вход по аккаунту


Chemically activated channels in muscle and spinal cord.

код для вставкиСкачать
Chemically Activated Channels
in Muscle and Spinal Cord
Meyer B. Jackson, PhD
The patch clamp can be used to record single chemically activated channel currents in a variety of cell culture
preparations. In the case of the y-aminobutyric acid (GABA) response in spinal cord cell culture, the channel i s C1-selective. C1- can be made to flow into or out of a cell by changing the direction of the electrochemical driving force for
C1-; as a result, positive or negative channel currents are produced. Channel currents generated by GABA and the
GABA agonists muscimol and pentobarbital have the same amplitude.
The kinetics of channel gating are studied by analyzing distributions of dwell times in conducting and nonconducting states. Such analyses of the GABA-activated channel and the acetylcholine-activated channel reveal that gating is
complex. More elaborate procedures of data analysis have been used in an attempt to elucidate detailed molecular
gating mechanisms.
Jackson MB: Chemically activated channels in muscle and spinal cord.
Ann Neurol 16(suppl):S5 2-S 58, 1984
Many questions about the electrical activity of neurons
can be reduced to questions about ionic channels in
nerve cell membranes. These integral membrane proteins confer upon a nerve cell its unique electrical properties. The difference between rapid-impulse activity
and slower modulation of membrane potential is then
explained in terms of the different modes of collective activity by populations of channels with different
intrinsic rates of transition between their open and closed
conformations. In this context, the anomalous electrical activity of neurons during seizures is produced
by channels that, under other conditions, give rise to
normal electrical activity.
Studies of ionic channels have been greatly aided by
the development of techniques to measure their currents directly. With a small extracellular patch electrode
(diameter = 1 pm) fabricated in a microforge from
a glass capillary, Neher and Sakmann 122) recorded
the current through single acetylcholine receptor
channels in denervated frog muscle. This development
permitted the direct visualization of pA (picoampere)
steps in membrane current caused by the opening
and closing of individual channels. Measurements of
elementary channel currents were previously possible
only in artificial bilayer membranes [lo].
The pace of technical advances has been rapid, with
the development of improved electrode-membrane contacts (gigaohm seals) 1341, excised patches [lZ, 141,
and ultra-low-noise instrumentation 12). In step with
these technical advances has come a proliferation of
applications of the patch electrode to innumerable
preparations 118). The recording of channel current from cells in culture is of particular importance
115, 17, 261. With the availability of cultures from
many different tissues 127) and the ease of patch
electrode recording from dissociated cells in culture,
the electrophysiological diversity of the nervous system
has been opened up to quantitative biophysical analysis. Cell cultures of the mammalian central nervous
system (CNS) 1281, including the hippocampus, should
be of value in permitting the study of the channels
involved in epileptic activity.
In this paper recent patch electrode studies of the yaminobutyric acid-activated (GABA-activated) channel
and the acetylcholine-activated channel are reviewed,
to illustrate how this technique can reveal channel
properties and elucidate gating mechanisms.
From the Department of Biology and the Mental Retardation Research Center, University of California, Los Angeles, CA 90024.
Address reprint requests to Dr Jackson, Department of Biology,
University of California, Los Angeles, 405 Hilgard Ave, Los
Angeles, CA 90024.
GABA-Activated Channels in Spinal Cord
The inhibitory action of GABA on the CNS is well
known. Cell cultures of the spinal cord, prepared from
mouse embryos [30), provide an accessible preparation
in which to probe the mechanism of GABA-induced
inhibition 12). These neurons in culture are not covered by other cells but are exposed to the bathing solution, so that a patch electrode can be positioned to be
in direct contact with the cell membrane. When a patch
electrode is filled with a dilute solution of GABA or
GABA agonist, channel currents such as those shown
in Figure 1 are recorded. Such channel currents were
Fig 1 . y-Amznobutyric acid (GABA) receptor channel cuwents
recorded from cultured mouse spinal neurons bathing in Hepes
i4-{2-bydroxyethyl}-1-piperazine-ethanestllfonicacid) bufmed
Earle's saline at room temperature (22°C). A 100 megaohm seal
was obtained. Data were filtered at 100 H z and sampled digitally at 1 kHz. (A) Channels were activated with a solution of
0.3 pki muscimol. (B)A 0.5 p M solution of G A B A was used.
