Bioelectricity of Cell Membranes

Content
Part I Electrical Properties of Membranes

The Origin of Membrane Potentials

Action Potentials

Membrane Currents

Part II Ion Channel Properties

Channel Kinetics

Ion Selectivity

Channel Blockers and Toxins

Part III Nerve Impulse Encoding and Propagation

Axonal Propagation and Myelination

Encoding of Action Potentials

Synaptic Transmission

Channels and Diseases

Bioelectricity deals with cell membrane transport processes that control the formation
and dissipation of ion gradients. Ion gradients store energy in form of an electrochemical
potential. This energy can be converted into other forms of energy. The electrochemical
potential is available to organisms for biosynthesis (photosynthesis and respiration), transport
of metabolites (absorption and secretion), mechanical work (bacterial flagella rotor,
swimming, crawling), and signaling processes (action potentials). Action potentials are a form
of information used by electrically excitable membranes to control the activity of cells
(calcium signaling, muscle contractility) and to support or suppress communication between
cells (release of chemical signaling molecules; hormones, neurotransmitters).

Part I. Electrical Properties of Membranes

The Origin of Membrane Potentials

It is important to understand the origin of membrane potentials. Biological membranes

are electrical insulators due to their phospholipid bilayer structure and are impermeable to
ions, unless specific ion channels are temporarily open. In real cells, several different ion
types, each with its own gradient contribute to this charge separation. Any ion that forms an
ion gradient across a membrane, and that is permeable contributes to the actual membrane

potential. At rest, most cells have a potential around -40 to -80mV indicating that they are
dominated by K or Cl permeability (see table).

TABLE

Ion distribution across mammalian skeletal muscle membrane

Ion

Intracellular

Extracellular

Gradient

Nernst Potential

Na+

12 mM

145 mM

12 fold

+67 mV

K+

155 mM

4 mM

Ca++

0.001 mM

1.5 mM

15,000 fold

+129 mV

Cl-

4 mM

123 mM

29

-90 mV

0.0026 fold (39

fold)

-98 mV

While ion gradients are the result of pumps that move ions across membranes
in an ATP dependent fashion (e.g. the Na/K-ATPase) charge separation is the result of
ion selective channels used for rapid signaling events or prolonged voltage
stabilization (resting conditions). In general, transport processes are described by
diffusion (diffusion coefficient D). For cell membranes diffusion is a two step process
of moving ions from one aqueous compartment to an other across and within the
hydrophobic core of the phospholipid bilayer. This is best characterized by the oilwater partition coefficient K (moving in and out of the bilayer) and the diffusion
coefficient D within the bilayer. It is important to realize that the entire gradient spans
only the thickness of a cell membrane, a distance of about 4-6 nm. Since measuring
these parameters is fairly difficult to measure, a more accessible parameter, the
permeability coefficient P, can be defined that includes both K and D. Ion gradients
can be mimicked, manipulated, and measured by electrodes attached to current
generators and volt meters by a type of experimentation summarily known as
electrophysiology. This can been done by modeling membranes as electrical circuits
with resistance, capacitance, and charge (as in a battery) and correlated this circuit
elements with structural features of cell membranes. Whatever the goal of an
investigation may be, the important point is to define properties that can be measured
experimentally. For membranes this includes the membrane potential (E) and
membrane currents (I). Form these parameters we can calculate resistance (R), or its
inverse the conductance (g), and permeability ratios. One basic equation, Ohm's law,
describes the relationship between these parameters. It has been derived from flux rate
analysis and, for charged particles:

E = RI or I = gE
Ohm's law
Concentration gradients of charged ions determine the membrane potential E
as outlined above and the two parameters can be related using the thermodynamic or
chemical equilibrium of transport phenomena. The equilibrium potential for a single
ion species across a thin membrane separating two compartments with unequal
concentration is described by the Nernst potential:
E(eq) = [RT/zF]ln{C(out)/C(in)}
Nernst equilibrium potential
R is the gas constant, T the actual temperature, z the number of charges of the
ion species, F the Faraday constant, and C(out) and C(in) the extracellular and
intracellular concentrations of the ion. Such a potential forms in the presence of a
semi-permeable membrane. Consider that an ion species, e.g. K+ always comes with
an anion neutralizing its charge. However, the two counter ions can be separated
physically by some distance without violating the electroneutrality requirement of a
salt solution. If the membrane separates an ion gradient, a potassium selective (semipermeable) membrane would form a diffusion potential across the thin partition. The
strength of this diffusion potential is captured by the Nernst equation, if the ion
selectivity is absolute, i.e., the membrane is permeable for potassium ions only, but
impermeable for the anion. Such absolute selectivities are never found in real cell
membranes and potential equations are more complex than indicated for a single ion
species. The most basic voltage equation that captures the relationship between ion
gradients and membrane potential of excitable membranes (in neurons and muscle
cells) is the Goldman-Hodgkin-Katz voltage equation, or GHK. It quantifies the
relationship between equilibrium voltage, permeabilities and concentrations of Na, K,
and Cl ions.
One particular situation arises in the presence of a charged macromolecule (=
poly-electrolyte) like a protein or nucleic acid on one side of the membrane. Due to
electroneutrality consideration and the impermeability of the macromolecules, the
latter affects the distribution of the smaller, diffusible ions (K, Na, Cl, Ca etc.)
attracting the counterion to its side, while repelling the ion with the same charge. In
the most simple case, a negatively charged macromolecule P on one side of a Na and
Cl permeable membrane will cause the Na concentration to be slightly higher and the
Cl concentration in the compartment with the macromolecule. The opposite side will

have an equal amount of Cl and Na ions, yet both ions will have a gradient of equal
but opposite strength such that:
Na(out)=Cl(out) & Na(in)=Cl(in)+zP(in)
Na(out)/Na(in) = Cl(in)/Cl(out) --------------Donnan ratio
The potential associated to this gradient is called Donnan potential or
equilibrium. It is important to keep in mind that actual biological transport processes
are not at their chemical equilibrium but rather follow a steady-state equilibrium and
are best described as homeostatic systems. Thus all formal analysis using Nernst
potentials are approximations because they assume equilibrium or near equilibrium
conditions. The actual processes can be estimated quite well from these
approximations. Another important concept to keep in mind when analyzing
bioelectric phenomena is the difference between macroscopic and microscopic
processes. Macroscopic systems involve a very large number of units in a system and
any quantification thereof describes averaged behavior. Concentration, voltage,
current, flux rate, temperature are macroscopic properties of systems. Having an
extraordinary good insight into molecular structures allows us to also analyze systems
at their microscopic level. This type of analysis describes the behavior of individual
molecules and proteins and their behavior is statistical and probabilities can be
assigned to individual events. A good example (discussed below) is the open
probability of an ion channel. Probabilities multiplied by the number of particles in a
system must equal the macroscopic behavior. Thus, structural presentation
(microscopic) of a biological system can be linked to experimental observation
(macroscopic) allowing complete physical description of mechanisms. Historically, it
is thus no surprise that in the 1940s when electrophysiologists begun studying nerve
potentials and currents readily proposed unit conductances (the quanta was already
well established in the physical chemists mind) that could be decoded from the noise
spectra of macroscopic currents. These unit conductances, it was understood, were
most likely the result of individual protein channels that could switch between open
and closed states. Experimental proof of the existence and structural features of these
proteins took several decades.

Action potentials
Cell membranes have stable potentials (resting potentials) that depend on the
gradients of permeable ions and in excitable cells can be induced to form selfpropagating, dynamic action potentials. An action potential can be induced when the
membrane potential changes electrotonically reaching a threshold needed to trigger an
action potential. Electrotonic potential changes are passive and characterized by the
time and space constants of the membrane and the membrane conductance does not
change as the potential changes.
TABLE Cell Membrane Potentials
Type

Propagation

Property

Information encoding

Electrotonic

Passive, graded

Time & space constant

Amplitude

Equilibrium*

Threshold

Action

Self-propagation Activation and

Potential

inactivation kinetics

Frequency
Duration

Non-equilibrium
No cell membrane potential is ever at chemical equilibrium, but at a steady-state
equilibrium.

