ACh dissociation from diliganded receptors, though the lifetime of individual openings was somewhat increased too. As well as producing long bursts, the mutant receptor also produced many very short openings, though these carry little current. The burst structure for the mutant receptor at low ACh concentration is unusual in that most long bursts appear to start in a very brief monoliganded open state that then usually binds another ACh molecule to produce a long diliganded activation. The first opening is so short that it will usually be missed (together with the shut time that follows it), so the true burst length is likely to be underestimated.
(Received 16 December 2004;
accepted after revision 17 February 2005;
first published online 24 February 2005)
Corresponding author D. Colquhoun: Department of Pharmacology, University College London, London WC1E 6BT, UK. Email: d.colquhoun{at}ucl.ac.uk
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
subunits, and single ß, 
, and subunits arranged around a central cation-conducting pore. There are two non-identical ACh-binding sites, situated at the interfaces between the and subunits, and between the and subunits (the and sites, respectively). In the adult mouse receptor, the two sites are usually assumed to be identical (Akk & Auerbach, 1996; Wang et al. 1997; Salamone et al. 1999; Akk, 2002), although there is some evidence that the ACh dissociation rates (if not the ACh affinity) are different at the two sites (Salamone et al. 1999). However, in adult human (Hatton et al. 2003), fetal mouse (Jackson, 1988; Zhang et al. 1995), and Torpedo (Sine et al. 1990) receptors the sites have been inferred to have different ACh binding properties. The distinction between the two sites is important when considering mutations that could affect one site more than the other. Mutations in the human receptor have been found to be responsible for slow channel congenital myasthenic syndromes (SCCMS), in which endplate currents decay abnormally slowly. Most patients show varying degrees of weakness that is particularly apparent in the ocular, shoulder, limb, cervical, wrist and finger muscles. The patients fatigue easily and may have breathing difficulties. There is usually degradation of the muscle end plates. Excessive accumulation of calcium has been detected in the endplates of both SCCMS patients (Engel et al. 1982) and a transgenic mouse model (Gomez et al. 2002). It is this calcium overload that is believed to be responsible for the degeneration of junctional folds and apoptosis at subsynaptic end plate nuclei.
Several mutations that underlie SCCMS have been studied in detail using single channel analysis. The primary effect of many mutations is to decrease the rate of agonist dissociation, e.g. N217K (Wang et al. 1997), G153S (Sine et al. 1995), L221F (Hatton et al. 2003), although other mutations, such as V249F (Milone et al. 1997), have been found that have effects largely, or entirely, on channel gating too.
Here we study the effects of a point mutation in the subunit that is responsible for a slow channel congenital myasthenia (Croxen et al. 2002). A single homozygous nucleotide point mutation of thymine to cytosine at position 233 of the mature subunit coding region results in the substitution of a proline residue for a leucine residue at position 78 in the protein. The position is within the N-terminal extracellular domain of the receptor but not within the generally accepted ACh binding sites. The mutation was identified originally in a 29-year-old female patient who failed to breathe following a general anaesthetic. Six to seven years earlier she had noticed muscle weakness. Subsequent examinations identified bilateral ptosis, limitation of eye movements, weakness of facial and hand muscles, and weakness of the neck, shoulders and hips. Electromyography showed an abnormally large decrement of response at 3 Hz stimulation. Antibodies to nicotinic receptors were not detected. One of her siblings reported a similar failure to breathe after administration of a general anaesthetic.
In order to understand properly the effects of mutations in the receptor, it is necessary to describe the single channel properties of the receptor in terms of the underlying rate constants that govern the receptor's functions (Colquhoun, 1998). This necessitates identification of a reaction mechanism that describes, to a sufficiently good approximation, the actual physical events that occur during receptor activation. Another reason why it is necessary to postulate a mechanism is that without a mechanism it is, in general, not possible to allow for the fact that brief openings and shuttings will be missed (Colquhoun & Hawkes, 1995). Some preliminary results from this study have been presented in partial form previously (Croxen et al. 2002; Shelley et al. 2003).