(From 117) 0 1982.)
seen in essentially every patch of membrane on the
nerve cell body 1171, indicating that the GABA receptor channel is distributed throughout the membrane of
the cell body, even if this distribution is not homogeneous [23.
In these experiments cells being tested were simultaneously impaled with a microelectrode filled with 3 M
of KC1, thus increasing the intracellular C1- content.
As a result, the opening of a C1-selective channel by
GABA produces a net inward current, as C1- moves
out of the cell. This pattern is consistent with the observation that the GABA response of a cell impaled
with a KCl electrode has a reversal potential of about
-20 mV [2]. In the study from which Figure 1 was
taken 1171, the seals between the electrode and membrane were poor by current standards. Consequently,
the background current noise was high, placing a limit
on the size and duration of the channel currents that
could be detected. With improved techniques 1121,
seals were increased by a factor of 50 to a resistance of
5 gigaohms or more, so that smaller channel currents
could be recorded and the potential under the tip of
the patch electrode could be controlled with the patch
clamp amplifier (Fig 2). With the patch electrode held
at a potential of -40 mV and no impalement by an
intracellular microelectrode, the C1- driving force is
inward, producing positive channel currents (Fig 2 , upper three records). These 0.5 pA current pulses could
not have been observed under previous recording conditions.
Making the potential of the patch electrode more
negative ( - 80 mV) causes these channel currents to be
larger but also activates other types of channels that are
voltage-sensitive (Fig 2 , lower two records). The largest
channel currents seen in these records are probably
CaZ+-activated K f channels, as judged by their size
and voltage dependence and by comparison with published work on other preparations 13, 18, 21, 381.
In the earlier study of GABA-activated channels
[17), other agonists were used and channel current
amplitudes were identical to those resulting from
GABA activation. Averaging the currents of many single-channel events and dividing by an estimated driving
force indicated that the channel has a conductance of
22 picosiemens in a situation in which the intracellular
C1- content is artificially high [ 17). Channels activated
by the GABA agonists muscimol (see Fig 1A) or ( - ) pentobarbital have the same conductance.
Acetylcholine-Activated Channels i n Muscle
The larger conductance of the nicotinic receptor channel makes its currents much easier to record. It has
been established and verified repeatedly that different
agonists of the nicotinic receptor produce channel currents with the same amplitude [4, 16, 18). Channel
currents recorded in the presence of local anesthetics
were broken into bursts of openings separated by brief
closures [24]. A quantitative analysis of these bursts
indicated that local anesthetics enter and block the
channel .
Uniform channel current amplitudes are expected
when the channel is activated by different agonists.
What was surprising in the case of the acetylcholine
receptor channel was that a classic antagonist, curare,
behaved as a very weak agonist in embryonic and cultured muscle and activated channel currents of the
same amplitude as those activated by agonists. Curareactivated channels differ from agonist-activated channels in that they are of much shorter mean duration
116, 21a, 361. Because curare occupies the receptor
binding site but produces a very weak response, it still
antagonizes the stronger responses produced by conventional agonists. The action of curare appears to be
under developmental control, since only embryonic
and cultured muscle respond, but not adult muscle {39].
Gating Kinetics: The Two-State
Model and Beyond
The concept of stable protein conformations with rapid
reversible transitions between them provides a useful
Jackson: Ionic Channels in Muscle and Spinal Cord
framework that clarifies many biophysical problems.
Thermodynamic studies of protein conformational
transitions, including reversible thermal denaturation,
support the concept of a two-state model with interconversions occurring in a cooperative (or all-or-none)
fashion {20, 37). The stepwise changes in membrane
current are a graphic representation of discrete conformations that have either a high conductance (the open
state) or a low conductance (the closed state). Typically,
these states endure for milliseconds, but the transitions
between states occur in 10 +s or less C121.
Kinetic analysis of the two-state model begins by
assigning rate constants to the opening and closing processes {9]:
where C denotes a closed channel, 0 denotes an open
channel, (Y is the rate of opening, and f3 is the rate of
Although step sizes are uniform (within the range of
experimental error) durations of the closed and open
states are clearly variable and distributed. Transitions
between states are dependent upon thermal fluctuations in energ-- and are equally likely to occur at any
time. In this context, the closing rate (p) has a molecular equivalent in the probability (pdt) that an open
channel will close in the infinitesimal time interval dt.
