Download A Novel Plant Major Intrinsic Protein in Physcomitrella patens Most

This article is published in Plant Physiology Online, Plant Physiology Preview Section, which publishes manuscripts accepted for
publication after they have been edited and the authors have corrected proofs, but before the final, complete issue is published
online. Early posting of articles reduces normal time to publication by several weeks.
A Novel Plant Major Intrinsic Protein in Physcomitrella
patens Most Similar to Bacterial Glycerol Channels1
Sofia Gustavsson, Anne-Sophie Lebrun, Kristina Nordén, Francxois Chaumont, and Urban Johanson*
Department of Plant Biochemistry, Centre for Chemistry and Chemical Engineering, Lund University,
SE–221 00 Lund, Sweden (S.G., K.N., U.J.); and Unité de Biochimie Physiologique, Institut des Sciences
de la Vie, Université Catholique de Louvain, B–1348 Louvain-la-Neuve, Belgium (A.-S.L., F.C.)
A gene encoding a novel fifth type of major intrinsic protein (MIP) in plants has been identified in the moss Physcomitrella patens.
Phylogenetic analyses show that this protein, GlpF-like intrinsic protein (GIP1;1), is closely related to a subclass of glycerol
transporters in bacteria that in addition to glycerol are highly permeable to water. A likely explanation of the occurrence of this
bacterial-like MIP in P. patens is horizontal gene transfer. The expressed P. patens GIP1;1 gene contains five introns and encodes
a unique C-loop extension of approximately 110 amino acid residues that has no obvious similarity with any other known protein.
Based on alignments and structural comparisons with other MIPs, GIP1;1 is suggested to have retained the permeability for
glycerol but not for water. Studies on heterologously expressed GIP1;1 in Xenopus laevis oocytes confirm the predicted substrate
specificity. Interestingly, proteins of one of the plant-specific subgroups of MIPs, the NOD26-like intrinsic proteins, are also
facilitating the transport of glycerol and have previously been suggested to have evolved from a horizontally transferred bacterial
gene. Further studies on localization and searches for GIP1;1 homologs in other plants will clarify the function and significance of
this new plant MIP.
The major intrinsic proteins (MIPs) form a large and
ancient protein family since members of the MIP
family are found in all kinds of organisms from the
three domains of life: bacteria, archaea, and eukaryotes. MIPs form channels or pores through membranes to facilitate passive transport of water and
other small polar molecules, such as glycerol, across
membranes. According to substrate specificity, MIPs
can be classified as aquaporins (AQPs; water channels)
or glycerol facilitators (glycerol intrinsic proteins
[GLPs]; Heymann and Engel, 1999). GLPs are also
referred to as aquaglyceroporins because they, in addition to glycerol, are permeable to water to varying
extent (Agre et al., 1998). High-resolution structures
are available for AQPs from mammalians (AQP0 and
AQP1) and Escherichia coli (AQPZ) along with a glycerol facilitator from E. coli (EcGlpF; Fu et al., 2000; de
Groot et al., 2001; Sui et al., 2001; Savage et al., 2003;
Gonen et al., 2004; Harries et al., 2004). Despite the
evolutionary distance and the different substrate specificities, the structures are remarkably similar. All
MIPs share the same basic topology of six transmembrane a-helices and a seventh transmembrane structure formed by two short helices entering the membrane
1
This work was supported by the Erik Philip-Sörensen Foundation and the Swedish Research Council for Environment, Agricultural Sciences, and Spatial Planning (FORMAS; grants to U.J.) and by
the Belgian Fund for Scientific Research and the Interuniversity
Attraction Poles Programme-Belgian Science Policy (grants to F.C.).
* Corresponding author; e-mail [email protected];
fax 46–46–2224116.
Article, publication date, and citation information can be found at
www.plantphysiol.org/cgi/doi/10.1104/pp.105.063198.
from opposite sides and connecting at the conserved
Asn-Pro-Ala (NPA) motifs. Plants have more genes
encoding different MIPs than any other type of organism, possibly reflecting the importance of differential
regulation of transport of water and small polar solutes in different tissues. In the model plant Arabidopsis (Arabidopsis thaliana), we have identified 35 genes
coding for full-length MIPs (Johanson et al., 2001).