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Human embryonic kidney cells (HEK-293), obtained from either the American Type Culture Collection (ATCC) or the European Collection of Cell Cultures (ECACC), were cultured as previously described (Hatton et al. 2003). For transfection using the calcium phosphate method they were plated onto polylysine-coated coverslips in 30 mm Petri dishes. For each Petri dish, 3 µg of total DNA was used. The DNA consisted of pcDNA3.1 plasmids (Invitrogen) containing inserts encoding the subunits 1, ß1, , , enhanced green fluorescent protein (eGFP) (Invitrogen), and the non-coding plasmid pcDNA3.1 in the ratio of 2: 1: 1: 1: 65: 30. The construction of the plasmids coding for the subunits has been previously described (Croxen et al. 2002). The eGFP and the non-coding pcDNA3.1 plasmids were a gift from Paul J. Groot-Kormelink (University College London). The DNA mix was added to 60 µl of 340 mM CaCl2. This mixture was in turn added dropwise to 66 µl of 2 x Hepes-buffered saline (280 mM NaCl, 50 mM Hepes, 2.8 mM Na2HPO4, pH 7.2 with NaOH). The solution was left for 1 min and then added to the Petri dish. Cells were then incubated overnight at 37°C in 5% CO2 and the medium replaced next morning. Cells were patched between 16 and 56 h following the transfection procedure.
Single channel recording
Steady state single channel recordings were made in the cell-attached configuration at 19°C, from cells that emitted green fluorescence. The extracellular solution contained 142 mM KCl, 10 mM Hepes, 5.4 mM NaCl, 1.7 mM MgCl2, 1.8 mM CaCl2, pH adjusted to 7.4 with KOH. Borosilicate pipettes (GC150F-7.5, Clark Electromedical) were pulled to give a resistance of 8–12 M
and their tips coated with 10% Sylgard (Dow Corning). Immediately before patching the pipettes were fire-polished and back-filled with extracellular solution containing ACh chloride. In our cell-attached recordings the membrane potential is unknown, so the holding potential was adjusted to produce openings of 6 pA, which required a membrane potential of approximately –100 mV. Currents were amplified using an Axopatch 200A (Axon Instruments) coupled to a CV 201AU headstage (Axon Instruments). Data were filtered initially at 10 kHz and recorded on digital audio tape (DAT) using a DTR-1204 digital tape recorder (Biologic Science Instruments). Before digitization data were filtered again at 5 kHz to 8 kHz using an 8th order Bessel filter (built in-house), giving a final –3 dB frequency of 4.8 kHz to 6.2 kHz. Data were digitized at 50 kHz to 80 kHz using a CED 1401 interface (Cambridge Electronic Design) and the CONSAM program (I. Vais and D. Colquhoun).
Data analysis
Following an initial inspection of the digitized single channel trace, the amplitudes and the duration of the open and closed states were measured using the SCAN program to give an idealized record. The transitions were fitted with convolved step-response functions using a least-squares criterion (Colquhoun & Sigworth, 1995). A resolution of 20–40 µs was imposed retrospectively on the results; stability plots (program EKDIST) were inspected before further analysis. The duration of an ‘open period’ was defined as the length of time for which the channel remained open (regardless of conductance), the period ending when a shut time longer than the imposed resolution occurs. Bursts were defined as a sequence of openings and shuttings that ends once a shut time longer than tcrit occurs. The tcrit values were chosen so as to ensure that an equal proportion of intervals from both the low and the high components were misclassified (Colquhoun & Sakmann, 1985). In some cases this was not possible due to the bisection algorithm employed not converging. In these cases tcrit was chosen so that an equal number of intervals from the low and high components were misclassified (Magleby & Pallotta, 1983; Clapham & Neher, 1984). The number of intervals fitted for each patch ranged from 2000 to 50 000. For a typical set of fits the mean number of transitions fitted per patch was 17 500 ± 2600.
The open period, shut time, and burst length histograms were fitted empirically with mixtures of n exponential probability density functions (pdfs) using EKDIST.
The microscopic rate constants in putative physical mechanisms for receptor activation were estimated using the HJCFIT program, which maximizes the likelihood of the entire sequence of open and shut times, with exact allowance for missed brief events (Hawkes et al. 1990; Hawkes et al. 1992; Colquhoun et al. 1996). For a complete description of the use of the HJCFIT program, and tests of its performance see Colquhoun et al. (2003), Hatton et al. (2003) and Burzomato et al. (2004). Scheme 1, shown in Fig. 4, and some variants of it were used to describe the receptor mechanism. This scheme describes a receptor with two different binding sites, with either of the two different monoliganded forms being able to open, as well as the diliganded form. For all the fits we made the customary assumption that the two binding sites are independent. In other words, we assume that binding of a molecule of ACh to the a-site does not affect binding of a second molecule of ACh to the b-site and vice versa. Using the notation in Fig. 4, this means imposing the following three constraints during fitting.
