Journal of Experimental Botany, Vol. 46, No. 283, pp. 199-209, February 1995
Journal of
Experimental
Botany
Reversible closing of water channels in Chara internodes
provides evidence for a composite transport model of the
plasma membrane
Tobias Henzler and Ernst Steudle1
Lehrstuhl Pflanzenokologie, University Bayreuth, D-95440 Bayreuth, Germany
Received 12 July 1994; Accepted 24 October 1994
Abstract
Key words: Anomalous osmosis, Chara, composite membrane, hydraulic conductivity, permeability coefficient,
reflection coefficient, water channels.
Treatment of internodal cells of Chara corallina with
the water channel blocker HgCI 2 caused a decrease of
the hydraulic conductivity of the membrane (Lp) by a
factor of three to four. In the presence of (practically)
Introduction
non-permeating solutes such as sugars or salts, the
In both plant and animal cells, highly specialized channels
osmotic responses were similar to those found in con(aquaporins) facilitate the transport of water across cell
trols, i.e. reflection coefficients (<rj were close to unity.
membranes
and contribute to most of the cell's hydraulic
However, when treating the internodes with osmotic
conductivity
(see books and reviews by Macey, 1984;
solutions of rapidly permeating lipophilic substances
Stein,
1986;
Finkelstein, 1987; Preston et al., 1992;
such as low molecular weight alcohols or acetone, the
Verkman,
1992;
Chrispeels and Maurel, 1994). Water
picture of biphasic pressure relaxations due to the
channels
are
proteins
which belong to an ancient family
exchange of both water and solutes changed considerof
integral
membrane
proteins, called the MIP family
ably. In the presence of HgCI 2 , reflection coefficients
(MIP
=
membrane
intrinsic
protein; Reizer et al., 1993).
were substantially reduced and even became negative
The
polypeptide
chains
of
aquaporins usually have
for some solutes (anomalous osmosis). Different from
molecular
weights
around
30
kDa
and span the membrane
reflection coefficients, permeability coefficients (PJ
six
times
to
form
the
water
pore
(Preston et al., 1992;
remained constant upon treatment. When using heavy
Chrispeels
and
Maurel,
1994).
Water
channels are reverswater (HDO) as a small hydrophilic solute which should
ibly
blocked
by
sulphydryl
reagents
such as mercurials
cross the membrane largely via water channels, results
(e.g.
/?-chloromercuribenzene
sulphonate,
/7-CMBS, or
were different: the reflection coefficient of HDO
mercuric
chloride,
HgCl
2; Preston et al., 1992; Maurel
increased and permeability decreased. Treating the
et al., 1993). In plants, water channels have been demoncells with 5 mM 2-mercaptoethanol to remove the merstrated in the tonoplast (H6fte et al., 1992), and also in
cury from transport proteins (water channels) reversed
the plasma membrane (Kammerloher and Schaffner,
changes in Lp and <rt. The results disagree with con1993). To date, the evidence that aquaporins are formed
ventional membrane models, which imply a homogenis
largely based on observations on frog oocytes. When
eous membrane. They are readily explained by a
in
vitro synthesized mRNA encoding for aquaporins is
composite model of the membrane in which proteininjected
into frog eggs, channel proteins are synthesized
aceous arrays with specific water channels are
which
increase
the osmotic permeability (hydraulic conarranged in parallel with lipid arrays. The latter
ductivity,
Lp)
of
the membrane 5-10-fold (Macey, 1984;
account for most of the permeability of the small
Preston
et
al.,
1992).
Using the technique of transcellular
organic test solutes used and are not affected by the
osmosis,
Wayne
and
Tazawa
(1990) observed a reversible
channel blocker. From the data, detailed information
decrease
of
Lp
of
the
internodes
of the characean
is obtained about transport properties of the arrays
Nitellopsis
obtusa
when
treated
with
/7-CMBS
which was
(water channels).
1
To whom correspondence should be addressed. Fax: +49 921 55 2564.
© Oxford University Press 1995
200
Henzler and Steudle
interpreted as a closing of water channels. Changes of
the hydraulic conductivity of plant cell membranes {Lp)
following treatment with mercurials have not yet been
measured using the cell pressure probe. As shown in this
paper, the latter technique also allows one to measure
solute permeability and reflection coefficients besides the
water, at least for isolated cells.
As aquaporins are quite selective for water, they may
close without affecting solute transport (Maurel et al.,
1993). On the other hand, the selective blockage of ion
channels or of transport systems for low molecular weight
organic solutes did not affect water permeability which
suggests no significant co-transport or drag of water with
these solutes in other transport systems (Maurel et al.,
1993), although there are exceptions (Wayne and Tazawa,
1990). It appears that aquaporins represent narrow channels, just sufficient to allow for the passage of individual
water molecules. Hence, it has been suggested that water
would have to move along the channels one by one in a
single file (Levitt, 1974; Finkelstein, 1987).
Classical pore models of membranes may be applied
to transport across water channels such as the frictional
models of Kedem and Katchalsky (1961) or Dainty and
Ginzburg (1963) which are derived from irreversible
thermodynamics. Besides the hydraulic conductivity {Lp)
and the solute permeability (permeability coefficient, PJ,
the selectivity of membranes expressed by the reflection
coefficient, a, would be important. In terms of the Dainty
and Ginzburg (1963) version of the frictional model, a,
would be given by:
V xP
K•
~LpxRT~f,w-
(1)
for a given solute denoted by subscript 's'. Here, V,=
partial molar volume of solute 's'; /1W = coefficient
expressing the frictional interaction between V and water
as they cross the membrane; / m = coefficient expressing
the friction between solute and membrane; K, = distribution coefficient of solute 's' between solution in the
membrane and bulk medium. The reflection coefficient is
a dimensionless number of < = 1 which provides a quantitative measure of membrane selectivity. It is seen from
Eq. (1) that for an ideally semipermeable membrane
a, = 1, and selectivity is maximal. In this case, the solute
permeability is zero (/*s = 0) and, as the solute is excluded
from the membrane, also/, w = 0. If there is a solute flow
and water and solute move separately across the membrane, e.g. by an independent diffusion across a lipid
layer,/„ will be zero as well. However, in this case, a,< 1,
because the second term on the right side of Eq. (1) will
not vanish. This term describes the contribution of solute
flow (numerator) in relation to the opposing water
(volume) flow (denominator) during osmosis in the presence of a permeant solute. For the solutes usually present
in cells, the contribution of solute flow to the overall
volumeflowwill be negligibly small because the permeability of the solutes in cells is lower than that of water by
several orders of magnitude. Hence, for these solutes
a, « 1 will hold, i.e. cells retain their solutes having a
low P,.
