Acta Physiol 2009, 195, 171–186
REVIEW
The lateral intercellular space as osmotic coupling
compartment in isotonic transport
E. H. Larsen,1 N. J. Willumsen,1 N. Møbjerg1 and J. N. Sørensen2
1 Department of Biology, August Krogh Institute, University of Copenhagen, Copenhagen, Denmark
2 Department of Mechanical Engin1eering, Technical University of Denmark, Lyngby, Denmark
Received 17 June 2008,
accepted 8 August 2008
Correspondence: E. H. Larsen,
Department of Biology, August
Krogh Institute, Universitetsparken
13, DK-2100 Copenhagen Ø,
Denmark.
E-mail: [email protected]
Abstract
Solute-coupled water transport and isotonic transport are basic functions of
low- and high-resistance epithelia. These functions are studied with the
epithelium bathed on the two sides with physiological saline of similar
composition. Hence, at transepithelial equilibrium water enters the epithelial
cells from both sides, and with the reflection coefficient of tight junction
being larger than that of the interspace basement membrane, all of the water
leaves the epithelium through the interspace basement membrane. The
common design of transporting epithelia leads to the theory that an osmotic
coupling of water absorption to ion flow is energized by lateral Na+/K+
pumps. We show that the theory accounts quantitatively for steady- and time
dependent states of solute-coupled fluid uptake by toad skin epithelium. Our
experimental results exclude definitively three alternative theories of epithelial solute–water coupling: stoichiometric coupling at the molecular level
by transport proteins like SGLT1, electro-osmosis and a ‘junctional fluid
transfer mechanism’. Convection-diffusion out of the lateral space constitutes the fundamental problem of isotonic transport by making the emerging
fluid hypertonic relative to the fluid in the lateral intercellular space. In the
Na+ recirculation theory the ‘surplus of solutes’ is returned to the lateral
space via the cells energized by the lateral Na+/K+ pumps. We show that this
theory accounts quantitatively for isotonic and hypotonic transport at
transepithelial osmotic equilibrium as observed in toad skin epithelium
in vitro. Our conclusions are further developed for discussing their application to solute–solvent coupling in other vertebrate epithelia such as small
intestine, proximal tubule of glomerular kidney and gallbladder. Evidence is
discussed that the Na+ recirculation theory is not irreconcilable with the wide
range of metabolic cost of Na+ transport observed in fluid-transporting
epithelia.
Keywords isotonic transport, Na+ recirculation theory, physical-mathematical modelling of epithelial transport, metabolic cost of active sodium
transport, solute-coupled fluid transport, toad skin epithelium.
Solute-coupled water transport is a basic function of
transporting epithelia. Associated herewith is the
capacity to produce a transportate that is isotonic
with the external solutions of identical composition.
Transepithelial movement of bulk water against an
adverse osmotic gradient (uphill) is another general
feature of transporting epithelia. Elucidating the mechanisms has been a major challenge for understanding
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
171
Solute-coupled water transport
Æ E H Larsen et al.
fluid handling by kidney, nutrient absorption by the
digestive system, the respiratory defence system of
upper airways, formation of cerebrospinal fluid and
the composition of secretions by exocrine glands.
Before dealing with the question of how the epithelium
at transepithelial osmotic equilibrium transports fluid
of adjusted tonicity, clarification of the nature of the
mechanism that couples transepithelial ion fluxes with
the flow of water is required. Four mechanisms have
been suggested and are still considered in the literature: (1) osmotic coupling in intraepithelial compartments (Curran 1960, Curran & MacIntosh 1962,
Diamond & Bossert 1967); (2) stoichiometric coupling
at the molecular level by specific transport proteins as
for example the Na+-glucose transporter, SGLT1 (Loo
et al. 1996, Meinild et al. 1998); (3) electro-osmotic
coupling in a channel-forming protein of tight junctions (Fischbarg et al. 2006), and (4) a paracellular
junctional fluid transfer mechanism (Hill & ShacharHill 2006). The primary aim of the present review is to
discuss previously published and new experiments on
the coupling mechanism in a tight epithelium. We
provide compelling evidence that the lateral intercellular space is the coupling compartment which
osmotically forces water through the epithelium energized by lateral Na+/K+ pumps. Our studies lead to the
more general conclusion that the three other coupling
mechanisms listed above (2–4) can be excluded definitively for this type of epithelium. Finally, our conclusions are developed for discussing implications for
solute-coupled fluid absorption by other vertebrate
epithelia.
clamp), or by the voltage response to a 10 lA current
pulse (current clamp). The current carried by the active
Na+ flux was measured as the short circuit current, ISC
(VT = 0) or expressed as the equivalent short-circuit
current defined by:
Eqv
ISC
¼ GT VT :
where VT is the spontaneous transepithelial potential
difference (IT ¼ 0)
In open circuit mode (IT = 0) the active Na+ flux is
accompanied by a flux of Cl) of similar magnitude.
With the transepithelial water flow (JV) measured
simultaneously, the osmotic concentration of the transported fluid could therefore be estimated by:
Eqv
2 ISC
=ðF JV Þ
Eqv
ISC
=F always tends to overestimate the active Na+ flux
at open circuit conditions, and the more so in
preparations with a large shunt resistance, i.e. a low
Cl) conductance. Hence, the true osmotic concentration of the transported fluid is less than the estimates
given by the method above. At standard physiological
conditions the overestimation amounts to 10–12%
(Larsen et al. 2007). Our major conclusion is that
the epithelium has the capability of producing an
isotonic as well as a hypotonic absorbate. Obviously,
this conclusion is uninfluenced by the overestimation
mentioned.
Physical–mathematical modelling of epithelial transport
The mathematical description of Larsen et al. (2002,
2006) was introduced in the present study with the
following modifications:
Method
Recording of water volume flow and ion transport
The epithelium of toad skin (Bufo bufo) was isolated
by collagenase according to the method of Willumsen
et al. (1992). Transepithelial fluid flow (JV), current
(IT, inward current positive) and voltage (VT, serosal
bath grounded) were recorded in preparations exposed
to amphibian Ringer’s solution on both sides (mm),
117.4 Na+, 2.0 K+, 1.0 Ca2+, 114 Cl), 2.4 HCO3),
5 CH3COO), 5 glucose with a nominal osmotic
concentration of 252.8 mOsm, and pH = 8.2 when
equilibrated with atmospheric air (Nielsen & Larsen
2007). The osmotic concentrations of the bathing fluids
were measured before and after each experiment using
a VAPRO-5520 vapour pressure osmometer (Wescor,
Logan, UT, USA) to check that the osmolarity of the
bathing solutions was not changed to any measurable
extent during the experiment. The transepithelial conductance (GT) was calculated by the response of the
clamping current to a 10 mV voltage pulse (voltage
172
Acta Physiol 2009, 195, 171–186
(1) In the Na+/K+ pump flux equations the reversal
potential of the pump (Epump
Rev ¼ 200mV) was
included:
!3 2
C cell
CKout
pump
max
Na
JNa ¼ JNa
out
cell
kout
kcell
K þ CK
Na þ CNa
JKpump
ðV m Epump
Rev Þ
2 pump
¼ JNa
3
ð1Þ
(2) The mitochondria-rich (MR) cells were modelled
as a separate pathway between the principal cells
conducting Cl) between the apical solution and
the lateral intercellular space with a voltage- and
time-dependent electrodiffusive permeability for
Cl).
(3) The set of equations were solved for time-dependent states, which will be discussed elsewhere.
Generally, the transport equations for water and
solutes are written as:
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
E H Larsen et al.
X
dV
¼
JV
dt
j
ð2Þ
SÞ X
dðVC
JS
¼
dt
j
ð3Þ
denotes the volume of the cell compartment or
where V
the lateral intercellular space, and JV and JS denote the
water and solute fluxes, respectively, through the
various membranes, with j indicating the membrane
considered (j = 1–5). In the steady case the left-hand
side is zero and the solution of the system of equations
proceeds as explained in Larsen et al. (2002, 2006).
When studying transients, however, the time-dependent
behaviour of Eqns 2 and 3 needs to be simulated. To
solve the equations in time we utilize second-order
accurate, three-point backward difference schemes
(Taylor expansion) as follows:
i X
1 h ðnÞ
ðnÞ
ðn1Þ þ V
ðn2Þ ¼
3V 4V
JV
2Dt
j
ð4Þ
i X
1 h ðnÞ
S Þðn1Þ þ ðVC
S Þðn2Þ ¼
3ðVCS ÞðnÞ 4ðVC
JS
2Dt
j
ð5Þ
where index n refers to time tn, and Dt is the time-step,
such that tn = tn)1 + Dt. Thus, the equations are solved
for all variables with index n at time t = tn, leaving the
remaining terms as known from the former time steps.