|
| (1) |
|
| (2) |
|
|
|
b) with a vector that implies that all bursts start in a specified state (Colquhoun & Hawkes, 1982), for example, for start in state 1:
|
|
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Figure 1 shows examples of the single channel currents recorded from wild-type muscle nicotinic receptors, and from receptors that contain the L78P mutation. It is obvious to the naked eye from these traces that receptors with L78P mutation produce bursts of openings that are often longer than those seen in wild-type. This suggests that receptors with this mutation will produce slowly decaying synaptic currents (see Fig. 10 and Discussion). In order to make this impression more quantitative, distributions of open period durations and of burst lengths were fitted empirically (i.e. without reference to any reaction mechanism) with mixtures of exponential pdfs (see Methods). Examples are shown in Fig. 2.
|
|
L78P receptors, with 100 nM ACh, could be fitted adequately with three, and in one case four, exponential components. The values of the individual time constants and their areas are shown in Table 1; the histograms and the fits are shown in Fig. 2. The L78P receptors contained a greater proportion of short openings compared with wild-type receptors, but an approximate doubling of the slowest time constant meant that the overall mean apparent open time duration was similar for L78P and wild-type receptors, being 1.17 ms and 1.15 ms, respectively.
|
L78P receptors could be fitted with four or five components; an example is shown in Fig. 2. The values found for the five component fits are shown in Table 2 and compared with wild-type receptors, both being measured with 100 nM ACh. The L78P receptors had a greater proportion (a1) of ‘short shut times’ (
1) compared with wild-type receptors. The proportion of ‘intermediate shut times’ (2 for the wild-type, and 2, 3, and 4 for the L78P receptors) was similar for both.
|
Burst length duration histograms were constructed for wild-type and L78P receptors. The bursts were defined by fitting the shut time duration histograms with the minimum number of components that described the data adequately. In the case of wild-type receptors this was always three components. For L78P receptors four or five components were required. In the cases where shut times could be described with three or four components then a tcrit value was specified so as to omit the slowest of the shut time components. In the cases where five shut time components were required then a tcrit was specified so as to omit the slowest two components as it was considered that the mean value of 4 (35 ms) was too slow to be included within a channel ‘activation’. This component and the ‘smear’ of shut times between 1 and 10 ms in Fig. 2D make the definition of tcrit for the mutant receptor somewhat ambiguous (according to the interpretation of the results given below, these intermediate shut times result largely because brief openings are missed, and are not present in the ideal distribution). Using these criteria gave a mean tcrit of 0.79 ms ± 41% for wild-type channels, and 3.63 ms ± 38% for L78P channels. Figure 3 and Table 3 show that the distributions of the burst length durations of wild-type and L78P receptors are very different. The mean burst length of wild-type receptors at 100 nM ACh was 1.41 ms ± 21%, surprisingly similar to the mean burst length duration of L78P receptors, 1.2 ms ± 32%. Although the L78P receptors visibly produce longer bursts than the wild-type receptors, the majority of the bursts from L78P receptors consisted of very short burst length durations (1
= 7 µs, area = 84%), and thus there is little change in the overall mean length of bursts. The very short bursts are expected to be mostly isolated single-liganded openings; because they are so short, they will have little influence on the endplate current during physiological activation of the receptors. Virtually all of the synaptic current is carried by multiple diliganded open periods occurring in quick succession, a ‘typical’ burst, because such bursts are much longer than monoliganded openings. Therefore the burst length durations were re-examined by looking at the distribution of only those bursts that contain two or more apparent open periods. The overall mean length of these bursts was eightfold greater for mutant than for wild-type. The wild-type receptor distributions were consistently fitted with a single exponential component, but L78P receptors required at least two clearly separated exponential components, shown in Table 4 and Fig. 3.
|
|
|
L78P mutation on the microscopic rate constants of activationFits with Scheme 1. To estimate the rate constants that determine channel activity, maximum likelihood fitting was done by the HJCFIT method. Initially the simplest mechanism (Scheme 1, Fig. 4) was used (see Methods).
No desensitized or ACh-blocked states were included in the scheme as the data were all obtained at low ACh concentrations. We are not investigating desensitization, and it has been shown that there is no need to include it in the reaction scheme in order to estimate the quantities of physiological relevance (Colquhoun et al. 2003). Several constraints were imposed on the rate constants. One of the forward association rates was fixed at a plausible value, because it is not known how many channels are in a cell-attached patch, so the absolute rate of channel activations cannot be known. Therefore it follows that at least one of the initial (binding of a first molecule of ACh) association rate constants cannot be determined. It has been shown (Colquhoun et al. 2003; Hatton et al. 2003) that the quantities of physiological interest can be estimated from low concentration records alone if an association rate constant (k+1a) is fixed. If the value at which it is fixed is incorrect, this will cause errors in the estimates of one monoliganded opening rate constant (ß1b) which is in any case ill-defined, and in one affinity constant (Ka). But an incorrect value for k+1a will not cause errors in the diliganded gating rate constants or in any of the dissociation rate constants (Colquhoun et al. 2003; Figs 3–5). Therefore k+1a was fixed at a value close to that found in other studies, 1 x 108 M–1 s–1 (Salamone et al. 1999; Hatton et al. 2003).