Action potentials, however, are the result of ion flow through voltage gated
channels. The number of open channels changes as the membrane potential changes
and thus the membrane conductance changes as well. This phenomenon is known as
rectification. As a result, the increasing number of open pores increases the ion flow
(current) disproportional to Ohm's law. And as the membrane potential is a function of
ion flow, and the ion flow a function of the membrane potential, cell membranes with
voltage gated ion channels form dynamic systems (here action potentials) that can
best described as feedback coupled loops that 'oscillate' and due to the spatial
distribution of channels behave like self-propagating waves of oscillating shifts of
membrane potentials. One of the most important feature of action potentials is the
kinetics of depolarization (making the membrane potential more positive by moving
Na+ through Na-channels into the cell) and hyperpolarization (making it more
negative by moving K+ through K-channels out of the cell; note that the sign of the
voltage always refers to the inside or cytoplasmic side of the cell membrane). The
kinetics of the action potential, how fast it depolarizes and hyperpolarizes and how

fast it can propagate along a cell membrane depends on two physical features of cell
membranes, i.e., its resistance Rm and capacitance Cm. Basically, the entire process
depends on the ability of certain membrane proteins to undergo conformational
changes in a changing electric field strength. This conformational changes as a
function of voltage is known as voltage gating and affects both activation (pore
opening) and inactivation (pore closure) of these channels. It is clear that both
processes are not independent of each other and form a dynamic feedforward and
feedback loop system driving the potential change like a wave along an axonal
membrane.
The Hodgkin cycle describes such a dynamic self-referential system as the Na
channel dependent upstroke of an action potential. Inward currents carried by sodium
ions cause the membrane to depolarize. The depolarization causes more Na-channels
to open increasing the sodium conductance which leads to more inward currents and
more depolarization. The action potential does not rise indefinitely for three reasons.
First, the activated Na-channels rapidly inactivate. Second, voltage sensitive Kchannels are being activated allowing K ions to flow outward, counter balancing the
influx of positive charges by moving positive charges out of the cell. Third, the Na
currents become weaker as the membrane potential depolarizes towards the Nernst
potential of Na which measures +55mV. There, the sodium current becomes zero and
reverses if the potential becomes even more positive. The Na current would contribute
to hyperpolarization of the membrane in conjunction with the K current. The passive
time course of charging/discharging a membrane is the result of its capacitance and
resistance. When applying a current pulse to a membrane by means of a
microelectrode, the resulting change in membrane potential (charging) follows an
exponential time course (natural logarithm) expressed as time constant tau which is
proportional to the resistance and capacitance.
= RC
The smaller the resistance (i.e. many ion channels open) and the smaller the
capacitance (local current circuit) the faster the charge/discharge kinetics.
V = V(eq)exp{-t/RC)
In the 1940s when the first electrophysiology experiments on giant axons from
squids were performed, the replacement of various ions quickly demonstrated that
both Na and K currents are responsible for action potentials. Over the next decades,

the role of ion channels in action potentials, and for that matter all membrane
potentials, has been well established. Besides artificially changing ion compositions
and concentrations showing the existence of several channel proteins with unique ion
selectivity, the use of pharmacological agents, mostly neurotoxins from snails, snakes,
spiders, fish and plants, and now also synthetic analogues has helped elucidate most
of the mechanisms underlying electrical excitability of neurons and muscle cells.

Membrane currents
Measuring membrane currents instead of potentials has been the way of
understanding mechanisms underlying action potentials. Membrane currents are the
result of opening ion selective channels which causes ions to flow across cell
membranes. This flow is spontaneous because all ion types are distributed unevenly
between cellular and extracellular compartments. In general, cell contain high loads of
K+, but low Na+ and Ca2+ ions, while extracellular fluids contain high Na+ and Ca2+
ions, but low K+ concentrations. Accordingly, ion gradients ranging from 10 to 10,000
fold depending on the ion species exist. When channels are activated, ions will always
start diffusing through the pores in either direction, although more ions will flow from
the high to the low concentration (down hill). These ion diffusion is an important part
of bioelectricity maintaining resting potentials and generating action potentials. It is
also used to couple the transport of secondary solutes that can be upconcentrated
inside or outside according to metabolic needs. Finally, ATP hydrolyzing pumps
reverse the flow of ions regenerating the gradients dissipated by the activity of
channels and secondary transporters. Summarily, chemical energy is used to maintain
the formation and use of membrane potentials and ion gradients. While almost all
transporters somehow involve the flux of ions across membranes, ion channels are
unique in their fast kinetics facilitating the flow of up to 10 million ions per second.
Pumps work at a roughly 10,000 fold slower rate. Despite the impressive flux rate
through ion channels, ion gradients are not dissipated quickly because ion channels
stay open only for milliseconds at a time. Prolonging their open state usually causes
sever stress on cells and eventual cell death. Small, ion channel forming peptides from
microorganisms but also animals are used as defense mechanism because they can
penetrate membranes of competitors or pathogenic microorganisms forming
permanent ion channels in host cell membranes destroying the cells. Examples are