The equations are solved together with the equations for
electro-neutrality and the compliance model. The solution of the total system of equations is carried out using
Newton’s method employing the technique described in
Larsen et al. (2002).
Results
General remarks
In this study, our experimental results are discussed
with reference to the theory of solute-coupled water
transport via osmosis in a lateral intercellular coupling
compartment. In some cases its theoretical mechanistic
behaviour is understood in non-mathematical intuitive
terms. For predicting less obvious quantitative behaviour we employ physical–mathematical modelling
programmed for in silico experiments.
In vertebrate epithelia the Na+/K+ pump is abundantly expressed in plasma membranes lining the lateral
intercellular space, which has been demonstrated for
frog skin (Mills et al. 1977), small intestine (Stirling
1972), gallbladder (Mills & DiBona 1978), urinary
bladder (Mills & Ernst 1975), exocrine glands (Ernst &
Mills 1977) and kidney proximal tubule. In the latter
mentioned epithelium plasma membranes of basolateral
Æ Solute-coupled water transport
infoldings display Na+/K+ pumps, as well (Kashgarian
et al. 1985, Pihakaski-Maunsbach et al. 2003). Figure 1a which shows the intraepithelial distribution of
Na+/K+ pumps in amphibian skin is reproduced from a
study employing [3H]-ouabain for localizing the pump
sites (Mills et al. 1977). It is noted that Na+/K+ pumps
are found exclusively in the plasma membranes lining
the lateral intercellular space, which is emphasized in
Figure 1b showing that the transepithelial active Na+
flux passes the lateral intercellular space (lis). With n
osmolytes the water flux, JV, across the barriers between
the lateral intercellular space and the bathing solutions
(tight junction, tj, and interspace basement membrane,
ibm) is given by:
"
JV ¼ LP RT
n
X
#
rj DCj DP
ð6Þ
j¼1
LP is the hydraulic conductance of tj or ibm, R the
universal gas constant, T the absolute temperature, rj
the reflection coefficients of tj or ibm, and DP and DCj
the hydrostatic pressure difference and the concentration difference, respectively, across the barrier considered. With identical solutions and similar hydrostatic
pressures on the two sides of the epithelium it is
immediately recognized that: (1) The direction of water
transport depends on the relative magnitude of the
reflection coefficients of the membranes delimiting lis;
in absorbing epithelia, rtj > ribm, which is considered
here. (2) Water flows into the epithelial cells from both
sides while all of the water entering the epithelium exits across the interspace basement membrane
(rtj > ribm).
Relationship between Na+ pumping and fluid transport
at transepithelial osmotic equilibrium
According to the theory it is hydrolysis of ATP by the
Na+/K+ ATPase in the plasma membrane lining the
lateral intercellular space (lis) that delivers the energy
for driving fluid through the epithelium. The steadystate Clis
Na is given by the balance between pump influx
across the lateral plasma membrane and leak fluxes
across the tight junction and the interspace basement
membrane. With constant leak permeabilities, as a first
approximation Clis
Na is expected to increase linearly with
the pump rate, and as Clis
Na (together with the concentration of a matching anion) provides the osmotic force
for water uptake, we expect, and find (Windhager et al.
1959, Curran 1960, Nielsen 1997, Nielsen & Larsen
2007) a linear relationship between active Na+ flux (JNa)
and volume flow (JV). This is shown in Figure 2 with
data obtained from 12 different preparations bathed on
both sides with Ringer’s solution, i.e. the osmotic
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
173
Solute-coupled water transport
Æ E H Larsen et al.
Acta Physiol 2009, 195, 171–186
(a)
(b)
Apical side
Interstitial side
Epithelial cell
H2O
H2O
Na+
H2O
Na+, Cl–
Cl–
H2O
P
H2O
+, Cl–
Na
ibm
lis
tj
lm
am
Epithelial cell
sm
H2O
Figure 1 Functional organization of transporting epithelia. (a) Localization of Na+/K+ pump sites in ventral pelvic skin of Rana
catesbeiana by [3H]ouabain labelling and autoradiography. The distribution of grains supports a model with the living cells of all
layers constituting a functional syncytium and the sodium ions being pumped into the lateral intercellular space before they enter the
serosal bath. Gr, granular layer; Sp, spiny layers; Ger, germinal layer. From Mills et al. (1977). (b) General model of solute-coupled
water transport of a fluid absorbing epithelium exposed on the two sides to physiological saline of identical composition. Water
enters the epithelial cell compartment from both sides and leaves the epithelium through ibm. This principle is independent of the
mechanisms by which solutes enter the epithelium. To avoid crowding of symbols, Na+ and Cl) enter the epithelium via two
different cells and K+ is omitted. The apical, lateral and serosal plasma membrane are indicated by am, lm and sm respectively. tj,
tight junction; ibm, interspace basement membrane; lis, lateral intercellular space.
concentrations on the two sides of the epithelium are
identical. The two fluxes decrease with similar time
courses after the addition of ouabain to the serosal bath
confirming the dependence of JV on the activity of the
Na+/K+ pump (see Fig. 3). Very likely, the slow time
course of inhibition reflects that ouabain has to diffuse
against an intraepithelial flow of fluid before binding to
the pumps in the plasma membranes lining the labyrinth
of intercellular spaces (cf. Fig. 1a).
It is easy to understand that as water is at thermodynamic equilibrium across the epithelium increasing
the hydraulic conductance of the plasma membranes or
the interspace barriers does not affect the rate of volume
transport provided the Na+ pump flux and Na+ leak
174
fluxes remain constant. Importantly, such an increase
results in a smaller steady-state Clis
NaCl , which is of
significant physiological importance as it reduces the
diffusion flux of Na+ across ibm and thereby the
hypertonicity of the fluid emerging from lis relative to
the tonicity of lis (Larsen et al. 2002). Similarly, it is
intuitively comprehended that with constant pump rate
and hydraulic conductances the ratio of JV/JNa, which is
a measure of the efficiency of coupling, is affected by the
leak permeabilities for Na+. For example, JV/JNa
decreases with increasing Na+ permeability of tj and
ibm because this manoeuvre leads to a smaller Clis
NaCl ,
that is the osmotic driving force of the water flux from
the outside solution into lis is being reduced.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
Figure 2 Steady-state relationship between the active Na+ flux
and the simultaneously measured rate of transepithelial water
transport by short-circuited toad skin epithelium. Data from
n = 12 preparations (Nielsen & Larsen 2007).
Figure 3 Stimulation of the active Na+ flux and fluid absorption by the b-adrenergic agonist isoproterenol (5 lm on serosal
side). Subsequent addition of the Na+/K+ pump inhibitor ouabain (100 lm on serosal side) results in slow inhibition of both
parameters. The experiment was performed at open-circuit
conditions with JNa estimated from the equivalent short-circuit
current. The indicated osmotic concentrations of the transEqv
ported fluid were estimated as 2 · ISC
=ðF JV Þ, see Method
section. Data from Nielsen & Larsen (2007).
Uphill water transport
The membrane-impermeable non-electrolyte sucrose
was used for reducing the water activity of the outer
bath (corneal side) of the epithelium (Nielsen & Larsen
2007). Figure 4a shows that while JNa stayed near its
control value prior to the osmotic challenge, the
simultaneously recorded JV was reduced stepwise in
response to a stepwise increase in the external sucrose
concentration. In this experiment the water flow
reversed at a transepithelial osmotic concentration
difference of )34 mOsm (Fig. 4b). From the slope of
the linear relationship and a molar water volume of
18 cm3 mol)1 H2O (20 C) we calculate an apparent
E H Larsen et al.
Æ Solute-coupled water transport
water permeability of, Pf = 8.6 · 10)3 cmÆs)1. Obviously, this is not the true water permeability of the
epithelium as the osmotic driving force for water flow
into the epithelium is given by the difference in osmotic
concentration between lis and the outside solution
(cf. Eqn 6 above). This is an unknown quantity which
has never been estimated precisely in experiments with
transporting epithelia, but it is supposedly much smaller
than the transepithelial osmotic gradient (Weinstein &
Stephenson 1981, Larsen et al. 2000).
An estimate of the ion concentrations of lis was
obtained by simulating the experiment discussed above.
The values of JNa (Fig. 5a), the X-axis intercept and
the slope of the regression line (Fig. 5b,
Pf = 12 · 10)3 cmÆs)1) are in fair agreement with the
experiment shown in Figure 4a,b as are all intracellular
observables (data not shown, but see Larsen et al.