The two agonist binding sites were assumed to be independent, that is binding of ACh to site a is unaffected by the occupancy of site b, and vice versa (see Methods, eqns (1) and (2)). Good fits to single channel data can be obtained with this assumption of independence (Akk et al. 1999; Salamone et al. 1999; Hatton et al. 2003). This is just as well, because if the two sites are not independent it is not possible to get reliable estimates of the rate constants at low ACh concentrations, unless the number of channels in a patch is known (Colquhoun et al. 2003).
Figure 5 shows the HJC distributions overlaid onto (not fitted to) the observed open and shut time histograms and for both wild-type and L78P channels. The observations appear to be described quite well by the mechanism and fitted rate constants. The most robustly estimated parameters are those that describe the properties of the diliganded receptor. The biggest effect (as might be expected from a mutation near the binding site) is a fourfold slowing of the total rate of ACh dissociation from the diliganded closed receptor, from 16 100 s–1
± 6% to 3910 s–1
± 9%. Although the L78P mutation has no substantial effect on the opening rate of the diliganded receptor (ß2) (47 900 s–1
± 9% compared to 59 100 s–1
± 12% for L78P), the rate of closing of the diliganded receptor (2) is decreased by about twofold, from 1620 s–1
± 6% to 691 s–1
± 20%. This means that the individual diliganded openings are about twice as long on average for the L78P receptors as they are for the wild-type. The reduction in 2 is reflected also as an increase in E2, the efficacy of the gating process for diliganded channels for L78P, from 30.3 ± 9% to 97.0 ± 14%. The decrease in dissociation rate means that the diliganded but shut channel ACh is relatively more likely to re-open, before the agonist dissociates, thus producing more openings in each channel activation (burst of openings). In addition, each individual opening is longer, on average, too.
Figure 5 shows also the plots of conditional mean open period duration (open periods that are adjacent to shut times in specified ranges) against apparent shut time (the mean of the shut times in each range). The predicted conditional means, calculated as in Colquhoun et al. (1996), agree quite well with the observed values, for both the wild-type and L78P receptors.
Heterogeneity. Despite the fact that the fit of Scheme 1 to a single concentration (Fig. 5) looks good, this simple scheme is not satisfactory, as previously shown (Salamone et al. 1999; Hatton et al. 2003). The best way to fit a mechanism is to fit simultaneously records made at a range of agonist concentrations. A single set of rate constants should be able to fit the data well at every concentration. This will work well only if the records are homogeneous, and that ideal is not always realized. In this study, heterogeneity was a sufficiently bad problem that we preferred not to do simultaneous fits, but rather to fit each concentration separately.
One problem arose from the fact that an obvious empirical observation, such as the time constant for the main component of apparent open times, was not as constant from one recording to the next as might have been hoped. Another problem arose from the observation of very brief openings in the periods, normally silent, between clusters of openings at high agonist concentrations (see Fig. 1 in Elenes & Auerbach, 2002; and Fig. 2 in Hatton et al. 2003). In native receptors at the endplate, very few such openings are seen, but they seem to have become increasingly common in our recombinant preparations. They are not predicted by any of the mechanisms that have been used so far, and at present we think they are most likely to originate from some sort of atypical channel that contaminates our records to varying extents (R. Lape & D. Colquhoun, unpublished observations).
The question arises of how to distinguish between poor fits that result because an inadequate reaction mechanism is being used, and poor fits that result from heterogeneity. Heterogeneity would be manifest as variability in the values of the fitted rate constants, but there is no obvious reason why the values should be correlated with agonist concentration. In fact, when Scheme 1 was fitted, a systematic change in the values of the rate ‘constants’ was found over the concentration range from 50 nM to 50 µM ACh, as shown in Fig. 6 (left column). At high ACh concentrations the estimates of both 2 and ß2 appeared to increase slightly. This anomaly suggests that Scheme 1, although it fits well at one concentration (Fig. 5) is not adequate to describe the whole concentration range, as already reported by Salamone et al. (1999) and by Hatton et al. (2003).
|
Maximum likelihood fitting of Scheme 2 with HJCFIT was tried on 29 patches, at a range of ACh concentrations. Of these, seven patches did not fit well, though these seven patches showed no obvious common properties in ACh concentration, temperature, date of the experiments or tcrit values. For the rest of the patches, as with Scheme 1 there was good agreement between the open and shut time duration histograms and the HJC distributions. The estimates of the diliganded parameters (channel opening and closing rate, and the total rate of ACh dissociation from the diliganded shut state) were similar for both wild-type and L78P to those obtained with Scheme 1.