Based on phylogenetic comparisons, plant MIPs are
classified into four subfamilies, plasma membrane
intrinsic proteins (PIPs), tonoplast intrinsic proteins
(TIPs), Nod26-like intrinsic proteins (NIPs), and smallbasic intrinsic proteins (SIPs). The PIPs and TIPs are in
general regarded as AQPs, whereas the NIPs have
been reported to transport both glycerol and water (for
review, see Johanson et al., 2001). The specificity of the
fourth subfamily is not yet known (Johanson and
Gustavsson, 2002). All four subfamilies have also been
identified in the primitive nonvascular moss Physcomitrella patens, suggesting that they are present in all
terrestrial plants (Borstlap, 2002). Phylogenetic analyses have made it likely that the NIPs originally originated from a horizontally transferred bacterial AQP
gene and were later adapted for glycerol transport
(Zardoya et al., 2002). The lack of GLPs in plants was
suggested to provide the driving force for the evolution of NIPs, although the physiological role of glycerol transport in plants is not completely clear.
Here, we report and characterize a novel plant MIP
with very low similarity to any other plant MIP but
a surprisingly high similarity to a subclass of bacterial
aquaglyceroporins. We suggest that this gene was also
introduced into plants via horizontal gene transfer as
proposed for the NIPs.
Plant Physiology Preview, www.aspb.org Ó 2005 American Society of Plant Biologists
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
1 of 9
Gustavsson et al.
RESULTS
Sequence of cDNA Clone and Genes
We have identified a novel plant MIP in P. patens
upon searching for MIPs by TBLASTN in the P. patens
expressed sequence tag (EST) database at http://
www.moss.leeds.ac.uk. Two nonoverlapping EST sequences, BU052189 and BQ826917, both originating
from the clone PPAS020308, were found. These ESTs
encode MIP peptides that are very unlike previously
known plant MIPs. Sequencing of PPAS020308 revealed a 1,602-bp-long cDNA insert with an open
reading frame of 1,113 bp. Alignments with other MIPs
support that this open reading frame contains the
complete coding sequence. To ensure that the cDNA
clone originated from P. patens and to study the gene
structure, the corresponding genomic sequence was
directly amplified from P. patens DNA with specific primers based on the cDNA sequence. Different primer
combinations resulted in overlapping PCR products,
which were sequenced with amplification primers and
internal specific primers. The sequences of the exons
and the cDNA are identical and confirm that the
cDNA clone originates from P. patens. The gene from
translation start to stop, including five introns, is
2,291 bp long. Intron lengths and positions relative to
transmembrane regions in the protein are presented in
Figure 1. The gene contains six exons and five introns,
which all have canonical dinucleotides GT and AG for
donor and acceptor sites. In the translated protein,
introns I and V are positioned after H2 and in H5,
respectively. Introns II, III, and IV are all located in the
sequence encoding the elongated C-loop (see below).
It has previously been noted that intron positions
are conserved within each of the four MIP subfamilies
in Arabidopsis (Johanson et al., 2001). It is therefore
worth mentioning that the position of the first intron is
situated in the exact site as the first plant PIP intron.
Part of an orthologous gene encoding the novel MIP
type, 421 bp including intron V, was also successfully
amplified and sequenced in the moss Funaria hygrometrica. Sequence comparison of the orthologous genes
revealed six indels and approximately twice the substitution frequency in the introns relative to the coding
sequences, 0.19 and 0.10, respectively (data not shown).
The majority of the substitutions in the coding region
are synonymous, reducing the effect in the protein to
five conservative replacements in the 54 amino acids
encoded by the sequenced region of the gene.
General Structure and Localization
By comparing this P. patens MIP sequence to AQPs
and glycerol channels, it is obvious that it belongs to the
superfamily of MIPs (Fig. 2). The highly conserved NPA
motifs that often are used as fingerprints of the MIP
family are indeed present, and the predicted topology of
six transmembrane helices agrees well with known MIP
structures. However, this MIP is clearly different from all
earlier identified plant MIPs and cannot be categorized
to any of the earlier identified subfamilies of MIPs in
plants (Johanson et al., 2001). Most surprisingly, when
performing BLASTP and TBLASTN searches, the most
significant hits are bacterial glycerol channels rather
than plant MIPs. In accordance with the proposed
nomenclature for plant MIPs, in the following we will
refer to this protein as GlpF-like intrinsic protein (GIP1;1;
Johanson et al., 2001).
The most striking feature of GIP1;1 is that the C-loop
has an extension of around 100 amino acid residues
compared to most other MIPs (112 residues compared
to the most similar sequences). The insert does not
show any obvious similarity with any known protein
in public protein or translated DNA databases. In the
structure of EcGlpF, the extension fits nicely at the very
top of the extracellular protrusion of the protein. The
inserted region consists of a large portion of charged
and polar residues, and according to the topology
prediction, the C-loop does not form any transmembrane helices.