An important conclusion from Eq. (1) is that, if there
is no common path for water and solute (flw=0), the
last term on the right side vanishes. The remaining
equation then either describes transport across a 'film
membrane' (e.g. a lipid bilayer) or across a membrane
containing strictly selective transport systems which do
not allow frictional interactions betweenflows.According
to the picture given above about membranes in which
water and solutes use different selective channels and
other transport systems, plasma membranes should largely refer to the latter type of membrane. However, the
combination of such a model with classical pore models
causes difficulties, since the second term on the right side
of Eq. (1) is usually small even for rapidly permeating
solutes having a high Pt (e.g. only up to about 0.1 for
Chara; Steudle and Tyerman, 1983). When water and
solutes do move separately and the interaction term
vanishes, measured reflection coefficients of permeating
solutes such as low molecular weight organic substances
are much lower than would be expected from Eq. (1).
Thus, provided that interactions between flows are small,
the frictional pore model which has been used as an
important basis for understanding couplings between
water and solute transport during the past 30 years
(Dainty and Ginzburg, 1964a, b; Tyerman and Steudle,
1982; Steudle and Tyerman, 1983; Steudle, 1989) fails to
interpret important experimental findings.
The reason for the failure is that the model implies a
homogeneous membrane. Because the plasma membrane
consists of arrays or regions of different composition and
structure which also exhibit different transport properties
(transport coefficients Lp, Ps, a j , this assumption may
not hold. Plasma membranes should be 'patchy' or exhibit
some kind of a mosaic structure. For such a structure,
again, basic models derived from thermodynamics of
irreversible processes apply, i.e. the concept of composite
membranes (Kedem and Katchalsky, 1963). This concept
shows that composite membranes may exhibit transport
properties {Lp, P,, a,) which are not simply additive of
those of the elements (membrane patches) they are composite of and to which 'homogenous concepts' such as
the lipid-pore model may apply. To date, composite
membrane models have not really been applied to cell
membranes although their composite structure is evident.
In this paper, we show for the alga Chara corallina
that the hydraulic conductivity is reversibly affected by
mercuric chloride indicating an opening or closing of
water channels. We demonstrate that changes in solute
permeability and of reflection coefficients associated with
Water channels in Chara 201
changes in Lp cannot be explained in terms of conventional models, but are readily explained by a composite
model of the plasma membrane. If the conclusions drawn
for Chara of a composite transport system also hold for
other plant cells, this should have considerable consequences for our understanding of water flow and
osmosis in plants. The application of simple membrane
models should be cautious.
To a good approximation, the entire process following a change
in the external concentration of permeating solute is given by
(Steudle and Tyerman, 1983; Steudle, 1993):
V(t)-Vo
P(t)-Po
a,A-n°Lp
[exp(-k,xt)-exp(-k,xt)].
(3)
Equation (3) was derived neglecting solvent drag, i.e. the
frictional interaction between water and solutes as they cross
the membrane. However, by comparing the analytical (Eq. (3))
Materials and methods
and numerical solutions (which included the solvent drag) with
experimental results, it was shown for Chara internodes that
Plant material
the theory excellently fitted the experimental results even for
Chara corallina was grown in artificial pond water (APW) rapidly permeating solutes with tow reflection coefficients
(composition irtmM: 1 NaCl, fl.r KCi, 0.1 CaCl2, O.t MgCl2)
(Rudinger et al., 1992). Solvent drag effects were negligible.
in tanks which contained a layer of natural pond mud. Tanks
For the solutes used in this paper, it held that k^»kt. Thus,
were placed in a greenhouse without additional illumination.
T\/2 (k,) was calculated from biphasic response curves at timeThe internodes used were 20 to 70 mm in length and 0.8 to intervals
where the first term in the brackets on the right side
1.0 mm in diameter.
of Eq. (3) had vanished. T\l2 (k,) was directly related to the
permeability coefficient of the solute (/",) given, i.e.:
Calculation of transport parametersfl-P,P., and a j
Transport parameters were calculated from 'hydrostatic' (Lp)
and 'osmotic' (Lp, P,, a,) experiments as previously described
(Steudle and Tyerman, 1983; Rfldinger et al., 1992; Steudle,
1993, 1994). In hydrostatic experiments, turgor pressure (P)
was rapidly changed with the aid of the cell pressure probe to
disturb the water flow equilibrium and to induce water flows
across the cell membrane. In osmotic experiments, the osmotic
pressure of the medium (APW) was changed. In response to
deflections, relaxations in cell turgor were recorded. Half-times
(77)2) or rate constants (kw) of water flow equilibration were
measured from P(t) curves on recorder strips. To evaluate
volume (water) flows from P(t) curves (pressure relaxations),
it was necessary also to measure the elastic coefficient of the
cells (elastic modulus, <r; <r = Vx dP/dVx VxAP/AV; K=cell
volume). Half-times of relaxations (7Y/2) are directly related to
the hydraulic conductivity (water permeability) of the cell (Lp)
by (Steudle, 1993, 1994):
Vxln(2)
(2)
LpxAx
Here, A =cell surface area and n' = osmotic pressure of cell sap.
A and V were obtained from the cell dimensions measured with
a caliper or using a stereo microscope, w' was calculated from
the initial cell turgor (Po) and the osmotic pressure of the
medium (APW), since />O = TT' — -n° (77° = osmotic pressure of the
medium).
Equation (2) was used for both hydrostatic and osmotic
pressure relaxations in the presence of non-permeating solutes.
In order to determine permeability (P,) and reflection (a,)
coefficients in the presence of permeating test solutes, 'biphasic'
osmotic pressure relaxations had to be analysed. There was a
first phase during which water flows tended to equilibrate. This
phase was rapid because of the high mobility (permeability) of
water. The 'water phase' (which also contained effects due to
solute movement) was followed by a 'solute phase' during
which the concentration gradient imposed across the membrane
equilibrated by permeation of solute V across the membrane.
The rate of the solute phase strongly depended on the nature
of the solute used. Solutes which were soluble in the lipid phase
of the membrane had short half-times; those which were polar
(ions, hydrophilic solutes) had very long half-times. During the
second phase, P(t) and V(t) were also exponential with time.
ln(2) Vln(2)
(4)
Thus, P, was also worked out. Depending on whether hypertonic
(ATT°>0) or hypotonic (An°<0) changes are made, Eq. (3)
described biphasic pressure relaxations with minima or maxima,
respectively. It is verified from Eq. (3) that the reflection
coefficient for a given solute 's' is derived from the maximum
change in turgor pressure (Po-P^n or Po — PmiJ and from the
change in the osmotic pressure of the medium, ATT°. The
reflection coefficient (a,) is given by (Steudle and Tyerman,
1983):
P —P min(max)
£+77'
(5)
The second term on the right side of this equation ((e + 7r')/t)
corrects for the shrinking or swelling of the cell and the third
term (exp(k, x t ^ , , ^ , ) for theflowof permeating solute during
the time necessary to reach the minimum (or maximum;
'min(maJi)/ •
Measurement of transport coefficients (Lp, P., and a j
An internode was freed from branches and from adjacent
internodes and placed in a glass tube (internal diameter: 3 mm)
where it rested in a slit to make the cell secure and to avoid
vibrations which would cause leakages during turgor measurements. Because of the rapid perfusion of the glass tube with
solution (APW or test solutions), the liquid around the cell was
vigorously stirred thus minimizing the thickness of external
unstirred layers (Steudle and Tyerman, 1983). Flow rates in the
tube were 0.15-0.2 m s"1. Thus, APW and test solutions could
be rapidly exchanged in times which were much shorter than
?T/2 or T\n.