The estimates of the rates of ACh dissociation from the wild-type open channels (k–2a(o) and k–2b(o), Schemes 2 and 5) were 86 s–1 and 154 s–1 for the two different sites. These are low compared with the estimate of the rate of channel closing (1660 s–1) and in the same range as those found for dissociation of ACh from the open state of mouse nicotinic receptors using somewhat different methods of single channel analysis (Grosman & Auerbach, 2001). These values imply that, from the open diliganded state, most transitions will involve closure of the pore, rather than dissociation of an agonist molecule. Dissociation from the open channel is much (roughly 40-fold) slower than from the shut channel. However Scheme 2 was still not satisfactory, for two reasons. Firstly, the estimates of the association rate constants for binding to the open channel were implausibly high, of the order of 1010 M–1 s–1, and secondly Fig. 6 (middle column) still shows some signs of (non-linear) dependence of the rate constant estimates on concentration.
Fits with Scheme 3.
In the light of the imperfections of fits with Scheme 2, we returned to the scheme used previously by Salamone et al. (1999) and by Hatton et al. (2003). This is shown as Scheme 3 (Fig. 4), and differs only from Scheme 1 only by the addition of a short lived shut state accessible from the diliganded open state. Maximum likelihood fitting with this scheme yielded estimates of the diliganded parameters that were similar to those found with Schemes 1 and 2, namely 2
= 1800 s–1, ß2
= 53 300 s–1, E2
= 29.1, total rate of dissociation = 14 900 s–1. Of 29 patches at a range of ACh concentrations, only two did not give good matches between the observed open time and shut time distributions and their corresponding HJC distributions. Moreover, the concentration dependence of 2 seen when fitting to Scheme 1 was not present (Fig. 6, right column). An example of a fit is shown in Fig. 7, and the averaged estimates of the rate constants are shown in Table 5.
|
1a
= 0.27 ms, compared with the mean open time for doubly occupied receptors which is 1/2
= 0.56 ms). The rate constant for entry into the putative short-lived shut state, A2RD, in Scheme 3 is estimated as ßD
= 96 s–1 for wild-type and ßD
= 16 s–1 for the mutated receptor, the former being similar to the values of 32–98 s–1 found by Salamone et al. (1999) and 65 s–1 found by Hatton et al. (2003). However, the rate constant, D, for leaving this state are around 10 times faster than we found earlier for human receptor (Hatton et al. 2003), and than others have found for mouse receptor (Salamone et al. 1999), so the equilibrium constant for entry into the A2RD state is consequently smaller; the reason for this difference is unknown. Once we have a realistic mechanism, and values for the rate constants, it should be possible to predict the behaviour of the receptor under any circumstances. The values in Table 5 were used to predict the time course of synaptic currents (Fig. 10) and also for calculations that are designed to cast light on the physical basis of the unusually well-separated time constants for the distribution of the length of bursts with more than one opening for the L78P receptor, that was shown in Fig. 3D. Bursts with more than one opening
The burst length distribution, unlike that of the apparent open time and shut time, is usually not greatly affected by the inability to detect short events (shorter than 30 µs here). The case of the mutant receptor, L78P, proved to be an exception to this rule, so the explanation for the shape of the burst length distributions shown in Fig. 3 was investigated by simulation of open and shut times at 100 nM ACh (with the program SCSIM), using the estimated rate constants from Scheme 3 for the L78P receptor (as in Table 5). The simulation method is explained in greater detail by Colquhoun et al. (2003). Although the burst length distribution is easily calculated in the ideal case of perfect resolution (Colquhoun & Hawkes, 1982; eqn (3.17)), it turned out in this case that the inability to detect brief events had an unusually important effect, so simulation was used. After simulation of the open–shut sequence, a resolution of 30 µs was imposed on the results, as in real experiments. The distributions of simulated open time duration, shut time duration, and of burst length duration, with a resolution of 30 µs, are shown in Fig. 8. The shut time distribution for the mutant receptor (Fig. 8B), like the observed distribution (Fig. 2D) shows a smear of shut times between 1 and 10 ms and an empirical fit gives a component with
= 12.5 ms, comparable with the observed 4
= 35 ms. This smear makes the selection of an appropriate tcrit somewhat ambiguous. However, calculation of the ideal shut time distribution from the same rate constants as used for simulation shows that no such intermediate shut times are predicted: virtually nothing is expected between a 0.2 ms component and a 90 ms component. The intermediate shut times that give rise to ambiguity in the definition of tcrit are not ‘genuine’, but arise because many short openings are undetected. For the definition of bursts from the simulated mutant data (Fig. 8C and D), we chose to use the same tcrit value of 3.63 ms that was used for the experimental results (Fig. 3C and D). The distributions of the length of all bursts (Fig. 8C), and the distribution of the length of bursts with more than one opening (Fig. 8D) are similar in shape to those found experimentally (Fig. 3C and D).