2007). In the computed example, at transepithelial
equilibrium the concentrations of the lateral space were
lis
lis
(mm): Clis
Na = 120.385, CK = 0.439 and CCl = 120.823
at a water flow into lis of 7.597 nL cm)2 s)1 (via
translateral and tight junction routes). From these
numbers the true apical water permeability is calculated
to be Pf = 2.57 · 10)1 cm s)1. This number, which is
about 20 times larger than that obtained in Figures 4b
and 5b, is identical to that calculated from the input
values of the model of the apical and lateral plasma
membranes in series with each other and in parallel with
the tight junction barrier (Pam
= 2.71 · 10)1, Plm
f =
f
tj
)1
)2
8.12 · 10 and Pf = 5.41 · 10 cm s)1 respectively).
tj
The assumption of Pam
f >Pf is in agreement with the
finding that antidiuretic hormone stimulates the cellular
water permeability (MacRobbie & Ussing 1961) by
incorporation of aquaporins in the apical plasma
membrane of the principal cells (Hasegawa et al.
2003). Hence, the major water pathway in this highresistance epithelium is supposedly translateral. Notably, similar computed results would have been obtained
tj
if Pam
f <Pf with the individual permeabilities adjusted to
provide the same apical water permeability as above.
The above example leads to the general conclusion
that the true water permeability of an epithelium with
solute-coupled water transport cannot be estimated by
recording the response of the steady-state water flow to
perturbation of the external osmotic concentration.
The concentration of the transported fluid
Figures 3 and 6a show that stimulation of JNa by the
b-adrenergic agonist isoproterenol results in a parallel
stimulation of JV. Likewise, inhibition of JNa with
amiloride is associated with inhibition of JV (Fig. 6a)
confirming that JV is depending on and coupled to JNa.
At open circuit conditions (IT = 0), the active flux of
Na+ generates a transepithelial electrical potential
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
175
Solute-coupled water transport
Æ E H Larsen et al.
(a)
Acta Physiol 2009, 195, 171–186
(b)
Figure 4 Uphill water transport by toad skin epithelium. (a) Effect on active Na+ flux and fluid transport of adding increasing
concentrations of sucrose (numbers on the graph in mm) to the solution bathing the outside of the epithelium at open-circuit
conditions. (b) Variation of steady-state rate of fluid transport with the difference in osmotic concentration between outside and
inside solution. DC = Cinside ) Coutside with excess osmotic concentration of the outside bath obtained by adding sucrose to the
Ringer’s solution. The direction of fluid flow was estimated to reverse at DCRev = )34 mOsm.
(a)
(b)
Figure 5 Model computations simulating the experiment shown in Figure 4a,b. (a) A non-permeant electroneutral solute was
added to the external ‘Ringer’s solution’ in steps of 5 mm with IT clamped at 0 (open circuited conditions). At each step, the
concentration was increased exponentially with a time constant of 5 s. Shown are the time courses of the active Na+ flux and the
rate of fluid transport given by the model. (b) Computed steady-state dependence of rate of fluid transport on transepithelial osmotic
concentration difference. DC = Cinside ) Coutside with excess osmotic concentration of the outside compartment obtained by adding
non-permeant electroneutral molecules on top of the ‘Ringer’s solution’ of the outside compartment [see (a)]. The direction of fluid
flow reverses at DCRev = )31 mOsm.
difference that drives Cl) inwardly at a rate similar to
that of Na+. As no other solutes pass the epithelium, the
osmotic concentration of the transported fluid can be
estimated by 2JNa/JV. The values thus obtained overestimate the true osmotic concentration of the transported
fluid by 10–12% prior to and following isoproterenol
stimulation (see Method section). The osmotic concentrations calculated by the above method are indicated in
Figures 3 and 6a with a summary given in Figure 6b.
On average, prior to hormone stimulation the tonicity
of the transported fluid tends to be hypertonic, whereas
the fluid becomes near-isotonic following isoproterenol
stimulation. With amiloride in the outside bath the
transepithelial active Na+ flux decreased relatively more
than the flux of water resulting in a new steady state
with a highly significant hypotonic transportate at the
transepithelial osmotic equilibrium (Fig. 6a,b and comments to Table 1 below). In other words, the trans-
176
ported fluid is significantly diluted (< 62 12 mOsm,
mean SEM, n = 7, P < 10)3) with respect to the
bathing solutions of Ringer’s strength (253 mOsm). As
discussed in Nielsen & Larsen (2007) and below, this
indicates that a fraction of the sodium ions pumped into
the lateral space is derived from the serosal bath, which
therefore does not contribute to the transepithelial net
uptake of Na+ and electrical charge flux.
ENaC inhibition by amiloride
The experiments shown above make it evident that the
volume flow is coupled to the transepithelial active flux
of Na+ (Figs 2, 3 and 6a). It turned out that these
variables could be dissociated temporarily following
fast block of the apical ENaC by amiloride, which was
associated with slow decrease in JV with a half time of
8 min (Fig. 7a). The rate at which JNa decays depends
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
(a)
E H Larsen et al.
Æ Solute-coupled water transport
(a)
(b)
(b)
Eqv
Figure 6 (a) Time course of ISC
=F (an estimate of JNa) and
rate of fluid transport in response to adding a b-adrenergic
agonist to the serosal bath and – subsequently – amiloride to
the outside bath. The upper graph shows the time course of
Eqv
2ISC
=ðF JV Þas an estimate of the osmotic concentration of
the transported fluid. Data from (Nielsen & Larsen 2007). As
Eqv
mentioned in the text 2ISC
=ðF JV Þ represents a 10–12%
overestimation of the true osmotic concentration prior to
amiloride treatment (Larsen et al. 2007). See also Table 1. (b)
Summary of results obtained by the experimental protocol
indicated in (a) (mean SEM, n = 7).
on how fast amiloride blocks ENaC. Essentially, this is
determined by the unstirred layer at the apical plasma
membrane. According to the theory, the time course of
JV inhibition depends on the rate at which the intracellular Na+ pool is emptied by the lateral Na+/K+ pumps.
The model computations shown in Figure 7b confirm
that with a physiological value of the cellular Na+ pool,
the half-time of the pool would be relatively slow,
T½ = 5.8 min. In the model, the Na+ pool of the
syncytium of principal cells prior to ‘amiloride’ treatment was 76 nmol cm)2, corresponding to a total cell
water volume of 7.2 lL cm)2 with Ccell
Na = 10.6 mm.
The reduced activity of the lateral Na+/K+ pumps
results in depletion of the NaCl pool of lis associated
with a decrease in the volume of lis from 516 to
353 nL cm)2 and a parallel decrease in its excess
Figure 7 Transient dissociation of the volume flow from the
active Na+ flux. (a) Time course of active Na+ flux and rate
of fluid absorption in short-circuited toad skin epithelium
following addition of 100 lm amiloride to the external solution. (b) Results given by the model with a protocol similar to
that of the experiment shown in (a). The apical Na+ permeability was reduced exponentially with a time constant of 10 s
to 1% of its control value prior to ‘amiloride’. The much
slower decrease in JV proceeds with a half time of 6 min.
hydrostatic pressure from 0.863 to 0.220 cmH2O.
These numbers cannot be verified experimentally. They
are mentioned here to show that with a hydraulic
conductance of ibm that is 1–2 orders of magnitudes
larger than that of tj, and with the experimentally
estimated compliance factor of epithelial plasma membranes (Spring & Hope 1979) the hydrostatic pressure
forcing fluid out of lis is expected to be small.
Water flow associated with a passive Cl) flux through
mitochondria-rich cells energized by transepithelial voltage
and current clamps
The apical plasma membrane of principal cells
is impermeable to Cl) even in preparations with
maximally stimulated transepithelial Cl) conductance
(Willumsen & Larsen 1986, Willumsen et al. 1992).
The major physiological flux of Cl) goes through the
MR cells. These bottle-shaped cells are located in the
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
177
Solute-coupled water transport
(a)
Æ E H Larsen et al.