bacterial Gramicidin A, bee venom mellittin, frog skin antimicrobial magainins, and
intestinal defensins, ionophoric peptides of animals that serve as a first line of defense
against pathogenic bacteria. The methods to study membrane currents are voltage
clamp (two electrodes) and patch clamp (one electrode) techniques. The latter allows
the measurement of currents through single channel units, while the former is used to
measure macroscopic currents, which are the result of the simultaneous activity of
hundreds to thousands of channels. The noise recorded in the early days of
electrophysiology indicated the presence of unit conductances which later have been
correlated with the presence of ion channels, the physical units of electrical
conductivity in cell membranes. Today, the activity of these channels, their
distribution and regulation is well described and one of the few protein systems that
can be studied at the single unit level. Accordingly, single channel recordings have
allowed the detailed characterization of kinetic properties and the opening and closing
of these proteins. High resolution structural analysis has corroborated the existence of
these channels, their channel structure, the gating mechanism, ion selectivity, voltage
sensing elements, and ligand binding sites (e.g. where neurotransmitters can bind and
activate the channel). The correlation between single channel behavior and
macroscopic currents is straight forward. The latter are the summation of single
channel activities such that the macroscopic conductance g is the product of the single
channel conductance times the number of channels in the membrane studied. The
macroscopic current is proportional to the product of the single channel conductance,
times the number of channels, times the open probability and the effective driving
force.
I = g(E-En) = NPo(E-En)
where I is the membrane current, N the number of channels, g the membrane
conductance, the single channel conductance, Po the open probability of the single
channel, E the membrane potential and En the Nernst potential of the ion species
selective for the channel. The single channel conductance is an intrinsic property of a
protein (channel) and differs with different ions used but is independent of the voltage
or ion gradient. Membranes with multiple copies of the same channel show multiple
distinct conductance levels of equal size. This observation has led to the concept of
the unit conductance, one of several physical parameters by which ion channels can
be distinguished (other parameters would be ion selectivity and pharmacological

profile). An other physical property that appears to be intrinsic to specific channel

type is the probability of the channel being open and closed. Voltage-gated channels
change their open probability with the membrane potential, a property known as
rectification. In ligand gated channels, this probability increases with the proper
ligand (e.g. neurotransmitter) bound to the channel. Yet other channels react to
changes in pressure or temperature and function as mechanosensing or temperature
sensing elements (e.g. in nociception, the perception of physiological pain).

Channel kinetics
The surprising observation is that voltage gating behavior found in current
recordings can be kinetically described (fitting an equation to the observed
macroscopic current of a Na current, for example) and correlated to structural
elements of these proteins. Thus activation and inactivation processes are the result of
structural changes in Na and K channels that response to changes in the electric field
strength of a changing membrane potential. The structural elements of these proteins
are referred to as voltage sensors that rely the stimulus to other structural components
called the activation and inactivation gates. Sensors and gates are small domains in
these channel proteins that shift their position affecting the quaternary structure of
these multi-subunit and multi-domain proteins. Some gates function like a diaphragm
while others literally swing in and out of an ion conducting pathway through the
protein allow ions to flow across the membrane when both the activation and
inactivation gates are open, but not when either one of the gates is closed. Careful
analysis of the kinetics and structural features of voltage-gated ion channels has
shown that they contain an activation gate made of four voltage sensors that control
the opening and closing of the pore, and one (Na-channel) or four (K-channel)
inactivation gates. The movement of the voltage sensors in these channels depends on
several positively charged amino acids that move in synchrony toward the outside and
inside surface of the membrane. The movement of these positive charges can be
measured electrophysiologically in the absence of mobile ions and is known as gating
current. The inactivation gate is known as 'ball and chain' module where the Nterminal domain of the channel binds to the intracellular opening of the pore blocking
ion flow. The different amino acid sequences of different Na and K-channels can
explain the differences in the gating behavior (i.e. kinetics) of these channels. Both