|
Conditional burst length distributions
In order to understand better the mechanisms underlying the structure of bursts of openings, we inspected the sort of bursts that are predicted conditional on a burst starting in each of the three open states. These predictions are based on Scheme 3, with the rate constants found in Table 5. They are calculated for perfect time resolution, as described by Colquhoun & Hawkes (1982), using the SCBST program.
At high ACh concentrations, almost all burst of openings will start in the diliganded open state. As shown in Table 6 and Fig. 9, bursts (activations) that start in the diliganded state will have for wild-type a mean length of 2.7 ms with 5.07 openings per burst, whereas for L78P receptors they will have a mean length of 42 ms with 91 openings per burst. It is these lengthened bursts that underlie the slow decay of the synaptic current, predicted in Fig. 10.
|
|
b, is calculated from the Q matrix, elements of which are the transition rate constants of the kinetic scheme (Colquhoun & Hawkes, 1982; eqn (3.2)). Values are given for wild-type and the L78P receptors in Table 6. Wild-type receptors. For the wild-type, 10% of bursts are predicted to start in the diliganded open state even at 100 nM ACh, and these contribute almost all of the 2.7 ms component in the overall burst length distribution. Bursts that start in the monoliganded a open state have a single principal component, with a time constant of 0.274 ms. This corresponds almost exactly to the predicted lifetime of 0.270 ms for a single monoliganded opening elicited from the a site, as expected, since such bursts are almost all single openings (mean of 1.01 openings per burst). Bursts that start in the monoliganded b state have three principal components. The predominant component has a time constant of 16.3 µs with 81% of the area. This is almost exactly the same duration as a single monoliganded opening from the b site (16.4 µs). The second component has a time constant of 0.32 ms and an area of 16%. The third component has a time constant of 2.7 ms and an area of 3%. The 2.7 ms bursts correspond to bursts that are initiated in the brief monoliganded b state and then proceed via shut states to a sequence of transitions between the diliganded shut and the diliganded open states.
L78P receptors.
For mutant channels, when no short events are missed, it is predicted that virtually no bursts start in the diliganded open state at 100 nM ACh. Most (88%) start in the monoliganded b state, and almost all of these bursts (98%) will lead to long bursts with time constants of 12 ms or 30.6 ms. The monoliganded open state b is so short that it will almost always be missed, and when that happens, the shut time that follows it will not be counted as part of the burst length either. After return from the open state b (RA*, Scheme 3 and Fig. 4) to the adjacent RA state, there will be an average of 1.6 ms before the next open state is entered, omission of which will cause substantial error in the measured burst length. The overall burst length distribution for L78P receptors contains also about 12% of shorter bursts (
= 0.16 or 0.46 ms) that are seen to originate from bursts that start in the monoliganded open state a, as shown in Table 6.
Predicted effects of the L78P mutation on synaptic currents
The values obtained for the microscopic rate constants can be used to calculate the synaptic current that would be produced at the muscle endplate. The SCALCS program was used to calculate the synaptic current using the rate constant estimates from Table 5, the transmitter time course being taken as a 0.1 ms pulse of 1 mM ACh (the results do not depend strongly on the duration of the pulse as long as it is brief). The predicted current is shown in Fig. 10, along with that predicted from the wild-type receptor rate constants.
The wild-type macroscopic current has a predominant decay time constant of 2.6 ms, similar to that found at actual end plates (Cull-Candy et al. 1979). The L78P receptors have a predominant decay time constant of 26.7 ms, 10-fold slower than the wild-type receptors. The predominant time constant of decay of a macroscopic current is expected to be the same as the slowest time constant of the burst length distribution, when the latter is measured at very low concentration (Colquhoun et al. 1997; Wyllie et al. 1998). The predicted decay is indeed close to this value (Fig. 9, Table 6).