S 5 4 Annals of Neurology
Fig 2. y-Aminobutyric acid IGABA) receptor channel currents
recorded from cultured mouse spinal neurons, as in Figure 1 . The
patch electrode waJ filled with 0.3 PM GABA. A 5 gigaohm
seal was obtained, reducing the noise level and making it possible
to increase the recording bandwidth to 500 Hz. Data were
sampled digitally at a frequency of 5 kHz. The pairs ofcalibration bars are 20 ms long and are separated by 1 PA. The upper
three tracings were made with the patch electrode clamped t o a potential of - 40 mV. The lower two tracings were made uith the
patch electrode clamped at - 80 mV.
This relationship leads to an exponential time dependence for the distribution of open times. Likewise, an
exponential distribution of closed times is expected
from two-state gating.
The earliest studies of chemically gated channels
verified this prediction 115, 24, 261 and thus appeared
to support the two-state model. As instrumentation
with wider bandwidths was developed, however, experiments with many chemically activated channels
demonstrated clearly that open times and closed times
were not exponentially distributed [5-7, 16-19).
Once the detection of short-duration events became
possible, it became apparent that the earlier, exponential distributions were only one slower component of a
more complex distribution.
Open-time (Fig 3A) and closed-time (Fig 3B) distributions derived from acetylcholine channel data from
cultured rat muscle are very well fit by a sum of two
Supplement to Volume 16, 1984
m 200
pretation. According to this explanation, a channel IS
activated prior to opening by the binding of agonist.
After a channel closes, it remains activated as long as
the agonist remains bound to the receptor. The fast
component in the closed-time distribution is thus the
reopening of a channel that has closed but has not
released its agonist. The slow component results from
the independent activation of any of the possibly hundreds of channels in the patch of membrane under the
electrode tip. Independence in the opening of different
acetylcholine channels in a patch of membrane is supported by the fact that a Poisson distribution describes
the frequency of multiple openings [ 2 3 } and is also
supported by comparisons of microscopic and macroscopic rates 1251.
In contrast to the distribution of closed times, the
double-exponential open-time distributions are more
difficult to interpret. Below are five different models
that would yield the appropriate open-time distributions. Closed-time distributions would be the sums of
two or more exponential functions, since all the models
include an activated agonist-receptor complex. In order
to distinguish among these various models, more
sophisticated methods of analysis are necessary. The
development of such methods is in its infancy, but a
few examples are given here.
Fig 3. (A)A distribution of open-state lifetimes of subeyldicholine-actzvated channels is plotted semilogarithmically (subeyldicholine concentration = 0.5 p ~ )The
. smooth curves are
the best-jitting sums of two exponentialfunctions. Arrow denotes
a time at which the probability densities ofthe two exponential
components are equal. The membrane potential was held at
- 90 mV during this experiment with two intracellular microelectrodes. (B} The distribution of closed-state intervalsfrom the
same experiment is plotted semilogarithmically for closed times of
leu than 50 m.r.
exponential functions. Double-exponential open-time
distributions have also been seen for the nicotinic receptor channel in muscle from many different species,
including frogs [ S ] , chickens [I, IS}, humans [ l b } ,
and mice (Jackson MB: unpublished data, 1983). These
distributions occur as well in other types of receptor
channels, including those of locust muscle (which are
glutamate-activated) Cb] and those of the mouse spinal
cord (which are GABA-activated) C17). Heterogeneous GABA binding sites in rat brain may be a factor in
the complexity of GABA receptor-channel kinetics
[29). Closed-time distributions that are double exponential have been seen in frog muscle 151, snake muscle 171, locust muscle 161, and mouse spinal cord (Jackson MB: unpublished data, 1983). In order to interpret
these findings, the basic concept of stable conducting
or nonconducting conformations and rapid interconversions is retained, but consideration of models with
more than two states is necessary.
Brief-duration closures have a widely accepted inter-
(Model 111)
(Model I V )
(Model V )
where R denotes the receptor, C denotes a closed
channel, 0 denotes an open channel, A denotes an
agonist molecule, and n denotes the number of agonist
Jackson: Ionic Channels in Muscle and Spinal Cord
Additional exponential components in a distribution
of dwell times result from additional stable states with
the same conductance. Although all of the above models have two open states, they differ in their number of
closed states. Models I and 111 have two closed states,
Models IV and V have three, and Model I1 has four. In
light of this variation, the mention of occasional closedtime distributions with three exponehtial components
from acetylcholine channel data in the frog is noteworthy [5]. This pattern has been confirmed in cultured
mouse muscle, with the observation of a third, very fast
component of closures, on a time scale of 50 to 100 p s
(Jackson MB: unpublished data, 1983).