Figure 1. Gene structure of P. patens GIP1;1. GIP1;1 is presented. Based on alignments and structural data from E. coli, GlpF
membrane spanning helices (H1–H6) and half spanning helices (HB and HE) are depicted as boxes. Corresponding helices in the
direct repeats are color coded. Intron positions I to V in P. patens GIP1;1 are indicated by green arrows and compared to general
intron positions of Arabidopsis PIPs (black), TIPs (orange), SIPs (magenta), and NIPs (cyan). Exact positions of introns in GIP1;1
are calculated from the first nucleotide of the initiation codon: intron I, 193 to 639; intron II, 894 to 1,072; intron III, 1,176 to
1,296; intron IV, 1,379 to 1,528; and intron V, 1,803 to 2,083.
Plant Physiology Preview
2 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Bacterial-Like Glycerol Channel in Mosses
Figure 2. Multiple protein alignment of GIP1;1 (PpGIP1;1), three bacterial GLPs type II, one bacterial GLP type I, AQPZ from
E. coli, NIP 1;1 from maize (the closest plant homolog to GIP1;1 found by BLAST searches), and bovine AQP1. Identical residues
are boxed and shaded in gray, and similar residues are shaded in light gray if present in more than four taxa. Membrane spanning
and half membrane spanning helices are indicated by boxes. A novel putative transmembrane region in B. halodurans is
indicated by a dashed box. Also indicated are the positions of the four residues in the constriction regions, R1 to R4. Residues in
GIP1;1 that are conserved among bacterial GLPs or only among type II GLPs are marked with crosses and stars, respectively.
Another odd feature of GIP1;1 is that it differs in
some amino acid positions that are very well conserved
among glycerol facilitators and even MIPs in general.
For example, in EcGlpF, H66, T72, F89, and Q93 have
been suggested to be involved in packing of the core
near the first NPA-box (Fu et al., 2000). However, in
GIP1;1, the corresponding residues are F66, A72, C89,
and E93, which all have properties that are very different from the conserved residues. In addition, the
loop preceding the first NPA-box is one amino acid
residue shorter in GIP1;1 than in most MIPs.
GIP1;1 has a relatively short N terminus, which is
unusual for plant MIPs and only found in the SIP
subfamily. In contrast, short N termini are common
among bacterial glycerol channels. The C terminus
of GIP1;1 is also relatively short but contains a putative phosphorylation site in resemblance to some
other plant MIPs that have been suggested to be regulated by phosphorylation (Johansson et al., 2000). How-
ever, the putative phosphorylation motif in GIP1;1,
RXXRXS, is not identical to phosphorylation motifs
in other plant MIPs. Only short motifs with high
probability of occurrence by chance are found when
GIP1;1 is scanned for motifs and patterns in the
PROSITE database (data not shown).
The most probable subcellular localization predicted by TargetP V1.0 is ‘‘other membrane’’ (score of
0.794), which excludes membranes from chloroplast
(0.007), mitochondria (0.020), and secretory pathway
(0.657). The PSORT result suggests plasma membrane
localization (0.600). Second best is the chloroplast
thylakoid membrane (0.421). Hence, the most probable
localization of GIP1;1 is in the plasma membrane.
Phylogenetic Analysis of GIP1;1
GIP1;1 belongs to the MIPs, but it clearly diverts from
any other plant MIP characterized to date. Instead, it is
Plant Physiology Preview
3 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Gustavsson et al.
more similar to bacterial glycerol transporters. To
clarify the origin of the GIP1;1 gene, phylogenetic analyses were performed. Bayesian, neighbor joining, and
parsimony methods were employed for analysis of
a multiple alignment of 67 taxa, including mammal,
plant, fungi, bacterial, and archaean MIP sequences.
These particular taxa were sampled to represent all the
plant MIP subfamilies, mammalian AQPs and GLPs
and a wide range of bacterial glycerol transporters.
The archaean sequence MthAQPM was used as an
outgroup. The ingroups form two major clades that are
well supported both by neighbor joining and Bayesian
methods (Fig. 3). The plant MIPs, except GIP1;1,
together with the bacterial and human water channels,
form one major clade from which the NIPs split off
first and then the bacterial AQPs. The second major
clade consists mainly of bacterial and human glycerol
transporters that branch into two subclades: type I
glycerol transporters together with the human aquaglyceroporins and type II glycerol transporters together with GIP1;1. The bootstrap value and the
posterior probability of the node joining type II glycerol transporters and GIP1;1 are 71% and 100%, respectively. This makes the clustering of GIP1;1 with
Figure 3. Bayesian 50% majority-rule consensus
tree. Posterior probabilities of each node are indicated. GIP1;1 (PpGIP1;1; marked by an asterisk) is
closely associated with the bacterial GLPs of type II.