(b)
(d)
Acta Physiol 2009, 195, 171–186
(c)
(e)
Figure 8 Summary of features of Cl) transport by the c-type mitochondria-rich (MR) cell of amphibian skin epithelium. (a)
Electron micrograph of an MR cell located between the uppermost layers of principal cells. (b) Steady-state current–voltage
curve of whole skin with Cl) currents (ICl) determined using the 36Cl) tracer technique (n = 5–13). The active Na+ flux was
eliminated by replacing Na+ of the outside bath with K+. Data from Willumsen & Larsen (1986). (c) Single MR cell recordings were
performed with the neck of the cell sucked into the tip of the patch electrode and applying transcellular voltage clamps with an
Axopatch-200B amplifier interfaced to a Digidata 1322A and using pClamp9 (Molecular Devices, Union City, CA, USA) for
recording and analysis. (d) Left, transcellular currents generated by the method described in (c); right, resulting steady-state ICl–V
relationship (leak currents subtracted) exhibits outward rectification similar to the whole skin preparations. (e) Single channel
recording of apical plasma membrane patch in inside-out configuration showing long open state with transitions to substates. Bath
pip
and pipette solutions of identical composition with Cbath
Cl ¼ CCl = 120 mm. 100 lm amiloride in pipette solution for blocking the
+
Na channels of the apical MR-cell plasma membrane. )Vpip = 20 mV.
outer region of the epithelium with the neck between
the outermost principal cells so that the apical plasma
membrane faces the subcorneal space like the ENaC
expressing apical plasma membrane of principal cells
(see Fig. 8a). The apical plasma membrane of the most
abundant type of MR cells, the c-type MR cell (Larsen
1991), displays a population of Cl) selective channels
that is controlled by the apical membrane potential. The
Cl) channels are slowly activated by membrane depolarization brought about by transepithelial hyperpolarization, which results in a strongly rectified steady-state
transepithelial Cl) current (Fig. 8b). The definitive proof
that this is a specific function of this minority cell type
was obtained by voltage clamp experiments with
isolated MR cells (Fig. 8c,d), which generated currents
accounting quantitatively for the macroscopic voltageactivated currents with respect to their magnitude and
time course (Larsen et al. 2001) confirming earlier
studies relating macroscopic Cl) fluxes to the density
of MR cells (Voûte & Meier 1978, Willumsen & Larsen
1986, Devuyst et al. 1990). The apical Cl) channels
exhibit a large single channel conductance of 150–
300 pS identified by noise analysis of voltage-activated
Cl) currents in whole-cell patch-clamp studies (Larsen
& Harvey 1994) and by single-channel studies on cell
attached and inside-out patches (Sørensen & Larsen
1996). In the open state the channel is noisy with several
substates (Fig. 8e). The channel has been observed to
open and close in a stepwise fashion. It has been
178
difficult to study and is not characterized at the
molecular level.
In the present context the above macroscopic features
of the MR cells are useful for an independent way of
testing the theory of the lateral space as an osmotic
coupling compartment. From Figure 8a it can be
concluded that a current flowing through the MR cells
passes lis before entering the serosal bath. Therefore,
following amiloride block of the transepithelial active
Na+ flux we should be able to generate a graded flow of
fluid in the inward direction by activating the apical Cl)
channels of MR cells by transepithelial voltage or
current clamping. The results shown in Figure 9a,b
confirm this notion; both during activation in voltage
clamp mode and at steady state during current clamps a
direct dependence of JV on JCl is observed. To a fair
extent these experimental observations are reproduced
by in silico experiments applying similar protocols
(Fig. 9c,d). The other important outcome of these
computations is that the Cl) concentration of the lateral
space is predicted to be so close to that of the bathing
solutions that it would be impossible to measure the
difference with available methods.
Discussion
Historically, epithelial fluid transport in the absence of
external driving forces for water was discovered more
than 100 years ago by Edward Warmouth Reid in his
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
E H Larsen et al.
(a)
(b)
(c)
(d)
Æ Solute-coupled water transport
Figure 9 Cl) flux-generated transepithelial volume flows. (a) Variation of JV with JCl during development of the depolarization
activated Cl) conductance of the apical membrane of mitochondria-rich (MR) cells of a toad skin epithelium by a transepithelial
voltage clamp from 0 to )50 mV (grounded serosal bath). The protocol was repeated five times as indicated by different colours.
(b) Steady-state relationship between the inward transepithelial Cl) flux through MR cells and transepithelial volume flow (data
from one preparation, mean SEM, n = 8–11 data points). The straight line is the regression line according to the equation, JV ¼
(1.98 0.15) · 10)3 · JCl + 0.68 0.22, R2 ¼ 0.993. (c) Relationship between volume flow and chloride flux given by the model
produced by shifting VT from 0 to )50 mV, thus simulating the experiments of (a). (d) Computed steady-state variation of JV with
the Cl) flux through MR cells in transepithelial current clamp mode at a constant chloride permeability of the MR cell pathway. The
steady-state Cl) concentration of lis is given for each clamping current.
in vitro studies on rabbit small intestine and frog skin.
Reid concluded: ‘It would appear [..] that it is possible
[..] to gain distinct, positive, evidence of vital absorptive
action in the intestine’ (Reid 1892a, his accentuations)
and, ‘there is present in the living skin of the frog a vital
absorptive force dependent upon protoplasmic activity
[…] and by virtue of this vital action the skin is actually
able to cause a stream of fluid to pass from its outer to
its inner surface’ (Reid 1892b). With his method and
interpretations Reid was far ahead of his time, and more
than half a century passed before similar in vitro studies
were performed on other transporting epithelia that
confirmed and generalized Reid’s findings, e.g. toad skin
(Kalman & Ussing 1955), rat small intestine (Curran &
Solomon 1957, Parsons & Wingate 1958, Curran
1960), proximal tubule of the glomerular nephron
(Windhager et al. 1959), gallbladder (Diamond 1962,
1964), exocrine glands (Thaysen & Thorn 1954) and
upper airways (Grubb et al. 1997).
The coupling mechanism
A wide range of experimental observations provide
strong evidence that the lateral space serves as the site
of solute–water coupling in the amphibian skin
epithelium.
(1) The dependence of water flow on active Na+
transport indicates that Reid’s vital absorptive action is
generated by Na+,K+-ATPase. This enzyme is localized
in the plasma membranes lining the lateral intercellular
space (Mills et al. 1977) (see Fig. 1a).
(2) The flow of water decreases following application
of the Na+/K+ pump inhibitor ouabain (Fig. 3).
(3) The linear relationship between the active flux of
Na+ and the rate of transepithelial water transport
follows logically when it is appreciated that lis is a pumpleak system. With constant leak permeabilities, Clis
Na
would increase linearly with the pump rate, whereby JV
must also increase linearly with the pump rate (Fig. 2).
(4) Stimulation of JNa via intracellular cAMPdependent protein kinase (PKA) signalling is paralleled
by an increase in JV (Figs. 3 and 6a,b).
(5) It was predicted that the coupling of the
transepithelial Na+ flux and JV becomes temporarily
dissociated by amiloride application because the time
course of JNa inhibition is governed by the rate at which
amiloride binds to the population of apical ENaC while
the time course of JV inhibition is governed by the
turnover of the intracellular Na+ pool. The experimental results confirmed this notion (Fig. 7).
(6) A further test of the theory came from the
experiments with voltage activation of the Cl) flux via
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
179
Solute-coupled water transport
Æ E H Larsen et al.
MR cells. We reasoned that if the lateral space
constitutes an osmotic coupling compartment we
should be able to generate an inward flow of water by
forcing Cl) into the space by an external electromotive
force. Our experiments made it plain that this reasoning
was correct (Fig. 9).
(7) With no further assumptions uphill water transport follows logically from the theory of a lateral
intercellular coupling compartment (Figs. 4 and 5).
(8) Our experimental results exclude the other current
theories on the mechanism of solute–solvent coupling in
epithelia. The flux of Na+ into the cellular compartment
of amphibian skin depends on apical sodium channels
with no contributions from Na+-dependent co-transport
mechanisms (Fuchs et al. 1977, Van Driessche &
Lindemann 1979). This rules out the idea of Na+
gradient-driven molecular water pumps as the mechanism of epithelial fluid absorption (Zeuthen et al.
2001). The demonstration of water flow in the shortcircuited epithelium with current flowing entirely
through the cellular compartment (Fig. 2) and the
dissociation of JV from the ion flux during non-steady
state (Fig. 7) rule out the theory of paracellular electroosmosis developed by Fischbarg et al. (2006). The
localization of Na+/K+ pumps in lateral plasma membranes (Fig. 1a), the demonstration of a Cl) fluxgenerated water flow (Fig. 9) and lack of evidence for
an ATP-driven ‘junctional fluid transfer mechanism’ in
any known epithelium make it obvious that the theory
of Hill & Shachar-Hill (2006) would have to be
abandoned as well.
Tonicity of transported fluid
It is often believed that the physiological solution to
isotonic transport is the development of a hydraulic
conductance of the epithelium that is relatively large
compared to the active flux of Na+. Our experimental
results and theoretical analyses indicate that it is not
quite so simple.