Unique to Model I1 is the requirement of two separate populations of receptor channel complexes that
gate independently. If this model were correct, one
might expect the currents produced by these two different molecular species to have different conductances, giving rise to current pulses of two different
characteristic amplitudes. In acetylcholine channel studies of cultured rat muscle, an attempt was made to
detect differences in unitary conductance between the
long-duration and short-duration openings. No correlation between conductance and open time was found,
indicating that if there are two populations of acetylcholine-activated channels giving rise to open states
that close with different rates, then these populations
have identical conductances [l9].
Using a different method, the concept of different
populations was also tested for the glutamate receptor
channel of locust muscle. Some patches of locust muscle membrane were found to contain only one glutamate receptor channel. This conclusion was arrived at
for a very long record that contained many individual
channel currents but in which two simultaneous events
never occurred. Open-time distributions were still the
sums of two exponential functions, providing a con-
S56 Annals of Neurology
Fig 4. Records of suberyldicholine-activatedchannel currents in
rat muscle display two successive openinxs, which are very probably reopenings of a single channel. The membrane potential was
held at - 90 mV. Data were sampled digitally at a frequency of
10 kHz, after filtering at 1 kHz. (A) Successive slow eoents. (Bi
Successivefast events. (C) Pairs consisting of one fast and one
slow event. ( D )A single record with one fast pair and one slou,
pair. The time interval between the t w o pairs in D is long
enough to consider each pair as an independent instance ofcharinelactivation. (From {19} by copyright permission ofthe Biophysical Society.)
clusive argument for the elimination of the twopopulation model in this case {6].
Other models can be tested by investigating whether
the open times of two successive openings of the same
channel are correlated. Different models pose different
predictions as to whether successive openings of the
same channel (which are separated by very short closed
times) should be similar in open time. Examples of
pairs of events with open times that are either both
short or both long are shown in Figure 4A, B, and D.
Figure 4C displays pairs of events with no apparent
correlation in open time. Pairs of correlated events predominated in the analysis of acetylcholine receptor
channel currents in cultured rat muscle. This outcome
suggests that different activated states give rise to
short-duration and long-duration openings. This result
is consistent with the observation of three exponential
closed-time distributions [ S ] and independently serves
to eliminate Models I and I11 {19}.
Long-term correlations in open time have also been
studied in records produced by reconstituted torpedo
acetylcholine receptor channels in bilayers [19a]. It was
concluded that open states were coupled, with a twoagonist model such as Model IV likely. Perhaps future
studies of the effect of agonist concentration on opentime distributions will put this model on firmer ground.
Supplement to Volume 16, 1984
Multiple Conductance Levels
Multiple states with the same conductance are revealed
by a detailed analysis of gating kinetics, but when conducting states have different unitary conductances, a
straightforward analysis of channel current amplitudes
reveals this directly. Hamill and Sakmann [133 discovered three different conducting states activated by
acetylcholine in cultured rat muscle. Two of the conducting states were comparable in conductance to the
two different channel types seen in denervated muscle
and end-plates IS, 32). These two open states never
interconverted directly and are likely to represent different populations. A third, smaller conducting state
only occurred contiguous with events that had one of
the two larger amplitudes; in other words, this state
could only be reached by means of another conducting
state. This low-conductance state probably represents a
conducting substate that is accessible to either of the
two putative channel populations. An unusual aspect of
this substate is that it only follows and never precedes
events of larger amplitude. Curiously, the opposite sequence was found when curare activated the acetylcholine receptor channel [363. The implications of
these observations have been emphasized by Sachs
[ 3 I}, with the suggestion that acetylcholine receptor
activation is not a stationary process and must be driven
by some energy source, since it appears to violate detailed balance.
Hamill and Sakmann [13} noted that each of the
three different conducting states has a different mean
lifetime. Kinetic analysis of the two larger conducting
species was carried out to show that open-time distributions of either conducting state are double exponential 1131. In chick muscle, the acetylcholine channel
frequently flickers with brief closures, but many of
these flickers are to a conducting substate with approximately 10% of the conductance of the fully open channel [I].