Archaean MthAQPM was used as outgroup.
Plant Physiology Preview
4 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Bacterial-Like Glycerol Channel in Mosses
the type II bacterial GLPs quite well supported and
confirms our initial classification of GIP1;1. Interestingly, the human aquaglyceroporins cluster together
with type I bacterial GLPs. However, with a posterior
probability and a bootstrap value of 52%, this grouping is poorly supported.
In a parsimony analysis of the 67 taxa, the resulting
trees were very unstable, and the topology of the trees
changed upon removal of some taxa (data not shown).
However, in an exhaustive parsimony search of 10 taxa
of the type II bacterial GLPs and GIP1;1, using EcGlpF
as outgroup, GIP1;1 appears at the top in all of the
resulting five trees (data not shown). GIP1;1 is associated with Clostridium tetani or Chloroflexus aurantiacus
GLPs in four trees and with C. aurantiacus and Oceanobacillus iheyensis GLPs in one tree. In the Bayesian
tree, the former association is supported by a posterior
probability of 96%. Although there are several differences between the trees from different methods, the
major groupings are similar and most importantly support that GIP1;1 belongs to the clade of type II bacterial
glycerol transporters rather than to other plant MIPs.
The divergency times of GIP1;1 and type II bacterial
GLPs and of NIPs and bacterial AQPZs were estimated to 1,038 and 1,135 million years, respectively,
using the nonparametric rate-smoothing method
(Sanderson, 2003).
Functional Analysis of GIP1;1 in Xenopus laevis Oocytes
In order to examine the permeability of GIP1;1, the
protein was heterologously expressed in X. laevis
oocytes. Oocytes were injected with water or in vitro
transcribed RNA coding for GIP1;1 or a positive control. Labeling of the oocyte proteins with [35S]-Met
confirmed that GIP1;1 was expressed in the oocytes
(Fig. 4A). The water permeability of oocytes that
expressed GIP1;1 was not significantly higher than
that of the negative control (Fig. 4B). In contrast, the
glycerol uptake was 4-fold higher in oocytes injected
with RNA encoding GIP1;1 than for the water-injected
negative controls (Fig. 4C). Although GIP1;1 is not as
permeable to glycerol as the positive control EcGlpF,
GIP1;1 is clearly a glycerol-specific channel.
DISCUSSION
GIP1;1 Belongs to the Bacterial Glycerol Transporters
of Type II
Based on sequence comparisons, glycerol transporters of bacteria are divided into two groups that have
been proposed to have different substrate specificity
(Hohmann et al., 2000; Froger et al., 2001). In the
phylogenetic tree in Figure 3, we refer to these groups
as types I and II GLPs. This classification reflects the
phylogeny of bacteria since type I GLPs are preferentially found in gram-negative bacteria, whereas type
II is commonly found in gram-positive bacteria
(Hohmann et al., 2000). In all our phylogenetic analyses,
Figure 4. Protein labeling and water and glycerol permeability measurements in X. laevis oocytes expressing GIP1;1, ZmPIP2;1, or GlpF.
A, In vivo labeled proteins in total membrane fraction prepared from
the oocytes. The arrowhead indicates the position of GIP1;1. This band
is not present in oocytes injected with water. The position of the
molecular mass markers is indicated. B, Osmotic water permeability
coefficient of oocytes injected with water or cRNA encoding GIP1;1 or
ZmPIP2;1. The positive control ZmPIP2;1 has high water permeability,
whereas the water permeability of GIP1;1 is not different from waterinjected oocytes. The data are expressed as means 6 SD of measurements from two independent experiments with the number of oocytes
indicated at each bar. C, Uptake of glycerol in oocytes injected with
water or cRNA encoding GIP1;1 or EcGlpF. The glycerol permeability
of GIP1;1 is significantly different from the water-injected negative
control (Student’s t test, P , 0.05) but not as high as in the positive
control EcGlpF. The data are expressed as the means 6 SD of measurements from two independent experiments with the number of oocytes
indicated at each bar.
Plant Physiology Preview
5 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Gustavsson et al.