As noted above, the fit to the experimental data (Fig. 3, Table 3) indicated somewhat longer bursts than were predicted by the fit. Possible reasons for this include (a) the ambiguity in the tcrit value, (b) the necessarily limited number of long bursts that are observed at 100 nM, which is reflected in the rather large errors for the time constants and area of the slowest components in Table 3, or (c) some inadequacy of the mechanism we have fitted.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Congenital slow channel myasthenic syndromes are caused by mutations in the nicotinic receptor that cause prolonged synaptic currents (e.g. Croxen et al. 2002; McConville & Vincent, 2002; Engel et al. 2003a, b). Many different mutations can cause SCCMS, but there appears to be surprisingly little correlation between the severity of the mutation on channel function, and the severity of the symptoms experienced (e.g. Croxen et al. 1997; Milone et al. 1997). Muscle weakness would necessarily result from degradation of the end plate region. This degradation is thought to be caused by excess Ca2+ entry into the muscle cytosol, which may activate the apoptotic machinery. The source of this Ca2+ could be direct entry through the muscle receptors (Costa et al. 1994), or via the large efflux of Ca2+ from the sarcoplasmic reticulum following the initiation of muscle action potentials by the depolarization of the end plate produced by the receptor activation.
The results here show that the L78P mutation causes ACh to produce very long activations of the channel, i.e. the channel reopens many times between the time when the first ACh molecule binds, and the time when the last ACh molecule dissociates. Fitting of the single channel results allows it to be predicted that the synaptic current will decay 10-fold more slowly than for wild-type receptors, though no direct measurements have been made on human endplates for this mutation. Nevertheless the symptoms of patients with this mutation are mild compared with those of some other cases of SCCMS. One contributory factor could be a compensatory increase in expression of the fetal 
subunit in damaged endplates (Engel et al. 1996; Ohno et al. 1997).
Mechanisms and rates
Single channel recordings could be fitted quite well with a standard mechanism (Scheme 3, Fig. 4), for the mutant L78P receptors as well as wild-type. Once the rate constants have been estimated it becomes possible to predict any single channel or macroscopic property of the receptor, whether at equilibrium or not.
The estimates obtained for the diliganded channel opening rate constant, ß2, for the human wild-type receptor are similar to those found previously in human and mouse muscle receptors. ß2 has been estimated in several studies to be 50 000–60 000 s–1 at 20°C at the mouse receptor (Wang et al. 1997; Salamone et al. 1999), and human receptor (Hatton et al. 2003). The estimate of ß2 is similar to that found for frog receptors (30 600 s–1 at 11°C; Colquhoun & Sakmann, 1985), bearing in mind the difference in temperature (the difference corresponds to a Q10 of about 2, a common value for rate constants: Gutfreund, 1995).
The rate constant for diliganded channel closing (2) for wild-type receptors was estimated as approximately 2000 s–1, in agreement with Hatton et al. (2003) who found 1850 s–1. Mouse receptors also appear to be similar to human receptors. Estimates for 2 in mouse receptors include 1300 s–1 (Akk et al. 1999), 1700 s–1 (Wang et al. 1997), and 2600 s–1 (Salamone et al. 1999). This is again similar to the value of 714 s–1 for frog muscle at 11°C (Colquhoun & Sakmann, 1985), if a Q10 of about 2 is assumed. The equilibrium gating constant (or ‘efficacy’), the ratio of opening to shutting rate constants, E2
=
ß2/2, for diliganded channel gating is 25–30 in these and other recent studies.
The total rates of dissociation of ACh from diliganded receptors (which can be measured more accurately than the two separate rates of dissociation from each site) are different in human and mouse. This study estimates this total rate (k–2a + k–2b), to be in the range of 14 900 s–1 (similar to 13 800 s–1 in Hatton et al. 2003) while estimates for the mouse receptor include 36 000 s–1 (Akk et al. 1999), 44 000 s–1 (Wang et al. 1997), and 49 000 s–1 (Salamone et al. 1999). Individually the two sites in both human and mouse have different dissociation rates, values of 18 000 s–1 and 43 000 s–1 are given for the mouse receptor (Salamone et al. 1999).