The cell pressure probe was introduced into the internodes
across the nodes. The oil/cell sap meniscus in the tip of the
probe served as a reference point during the measurements.
Elastic moduli (e) were determined by rapidly changing the cell
volume with the aid of the probe. Volume changes (A V) were
calculated from the shift of the meniscus and from the diameter
of the capillary. Changes in turgor (AP) were read from the
chart recorder. Since t depended on cell turgor, the changes in
202
Henzler and Steudle
turgor used to evaluate e were no larger than ±0.15 MPa at
full turgor (0.55-0.70 MPa).
In order to determine the 'hydrostatic Lp', step changes of
turgor were produced with the aid of the probe, but the position
of the meniscus kept constant after the change. This resulted in
pressure relaxations and, depending on the direction of the
change, endosmotic or exosmotic water flows. Using Eq. (2),
Lp was calculated from the 77/2 (kw) of relaxations. During
'osmotic' experiments, ethanol, acetone, formamide, and
dimethylformamide were used as permeating, and mannitol and
NaCl as practically non-permeating solutes. With four cells,
heavy water (HDO) was also used as an osmoticum which is
expected to have a permeability close to that of light water
(H2O) and a reflection coefficient close to zero. The external
medium was rapidly replaced by a test solution containing the
test solute at a known concentration. Relaxations were again
measured to obtain 77/2 ar) d T\jz. Except for HDO where
external concentrations were about 2 M, the external concentrations used during the osmotic experiments ranged between
50-200 raM. This ensured that the concentration dependence
of Lp was negligible when this parameter was evaluated from
osmotic pressure relaxations (Steudle and Tyerman, 1983).
Effects of HgCI2 on water and solute transport
First, Chora internodes were measured in the absence of the
water channel blocker HgCl2. Then they were treated for
10-15 min with 50 /xM HgCl2 in APW to induce changes in Lp,
solute permeability and reflection coefficients. Cells treated in
this way revealed a stable turgor for about 1 h and then steadily
lost turgor. However, when the mercury was removed from the
cells about 30 min after the treatment using a 4-5 mM solution
of 2-mercaptoethanol in APW, internodes remained turgid for
about 3 h. When changes in Lp, P,, and a, were determined,
the parameters were first measured in the absence of HgCl2 and
then right after the treatment in which test solutions also
contained 50 fiM HgCl2. These measurements were immediately
followed by a 5 min treatment with 2-mercaptoethanol and
measurements of Lp, P, and a, were repeated to demonstrate
whether or not the changes induced by HgCl2 treatment were
reversible. To indicate that cells maintained semi-permeability
with respect to the major solutes (salts) in the vacuoles, it was
verified that cell turgor remained constant during the experiments which lasted for about 2 h with individual cells.
Composite membrane model: parallel an~angement of elements
(Kedem and Katchalsky, 1963)
This model formed the basis of interpreting the results. In a
membrane composed of two different parallel arrays 'a' and 'b'
which exhibit different water and solute permeability (transport
coefficients Lp* and Lpb and P* and /*, respectively) as well as
different reflection coefficients (<rj and ab), the hydraulic
conductivity will be given by:
= ym x Lp' + yb x Lpb
(6)
b
where y* and y denote the fractional contributions of 'a' and
'b' to the overall surface area ( / + y b =l). Equation (6)
straightforwardly states that the elements of the composite
membrane will contribute to the overall Lp according to their
hydraulic conductances. The specific reflection coefficients of
the elements (a* and <jb) will contribute to the overall a,
according to their relative hydraulic conductances, i.e.:
'xLpb
Lp
(7)
The overall solute permeability (P,) will be larger than expected
from the contribution of the arrays to the overall surface area
as one would think in analogy to Eq. (6). The difference is due
to solvent drag effects:
, = / X I»,+yb X Pj
Lp*xLpb
x RTJL-——
Lp
X C,
(8)
The last term on the right side of the equation represents the
solvent drag which is proportional to the square of the difference
between reflection coefficients, c, is the mean concentration of
's' in the membrane. It can be seen from Eq. (8) that the
solvent drag term will vanish, if (aj —ob) = 0 (homogeneous
membrane) or if either pathways 'a' or 'b' are blocked for water
(Lp*=0or Lpb = 0).
Equations (6) to (8) have been used to interpret the results
in terms of a composite membrane model from experiments in
which water channels were either blocked or open. In principle,
by blocking water channels (array 'a' in Eqs (6) to (8)), either
the number of channels (ym), their water permeability (Lp*), or
both could have been reduced. However, it is thought that
blocking off a water channel is an all-or-none reaction and that
channels may be titrated by adding blocking agents (Macey,
1984). It can be seen from Eqs (6) and (7) that a blocking of
channels (pathway 'a') would tend to shift the overall a, to a
value closer to a,. If channels were completely blocked, the
system should assume the reflection coefficient of path 'b' and
an overall hydraulic conductivity of yb x Lpb.
If channels are ideally selective for water and only the parallel
lipid array (array 'b') is permeable to the solute, the solute
permeability of the membrane will be given by yb x /* (Eq. (8)).
This simplifying picture would assume that proteinaceous arrays
different from the water channels would be impermeable for
water. If selective water channels are blocked, this should not
affect the overall P,.
Table 1. Half-times of water exchange (T7/J of internodes o/Chara corallina ay obtained from relaxations of cell turgor in hydrostatic
experiments (Eq. (2))
TA denotes control measurements and TB measurements after treating the cells with 50 ^M HgCl2 for 10-15 min. T c values were obtained after a
5 min treatment with 5 mM 2-mercaptoethanol to remove the mercury from the cell membrane. Data are means from 15 or 19 individual cells±SD.
It can be seen that (on average) TB/T A was 3.5. According to Eq. (2), the increase in 77/2 was due to a decrease of the hydraulic conductivity (Lp)
by the same factor, i.e. Lp was reduced by 71%. The rauo of TA/TC was 0.93 (on average) indicating that changes were completely reversible after
treatment with 2-mercaptoethanol.
Half-time of water exchange, 77* M
TA
n
TB
n
Tc
n
TB/TA
TB/TC
TJTC
3.6 ±1.0
19
12.3±3.8
19
3.8 ±1.0
15
3.5 ±0.9
3.3±0.8
0.93 ±0.2
Water channels in Chara
Results
Typical relaxations of cell turgor in the presence of
hydrostatic driving forces (hydraulic water flows) are
shown in Fig. 1. It can be seen from the figure that a
10-15 min treatment with HgCl2 caused the half-time of
water exchange to increase by a factor of 2.5 for this cell.
