Aquaporin homologues in plants and mammals transport ammonia
Thomas P. Jahna,*, Anders L.B. Møllera, Thomas Zeuthenb, Lars M. Holmb, Dan A. Klærkeb,
Brigitte Mohsinc, Werner K€ uhlbrandtc, Jan K. Schjoerringa
aPlant Nutrition Laboratory, Department of Agricultural Sciences, The Royal Veterinary and Agricultural University,
Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
bDepartment of Medical Physiology, The Panum Institute, University of Copenhagen, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark
cDepartment of Structural Biology, Max Planck Institute of Biophysics, Marie-Curie Strasse 13-15, D-60439 Frankfurt am Main, Germany
Received 10 June 2004; revised 30 July 2004; accepted 3 August 2004
Available online 12 August 2004
Edited by Peter Brzezinski
tant deficient in ammonium (NHþ
wheat (Triticum aestivum) TIP2 aquaporin homologues were
isolated that restored the ability of the mutant to grow when 2
4 was supplied as the sole nitrogen source. When
expressed in Xenopus oocytes, TaTIP2;1 increased the uptake of
4analogues methylammonium and formamide. Furthermore,
expression of TaTIP2;1 increased acidification of the oocyte-
bathing medium containing NHþ
diffusion through the aquaporin. Homology modeling of Ta-
TIP2;1 in combination with site directed mutagenesis suggested
a new subgroup of NH3-transporting aquaporins here called
aquaammoniaporins. Mammalian AQP8 sharing the aquaam-
moniaporin signature also complemented NHþ
ciency in yeast.
? ? 2004 Federation of European Biochemical Societies. Published
by Elsevier B.V. All rights reserved.
Using functional complementation and a yeast mu-
4) transport (Dmep1–3), three
in accordance with NH3
Keywords: Ammonia; Aquaporin; Homology modeling;
The ammonium ion (NHþ
are the primary substrates for the synthesis of amino acids, and
are important for all living organisms. NHþ
millimolar levels within cells . However, in humans, high
levels of exogenous NHþ
4inhibit insulin release [2,3], cause
metabolic acidosis and renal failure [4,5], and have been linked
to Alzheimer’s disease  and hepatic encephalopathy . In
plants, high levels of cytoplasmic NHþ
cling across the plasma membrane  and NH3volatilization
from leaves .
Preston et al.  demonstrated that expression of CHIP28,
now called aquaporin 1 (AQP1), in frog oocytes created pores
in the plasma membrane, which specifically increased water
permeability. Diffusion through the lipid bilayer seems not to
4) and its conjugated base (NH3)
4can accumulate to
4result in a futile recy-
be the only pathway for gaseous compounds either, given that
membranes in several tissues and cell types have low perme-
abilities for CO2and NH3. This view has very recently been
supported by the finding that CO2limited growth correlated
with NtAQP1 expression in tobacco .
4uptake at low extracellular concentration in plants and
yeast is catalyzed by members of the ammonium transporter/
methylammonium permease (AMT/Mep) family [11,12]. These
transports have Kmvalues ranging from 0.5 to 40 lM NHþ
. Yeast (Saccharomyces cerevisiae) mutants, defective in
Mep homologues, were earlier used to clone and characterize
AMT homologues from plants and humans by functional
complementation [11,14]. Yet, no specific NHþ
porter, operating at elevated NHþ
been isolated or characterized in any organism.