The efficiency with which the steady-state ion flux
drives water through the epithelium varies with the ion
species (Figs 2 and 9b) and the physiological state of the
preparation (Fig. 6b). For example, the slopes of
7.7 · 10)3 nL H2O pmol)1 Na+ (Fig. 2) and
2.0 · 10)3 nL H2O pmol)1 Cl) (Fig. 9b) correspond
to coupling ratios of 428 H2O/Na+ and 110 H2O/Cl)
respectively. A simple interpretation of the pump-leak
model of the lateral space would suggest that the leak
permeabilities of the barriers delimiting lis are different
for the two ions. As the unidirectional Cl) fluxes via the
voltage-activated MR cells obey the flux-ratio equation
(Larsen 1991), the activation of the Cl) flux into the
lateral space inevitably leads to a simultaneous activation of the leak permeability for Cl). A similar type of
180
Acta Physiol 2009, 195, 171–186
interpretation would be quite insufficient for explaining
the different experimental results of Figure 6b with
mean ratios of 304 (control), 487 (isoproterenol) and
1787 H2O/Na+ (isoproterenol + amiloride) respectively. The clue comes from the observation that the
epithelium has the capacity to generate an isotonic as
well as a hypotonic absorbate.
Diffusion–convection of solutes at the exit of lis
creates a fundamental problem of isotonic transport,
which is conveniently analysed by considering Hertz’s
diffusion-convection equation (Larsen et al. 2000):
lis
serosa
ibm
JSibm
exp½JVibm ð1 ribm
ibm CS CS
S Þ=PS ¼
ð1
r
Þ
S
ibm
1 exp½JVibm ð1 ribm
JVibm
S Þ=PS ð7Þ
where JSibm and JVibm are the solute and volume flux across
serosa
the boundary, Clis
are the solute concentraS and CS
tions on the two sides of the boundary, while ribm
and
S
Pibm
are
the
reflection
coefficient
and
solute
permeabiS
lity respectively. As a numerical example we will
consider the rat proximal tubule with JVibm 65 nL cm)2 s)1 and Ribm = 0.6 W cm2 corresponding
to Pibm
1.6 · 10)3 cm s)1 (Frömter 1979, Weinstein
S
2008). With ribm
= 10)3, Cserosa
= 300 mm and
S
S
lis
CS = 301.5 mm, Eqn 7 predicts a concentration of
the fluid emerging from lis, JSibm =JVibm = 337.6 mm,
which is hypertonic relative to lis and the serosal bath
by 12.5% and should be compared to the hardly
detectable difference in solute concentrations on the
two sides of the boundary. In other words, a large Pibm
S
relatively to JVibm results in a diffusion flux that is
comparable to the convection flux within an order of
magnitude, implying that the solute concentration of
the appearing fluid is larger than that of lis. With x = 0
at the tight junction and L being the length of the
lateral space Diamond & Bossert (1967) assumed as
boundary condition in their standing-gradient theory
that, [dCS/dx]x=L = 0 (see Fig. 10a). When fulfilled the
diffusion flux at the boundary is zero and the emerging
fluid truly isotonic.
In the Na+-recirculation theory (Nedergaard et al.
1999) it is realized that the diffusion flux cannot be zero
and, therefore, the solute concentration of the fluid
emerging from lis tends to be hypertonic. The above
boundary condition is here replaced by the assumption
that isotonic transport is obtained by a simultaneous
solute flux from the serosal bath via the epithelial cells
into lis (Fig. 10b). Accordingly, the Na+/K+ pumps in
the lateral plasma membrane play a dual role in fluid
transport: (1) they make lis hypertonic and hyperbaric,
and (2) they pump into lis the sodium ions entering the
cells from the serosal bath thus ‘bringing back the
diffusion component to the coupling compartment’.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
(a)
H2O
Na+
Na+
E H Larsen et al.
H2O
dC/dx = 0
P
Na+
H2O
Standing-gradient
theory
P
Na+
H2O
(b)
H2O
Na+
Na+-recirculation
theory
Na+
H2O
Na+
H2O
Na+
H2O
H2O Na+
P
P
H2O Na+
Na+
Na+
H2O
Na+
Figure 10 Two theories of truly isotonic transport. (a) The
standing-gradient theory. By assuming the Na+/K+ pumps to be
displayed near the tight junction, the water channels in the
lateral plasma membrane along the entire lateral intercellular
space and with appropriate choice of model variables, the
condition of zero solute concentration gradient at the interface
between the lis and serosal bath is obtained (Diamond &
Bossert 1967). (b) The Na+ recirculation theory. The boundary
condition of (a) is replaced by the assumption that a regulated
back flux of ions through the serosal plasma membrane, driven
by the lateral Na+/K+ pumps, adjusts the composition of the
absorbed fluid to the demanded isotonicity (Nedergaard et al.
1999).
In order for this mechanism to be robust over a wide
range of physiological conditions we have to assume
that the back flux of Na+ (the recirculation flux) is
regulated for achieving the demanded osmotic concentration of the absorbate. Based upon the observation
that JV decreased in response to serosal bumetanide
application (Nielsen & Larsen 2007) we hypothesized
that the recirculation pathway of the serosal membrane
is the NKCC transporter, which was discovered in
Æ Solute-coupled water transport
studies on the regulation of volume and intracellular Cl)
concentration of frog skin principal cells (Dörge et al.
1985, 1989, Ussing 1985).
Our detailed experimental analyses of ion concentrations, membrane potentials and membrane conductances have provided sufficient data for modelling with
reasonable confidence the amphibian skin epithelium in
the resting state and during isoproterenol stimulation
and amiloride treatment (see above and Larsen et al.
2007). Here we will focus on the most interesting part
of the experimental results, which were obtained with
amiloride resulting in a hypotonic absorbate (Fig. 6a,b).
The computed results shown in Table 1 are compared
with observed physiological variables obtained with
double-barrelled microelectrodes in isolated voltageclamped epithelia mounted in a mini-Ussing chamber
on a microscope table (Larsen et al. 1992, Willumsen
et al. 1992) and a determination of the Na+ transport
pool with the isotope tracer technique (Nielsen 1982).
Our computations confirm that a recirculation flux
estimated from the intracellular Cl) concentration
together with experimentally obtained serosal plasma
membrane potential and Cl) conductance would generate a transepithelial fluid absorption of a magnitude
and tonicity comparable to the measured observables.
Applicability to other epithelia
The coupling mechanism. All experimental evidence
discussed in this paper shows that fluid absorption in the
absence of transepithelial driving forces for water is
brought about by osmotic coupling in the lateral
intercellular compartment by a mechanism essentially
conceived by Peter Curran in his studies on mammalian
small intestine (Curran 1960). The similar distribution
of the Na+/K+ pumps in all studied transporting epithelia
of vertebrates (Stirling 1972, Mills & Ernst 1975, Ernst
& Mills 1977, Mills et al. 1977, Mills & DiBona 1978,
Table 1 Experimental and computed biophysical parameters of the amphibian skin epithelium treated with 5 lm isoproterenol on
the serosal side and 100 lm amiloride on the outside. The model is that of Figure 7b, but here computed for open circuit conditions
P net
Eqv
Ji
2ISC
Eqv
cell
cell
VT
GT
ISC
C cell
C
C
V
J
cell
V
Na
K
Cl
FJV
JV
)2
)2
)2 )1
(mV)
(mS cm ) (lA cm ) (mm)
(mm)
(mm)
(mV)
(nL cm s ) (mOsm)
(mOsm)
Exp.
)1.5 0.7* 1.5 0.2* 1.5 0.8* 1.6 0.1 145 8 51 3§ )99 3§ 0.50 0.15* 62 12*
Model
)1.0
1.5
1.5
2.0
140
53.2
)96.5
0.43
69–
n.d.
24–
*Nielsen & Larsen (2007).
Nielsen (1982).
Larsen et al. (1992).
§
Willumsen et al. (1992).
–
The true osmotic concentration of the computed absorbate (24 mOsm) is significantly smaller than that calculated from the
equivalent short circuit current (69 mOsm). The difference reflects the fact mentioned in the Method section that the active Na+ flux
calculated from the equivalent short-circuit current is an overestimation of the active Na+ flux at open circuit conditions.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
181
Solute-coupled water transport
Æ E H Larsen et al.
Kashgarian et al. 1985, Pihakaski-Maunsbak 2003)
suggests that they share the same basic coupling mechanism. The abundant expression of Na+/K+ pumps in
projections from the basal plasma membrane in the
proximal tubule of the kidney (Maunsbach & Boulpaep
1991) suggests that this extracellular labyrinth constitutes a parallel coupling compartment. Unlike lis with
highly conductive tight junctions (Frömter 1979) the
basal infoldings are closed to the luminal solution
indicating that a relatively large flow of water takes a
translateral route via aquaporins (Nielsen et al. 2002).