Since the cell geometry (V/A ratio) as well as the elastic
modulus of the cell and the cell's osmotic pressure
remained constant (data not shown), this means that the
hydraulic conductivity (Lp) decreased by the same factor
(Eq. (2)). Table 1 indicates that, for all cells, the treatment
with HgCl2 caused an average reduction of a factor of
3.5. It is seen from Fig. 1 and Table 1 that the reduction
in water permeability was reversed by treating the cells
T^-50.0.
01020304060
0
120 240 360 460 600 720
Q_ OS
0 2 0 4 0 6 0 8 0 1 0 0
0
120 240 380 480 600
o
p
© after 5 mM 2-mercaptoethanol treatment
0.8
0.7
06
i 188 mM acetone
03
0
10 20 30 40
120
240
360
480
800
720
840
Time, t [s]
Fig. 1. Water and solute transport in Chara corallina: effects of the
water channel blocker HgCl2- The effect of treating an isolated internode
with a 50 fiM solution of HgCl2 in APW is shown as measured with
the cell pressure probe. Hydrostatic and osmotic turgor pressure
relaxations were first (A) measured in the absence of the channel
blocker. The osmoticum acetone was added to induce exosmotic water
flow and was then removed again after equilibration (see arrows). (B)
Treatment with HgCl2 for 10 min resulted in an increase of the halftime of water exchange (7"7/2) by a factor of 2.5 (on average), i.e. the
hydraulic conductivity (Lp) decreased by the same factor (Eq. (2)). In
osmotic experiments, reflection coefficients (cy) decreased from 0.18 to
— 0.16 (on average), i.e. the cell exhibited anomalous osmosis after the
treatment in the presence of acetone. The half-time of solute exchange
(^1/2) (permeability coefficient /"„ Eq. (4)) remained constant. (C) It
can be seen that the effects of the channel blocker were reversible upon
treatment with 5 mM 2-mercaptoethanol for 5 min which removed the
mercury from the membrane. Note the different scales on the time axes.
203
with 5 mM 2-mercaptoethanol which again caused the
half-times to assume values close to the original ones.
Thus, the effect of mercury was completely reversible
under these conditions. Water channels in the Chara cell
membrane which close in the presence of HgCl2, opened
again when mercuric chloride was removed. Similar effects
have been demonstrated for Xenopus oocytes (Preston
et al., 1992). With the latter, a largely reversible reduction
by a factor of 5 to 10 was found which is somewhat
larger than that found for Chara. However, for the frog
eggs, no combined measurements of hydraulic conductivity, reflection coefficients, and permeability coefficients
could be made which would have allowed a much deeper
insight into mechanisms and pathways for water and
solutes (see Discussion).
When using non-permeating and permeating solutes in
osmotic experiments, the hydraulic conductivity was
reduced in the presence of HgCl2 by an amount similar
to that found in hydrostatic experiments (data not
shown). More obvious were the dramatic changes in the
reflection coefficients of some of the solutes. When using
(practically) non-permeating substances such as mannitol
or NaCl, a, did not change and remained close to unity.
However, in the presence of permeating solutes, the
absolute value of the reflection coefficient was substantially reduced. Using acetone as the osmotic solute, it is
shown in Fig. 1 that the reflection coefficient decreased
when water channels were closed, but re-attained its
original value when the cells were treated with
2-mercaptoethanol, thus removing the mercury from the
membranes. For lipophilic solutes (ethanol, acetone)
which have low reflection coefficients (a, = 0.28 and 0.13,
respectively), the reduction resulted in negative a, values
(Fig. 1 and Table 2). This means that the treatment
caused a reversible change from regular to anomalous
osmosis. When treated with hypertonic solutions, turgor
increased and cells took up water instead of shrinking.
Reflection coefficients of small hydrophilic solutes such
as formamide and dimethylformamide (DMF) also
decreased upon treatment with HgCl2. For example, the
reflection coefficient of formamide decreased from 0.97
to 0.83 upon treatment and that of the more lipophilic
DMF from 0.54 to 0.21 (Table 2). This indicated that
there was either still some permeability across the lipid
part (perhaps, more likely) or that there were special
transporters for these solutes in the membrane not affected
by the treatment. According to the composite membrane
model, both possibilities would result in a decrease of a,
upon closing the water channels (Eq. (7)). Solutes which
were practically non-permeating (NaCl and mannitol)
did not change their a,. Obviously, the a, values of lipid
and proteinaceous arrays were both close to unity so that
the overall reflection coefficient remained constant when
closing water channels (and eliminating the contribution
of this path according to Eq. (7)).
204
Henzler and Steudle
Table 2. Reflection (a,) and permeability (P,) coefficients of permeating (acetone, ethanol, dimethylformamide (DMF), andformamide)
and of non-permeating (mannitol and NaCl) solutes for internodes o/Chara corallina as obtained from osmotic pressure relaxations
(osmotic concentration: 80-200
mOsmolkg'1)
Values represent means of different cells ±SD. As in Table 1, superscript A denotes control measurements and B determinations following a
10-15 min treatment with 50 ^M HgCl2. Ratios of Pf/Pf were calculated from values of individual cells ± propagated error. The data show that a,
decreases substantially upon treatment with HgCl2. For rapidly permeating solutes (acetone and ethanol) even negative values of erf were obtained
(anomalous osmosis). Different from a, and Lp (see Table 1), permeability coefficients did not change significantly upon treatment with HgCl2.
No. of cells
3
9
3
3
1
1
Test
solute
Acetone
Ethanol
DMF
Formamide
Mannitol
NaCl
Permeability coefficient, P,
[10" 6 m s"1]
Reflection coefficient,
[1]
o,
o?
of
0.13±0.04
0.28±0.06
0.54±0.10
0.97 ±0.02
0.97
0.99
-0.16 + 0.04
-0.074 ±0.074
0.21 ±0.04
0.83 ±0.03
0.96
1.0
For permeating solutes such as ethanol or acetone,
osmotic responses were biphasic as expected from the
theory (see Material and methods section). From the rate
constant of the solute phase, the permeability coefficient
of the membrane was worked out for the given solute
(P,; Eq. (4)). Different from the reflection coefficient,
there was no change in the permeability coefficient upon
treatment with the blocker despite the fact that water
channels did close (Fig. 1; Table 2). For a given cell and
solute, the rate constant of solute exchange (ks) was the
same regardless of whether the cell exhibited regular or
anomalous osmosis.
As long as solute diffusion occurs mainly across the
lipid layer or an array different from HgCl2-dependent
water channels, the picture of a composite membrane
system depicted so far explains the experimental findings
reasonably well. However, if the model applies, it would
also predict that a solute which would largely use water
channels to cross the membrane would exhibit the opposite properties, i.e. an increase of the reflection coefficient
ifCTJ< aj", and a decrease of the permeability coefficient
when closing water channels. Such behaviour could,
of course, also result from the conventional pore model
(see Discussion).
In order to cross the membrane via water channels, the
diameter of solute molecules would have to be similar to
that of water. The solute should also be polar (as water).