Investigators are now finding other chemically gated
channels with multiple conducting states. The glycine
receptor channel in spinal cord cell culture has at least
two conducting levels { I l l . The K' channel, which
closes following the application of serotonin to Aplysza
neurons, also has a conducting substate [33}.
At the present time there is no consensus as to the
gating mechanism of the acetylcholine receptor channel, although a number of schemes have been eliminated. Methods of analysis such as those described here
may eventually converge on a single model for the
activation of the nicotinic receptor channel of skeletal
muscle. Such a model will have to include states with
similar as well as different conductance. Thus, it will be
more complicated than the model shown in Figure 4.
In the case of the GABA-activated channel, in which
different classes of drugs (such as benzodiazepines
and barbiturates) modify the kinetics of channel gating
1351 and alter the binding of GABA to its receptor
(see paper by Olsen et al in this supplement), models of substantial complexity will be necessary.
1. Auerbach A, Sachs F: Flickering of a nicotinic ion channel to a
subconductance state. Biophys J 42:l-10, 1983
2. Barker JL, Ransom BR: Amino acid pharmacology of mammalian central neurons grown in tissue culture. J Physiol (Lond)
280:331-354, 1978
3. Barrett J N , Magleby KL, Pallotta BS: Properties of single calcium-activated potassium channels in cultured rat muscle. J
Physiol ( h n d ) 331:211-230, 1983
4. Colquhoun D, Dionne VE, Steinbach JH, Stevens CF: Conductance of channels opened by acetylcholine-like drugs in muscle
end-plate. Nature 253:204-206, 1975
5. Colquhoun D , Sakmann B: Fluctuations in the microsecond
time range of the current through single acetylcholine receptor
ion channels. Nature 294:464-466, 1981
6. Cull-Candy SG, Parker I: Rapid kinetics of single glutamate
receptor channels. Nature 295:410-412, 1982
7. Dionne VE, Liebowitz MD: Acetylcholine receptor kinerics: a
description from single channel currents at snake neuromuscular
junctions. Biophys J 39253-261, 1982
8. Dreyer F, Muiler K-D, Peper K, Sterz R: The m. omohyoideus
of the mouse as a convenient mammalian muscle preparation. A
study of junctional and extrajunctional acetylcholine receprors
by noise analysis and cooperariviry. Pfluegers Arch 367:1 15122, 1976
9. Ehrenstein G, Blumenrhal R, Latorre R, Lecar H: l n e t i c s of the
opening and closing of individual excitability-inducing material
channels in a bilayer. J Gen Physiol 63:707-721, I974
10. Ehrenstein G, Lecar H: Electrically gated ionic channels in lipid
bilayers. Q Rev Biophys 10:l-34, 1977
11. Hamill OP, Bormann J, Sakmann B: Single glycine-activated
chloride channels in spinal cord neurons display anomalous selectivity for anions. Biophys J 41:388a, 1983
12. Hamill OP, Marty A, Neher E, et al: Improved patch-clamp
techniques for high resolution current recording from cells and
cell-free membrane patches. Pfluegers Arch 39135-100,
13. Hamill OP, Sakmann B: Multiple Conductance states of single
acetylcholine receptor channels in embryonic muscle cells. Nature 2941462-464, 1981
14. Horn R, Patlak JB: Single channel current from excised patches
of muscle membrane. Pmc Natl Acad Sci USA 77:6930-6934.