Simultaneous deletion of MEP1, MEP2 and MEP3(Dmep1–
3) renders yeast dependent on relatively high concentrations of
4(>5 mM at pH 5.5) when supplied as the sole nitrogen
source . This prompted us to use this yeast mutant in an
approach to identify other transporters potentially involved in
Here, we have identified members of the aquaporin super-
family in plants and human that transport NHþ
aquaammoniaporins. We show that substrate specificity of the
aquaammoniaporins is correlated with substitutions within the
constriction region of the channels providing a larger pore
diameter. These substitutions are conserved in both plant and
4/NH3 concentrations, has
2. Materials and methods
2.1. Yeast strain and growth
The yeast strain Saccharomyces cerevisiae 31019b (MATa, ura3,
mep1D, mep2D::LEU2 mep3D::kanMX2)  was transformed with a
wheat root cDNA library in pYES2  by electroporation as de-
poration.htm). Transformants were grown on synthetic medium with
2% glucose or galactose, 50 mM succinic acid/Tris base, pH 5.5 (if not
indicated otherwise), 0.7% yeast nitrogen base w/o amino acids and
4(Difco) supplemented with 0.1% proline or different concentra-
tions of (NH4)2SO4as the sole nitrogen source. Results shown from
complementation analysis in yeast are representatives from more than
15 independent assays.
2.2. Oocyte expression and uptake of radiotracer
TaTIP2;1 and HsAQP1 cDNAs were cloned into an oocyte ex-
pression vector containing the 50- and 30-untranslated regions for
*Corresponding author. Fax: +45-3828-3460.
E-mail address: email@example.com (T.P. Jahn).
Abbreviations: AMT, ammonium transporter; AQP, aquaporin; GlpF,
glycerol facilitator; Mep, methyl ammonium permease; MIP, major
intrinsic protein; NIP, nodulin-like intrinsic protein; PIP, plasma
membrane intrinsic protein; TIP, tonoplast intrinsic protein
0014-5793/$22.00 ? 2004 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
FEBS 28735 FEBS Letters 574 (2004) 31–36
Xenopus laevis beta-globin and a poly-A segment. RNA preparation
and injection of oocytes were as described earlier . After injection
of mRNA, the oocytes were kept at 19 ?C for 5–6 days in Kulori (90
mM NaCl, 1 mM KCl, 1 mM MgCl2, 1 mM CaCl2and 5 mM HEPES-
Tris, pH 7.4) before further procedures. Measurements of extracellular
pH were done with an extracellular pH-electrode (Radiometer, Co-
penhagen, Denmark) in 70 mM Naþand 20 mM NHþ
HEPES/Tris. For each measurement, 20 oocytes were added abruptly
to a well stirred bathing medium while recording the pH. Uptake
studies were done essentially as described . The uptake medium
contained 20 mM methylammonia or formamide supplemented with
14C-methylammonia (Pharmacia) or14C-formamide (American Radi-
olabeled Chemicals, St. Louis, MO) to a final activity of 4 lCi/ml.
Uptakes were performed at room temperature. Oocytes were subse-
quently washed in ice-cold Kulori and radiotracer uptake was mea-
sured in a scintillation counter (Packard Tri-Carb).
4in 5 mM
Homology modeling of TaTIP2;1 was performed using the structure
of bovine AQP1 (Protein Data Bank Accession No. 1j4n.pdb) as the
template  and the program MODELLER6v.2 . Graphics were
generated using RasWin Molecular Graphics, Windows version 2.6-
ucb,?1993–1995 R.Sayle and software from SYBYL?6.9 Tripos Inc.,
1699 South Hanley Rd., St. Louis, Missouri, 63144, USA.
2.4. Mutagenesis of TaTIP2;1
Mutations in TaTIP2;1 were introduced by whole plasmid PCR
using complementary primers, including the respective mutations. The
H64F mutation was introduced using primers GCC ATA TGC TTC
GGG TTT GGG CTG T and GCC CAA ACC CGA AGC ATA
TGG CCA C, the I184H mutation using primers GGC GCC AAC
CAC CTC GTG GCC and GCC GGC CAC GAG GTG GTT GGC,
and the G193C mutation using primers TTC TCC GGC TGC AGC
ATG AAC CCT GCA C and CAG GGT TCA TGC TGC AGC CGG
AGA AGG G. The PCR reactions were digested with DpnI prior to
transformation into Escherichia coli to digest the methylated template
plasmid DNA. All sequences were verified by plasmid sequencing.