Nevertheless, another coupling mechanism has
been suggested (Zeuthen et al. 2001). Zeuthen and
co-workers assume that the steady-state isotonic water
transport across the brush border of the small intestine
and kidney proximal tubule is divided into two components of which one is generated by a stoichiometric
coupling to Na+ and glucose fluxes via SGLT1, and the
other is due to osmosis via a water channel in the SGLT1
protein of the small intestine or the apical AQP1 water
channel of the proximal tubule. Common to these two
models is the assumption that the epithelial cell becomes
hyperosmotic because of the stoichiometry of 264
H2O : 2 Na+ : 1 glucose of hSGLT1. It is true that with
a volume of 18 cm3 mol)1 H2O and the above
stoichiometry the coupled fluxes of the three substrates
represent a hyperosmotic transportate (631 mOsm). It is
incorrect, however, to assume that at steady state the cell
water is hyperosmotic. On the contrary, the active
transport of water into the epithelial cell results in a
steady state where water is above thermodynamic
equilibrium, i.e. the cell water becomes hyposmotic
relative to the bathing solutions. This implies that if a
water channel is displayed in the brush border membrane together with the above SGLT1 mechanism water
recirculates across the apical plasma membrane. Hence
the overall water uptake from the luminal solution
becomes smaller than the supposed Na+ gradient-driven
water flux through SGLT1, which is in contrast to
Zeuthen and co-workers’ assumption (discussed with
quantitative details in Larsen et al. 2002).
It has been further argued that an osmotic mechanism
has to be abandoned or is questionable because
attempts to detect a difference in osmotic concentration
between lis and the bathing solutions have failed
(Zeuthen 1983, Ikonomov et al. 1985). This argument
is significantly weakened by our computations showing
that with a relatively large ratio of apical water
permeability and active Na+ flux the excess steady-state
lateral intercellular ion concentrations would be too
small to be detected with available methods. Our
experiments and computations with Cl) as the driving
ion lead to the same conclusion. In connection with this,
it is recalled that at a given rate of active Na+
absorption, a large water permeability of the barriers
182
Acta Physiol 2009, 195, 171–186
between lis and the luminal solution inevitably results in
small steady-state osmotic concentration of lis. For
example, in the highly water-permeable proximal
tubule of the rat kidney the excess osmotic concentration was estimated to be less than 1 mOsm (Larsen
et al. 2006). As this conclusion is independent of the
apical mechanism of Na+ absorption, it applies generally, including all reabsorbing segments of the proximal
tubule in vertebrate glomerular kidneys.
We conclude that all experimental evidence points to
solute–water coupling by osmosis in an intracellular
coupling compartment energized by lateral Na+/K+pumps.
The tonicity of the absorbate. Hence the puzzles posed
by isotonic transport should be attacked by realizing
that the fluid leaving the intraepithelial osmotic coupling compartment always tends to be hypertonic relative
to the fluid in the coupling compartment and to the
bathing solutions of identical composition and, as
discussed above, this would be compensated for by
regulated Na+ recirculation. The separation of cellular
and paracellular unidirectional Na+ fluxes by the presteady-state flux-ratio method (Sten-Knudsen & Ussing
1981) allowed Nedergaard et al. (1999) to estimate the
Na+ recirculation flux in the small intestine. In this
epithelium with low, if any, expression of aquaporins in
the brush border membrane the experimental results
indicated that about 65% of the Na+ pumped into the
lateral space is derived from the serosal bath. Subsequent modelling of the small intestine predicted that this
relatively water-tight epithelium would build up an
osmotic concentration in the lateral space of 6–7 mOsm
above that of the bathing solutions, and that the
resulting fairly large diffusion component of the Na+
flux emerging from the lateral space would have to be
compensated for by a recirculation flux of 63% if
isotonic transport is demanded (Larsen et al. 2002).
With active transport of water across the apical plasma
membrane via SGLT1 assuming a stoichiometry of 210
H2O : 2 Na : 1 glucose (Loo et al. 1996), the necessary
recirculation for isotonic transport decreased just by a
small number, from 63% to 56% (Larsen et al. 2002).
At the other end of the scale, by in silico experiments it
was estimated that truly isotonic transport in the highly
water-permeable rat proximal tubule demands less than
5% Na+ recirculation for ‘correcting’ an otherwise
3% hypertonic absorbate (Larsen et al. 2006).
Initial segments of the mammalian proximal tubule
pose an interesting problem because reabsorption of
Na+ is here accompanied by HCO3) (Weinstein 2008).
If further experimental studies indicate that the necessary fine adjustment of the tonicity of the tubular
reabsorbate rely on ion recirculation, different molecular mechanisms would have to serve this function in
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
E H Larsen et al.
different segments of the proximal tubule depending on
whether Cl) or HCO3) constitutes the dominating
anion accompanying Na+.
The overall conclusion is that the theory of osmotic
coupling of water and Na+ fluxes in intraepithelial
lateral spaces accounts for the water transport as well as
for other essential experimental observations in the
small intestine (Larsen et al. 2002) and kidney proximal
tubule (Larsen et al. 2006). The demand on Na+
recirculation for isotonic transport follows logically
from the physics of convection–diffusion of Na+ out of
the lateral space, which tends to make the Na+
concentration of the emerging fluid larger than that of
the coupling compartment. Further to this, the theory
considering Na+ gradient-driven apical water uptake by
a SGLT1 protein fails to predict the essential isotonic
transport in these two epithelia.
Metabolic cost of solute-coupled water transport
At first sight, one might object to the idea of Na+
recirculation by pointing out that the energetic efficiency would be too low for serving important energyconsuming body functions. Actually, the theory seems
to contradict the observation that in fluid-transporting
epithelia the ratio of Na+ transport and associated O2
consumption may be larger than that expected for the
Na+/K+ pump itself, which is 18 Na+/O2 (this ratio is
calculated from a pump stoichiometry of 3 mol Na+/
mol ATP and a mitochondrial ATP/O2 ratio of 6).
Table 2 presents a large number of such data. It is seen
that efficiencies both above and below the generic ratio
have been obtained. Any theory of isotonic transport
would have to deal with these numbers in a logical and
coherent way.
Æ Solute-coupled water transport
The statistical means in the two classical studies of
amphibian skin preparations are almost identical to
the ratio given above (Table 2). It is striking though
that both studies report a significant variation of this
quantity spanning from 12 to 40 Na+/O2 (Zerahn
1958) and from 6 to 35 Na+/O2 (Leaf & Renshaw
1957). This fairly large – and admittedly surprising –
range of metabolic cost of active Na+ transport was
confirmed by two more recent studies on high
resistance epithelia (Vieira et al. 1972, Nellans &
Finn 1974). In frog skin made ‘leaky’ by exposing the
outside of the epithelium to a high osmotic concentration, Ussing (1966) measured a ratio of 12–14.4
Na+/O2. Studies on the mammalian kidney electrolyte
and energy metabolism invariably reported values
significantly above 18 Na+/O2 (Deetjen & Kramer
1961, Lassen & Thaysen 1961, Lassen et al. 1961,
Thurau 1961, Torelli et al. 1966). Two studies on
rabbit gallbladder found similar high efficiencies of
Na+ transport, 24 Na+/O2 (Martin & Diamond 1966)
and 30 Na+/O2 (Frederiksen & Leyssac 1969), while
dilution of the external baths resulted in a drop of
this ratio in the gallbladder to 13 Na+/O2 (Frederiksen & Leyssac 1969).
While recirculation of Na+ would account for the
lower-range values, a number of experimentally verified
mechanisms by which sodium ions bypass the Na+/K+
pump account for ratios above 18 Na+/O2. These are:
(1) Na+ uptake via the paracellular route by convection
(solvent drag) in tight junction of the proximal tubule
(Frömter et al. 1973, see Larsen et al. 2006 for a
quantitative treatment), (2) cellular Na+ uptake via the
basolateral Na+/[HCO3)]n > 1 cotransporter of the
upper segments of the kidney proximal tubule (Boron
& Boulpaep 1983, Alpern & Chambers 1987,
Table 2 Metabolic cost of active sodium transport in high- and low-resistance epithelia
Species
Preparation
mol Na+/mol O2
Range
Reference
R. temporaria
R. esculenta
B. bufo
R. temporaria
B. bufo
R. pipiens
B. marinus
Rabbit
Dog
Dog
Dog
Rabbit
Rabbit
Rabbit
Rabbit
Skin in vitro
17.8
12–40
Zerahn (1958)
Skin in vitro
‘Leaky’ skin in vitro
Skin in vitro
Urinary bladder in vitro
Kidney slices
Kidney in vivo
Kidney in vivo
Kidney in vivo
Kidney in vivo
Gallbladder in vitro
Gallbladder in vitro
Gallbladder in vitro
with diluted baths
18.2 2.3
No mean
No mean
15
25
28
30
28
30
25
32
13
5.8–35.4
12–14.4
7.1–30.9
5–30
–
21–35
28.6–32
Leaf & Renshaw (1957)
Ussing (1966)
Vieira et al. (1972)
Nellans & Finn (1974)
Lassen & Thaysen (1961)
Lassen et al. (1961)
Thurau (1961)
Deetjen & Kramer (1961)
Torelli et al. (1966)
Martin & Diamond (1966)
Frederiksen & Leyssac (1969)
Frederiksen & Leyssac (1969)
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
183
Solute-coupled water transport
Æ E H Larsen et al.