15. Jackson MB, Lecar H: Single post-synaptic channel currents in
tissue-cultured muscle. Nature 282:863-864, 1970
16. Jackson MB, Lecar H, Askanas V, Engel WK. Single cholinergic receptor channel currents in cultured human muscle. J
Neurosci 2:1465- 1473, 1982
17. Jackson MB, Lecar H, Marhers DA, Barker J L Single channel
currents activated by GABA, muscimol, and ( - ) pentobarbital
in cultured mouse spinal neurons. J Neurosci 2:889-894, 1982
18. Jackson MB, Lecar H , Morris CE, Wong BS: Single channel
recording in excitable cells. In Barker JL, McKelvey JF (eds):
Current Methods in Cellular Neurobiology. New York, Wiley,
1983, pp 61-97
19. Jackson MB, Wong BS, Morris CE, et al: Successive openings of
the same acetylcholine receptor-channel are correlated in their
open-times. Biophys J 42:109-114, 1983
19a. Labarca P, Lindstrom J, Montal M: The acetylcholine receptor
Jackson: Ionic Channels in Muscle and Spinal Cord
channel from Torpedo ralifornica has two open states. J Neurosci
4:501-507, 1984
20. Lumry R: Some fundamental problems in the physical chemistry
of protein behavior. In King TE, Klingenberg M (eds): Electron
and Coupled Energy Transfer in Biologicd Systems. New York,
Marcel Dekker, 1971, pp 1-116
21. Many A: Ca-dependent K channels with large unitary conductance in chromaifin cell membranes. Nature 291:497-500,1981
2la. Morris CE, Wong BS, Jackson MB, Lecir H: Single-channel
currents activated by curare in cultured embryonic rat muscle. J
Neurosci 3:2525-253 1, 1983
22. Neher E, Sakmann B: Single channel currents recorded from
membrane of denervated frog. Nature 260:799-802, 1976
23. Neher E, Sakmann B, Steinbach JH: The extracellular patch
clamp: a method for resolving currents through individual open
channels in biological membranes. Pfluegers Arch 375:2 19228, 1978
24. Neher E, Steinbach JH: Local anaesthetics transiently block currents through single acetylcholine-receptor channels. J Phgsiol
( b n d ) 277:153-176, 1978
25. Neher E, Stevens C F Conductance fluctuations and ionic pores
in membranes. Annu Rev Biophys Bioeng 6:345-381, 1977
26. Nelson DJ, Sachs F Single ionic channels observed in tissuecultured muscle. Nature 2822361-863, 1979
27. Nelson PG, Lteberman M: Excitable Cells in Tissue Culture.
New York, Plenum, 1981
28. Nelson PG, Neale EA, Macdonald RL: Electrophysiological and
structural studies of neurons in dissociated cell cultures of the
central nervous system. In Nelson PG, Lieberman M (eds): Excitable Cells in Tissue Culture. New York, Plenuin, 1981, pp
Annals of Neurology
29. Olsen RW, Bergmdn MO, Van Ness PC, et al: y-Aminobutyric
acid receptor binding in mammalian brain: heterogeneity of
binding sites. Mol Pharmacol 19:217-227, 1980
30. Ransom BR, Neale E, Henkart M, et al: Mouse spinal cord in
cell culture. I. Morphology and intrinsic neuronal electrophysiologic properties. J Neurophysiol 40:1132-1150, 1977
31. Sachs F Is the acetylcholine receptor a unit conductance channel? In Neher E, Sakmann B (eds): Single Channel Recording in
Biological Membranes. New York, Plenum, 1983
32. Sakmann B: Acetylcholine-induced ionic channels in rat skeletal
muscle. Fed Proc 37:2654-2659, 1978
33. Siegelbauin SA, Camardo JS, Kandel ER: Serotonin and cyclic
AMP close single K + channels in Ap&u sensory neurons. Nature 299:413-417, 1982
34. Sigworth FJ, Neher E: Single Na' channel currents observed in
cultured rat muscle cells. Nature 287:447-449, 1980
35. Study RE, Barker J L Diazepam and ( - )pentobarbital: fluctuation analysis reveals different mechanisms for potentiation of
gamma-aminobutyric acid responses in cultured central neurons.
Proc Natl Acad Sci USA 78:7180-7184, 1981
36. Trautman A: Curare can open and block ionic channels associated with cholinergic receptors. Nature 298:272-275, 1982
37. Tsong TY, Hearn RP, Wrathall DP, Sturtevant JM: A
calorimetric study of thermally induced conformational transitions of ribonuclease A and certain of its derivatives. Biochemistry 9:2666-2677, 1970
38. Wong BS, Lecar H , Adler M: Single calcium-dependent potassium channels in clonal anterior pituitary cells. Biophys J
391313-320, 1982
39. Ziskind L, Denis JJ: Depolarizing effect of curare on embryonic
rat muscles. Nature 276622-623, 1978
Supplement to Volume 16, 1984
Без категории
Размер файла
633 Кб
channel, spina, muscle, chemical, cord, activated
Пожаловаться на содержимое документа