3. Results and discussion
3.1. Cloning of ammonium transporters
The yeast strain 31019b (Dmep1–3)  was transformed
with a wheat root cDNA library in pYES2  and around
100000 primary transformants were selected. Transformants
were plated on media with 2 mM NHþ
source. Out of a total of about 100 yeast transformants ana-
lyzed, 70% of the clones were yeast mutants transformed with
two different wheat AMT homologues. The remaining 30%
were all transformed with one of the three highly similar wheat
tonoplast-intrinsic protein homologues (TIP2), members of
the aquaporin super-family in plants.
After re-transformation, yeast cells expressing TIP2s grew
well at concentrations of 2 mM NHþ
4almost no difference was seen compared to the
control transformed with an empty vector (Fig. 1A). At an
concentration of 10 mM, however, the yeast cells
transformed with wheat TIP2s grew better than those trans-
formed with a high-affinity transporter of the AMT family,
4as the sole nitrogen
4and above, while at 0.2
Fig. 1. Complementation by AMT1 and TIP2 of the Dmep1–3 deletion in yeast. (A) Yeast mutant Dmep1–3 (31019b) transformed with TaAMT and
TaTIP2 homologues (Accession Nos. AY525637,AY525638,AY525639, AY525640, and AY525641)were spotted onto galactose-containing medium
with either proline or different concentrations of NHþ
was recorded after 6 days at 30 ?C. (B) 31019b transformed with either TaAMT1;1, TaTIP2;1 or empty pYES2 (10 ll OD66010?2) were spotted onto
galactose-containing medium with either proline or different concentrations of NHþ
4as nitrogen source (10 ll OD600nm10?2and 10?5; upper and lower spot, respectively). Growth
4at various pH. Growth was recorded after 5 days at 30 ?C.
T.P. Jahn et al. / FEBS Letters 574 (2004) 31–36
suggesting that NHþ
finity but high capacity had been identified.
4/NH3 specific transporters with low af-
3.2. TIP2 facilitates diffusion of NH3
Growth of yeast expressing TaTIP2;1 on NHþ
strongly pH-dependent, with progressively better growth at
increasing pH (Fig. 1B). The same was true for the control
31019b transformed with an empty vector pYES2. However, at
all pH values tested, the strain expressing TaTIP2;1 grew
much better than the control at comparable NHþ
tion. Increasing the pH by one unit strongly reduced the re-
quirement of NHþ
in the medium, both on solid media
(Fig. 1B) and in liquid cultures (data not shown). The data
thus strongly suggested that NH3, rather than NHþ
nitrogen species being transported by the TIP2s.
Addition of Xenopus oocytes to a well stirred bathing me-
dium containing 20 mM NHþ
4resulted in a continuous acidi-
fication of the medium, in line with the interpretation that
NH3diffused into the oocyte, leaving Hþin the external me-
dium (Fig. 2A). The pH of the buffer solution was stable in the
absence of oocytes and within the time of measurements pre-
sented, in line with the interpretation that pH changes were
due to NH3diffusion into the oocytes rather than its evapo-
ration from the buffer solution (not shown). Acidification was
significantly increased after injection with TaTIP2 mRNA
compared to control oocytes injected with water (Fig. 2A;
Table 1). Expression of human AQP1 did not increase NHþ
induced acidification compared to water injected controls, al-
though water transport could be demonstrated for both Ta-
TIP2;1 and HsAQP1 mRNA injected oocytes (Table 1).
Recordings of cytoplasmic pH, used earlier in attempts to
measure NH3 transport into oocytes , were strongly de-
pendent on the particular position of the electrode within the
oocyte and therefore less reproducible than pH measurements
of the bathing medium (data not shown).