Maunsbach et al. 2000), and as the in vivo studies were
carried out on whole kidneys, (3) paracellular electrodiffusion in thick ascending limb of Henle’s loop. The
point here is that the apical NKCC2 co-transporter in
series with the basolateral Na+/K+ pump in the thick
ascending limb of loop of Henle imply a ratio of 36
Na+/O2 if the paracellular shunt current is carried
exclusively by Na+ (Greger 1985). About 25% of the
primary urine is reabsorbed in this segment; hence the
high efficiency of this segment affects measurably the
overall energy expenditure of the kidney. With the
demand of a very small recirculation of Na+ in the
proximal tubule together with the above mechanisms in
the kidney the overall energetic efficiency of Na+
reabsorption exceeds significantly 18 Na+/O2 (Larsen
et al. 2006). Likewise, model simulation of gallbladder
with a ratio of 25 Na+/O2 when exposed to 300 mOsm
Ringer’s solution predicted a decrease to 12 Na+/O2
when the external baths were diluted to 50 mOsm.
This was caused by a large decrease in the paracellular
convection flux relative to the decrease in the recirculation flux required for isotonic transport (Larsen et al.
2000).
In conclusion, the Na+ recirculation theory with the
lateral intercellular space as the coupling compartment
of solute-coupled water absorption in a logical way
handles metabolic costs of active Na+ transport below
that of the Na+/K+ pump. Metabolic costs exceeding
that of 18 Na+/O2 of the Na+/K+ pump as observed in
both high- and low-resistance epithelia follow directly
from the observation that a varying fraction of
the Na+,K+-ATPase-driven transepithelial Na+ uptake
bypasses the Na+/K+ pump.
Conflict of interest
There are no conflicts of interest for our work.
Technical assistance in the laboratory by Mrs Pernille Roshof is
greatly appreciated. Thanks are also due to Mr Arne Nielsen
for fabrication and servicing of equipment. Our laboratory at
the University of Copenhagen is supported by frame grant
272-05-0417 from the Danish Natural Science Research
Council and by the Carlsberg Foundation.
References
Alpern, R.J. & Chambers, M. 1987. Basolateral membrane Cl/
HCO3 exchange in the rat proximal convoluted tubule.
Na-dependent and -independent modes. J Gen Physiol 89,
581–598.
Boron, W.F. & Boulpaep, E.L. 1983. Intracellular pH regulation in the renal proximal tubule of the salamander. Basolateral HCO3) transport. J Gen Physiol 81, 53–94.
Curran, P.F. 1960. Na, Cl, and water transport by rat ileum
in vitro. J Gen Physiol 43, 1137–1148.
184
Acta Physiol 2009, 195, 171–186
Curran, P.F. & MacIntosh, J.R. 1962. A model system for
biological water transport. Nature 193, 347–348.
Curran, P.F. & Solomon, A.K. 1957. Ion and water fluxes in
the ileum of rats. J Gen Physiol 41, 143–168.
Deetjen, P. & Kramer, K. 1961. Die Abhängigheit des O2Verbrauchs der Niere von der Na-Rückresorption. Pflügers
Arch Gesamte Physiol Menschen Tiere 273, 636–650.
Devuyst, O., Beaujean, V. & Crabbé, J. 1990. Effects of
environmental conditions on mitochondrial-rich cell density
and chloride transport in toad skin. Pflugers Arch 417, 577–
581.
Diamond, J.M. 1962. The mechanism of solute transport by
the gall-bladder. J Physiol 161, 474–502.
Diamond, J.M. 1964. The mechanism of isotonic water
transport. J Gen Physiol 48, 15–42.
Diamond, J.M. & Bossert, W.H. 1967. Standing-gradient
osmotic flow. A mechanism for coupling of water and solute
transport in epithelia. J Gen Physiol 50, 2061–2083.
Dörge, A., Rick, R., Beck, F.-X. & Thurau, K. 1985. Cl
transport across the basolateral membrane in frog skin epithelium. Pflugers Arch 405 (Suppl. 1), S8–S11.
Dörge, A., Beck, F.-X., Wienecke, P. & Rick, R. 1989. Cl
transport across the basolateral membrane of principal cells
in frog skin. Miner Electrolyte Metab 15, 155–162.
Ernst, S.A. & Mills, J.W. 1977. Basolateral plasma membrane
localization of ouabain-sensitive sodium transport sites in
the secretory epithelium of the avian salt gland. J Cell Biol
75, 74–94.
Fischbarg, J., Diecke, F.P., Iserovich, P. & Rubashkin, A. 2006.
The role of the tight junction in paracellular fluid transport
across corneal endothelium. Electro-osmosis as a driving
force. J Membr Biol 210, 117–130.
Frederiksen, O. & Leyssac, P.P. 1969. Trancellular transport of
isosmotic volumes by the rabbit gall-bladder in vitro.
J Physiol 201, 201–224.
Frömter, E. 1979. The Feldberg lecture 176: Solute transport
across epithelia: what can we learn from micropuncture
studies on kidney tubules? J Physiol 288, 1–31.
Frömter, E., Rumrich, G. & Ullrich, K.L. 1973. Phenomenologic
description of Na+, Cl) and HCO3) absorption from proximal
tubules of the rat kidney. Pflugers Arch 343, 189–220.
Fuchs, W., Larsen, E.H. & Lindemann, B. 1977. Current–
voltage curve of sodium channels and concentration dependence of sodium permeability in frog skin. J Physiol 267,
137–166.
Greger, R. 1985. Ion transport mechanisms in thick ascending
limb of Henle’s loop of mammalian nephron. Physiol Rev
65, 760–797.
Grubb, B.R., Schiretz, F.R. & Boucher, R.C. 1997. Volume
transport across tracheal and bronchial airway epithelia
in a tubular culture system. Am J Physiol 273, C21–
C29.
Hasegawa, T., Tanii, H., Suzuki, M. & Tanaka, S. 2003.
Regulation of water absorption in the frog skins by two
vasotocin-dependent water-channel aquaporins, AQP-h2
and AQP-h3. Endocrinology 144, 4087–4096.
Hill, A.E. & Shachar-Hill, B. 2006. A new approach to epithelial isotonic fluid transport: an osmosensor feedback
model. J Membr Biol 210, 77–90.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Acta Physiol 2009, 195, 171–186
Ikonomov, O., Simon, M. & Frömter, E. 1985. Electrophysiological studies on lateral intracellular spaces of
Necturus gallbladder epithelium. Pflugers Arch 403, 301–
307.
Kalman, S.M. & Ussing, H.H. 1955. Active sodium uptake by
the toad and its response to the antidiuretic hormone. J Gen
Physiol 38, 361–370.
Kashgarian, M., Biemesderfer, D., Caplan, M. & Forbush, B.,
III. 1985. Monoclonal antibody to Na,K-ATPase: immunocytochemical localization along nephron segments. Kidney
Int 28, 899–913.
Larsen, E.H. 1991. Chloride transport by high-resistance
heterocellular epithelia. Physiol Rev 71, 235–283.
Larsen, E.H. & Harvey, B.J. 1994. Chloride currents of single
mitochondria-rich cells of toad skin epithelium. J Physiol
478, 7–15.
Larsen, E.H., Willumsen, N.J. & Christoffersen, B.C. 1992.
Role of proton pump of mitochondria-rich cells for active
transport of chloride ions in toad skin epithelium. J Physiol
450, 203–216.
Larsen, E.H., Sørensen, J.B. & Sørensen, J.N. 2000. A mathematical model of solute coupled water transport in toad
intestine incorporating recirculation of the actively transported solute. J Gen Physiol 116, 101–124.
Larsen, E.H., Kristensen, P., Nedergaard, S. & Willumsen, N.J.
2001. Role of mitochondria-rich cells for passive chloride
transport, with a discussion of Ussing’s contribution to our
understanding of shunt pathways in epithelia. J Membr Biol
184, 247–254.
Larsen, E.H., Sørensen, J.B. & Sørensen, J.N. 2002. Analysis of
the sodium recirculation theory of solute-coupled water
transport in small intestine. J Physiol 542, 33–50.
Larsen, E.H., Møbjerg, N. & Sørensen, J.N. 2006. Fluid
transport and ion fluxes in mammalian kidney proximal
tubule: a model analysis of isotonic transport. Acta Physiol
(Oxf) 187, 177–189.
Larsen, E.H., Møbjerg, N. & Nielsen, R. 2007. Application of
the Na+ recirculation theory to ion coupled water transport
in low- and high resistance osmoregulatory epithelia. Comp
Biochem Physiol A Mol Integr Physiol 148, 101–116.