types of channels are encoded for by many different genes. The voltage gated Kchannels comprise a family of at least 50 different isoforms ranging from delayed
rectifier (slow activation, very slow inactivation) to the A-type K-channel (fast
activation, fast inactivation). The kinetic analysis of single channel recordings has
shown that channels, once activated, rapidly switch between an open and closed state,
before they inactivate and stay silent until a new trigger activates them. These
transitions between conducting and non-conducting states in an active conformation
are randomly distributed and are well described by stochastic models. A stochastic
model predicts that an event such as an opening transition is independent of its
previous event (a closing). No definite mechanism for this opening and closing
transitions have been found. Likely explanations are rapid conformational changes
due to intrinsic internal motion of atoms or even kinetic models of oscillating changes
of ion flow as the result of dynamic feed back loops between local fixed and mobile
charges. The local transmembrane potential is affected by nearby mobile charges,
while the number of mobile charges is affected by local surface potentials. As a result,
local ion concentrations can drop temporarily below the resolution limit of modern
electrophysiology equipment (about 0.1pA).

Ion Selectivity
All ion channels show selectivity and prefer certain ions while rejecting
others. No channel has an absolute selectivity for a single ion species. Three basic
types of selectivity mechanisms are found. First, channels can be size selective where
the size refers to the hydrated ion. Few channels depend solely on this size selectivity,
but noted exception are the connexin forming gap junction channels bridging the
double membrane structure of adjacent cells and mitochondrial and bacterial porins
which serve as molecular sieves with a working cutoff size of 600 Dalton Gap
junctions have a cutoff size selectivity around 1,000 Daltons. Thus, small metabolites
like ions, sugars, nucleotides and amino acids can freely diffuse between neighboring
cells. This process is known as metabolic coupling. Second, channels can be cation or
anion selective using mechanisms of electrostatic screening. A well understood
example is the nicotinic acetylcholine receptor which has a ring of negatively charged
glutamate residues on each side of the channel attraction cations but repelling chloride
ions. Third, most ion channels are ion selective such that they can discriminate

between sodium, potassium, or calcium ions, all cations. Obviously, a simple

electrostatic model cannot account for this selectivity. The high resolution structure of
a bacterial potassium channel (KcsA; non-voltage gated) with a conserved pore
structure with voltage gated K-channels shows that the difference in dehydration
energy of hydrated K+ and Na+ ions favors the binding of K+ ions in what is called
the selectivity filter. This structure comprises a narrow portion of the pore that
requires ions to lose their hydration shell. The protein surface mimics the hydration
shell with a series of backbone carbonyls pointing toward the channel center. These
carbonyls structurally mimics the oxygen binding states of water molecules in a
hydration shell. Nicotinic acetylcholine receptor pores are wider and both K and Na
ions can diffuse across the pore without stripping off their hydration shell.

Synaptic Transmission
Action potentials are generated on cell bodies where they form contact with
innervating cells through synaptic buttons. These buttons are cell extensions that
synthesize and store chemical neurotransmitters in small organellar vesicles. Upon an
action potential stimulus arriving via the axonal membrane, voltage gated calcium
channels trigger a signaling cascade using calcium ions that trigger membrane fusion
between the storage vesicles and the presynaptic membrane. An exocytotic event
releases the stored chemicals into the synaptic cleft where the rapidly diffuse towards
and bind to ligand-gated receptors. Receptor activation triggers a post-synaptic
membrane current in the receiving cell charging the potential to different levels. Thus
the electrical signal is transmitted chemically between neurons and neurons and
muscle cells. Synapses can stimulate both depolarizing and hyperpolarizing currents
depending on the neurotransmitter and their respective receptor used. Activating
cation selective ion channels causes excitatory (depolarizing) post-synaptic potentials
(EPSP) while activation of anion selective ion channels causes inhibitory
(hyperpolarizing) post-synaptic potentials (IPSP). Chemical synapses produce socalled miniature endplate potential (called this way because they have first been
characterized on the endplate regions of neuromuscular junctions) that are the result
of a single pulse of exocytotic event. Each vesicle releasing neurotransmitter contains
more or less the same amount of ligands activating a certain number of post-synaptic
receptors which results in a uniform depolarizing pulse - the miniature endplate