Such solutes are not easy to find, but heavy or tritiated
water are candidates. In this study, we used heavy water
(HDO) as an osmotic solute in about 2 M solution.
Heavy water should have transport properties quite similar to those of light water (H2O) and should use water
channels rather than the lipid arrays to cross the membrane. For HDO, the reflection coefficient of the water
channel array should be smaller than that of the lipid
array. Typical results of the action of HDO as an osmotic
solute are shown in Fig. 2. It can be seen that, in controls,
the reflection coefficient was quite low (a, = 0.0034, on
4.1 ±0.2
2.7 ±1.0
1.2±0.2
0.9 ±1.0
0.75-
P.B
Pf/Pf
3.7±0.2
2.9 ±1.0
1.6±1.0
1.6±1.0
0.9 ±0.06
1.2 ±0.6
1.3 ±0.4
1.7±0.8
control
+ 22 M HDO
0.70-
\
a>
S=
0.65 0
- 22 M HDO
Cl - 0.0048
X. - 21.3 »
20 40 60 80 100 120 140 160 180 200 220 240
0.74-
50 ^.M HgClj treatment
O
• 2 0.70-
+ 22 M HDO
S
0.64
60
C -
39.3 •
a, -
o.ooeo
120
180
- 2-2 M HDO
240
300
380
420
480
Time, t [s]
Fig. 2. Osmotic experiments with heavy water. The effects of the water
channel blocker HgCl2 on biphasic osmotic pressure relaxations of a
Chara internode caused by 2.2 M solutions of heavy water (HDO) are
shown. As in Fig. 1, exosmotic and endosmotic relaxations were
measured. (A) In the control, half-times of water exchange (Tfp) were
similar to those obtained in hydrostatic and osmotic pressure relaxations
with other solutes (see Fig. 1 and Table 1). (B) Treatment with 50 ^M
HgCl2 for 10 min resulted in an increase of 77/j by an average factor
of 2.5 (decrease of hydraulic conductivity (Lp) by the same factor) and
in an increase of T\fl by a factor of 2.1 (on average; decrease of the
premeability coefficient, Pd, by the same factor; Eq. (4)). The reflection
coefficient (a,) increased by a factor of 1.4 upon treatment which was
opposite to the findings for lipophilic solutes (Fig. 1).
average; Table 3), but was still accurately measurable
with the probe. This value agreed with data given by
Dainty and Ginzburg (1964ft; see Discussion).
In cells treated with 50 fiM HgCl2, reflection coefficients
of HDO increased by a factor of 2.0 or 100% (on average).
Water channels in Chara
205
Table 3. Reflection ( a j and permeability (P.) coefficients of heavy water (HDO) for internodes of Chara corallina as obtained from
biphasic pressure relaxations (Fig. 2)
Values of reflection coefficients were corrected for solute flow (Eq. (5)). Data are means from four different cells±SD. Means for individual cells
were calculated from two determinations. Ratios of o?/o? and of Pf/Pf indicate the effect of a 10-15 min treatment with 50jiM HgCl 2 (B) which
causes the reflection coefficient of HDO to double and the permeability coefficient to decrease by a factor of 1.8 (on average) as compared with the
control (A). Values of of denote reflection coefficients after a treatment with 5 mM 2-mercaptoethanol following the exposure to HgCl2.
Reflection coefficient, o,
[1]
Mean
SD
Permeability coefficient, P,
[lO-'ms"']
°i
a?
of
0.0034
0.0007
0.0070
0.0009
0.0039
0.0006
This was opposite to the finding with the other permeating
solutes, but was expected from both the composite membrane model and from the traditional pore model. The
permeability coefficient of HDO decreased by a factor of
1.8 upon treatment (Table 3). This is also expected and,
again, quite different from the results with other rapidly
permeating solutes where P, remained constant. Changes
in the a, and P, (Pd) of heavy water were reversible
(Table 3).
When comparing the diffusional permeability of heavy
water (Pd) as obtained from the solute phase of biphasic
pressure relaxations (Fig. 2) with the hydrostatic or
osmotic Lp, it was found that PA was much smaller than
expected. In order to compare both quantities, Lp was
transformed into an 'osmotic permeability', Pr, by:
2.0
0.6
P?
P."
P?/P»
7.7
3.0
4.3
2.0
1.8
0.2
the lipid bilayer (most likely) or in other proteinaceous
arrays such as ion pumps or channels or other transport
systems for specific solutes. This assumes that the latter
would allow some slippage of water besides the solutes
and also that they are not blocked by HgCl2. Experiments
in which water channel proteins from plants (tonoplast
intrinsic protein from Arabidopsis thaliana-y-TIP; Maurel
et al, 1993) and human red blood cells (CHIP28; Preston
et ah, 1992) were expressed in frog (Xenopus) oocytes
have shown that channels were specific for water and did
not transport ions or organic solutes. On the other hand,
transport proteins for specific solutes such as ion pumps
or the bacterial protein GlpF which facilitates the transport of glycerol did not affect the water permeability
(Maurel et al, 1993). From the results of Wayne and
Tazawa (1990) and from our results, it appears that
RT
characean cells behave similarly. Obviously, the HgCl2
(9)
treatment closes aquaporins in the membrane which act
where Vw = molar volume of water. If water flow across as selective valves for water. However, it cannot be
completely excluded that some of the water also moved
the membrane were diffusional in nature, i.e. if water
molecules moved independently of each other and not across ion channels and other transporters, provided that
these transport proteins were not completely selective to
across pores, Pr=Pd should hold. On the other hand, if
solutes. In this case, a blockage by HgCl2 would also
Pf>Pd, this may indicate some viscous flow across pores.
result in a decrease of Lp. However, if such a slippage of
However, this argument for the presence of pores may be
water would take place, it should have affected only water
questioned because Pd could be substantially underestiand not the passage of the small organic test solutes since
mated by the existence of unstirred layers. In the present
P, remained constant. This is unlikely. Therefore, we
experiments, the medium was vigorously stirred and
conclude that the contribution of transport proteins
unstirred layers should predominantly occur within the
different from water channels to the overall water flow
internodes. P{ and Pd differed by more than an order of
magnitude (P{/Pdx20) which is hard to explain solely by was small.
Since the transport of lipophilic solutes was not affected
unstirred layer effects. On the other hand, Pd was strongly
by the HgCl2 treatment, the structure of lipid arrays
affected by the HgCl2 treatment (see above) which also
appeared to remain unchanged. The small organic test
indicated that Pd at least was not completely dominated
solutes crossed the membrane largely by diffusion across
by internal unstirred layers (see Discussion).
lipid arrays so that their P, was not affected by
the treatment. For red blood cells, the closure of
Discussion
water channels with /7-chloromercuribenzene-sulphonate
(/?-CMBS) caused the osmotic water permeability of the
Provided that in Chara all water channels close upon
cells (P{) to assume a value comparable to that of the
treatment with HgCl2, the results indicate that about 3/4
diffusional permeability (Pd) also measured after closing
of the water transport across the plasma membrane is
(Macey et al., 1972; Moura et al, 1984). This suggested
mediated by specific water channels. The remaining 1/4
that most of the water flow was across water channels
of the hydraulic conductance would then reside either in
206
Henzler and Steudle
and the lipid arrays contributed only 10% to the total
water permeability. In Chara, the situation was different.