To directly demonstrate uptake as dependent on TaTIP2s,
Xenopus oocytes injected with either water or TaTIP2;1
mRNA were exposed to 20 mM14C-methylammonium or 20
mM14C-formamide (Fig. 2B and C). Methylammonium is a
well-known analogue of NHþ
4, which has been used extensively
in uptake studies to characterize Mep/AMT homologues .
Both methylammonia and formamide influx into Xenopus
oocytes were increased following expression of TaTIP2;1. The
initial flux of formamide increased to about 250%. Influx of
methylammonia was less affected by expression of TaTIP2;1,
but still evident (Fig. 2C).
4, was the
3.3. Homology modeling of TaTIP2;1
Aquaporins form a super-family of channel proteins with a
common primary structure characterized by a tandem repeat
of two similar halves. Each half contains three transmembrane
helices and a membrane embedded sequence with an abso-
lutely conserved asparagine involved in formation of the
channel pore (reviewed in ). Aquaporin homologues with
different substrate specificities have recently been crystallized
and structural data are available from X-ray diffraction for
AQP1, a mammalian aquaporin , and the glycerol facili-
tator (GlpF)  and the aquaporin (AqpZ)  from E. coli,
all at high resolution. All structures are surprisingly similar.
However, the aqueous pore of GlpF is slightly wider and key
hydrophobic residues allow the passage of the carbon back-
bone of glycerol .
Based on a classification by Heymann and Engel , TIPs
are structurally closer related to water-specific aquaporins
than aquaglyceroporins. We used the X-ray structure of bo-
vine AQP1 to create a homology model of TaTIP2;1. The
overall sequential homology between AQP1 and TaTIP2;1 is
31%. The model was overlaid with the AQP1 structure and
the overlay was subsequently rotated in order to examine
differences in the constriction region of the channels (Fig. 3A
and B), which are most likely to affect substrate specificity
. In the model of TaTIP2;1, F58 of AQP1 is substituted by
Fig. 2. Addition of NHþ
methylammonia and14C-formamide to Xenopus oocytes expressing
TaTIP2;1 and HsAQP1. (A) Oocytes injected with either water, Ta-
TIP2;1 mRNA or HsAQP1 mRNA were bathed in a weakly buffered
medium and the pH was recorded in the presence of 20 mM NH4Cl.
Proton fluxes are given in Table 1. (B, C) Oocytes injected with either
water (d) or TaTIP2;1 mRNA (j) were bathed in a medium con-
taining either (B)14C-methylammonium or (C)14C-formamide. Re-
sults from n ¼ 15 independent measurements are presented.
4/NH3 and their radiotracer analogues14C-
T.P. Jahn et al. / FEBS Letters 574 (2004) 31–36
histidine, H182 is substituted by isoleucine and C191 by
glycine. These three substitutions cause a widening of the
constriction region. Taking into consideration that it is the
carbonyl oxygen of C191 (bovine AQP1)  that participates
in the coordination of the water molecule, the substitutions
also increase the overall hydrophobicity of the constriction
3.4. Mutating H64, I184 and G193 by site directed mutagenesis
negatively affects yeast complementation
Human AQP1 was earlier suggested to transport NH3
when expressed in Xenopus oocytes  but failed to do so in
our study. We therefore expressed in yeast HsAQP1 and
HsAQP8. Both isoforms are only about 30% homologous to
TaTIP2;1. However, AQP8 shares the identical residue sub-
stitutions in the constriction region with TaTIP2. HsAQP1
did not complement the growth defect, while HsAQP8 sup-
ported growth on NHþ
4to a similar extent like TaTIP2s
(Fig. 4). In order to investigate the importance of H64, I184
and G193 in the transport of NHþ
tagenesis was employed (Fig. 4). When H64 was changed into
phenylalanine, the corresponding residue of mammalian
AQP1, the resulting mutant Tatip2;1 H64F no longer sup-
ported growth of the yeast mutant on NHþ
I184 by histidine significantly reduced growth of the yeast
mutant on NHþ
4compared to wt TaTIP2;1. Mutating G193
4/NH3, site directed mu-
into cysteine did not affect yeast complementation. However,
yeast transformedwith the
I184H,G193C displayed a similar phenotype as the yeast
transformed with Tatip2;1 H64F or the empty vector pYES2
(Fig. 4). Interestingly, H182 together with the carbonyl oxy-
gen of C191 has been shown to coordinate a water molecule
in AQP1  and may therefore be crucial to restrict for the
passage of H2O molecules through the channel pore in
AQP1. Point mutations employed here are unlikely to inter-
fere with either the folding or the targeting of the mutant
enzymes compared to wt TaTIP2;1 when expressed in yeast,
because these mutations substitute residues in TaTIP2;1 with
residues well conserved in other aquaporin homologues in the
respective positions. This is further supported by the obser-
Fig. 4. Substitutions within the constriction region appear to be critical
for NH3transport through aquaporins. Mammalian HsAQP1 and 8
and plant aquaporin TaTIP2;1 as well as mutants of TaTIP2;1 were
expressed in 31019b. The empty vector pYES2 was used as control.