potential (MEPP). Usually, neurotransmitters from dozens of vesicles are released

resulting in the local summation of many MEPPs eventually reaching threshold of the
post-synaptic membrane that causes voltage gated Na-channels to open. The chemical
transmission is complete. Due to the nature of this quantized neurotransmitter release
and a very large number of post-synaptic receptors clustered in the synaptic region
high frequency signals from stimulating neurons result in temporal or spatial
summation of MEPPs when a neuron is innervated by several synapses coming from
the same or different neurons, respectively. Summation can also be the result of
balancing excitatory and inhibitory synapses such that several neurons can control the
firing pattern of a receiving neuron which integrates the signals coming from different
regions of the body/brain. The complexity of neuronal signaling using combinations
of excitatory and inhibitory synapses is indeed great and the numbers of active
synapses or how strongly they can signal can be modified by metabolic feed back
mechanism. This variability is known as synaptic plasticity and is invoked in models
of learning and memory. Some synapses are electrically rather than chemically
coupled. These electrical synapses are composed of large gap junction plaques that
contain thousands of gap junction channels. These channels are cell-to-cell channels
spanning across two adjacent cell membranes directly connecting the cytoplasmic
compartments of two neighboring cells. Gap junctions are found in all cells that are
interacting with neighboring cells, not just neurons. Their major function is metabolic
coupling of dozens of cells that can rapidly exchange metabolites without using extracellular exchange routes. A second function of gap junction is the electrical coupling
of cell membranes. Thus, action potentials generated on one cell can rapidly spread to
neighboring cells without extracellular chemical signaling molecules. Electrical
communication is an order of magnitude faster than chemical synaptic transmission.
The mechanisms of metabolic and electrical coupling are not completely understood
particularly regarding the ability of many gap junctions to promote unidirectional
information flow.

Channels and Diseases

Malfunctioning channels can be the cause of many diseases. Voltage gated ion
channels contribute to arrhythmias in the heart because of their involvement in
shaping action potentials and firing patterns of neurons and muscle cells.

Understanding diseases requires a step up from individual cells to larger tissue and
organ systems. Arrhythmias particularly are not a property of single cells but the
entire heart pumping in carefully controlled time periods. Electrocardiograms capture
the propagation of currents flowing across membranes in different parts of the heart
during the pumping cycle. The EKG starts with the P-wave as the result of
depolarization of atrial cells followed by the QRS-complex measuring the current
spreading through the ventricles. The cycle ends with the T-wave that is characteristic
of the repolarization of the ventricle. The Long QT syndrome is diagnosed as an
abnormal extension of the QRS complex to the end of the T wave in an
electrocardiogram (extracellular recording) and that most cases are caused by
mutations in a subunit of the ventricular K-channel I(Ks). In this cardiac membranes,
I(Ks) is a heteromeric channel mixture composed of both delayed rectifier and shakertype (transient A-currents) subunits giving rise to the characteristic firing pattern of
ventricular membranes. Delayed repolarization of the ventricle increases the
refractory period of the ventricular myocardium delaying its subsequent
depolarization. The wave of excitation may then 'pursue a distinctive pathway around
a focal point in the myocardium. A variety of myotonias (dysfunctional coordination
of muscle contraction-elongation) can be traced to defects in voltage gated chloride
channels. These channels control the time interval between action potentials since
inward

chloride

currents

(due

to

their

negative

charge)

contribute

to

hyperpolarization. When chloride channels are defective or blocked, slow afterhyperpolarization potentials disappear and cells begin to fire rapidly during a
depolarizing inward current that can lead to muscle stiffness. Gap junctions are found
in large numbers in heart myocardium that allows the rapid spreading of electrical
signals from neuromuscular junctions across large muscle fibers. They are also found
in myelin sheets where they contribute to electrical coupling and K+ current circuit of
action potential propagation. Mutations in connexins (the proteins that make gap
junction channels) cause demyelination and are known to affect propagation behavior
of peripheral neurons. A known peripheral neuropathy related to connexin mutations
is Charcot Marie Tooth syndrome progressive degeneration of the muscles in the foot,
lower leg, hand, and forearm. Some forms of hearing loss are also related to connexin
mutations. In both diseases, the lack of gap junction coupling between cells causes a
disruption of local K+ currents essential for the maintenance of electrical excitability
of neurons or hair cells.