Although the L78P receptor produces longer bursts of openings than wild-type, and hence slower synaptic current decay, it also produces many very short openings. Empirical fitting suggested a mean length of 9 µs for these (Table 1) though fitting suggested the openings of the b monoliganded state are even shorter (about 4 µs). A similar discrepancy was seen for homomeric glycine receptors (Burzomato et al. 2004), and it is not surprising since the HJC distribution of observed open times would not be expected to be fitted by exponentials for such short events (Hawkes et al. 1990) because they are too short for the asymptotic distribution, which is a mixture of exponentials, to be valid (Hawkes et al. 1992). Our fits suggest that for the mutant receptor, most bursts will start in this very short-lived open state at low (100 nM) ACh concentrations, though most of these go on to reach the diliganded state and so generate long pathogenic bursts. Most openings that are so short will be missed, so there is an unusually large effect of missed events on the observed burst length distribution, a fact that emphasizes the importance of the exact missed event method used in HJCFIT.
The main reason for the increased burst length of L78P receptors is that the total dissociation rate for diliganded receptors is reduced fivefold, from 14 900 s–1 to 2840 s–1 (Table 5), the rest of the effect coming from 1.7-fold longer openings (reduced 2) and a 1.4-fold increase in opening rate (ß2) in the mutant. It is, perhaps, surprising that a mutation so far from the pore region should affect gating at all. However it must be remembered that the steps labelled as ‘binding’ in the mechanism (Scheme 3) that was fitted actually include everything that occurs before the shut-open conformation change. If there were a substantial change in conformation of the receptor before the channel opened, then it would not be surprising if such a conformation change were affected by mutations in the extracellular region. Exactly this sort of preopening conformation change provides an elegant explanation of observations with glycine receptors (Colquhoun & Sivilotti, 2004; Burzomato et al. 2004), and may also occur in nicotinic receptors (Chakrapani et al. 2004; R. Lape & D. Colquhoun, unpublished observations).
Structure and function
The position of L78P within the receptor can be guessed by examining the homologous acetylcholine binding protein (AChBP) for which a crystal structure has been obtained (Brejc et al. 2001; Celie et al. 2004). Alignment of the human subunit and the AChBP shows that the position L78 aligns with V74 in the AChBP, as shown in Fig. 11.
|
L78P can affect gating, despite being so far from the ion-conducting pore.
|
subunit would lead to effects on only one binding site. However, effects on both binding sites are seen; dissociation rates are decreased 4-fold at the a site and 42-fold at the b site. This is entirely possible because the subunit lies between two subunits in the arrangement of subunits around the pore. Thus, seen from the extracellular face of the membrane, changes in the subunit could be transmitted clockwise to the neighbouring subunit forming the binding site, or anticlockwise to the neighbouring subunit forming the binding site. Unless a mutation is directly in the binding site of the subunit it is impossible to predict with any accuracy whether the mutation affects the or the binding sites or associated gating steps.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Akk
G
&
Auerbach
A (1996). Inorganic monovalent cations compete with agonists for the transmitter binding site of nicotinic acetylcholine receptors. Biophys J
70, 2652–2658.
Akk
G, Zhou
M
&
Auerbach
A (1999). A mutational analysis of the acetylcholine receptor channel transmitter binding site. Biophys J
76, 207–218.
Beato
M, Groot-Kormelink
PJ, Colquhoun
D
&
Sivilotti
LG (2004). The activation mechanism of 1 homomeric glycine receptors. Journal of Neuroscience
24, 895–906.
Brejc K, van Dijk WJ, Klaassen RV, Schuurmans M, van Der Oost J, Smit AB & Sixma TK (2001). Crystal structure of an ACh-binding protein reveals the ligand-binding domain of nicotinic receptors. Nature 411, 269–276.[CrossRef][Medline]
Burzomato
V, Groot-Kormelink
PJ, Colquhoun
D, Beato
M
&
Sivilotti
LG (2004). Single channel behaviour of heteromeric 1ß glycine receptors: an attempt to detect a conformational change before the channel opens. J Neurosci
24, 10924–10940.
Celie PH, Rossum-Fikkert SE, van Dijk WJ, Brejc K, Smit AB & Sixma TK (2004). Nicotine and carbamylcholine binding to nicotinic acetylcholine receptors as studied in AChBP crystal structures. Neuron 41, 907–914.[CrossRef][Medline]
Chakrapani
S, Bailey
TD
&
Auerbach
A (2004). Gating dynamics of the acetylcholine receptor extracellular domain. J General Physiol
123, 341–356.
Clapham DE & Neher E (1984). Substance P reduces acetylcholine-induced currents in isolated bovine chromaffin cells. J Physiol 347, 255–277.