In the closed state, Pt/Pdx20 as compared with a
P f /P d «27 in the control. This suggests that there was
still some transport across pores after closing or, more
likely, that the differences were caused by internal
unstirred layer effects which affect Pa but hardly Pt (see
below). Because of the cell size, effects of internal unstirred
layers would be much more important in Chara (cell
diameter: 1 mm) than in red blood cells (cell diameter:
proposed, the mechanisms of which are not yet known,
but may be different from those observed for ion channels.
It has to be noted that the internodes remained intact
during the experiments and that reversibility of changes
in transport was demonstrated. It is noteworthy that
HgCl2 had a stronger effect than Hg(NO3)2 which was
also tested. This is, perhaps, due to the fact that HgCl2,
when dissolved in water, remains largely undissociated as
compared with Hg(NO3)2. The uncharged molecule may
cross the cell wall with its fixed negative charges more
easily and may not bind to the negatively charged surface
of the membrane. Rather, it could reach S-H groups of
Considering the changes in the hydraulic conductivity
membrane proteins (channels) at the membrane surface
upon blockage, an upper limit of the water permeability
or somewhat deeper in the membrane to cause conformaof the lipid arrays of the Chara membrane can be
tional changes and a closure of channels. This may explain
calculated from Lp values. Using an Lp of
why HgCl2 was effective at much lower concentrations
5.6 x 10~ 7 ms~ 1 MPa" 1 for treated cells, Pt would be
(50 fiM) than other mercurials used to block water chan7.6xlO" 5 ms~ 1 which is somewhat larger than that of
nels. For example, Wayne and Tazawa (1990) used
red blood cells (1.8 x 10" s ms" 1 ; Moura et al., 1984) or
/7-CMBS, a substance which would be present as an anion
of frog oocytes treated with HgCl2 (2xlO" 5 ms" 1 ;
at usual pH, to study effects on the water permeability of
Preston et al., 1992). As the hydraulic conductivity of
internodal cells of Nitellopsis obtusa. They used concentracharacean internodes and red blood cells is larger than
tions of up to 1 mM and found a reduction of the overall
that of most of the other plant and animal species
Lp of only 27% which was, perhaps, due the use of an
investigated so far (House, 1974; Finkelstein, 1987; Stein,
ionic compound rather than of HgCl2. The same may
1986; Steudle, 1989), the contribution of the lipid bilayer
in Chara would still be sufficient to account for a reason- refer to the use of p-CMBS in studies with animal cells
where concentrations necessary to close water channels
able Lp as compared with other plant and animal species.
The high absolute value of Lp of intact Chara corallina were also in the millimolar range. Other S-H reagents
such as iV-ethylmaleimide had no or very little effect when
may either result from a large amount of water channels
used with Chara in our experiments as in the experiments
in the membrane or from a high hydraulic conductance
of Wayne and Tazawa (1990).
of individual channels. It correlates with somewhat pecuIn general, a decrease of a, would be correlated with
liar hydraulic properties of these species in that Lp
an increase of P, and a decrease in Lp (Eq. (1)). As the
depends on both concentration and pressure (Dainty and
membrane becomes tighter, a, increases. In a homogenGinzburg, 1964a; Kiyosawa and Tazawa, 1972; Steudle
eous membrane, this is due to the fact that permeating
and Tyerman, 1983). Both phenomena could be related
solutes would contribute to the overall volume flow
to water channels, i.e. there could be some pressure- and
besides the water and hence:
concentration-dependent 'gating' of the channels (Steudle
and Henzler, unpublished). In this context, it is interesting
that in Chara corallina, Lp and a, decrease with increasing
(10)
external concentration. Most importantly, these changes
°'
LpxRT
occur at a constant solute permeability as during the
in the absence of a frictional interaction between water
treatment with HgCl2 (Steudle and Tyerman, 1983).
and solutes as they cross the membrane (see Eqs (1) and
There is evidence that the expression of water channels
(9)). For Chara, the absolute value of the second term
in plant cell membranes is under genetic control which
on the right side of Eq. (10) is only between 0.05 and
may allow plants to adapt to the water regime in their
0.08 for the rapidly permeating solutes used in this study
environment (Yamaguchi-Shinozaki et al., 1992). Since
(acetone, ethanol; Steudle and Tyerman, 1983) which can
there is a substantial variability in the hydraulic conducnot account for control values of a, of 0.13 and 0.28,
tivity of plant cells (Lp) which ranges over two orders of
respectively, and a, should be 0.95 and 0.92, respectively.
magnitude (Steudle, 1989), we have to assume that plasma
The second term on the right side of Eq. (10) also can
membranes differ considerably in the density and/or connot account for the small and even negative a, of cells
ductance of water channels. In addition, Lp may depend
treated with HgCl2. When Lp decreased by a factor of
on a number of internal and external factors such as
3.5 after the treatment, this would still result in a a, of
turgor, external concentration, salinity, drought, and tem0.85 or 0.76 rather than of -0.16 and -0.074, respectperature. It would be understandable if the opening and
ively (Table 2).
closing of water channels is triggered by these factors.
Alternatively, low a, values are explained by a considerSome kind of a 'gating' of water channels would be
Water channels in Chara
able interaction between water and solutes in membrane
pores according to the lipid-pore model (Eq. (1)).
However, the interpretation results in serious contradictions: (i) it is hard to explain how a lowering of Lp and,
hence, of the diameter of membrane pores could result in
a decrease of <J, (i.e. of selectivity) at constant P,. (ii) It
is not understandable why molecules such as formamide
or dimethylformamide, which are comparable in size to
acetone and ethanol, exhibit a, values so much higher
than these substances. Undoubtedly, the low o, and high
P, of acetone and ethanol are due to a rapid transport
across the lipid film. Rapid transport of lipophilic solutes
across the lipid layer should be correlated with a very
low or even negative ab (Eq. (It))). Thus, attempts to
explain the results in terms of conventional pore models
derived for homogeneous membranes fail.
For heavy water which should largely use the channel
path, Pd decreased at increasing a,. This result would be
in line with both the composite membrane and the lipidpore model. However, the pore model fails to explain
absolute values of transport coefficients of small organic
solutes as well as effects of the channel blocker HgCl2 on
transport coefficients of organic solutes and HDO.
In the composite membrane model, the lipid and water
channel arrays would contribute to the overall hydraulic
conductivity according to their hydraulic conductance
(Eq. (6)). Transport of low molecular weight organic
solutes would be largely across the lipid phase (Eq. (8))
and would depend on the solubility of solutes in
this phase. Thus, it would be expected that
/>, (acetone) > P,(ethanol) > i>, (dimethylformamide) > P,
(formamide) >P8(mannitol, NaCl) as found. On the other
hand, reflection coefficients would represent a mixed value
of the reflection coefficients of lipid and water channel
arrays. Reflection coefficients of lipid arrays should be
small or even negative for rapidly permeating solutes such
as acetone and ethanol. In this case, the second term on
the right side of Eq. (10) would become larger than unity.