Transformants (OD600nm 10?2and 10?5; upper and lower spot, re-
spectively) were spotted onto galactose-containing medium with either
proline or different concentrations of NHþ
was recorded after 6 days at 30 ?C.
4as nitrogen source. Growth
Fig. 3. (A) Overlay of the structure of bovine AQP1 (blue) with the
model of TaTIP2;1 (green). The highly conserved NPA sequences are
shown in yellow. (B) View through the constriction region of the
channels from the extra-cytoplasmic face; residues from TaTIP2;1 are
in front and labelled black. Residues from AQP1 are pale and in the
background and labelled in grey. The position of the water molecule
coordinated by H182 and the carbonyl oxygen of G192 in the structure
of AQP1 is indicated as a red ball.
Medium acidification expressed as Hþfluxes (JHþ[10?11mols?1oocyte?1]) for oocytes bathed in solutions with different NH3concentrations and pH
Ta tip2;1Hs aqp1Native
JHþwere calculated from the changes in pH obtained in experiments as those shown in Fig. 2A and the buffer capacity of the bathing medium. pHe,
external pH; Lp, hydraulic conductivity.
*Significantly different from water injected oocytes at the 0.001 probability level. NH3concentrations were calculated from the pKa of NHþ
T.P. Jahn et al. / FEBS Letters 574 (2004) 31–36
vation that neither of the single mutations I184H and G193C
prevent ammonia transport. However, the double mutation
I184H; G193C strongly impaired complementation on NHþ
when expressed in yeast. The data therefore strongly suggest
that H64, I184 and G193 in TaTIP2 are important for the
transport of NHþ
In addition to water, members of the aquaporin super-
family have been shown to transport different substrates.
These have been classified into orthodox aquaporins trans-
porting H2O with high specificity and aquaglyceroporins that
in addition to H2O also transport glycerol and urea [25,26].
AQP9 was recently shown to transport an exceptionally wide
range of different substrates  and may thus represent a
member of a separate group. Data presented here provide
evidence for another sub-family within the aquaporin super-
family that facilitates diffusion of NH3across biomembranes.