Lassen, U.V. & Thaysen, J.H. 1961. Correlation between
sodium transport and oxygen consumption in isolated renal
tissue. Biochim Biophys Acta 47, 616–618.
Lassen, N.A., Munck, O. & Thaysen, J.H. 1961. Oxygen
consumption and sodium reabsorption in the kidney. Acta
Physiol Scand 51, 371–384.
Leaf, A. & Renshaw, A. 1957. Ion transport and respiration of
isolated frog skin. Biochem J 65, 82–90.
Loo, D.D.F., Zeuthen, T., Chandy, G. & Wright, E.M. 1996.
Cotransport of water by the Na+/glucose cotransporter. Proc
Natl Acad Sci USA 93, 13367–13370.
MacRobbie, E.A.C. & Ussing, H.H. 1961. Osmotic behaviour
of the epithelial cells of frog skin. Acta Physiol Scand 53,
348–365.
Martin, D.W. & Diamond, J.M. 1966. Energetic of coupled
active transport of sodium and chloride. J Gen Physiol 50,
295–315.
Maunsbach, A.B. & Boulpaep, E.L. 1991. Immunoelectron
microscope localization of Na,K-ATPase in transport path-
E H Larsen et al.
Æ Solute-coupled water transport
ways in proximal tubule epithelium. Micron and Microscopy
Acta 22, 55–56.
Maunsbach, A.B., Vorum, H., Kwon, T.H., Nielsen, S., Simonsen, B., Choi, I., Schmitt, B.M., Boron, W.F. & Aalkjaer, C. 2000. Immunoelectron microscopic localization of
the electrogenic Na/HCO3 cotransporter in rat and ambystoma kidney. J Am Soc Nephrol 11, 2179–2189.
Meinild, A.-K., Klaerke, D.A., Loo, D.D.F., Wright, E.M. &
Zeuthen, T. 1998. The human Na+-glucose cotransporter is
a molecular water pump. J Physiol 508, 15–21.
Mills, J.W. & DiBona, D.R. 1978. Distribution of Na+-pump
sites in the frog gallbladder. Nature 271, 273–275.
Mills, J.W. & Ernst, S.A. 1975. Localization of sodium pump
sites in frog urinary bladder. Biochim Biophys Acta 375,
268–273.
Mills, J.W., Ernst, S.A. & DiBona, D.R. 1977. Localization of
Na+-pump sites in frog skin. J Cell Biol 73, 88–110.
Nedergaard, S., Larsen, E.H. & Ussing, H.H. 1999. Sodium
recirculation and isotonic transport in toad small intestine.
J Membr Biol 168, 241–251.
Nellans, H.N. & Finn, A.L. 1974. Oxygen consumption and
sodium transport in the toad urinary bladder. Am J Physiol
227, 670–675.
Nielsen, R. 1982. Effect of ouabain, amiloride, and antidiuretic
hormone on the sodium-transport pool in isolated epithelia
from frog skin (Rana temporaria). J Membr Biol 65, 221–226.
Nielsen, R. 1997. Correlation between transepithelial Na+
transport and transepithelial water movement across isolated
frog skin (Rana esculenta). J Membr Biol 159, 61–69.
Nielsen, R. & Larsen, E.H. 2007. Beta-adrenergic activation of
solute coupled water uptake by toad skin epithelium results
in near-isosmotic transport. Comp Biochem Physiol A Comp
Physiol 148, 64–71.
Nielsen, S., Frokiaer, J., Marples, D., Kwon, T.H., Agre, P. &
Knepper, M.A. 2002. Aquaporins in the kidney: from molecules to medicine. Physiol Rev 82, 205–244.
Parsons, D.S. & Wingate, D.L. 1958. Fluid movements across
the wall of rat small intestine in vitro. Biochim Biophys Acta
30, 666–667.
Pihakaski-Maunsbach, K., Vorum, H., Locke, E.M., Garty, H.,
Karlish, S.J. & Maunsbach, A.B. 2003. Immunocytochemical localization of Na,K-ATPase gamma subunit and CHIF
in inner medulla of rat kidney. Ann N Y Acad Sci 986, 401–
409.
Reid, E.W. 1892a. Preliminary report on experiments upon
intestinal absorption without osmosis. Br Med J 1, 1133–
1134.
Reid, E.W. 1892b. Report on experiments upon ‘absorption
without osmosis’. Br Med J 1, 323–326.
Sørensen, J.B. & Larsen, E.H. 1996. Heterogeneity of chloride
channels in the apical membrane of isolated mitochondriarich cells from toad skin. J Gen Physiol 108, 421–433.
Spring, K.R. & Hope, A. 1979. Fluid transport and the
dimensions of cell and interspaces of living Necturus gallbladder. J Gen Physiol 73, 287–305.
Sten-Knudsen, O. & Ussing, H.H. 1981. The flux ratio equation
under nonstationary conditions. J Membr Biol 63, 233–242.
Stirling, C.E. 1972. Radioautographic localization of sodium
pump sites in rabbit intestine. J Cell Biol 53, 704–714.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
185
Solute-coupled water transport
Æ E H Larsen et al.
Thaysen, J.H. & Thorn, N.A. 1954. Excretion of urea, potassium and chloride in human tears. Am J Physiol 178, 160–164.
Thurau, K. 1961. Renal Na-reabsorption and O2-uptake in
dogs during hypoxia and hydrochlorothiazide infusion. Proc
Soc Exp Biol Med 106, 714–717.
Torelli, G., Milla, E., Faelli, A. & Costantini, S. 1966. Energy
requirement for sodium reabsorption in the in vivo rabbit
kidney. Am J Physiol 211, 576–580.
Ussing, H.H. 1966. Anomalous transport of electrolytes and
sucrose through the isolated frog skin induced by hypertonicity of the outside bathing solution. Ann N Y Acad Sci 137,
543–555.
Ussing, H.H. 1985. Volume regulation and basolateral
co-transport of sodium, potassium, and chloride ion in frog
skin epithelium. Pflugers Arch 405 (Suppl. 1), S2–S7.
Van Driessche, W. & Lindemann, B. 1979. Concentration
dependence of currents through single sodium-selective pores
in frog skin. Nature 282, 519–520.
Vieira, F.L., Caplan, S.R. & Essig, A. 1972. Energetics of sodium transport in frog skin. I. Oxygen consumption in the
short-circuited state. J Gen Physiol 59, 60–76.
Voûte, C.L. & Meier, W. 1978. The mitochondria-rich cell of
frog skin as hormone-sensitive ‘‘shunt-path’’. J Membr Biol
40, 151–165.
Weinstein, A.M. 2008. Sodium and chloride transport. Proximal nephron. In: R.J. Alpern & S.C. Hebert (eds) Seldin and
186
Acta Physiol 2009, 195, 171–186
Giebisch’s The Kidney. Physiology and Pathophysiology, pp.
793–847. Academic Press, Elsevier, New York.
Weinstein, A.M. & Stephenson, J.L. 1981. Models of coupled
salt and water transport across leaky epithelia. J Membr Biol
60, 1–20.
Willumsen, N.J. & Larsen, E.H. 1986. Membrane potentials
and intracellular Cl) activity of toad skin epithelium in
relation to activation and deactivation of the transepithelial
Cl) conductance. J Membr Biol 94, 173–190.
Willumsen, N.J., Vestergaard, L. & Larsen, E.H. 1992. Cyclic
AMP-and b-agonist-activated chloride conductance of a toad
skin epithelium. J Physiol 449, 641–653.
Windhager, E.E., Whittembury, G., Oken, D.E., Schatzmann,
H.J. & Solomon, A.K. 1959. Single proximal tubules of the
Necturus kidney. III. Dependence of H2O movement on
NaCl concentration. Am J Physiol 197, 313–318.
Zerahn, K. 1958. Oxygen consumption and active sodium
transport in isolated amphibian skin under varying experimental conditions. Dissertation. University of Copenhagen
Universitetsforlaget Aarhus, 1–40.
Zeuthen, T. 1983. Ion activities in the lateral intracellular
spaces of gallbladder epithelium transport at low external
osmolarities. J Membr Biol 76, 113–122.
Zeuthen, T., Meinild, A.K., Loo, D.D., Wright, E.M. & Klaerke,
D.A. 2001. Isotonic transport by the Na+-glucose cotransporter
SGLT1 from humans and rabbit. J Physiol 531, 631–644.
2009 The Authors
Journal compilation 2009 Scandinavian Physiological Society, doi: 10.1111/j.1748-1716.2008.01930.x
Osmotic and Ionic Regulation – Department of Biology - University of Copenhagen
http://www1.bio.ku.dk/english/research/fys/ehl/[19-03-2012 12:21:25]