Reflection coefficients should be rather large for the water
channel array which is selective for water. Thus, by
closing the water channels (array 'a'), a, tends to approach
the reflection coefficient of the lipid array (ab in Eq. (7))
as found. The model predicts that for solutes which move
across the lipid array, a, should decrease when closing
the water channels. This is so because the contribution
of array 'a' (water channels) having a high aj would
become small or would completely vanish. For heavy
water (HDO), the situation is reversed. It is predicted
that a, increases upon closure as this solute should mainly
use the water channel path having a rather small oj for
this array and a relatively large ab for the lipid array.
The model predicts also that Pd would decrease upon
closure of channels. Both predictions have been verified
experimentally which supports the model.
In the literature, there are examples (giant algae and
207
higher plants) of low and even negative a, values for
solutes such as ethanol or acetone etc. It is interesting
that negative reflection coefficients for these solutes correlate with low Lp values (Gutknecht, 1968; Tyerman and
Steudle, 1982).
If we assume that the contribution of lipid arrays to
the overall water transport is constant and not affected
by HgCl2 (i.e. that yb x Lpb is constant), Eqs (6) and (7)
can be used to evaluate a\, i.e. the reflection coefficient
of water channels. For a given solute, Eqs (6) and (7)
yield:
t=
a\ - ybx Lpb x (a° - ab) X —.
Lp
(11)
Thus, plotting the measured quantities a, and l/Lp against
each other yields the reflection coefficient of water channels (aj) from the intercept with the a, axis and yb x Lpb
(aj — ab) from the slope of the resulting straight line. As
oj > ab should hold for all solutes except HDO (selective
water channels), we expect that the slope is negative (for
HDO positive). Examples of plots according to Eq. (11)
are shown in Fig. 3 for two solutes (ethanol and dimethylformamide, DMF). The figure indicates that the channel
aj (intercept on the a, axis) was surprisingly small for
the permeating solutes used (ethanol: a" = 0.40; DMF:
aJ = 0.65), although, if channels were selective for water,
one would expect a oj«1 even for rapidly permeating
solutes. The finding of a low oj (despite a rather low P\)
would be understandable if there were still some 'slippage'
of small solutes in the water channels. This would result
in some kind of a singlefiletransport of water and solutes
across the channels provided that the pores are so narrow
such that water and small solute molecules could not pass
each other within the pore. In this case, solutes (although
rarely passing the pores) would drag along a lot of water
thus reducing the reflection coefficient by an efficient
coupling between solute and water flow (Finkelstein,
1987). The effect should increase with an increasing
number of water molecules within the pore, i.e. with its
length (see below; Steudle and Henzler, unpublished).
Thus, a low aj; despite a low P\ for the small solutes is
possible. The example demonstrates that, by simultaneously measuring the changes of different transport properties {Lp, P,, a,), valuable information about physical
properties of water channels is obtained. These properties
have not yet been measured with other systems and
techniques because of experimental difficulties in determining different transport properties and their changes
simultaneously and with sufficient accuracy. It is surprising that the composite membrane model, which has been
known for a long time (Kedem and Katchalsky, 1963),
has never been employed to describe the transport across
membranes containing water channels.
The picture of water channels as single-file pores is in
208
Henzler and Steudle
0.4-
Ethanol
r1-
03-
0JB2
020.1 -
'O
0
0
05
2JS
-0.1 -
o
o .02A
o
"o
55
DC
DMF
0.6-
r*-0.85
0.402-
0
OS
1
1.5
2
25
3
VLp-10' 6 [s-m'-MPa]
Fig. 3. Plot of reflection coefficient ( a j versus the inverse of hydraulic
conductivity (\jLp). The relation is shown for different Chara internodes
and two permeating solutes (A: ethanol: 5 cells; B: dimethylformamide
( D M F ) : 3 cells). The plot yielded a straight line as expected from
Eq. (11). From the intercept with the a, axis, reflection coefficients of
water channels, uj, are obtained for ethanol (CTJ = 0 . 4 0 ) and for DMF
(oj = 0.65). r* = correlation coefficient.
line with the finding that Pd values found for HDO were
much smaller than the osmotic permeability, P{, calculated
from the measured osmotic or hydrostatic Lp. If we
compare the absolute values of the osmotic permeability,
P{, calculated from Lp (Eq. (9)) with the Pd measured
from the solute phase in osmotic experiments with HDO,
we find that P{ was larger by a factor of 27. To some
extent, the difference may be explained by effects of
internal unstirred layers which would tend to reduce the
measured Pd (see above). However, unstirred layers can
not completely account for the difference, because Pd was
substantially decreased by HgCl2 treatment and externally
applied HgCl2 should have very little (if any) effect on
internal unstirred layers. Hence, we conclude that there
should be a substantial difference between P{ and Pd,
although, at present, it would be hazardous to quantify
this because of unstirred layer problems. Provided that
water channels behave like single-file pores, the ratio of
'real' P{/Pd would represent the number of water molecules lined up in the water channel (Levitt, 1974). For red
blood cells, where unstirred layer effects are small because
of the much smaller cell diameter, the number has been
estimated to be about 10 water molecules per pore
(Finkelstein, 1987).
To our best knowledge, the heavy water experiments
provide the first data of a, and Pd for isotopic water
which have been measured in the same experiment. The
reflection coefficient of HDO was close to zero. It was
similar to values measured by Dainty and Ginzburg
(19646). From transcellular osmosis experiments, these
authors estimated a a, = 0.0061 to 0.0033 when using
HDO solutions at concentrations between 4 and 20 M.
Dainty and Ginzburg (19646) found that a, decreased
with increasing HDO concentration. This may be interpreted by unstirred layer effects in the experimental setup of Dainty and Ginzburg (19646), but also with
difficulties in applying linear force/flow relations of irreversible thermodynamics which are derived for dilute
solutions. Different from the transcellular osmosis technique, the a, values obtained with the cell pressure probe
should incorporate a smaller contribution from unstirred
layers because (i) we corrected for solute flow (Eq. (5))
which also involved some correction of internal unstirred
layers and (ii) the concentrations used to induce osmotic
waterflowsby externally applied HDO were smaller (due
to the higher sensitivity of the cell pressure probe).
Because of unstirred layers, both a, and Pd values should
be regarded as lower limits. However, despite these difficulties, the conclusions drawn from the qualitatively
different effects of HgCl2 treatment on the permeability
and reflection coefficients of organic solutes as compared
with HDO remain valid.