A database search revealed that many other MIPs, in par-
ticular in plants (TIPs and nodulin-like intrinsic proteins
(NIPs)), contain isoleucine or valine and glycine or alanine in
the respective positions (182 and 191 in bovine AQP1). In the
corresponding position of F58 (bovine AQP1), NIPs are
substituted by tryptophan, leucine or alanine, which would
render the constriction region even more hydrophobic as
compared to TIP2 and AQP8. Both the dipole moment of the
NH3molecule (1.49 vs. 1.85 D) and the dielectric constant of
liquid NH3(22 vs. 80) are considerably lower than the corre-
sponding values for water. A more hydrophobic environment
at the constriction region may therefore allow the transport of
NH3and the slightly larger size of the ammonia molecule may
require a larger pore diameter. The exchange of His from
position 182 to position 58 (in the sequence of bovine AQP1)
affects the arrangement of hydrogen bond acceptors and do-
nors in the restriction region and is likely to be a major factor
TIPs represent a group of aquaporin homologues in plants,
which are expressed in vacuolar membranes . Recently,
AtTIP2;1 expression in Arabidopsis roots was shown to be
upregulated in response to nitrogen starvation . One pos-
sible explanation is that TIP2 functions in remobilization of
4/NH3from vacuoles under starvation. However, TIP2 or
other TIPs may also be important for NHþ
The major intrinsic protein (MIP) family in Arabidopsis has
recently been classified into eight different groups according to
residues lining the constriction region of the channel pore ,
a classification partially overlapping with the traditional clas-
sification grouping plant MIPs according to localizations such
as TIP for tonoplast  and PIP for plasma membrane 
localized isoforms. For instance, all Arabidopsis PIPs share a
F/H/T/R signature, suggesting water specificity, while NIPs
and especially TIPs are characterized by a number of various
different signatures. Screening a wheat root cDNA library,
here, only TIP2s characterized by a H/I/G/R signature were
identified that are able to transport ammonia. The present
paper therefore supports the view that the constriction region
of MIPs represents a major factor for selectivity of the channel
proteins and that a functional classification may be more ap-
propriate than classification according to localization, as sug-
gested earlier .
Interestingly, TIP2 was recently identified by mass-spec-
trometry in purified peribacteroid membranes from Lotus .
Thus, TIP2, but also NIPs, may be responsible for the trans-
port of NH3 from nitrogen-fixing symbionts to the plant.
Ammonia permeablility through the peribacteroid membrane
was earlier shown to be reversibly sensitive to the channel
blocker mercury .
AtTIP2;1 was recently identified by mass spectrometry in
plasma membrane preparations of Arabidopsis (Erik Alexan-
dersson, personal communication) and was found to be local-
ized to the plasma membrane when transiently overexpressed
with a GFP tag in Arabidopsis protoplasts . It is therefore
tempting to speculate that TIP2 may be involved in low affinity
and high capacity uptake at the plasma membrane.
Acknowledgements: We thank Bruno Andr? e for the generous gift of the
yeast strain 31019b and Julian Schroeder for the wheat root cDNA
expression library in pYES2. This work was supported by grants from
the Danish Agricultural and Veterinary Research Council (23-01-0136)
to TPJ, from the EU-FP5 (SUSTAIN; No. QLKS-CT-2001-01461) to
JKS and from the Lundbeck Foundation and the Danish Medical
Research Council to T2 and LMH.
 Britto, D.T., Siddiqi, M.Y., Glass, A.D.M. and Kronzucker, H.J.
(2001) Proc. Natl. Acad. Sci. USA 98, 4255–4258.
 Sener, A. and Malaisse, W.J. (1980) Diabetes Metab. 6, 97–101.
 Sener, A., Hutton, J.C., Kawazu, S., Boschero, A.C., Somers, G.,
Devis, G., Herchuelz, A. and Malaisse, W.J. (1978) J. Clin. Invest.
 Wall, S.M. (1997) Am. J. Physiol. 273, F857–F868.
 Watts 3rd, B.A. and Good, D.W. (1994) J. Gen. Physiol. 103, 917–
 Zhou, B.-G. and Norenberg, M.D. (1999) Neurosci. Lett. 276,
 Husted, S. and Schjoerring, J.K. (1996) Plant Physiol. 112, 67–74.
 Preston, G.M., Carroll, T.P., Guggino, W.B. and Agre, P. (1992)
Science 256, 385–387.
 Cooper, G.J., Zhou, Y., Bouyer, P., Grichtchenko, I.I. and Boron,
W.F. (2002) J. Physiol. 542, 17–29.