In conclusion, the action of the water channel blocker
HgCl2 on the water and solute transport of Chara cells
indicates that water channels are responsible for most of
the water transport across the plasma membrane of this
alga. As the hydraulic conductivity of the membrane (Lp)
is reduced by blocking water channels, reflection coefficients of the membrane for permeating solutes decrease
substantially at constant solute permeability. The result
can not be explained in terms of conventional models
which imply homogeneous membranes. It is, however,
readily explained by a composite model of the plasma
membrane. According to the model, the different arrays
(proteinaceous and lipid arrays) contribute to the overall
reflection coefficient according to their hydraulic conductance and, hence, considerable changes in a, occur when
water channels are blocked. Thus, if there are rapidly
permeating lipophilic solutes exhibiting a negative o* for
the bilayer, this results in a negative overall a, and
anomalous osmosis when water channels are blocked.
Experiments with heavy water, a substance which uses
the same path as light water to cross the membrane,
result in opposite effects. They are in line with the
composite membrane model. It is suggested that effects
of turgor pressure or of external concentration which are
Water channels in Chara
known to affect the Lp of characean cells could be
mediated by some gating of water channels. We are
currently exploring the latter effects in more detail using
the cell pressure probe.
Acknowledgements
The authors are indebted to Dr J.B. Passioura, CSIRO, Division
for Plant Industry, Canberra, Australia, for reading the
manuscript and to Burkhard Stumpf for expert technical
assistance.
References
Chrispeels MJ, Maurel C. 1994. Aquaporins: the molecular
basis of facilitated water movement through living plant cells.
Plant Physiology 105, 9-15.
Dainty J, Ginzburg BZ. 1963. Irreversible thermodynamics and
frictional models of membrane processes, with particular
reference to the cell membrane. Journal of Theoretical Biology
5, 256-65.
Dainty J, Ginzburg BZ. 1964a. The measurement of hydraulic
conductivity (osmotic permeability to water) of internodal
characean cells by means of transcellular osmosis. Biochimica
et Biophysica Ada 79, 102-11.
Dainty J, Ginzburg BZ. 19646. The reflection coefficient of
plant cell membranes to certain solutes. Biochimica et
Biophysica Ada 79, 129-37.
Finkelstein A. 1987. Water movement through lipid bilayers,
pores and plasma membranes. Theory and reality.
Distinguished lecture series of the society of general physiologists. Vol. 4. New York: John Wiley & Sons.
Gutknecht J. 1968. Permeability of Valonia to water and solutes:
apparent absence of aqueous membrane pores. Biochimica et
Biophysica Ada 163, 20-29.
Hofte H, Hubbard L, Reizer J, Ludevid D, Herman EM,
Chrispeels MJ. 1992. Vegetative and seed-specific isoforms of
a putative solute transporter in the tonoplast of Arabidopsis
thaliana. Plant Physiology 99, 561-70.
House CR. 1974. Water transport in cells and tissues. London:
Edward Arnold.
Kammerloher W, Schaffner AR. 1993. PIP—an A. thalliana
plasma membrane MIP homologue cloned by expression in
mammalian cells (185 No. 185). Fifth International conference
Arabidopsis research, Ohio State University, Columbus,
OH, USA.
Kedem O, Katchalsky A. 1961. A physical interpretation of the
phenomenological coefficients of membrane permeability.
Journal of General Physiology 45, 143-79.
Kedem O, Katchalsky A. 1963. Permeability of composite
membranes. Part 2. Parallel arrays of elements. Transactions
of the Faraday Society 59, 1931-40.
Kiyosawa K, Tazawa M. 1972. Influence of intracellular and
extracellular tonicities on water permeability of characean
cells. Protoplasma 74, 257-70.
Levitt DG. 1974. A new theory of transport for cell membrane
209
pores. I. General theory and application to red cell. Biochimica
et Biophysica Ada 373, 115-31.
Macey RI. 1984. Transport of water and urea in red blood
cells. American Journal of Physiology 246, C195-C203.
Macey RI, Karan DM, Farmer REL. 1972. Properties of water
channels in human red cells. In: Kreuzer F, Siegers JFG, eds.
Biomembranes. Vol. 3. Passive permeability of cell membranes.
New York: Plenum Press, 331-40.
Maurel C, Reizer J, Schroeder JL, Chrispeels MJ. 1993. The
vacuolar membrane protein y-TIP creates water specific
channels in Xenopus oocytes. EMBO Journal 12, 2241-7.
Moura TF, Macey RI, Chien DI, Karan D, Santos H. 1984.
Thermodynamics of all-or-none water channel closure in red
cells. Journal of Membrane Biology 81, 105-11.
Preston GM, Jung JS, Goggino WB, Agre P. 1992. The
mercury-sensitive residue at cysteine 189 in the CHIP28
protein. Science 256, 385-7.
Reizer J, Reizer A, Saier Jr MH. 1993. The MIP family of
integral membrane channel proteins: sequence, comparisons,
evolutionary relationships, reconstructed pathways of evolution, and proposed functional differentiation of the two
repeated halves of proteins. Critical Reviews of Biochemistry
and Molecular Biology 28, 235-57.
Rudinger M, Hierling P, Steudle E. 1992. Osmotic biosensors.
How to use a characean internode to determine the alcohol
content of beer. Botanica Ada 105, 3-12.
Stein WD. 1986. Transport and diffusion across cell membranes.
Orlando: Academic Press.
Steudle E. 1989. Water flows in plants and its coupling with
other processes: an overview. Methods of Enzymology 174,
183-225.
Steudle E. 1993. Pressure probe techniques: basic principles and
application to studies of water and solute relations at the
cell, tissue, and organ level. In: Smith JAC, Griffith H, eds.
Water deficits: plant responses from cell to community. Oxford:
BIOS Scientific Publishers, 5-36.
Steudle E. 1994. The regulation of plant water at the cell, tissue,
and organ level: role of active processes and compartimentation. In: Schulze ED, ed. Flux control in biological systems:
from enzymes to populations and ecosystems. San Diego:
Academic Press, 237-99.
Steudle E, Tyerman SD. 1983. Determination of permeability
coefficients, reflection coefficients and hydraulic conductivity
of Chara corallina using the pressure probe: effects of solute
concentrations. Journal of Membrane Biology 75, 85-96.
Tyerman SD, Steudle E. 1982. Comparison between osmotic
and hydrostatic water flow in a higher plant cell: determination of hydraulic conductivities and reflection coefficients in
isolated epidermis of Tradescantia virginiana. Australian
Journal of Plant Physiology 9, 461-79.
Verkman AS. 1992. Water channels in cell membranes. Annual
Review of Physiology 54, 97-108.
Wayne R, Tazawa M. 1990. Nature of the water channels in
the internodal cells of Nitellopsis. Journal of Membrane
Biology 116, 31-9.
Yamaguchi-Shinozaki K, Koizumi M, Urao S, Sbinozaki K. 1992.
Molecular cloning and characterization of 9 cDNAs for genes
that are responsive to desiccation in Arabidopsis thaliana:
sequence of one cDNA clone that encodes a putative
transmembrane channel protein. Plant Cell Physiology 33,
217-24.