 Uehlein, N., Lovisolo, C., Siefritz, F. and Kaldenhoff, R. (2003)
Nature (London) 425, 734–737.
 Ninnemann, O., Jauniaux, J.-C. and Frommer, W.B. (1994)
EMBO J. 13, 3464–3471.
 Marini, A.M., Vissers, S., Urrestarazu, A. and Andr? e, B. (1994)
EMBO J. 13, 3456–3463.
 Gazzarrini, S., Lejay, L., Gojon, A., Ninnemann, O., Frommer,
W.B. and von Wir? en, N. (1999) Plant Cell 11, 937–948.
 Marini, A.-M., Matassi, G., Raynal, V., Andr? e, B., Cartron, J.-P.
and Cherif-Zahar, B. (2000) Nature Gen. 26, 341–344.
 Marini, A.-M., Soussi-Boudekou, S., Vissers, S. and Andr? e, B.
(1997) Mol. Cellular Biol. 17, 4282–4293.
 Schachtman, D.P. and Schroeder, J.I. (1994) Nature (London)
 Zeuthen, T. and Klærke, D.A. (1999) J. Biol. Chem. 274, 21631–
 Sui, H., Han, B.-G., Lee, J.K., Walian, P. and Jap, B.K. (2001)
Nature (London) 414, 872–878.
 Sali, A. and Blundell, T.L. (1993) J. Mol. Biol. 234, 779–815.
 Nakhoul, N.L., Herings-Smith, K.S., Abdulnour-Nakhoul, S.M.
and Hamm, L.L. (2001) Am. J. Ren. Physiol. 281, F255–F263.
 Heymann, J.B. and Engel, A. (2000) J. Mol. Biol. 295, 1039–1053.
 Fu, D., Libson, A., Miercke, L.J.W., Weitzman, C., Nollert, P.,
Krucinski, J. and Stroud, R.M. (2000) Science 290, 481–486.
 Savage, D.F., Egea, P.F., Robles-Calmeneras, Y., O’Conell III,
J.D. and Stroud, R.M. (2003) PloS Biology 1, 334–340.
 Unger, V.M. (2000) Nat. Struct. Biol. 7, 1082–1084.
T.P. Jahn et al. / FEBS Letters 574 (2004) 31–36
Sasaki, S. (1998) Biochem. Biophys. Res. Commun. 244, 268–274.
 Borgnia, M., Nielsen, S., Engel, A. and Agre, P. (1999) Annu.
Rev. Biochem. 68, 425–458.
 Carbrey, J.M., Gorelick-Feldman, D.A., Kozono, D., Praetorius,
J., Nielsen, S. and Agre, P. (2003) Proc. Natl. Acad. Sci. USA 100,
 Fragne, N., Maeshima, M., Sch€ affner, A.R., Mandel, T., Mati-
noia, E. and Bonnemain, J.-L. (2001) Planta 212, 270–278.
 Liu, L.H., Ludewig, U., Gassert, B., Frommer, W.B. and von
Wir? en, N. (2003) Plant Physiol. 133, 1220–1228.
 Wallace, I.S. and Roberts, D.M. (2004) Plant Physiol. 135, 1059–
 Johnson, K.D., Herman, E.M. and Chrispeels, M.J. (1989) Plant
Physiol. 91, 1006–1013.
 Kammerloher, W., Fischer, U., Piechottka, G.P. and Chaffner,
A.R. (1994) Plant J. 6, 187–199.
 Kirch, H.H. and Bohnert, H.J. (1999) Trends Plant Sci. 4, 86–88.
 Wienkoop, S. and Saalbach, G. (2003) Plant Physiol. 131, 1080–
 Niemietz, C.M. and Tyerman, S.D. (2000) FEBS Lett. 465, 110–
T.P. Jahn et al. / FEBS Letters 574 (2004) 31–36