GIP1;1 clusters with the type II GLPs. The bacterial
glycerol transporter best characterized so far is EcGlpF,
which belongs to type I glycerol transporters. EcGlpF
has been shown to be relatively permeable to glycerol
and other linear alcohols but less permeable to water
(Heller et al., 1980; Maurel et al., 1994; Borgnia and
Agre, 2001). Furthermore, the three-dimensional structure of EcGlpF is known at a resolution of 2.2 Å (Fu
et al., 2000). The only member of type II glycerol transporters functionally characterized is GlaLlac from
Lactococcus lactis. In contrast to EcGlpF, GlaLlac has
been shown to be functional both as a water channel
and as a glycerol transporter by complementation
studies in E. coli mutants and by studies of water and
glycerol uptake in X. laevis oocytes after heterologous
expression (Froger et al., 2001; Thomas et al., 2002). At
several amino acid residue positions, GIP1;1 shares
residues with both type I, type II, and other glycerol
transporting MIPs, such as NIPs and human aquaglyceroporins, that are not conserved among most other
MIPs (Fig. 2). Some of these residues have been suggested to interact with the substrate/substrates (Fu
et al., 2000). T137 and G195 (numbers are referring to
residues in EcGlpF) are two examples of such residues.
There are also positions that are conserved within the
type I or type II classes, for example, residues corresponding to P141 and P196 in EcGlpF, both suggested
to be involved in glycerol binding (Fu et al., 2000). In
type II and in GIP1;1, the corresponding conserved
residues are I243 and G309. In general, GIP1;1 is more
similar to glycerol transporters of type II than type I.
Constriction Region in GIP1;1 and Substrate Specificity
In addition to the extended C-loop, there are several
other deviations from previously identified MIPs. In
EcGlpF, H66, T72, F89, and Q93 have been suggested
to participate in packing of the core next to the first
NPA-box. These amino acid residues are very well
conserved throughout all MIPs with few exceptions.
The corresponding residues in GIP1;1, F66, A72, C89,
and E93, suggest that correlated replacements have
resulted in an alternative packing core in GIP1;1.
These changes might also modify the properties of
the pore in this region, but the exact consequences on
the substrate specificity are hard to predict.
According to available high-resolution structures,
the amino acid residues in the constriction region form
the narrowest part of the channel and directly interact
with the substrate. The constriction regions of some of
the discussed MIPs are summarized in Table I. In
AQP1, amino acid residues F58, H182, C191, and R197
form the constriction region (numbering refers to
bovine AQP1). A nitrogen in the His side chain makes
a hydrogen bond to one of the water molecules in the
channel. The same water is also bonded to the side
chain of the Arg. The Phe orients the water molecule
and thereby enables the hydrogen bonding from water
to H182 and R197 (de Groot and Grubmuller, 2001).
The Cys interacts with the water molecule by hydro-
Table I. Amino acid residues forming the constriction region in
relevant MIPs
MIP
R1a
R2
R3
R4
BtAQP1
EcGlpF
GLALlac
GIP1;1
GLP type IIc
F
W
Y
F
W
Hb
G
V
V
G
C
F
P
P
Y
R
R
R
R
R
a
RX, Residue position in the constriction region corresponding
to amino acid residues 58, 182, 191, and 197 in bovine AQP1
b
c
(BtAQP1).
More polar residues are shown in bold.
GLP type
II refers to consensus residues in this phylogenetic group.
gen bond from the backbone carbonyl group (Sui et al.,
2001).
In EcGlpF, the corresponding amino acid residues
are W48, G191, F200, and R206 (Fu et al., 2000; de
Groot and Grubmuller, 2001; Stroud et al., 2003). Here,
the Arg side chain creates hydrogen bonds with two of
the substrate hydroxyl groups. The backbone of the
glycerol molecule is wedged tightly against a hydrophobic corner created by the Trp and the Phe. To make
space for the bulky Phe, the His is replaced by a Gly.
Although the EcGlpF pore is larger than the pore of
AQP1, the passage of water is relatively small. One
reason for this is the hydrophobicity of the constriction
region, which reduces the possibility of water molecules to pass (Sui et al., 2001).
In GlaLlac from L. lactis, which is an aquaglyceroporin with ability to facilitate both glycerol and water
transport, the constriction region is Y49, V223, P232,
and R238 (Froger et al., 2001; Thomas et al., 2002). Replacement of W48 and F200 (EcGlpF) by the less
hydrophobic Y49 and P232 (GlaLlac) is suggested to
result in a channel with a constriction region that is
more hydrophilic than the EcGlpF pore. Water can
thereby pass through the channel. Since there is no
bulky His in the GlaLlac constriction region, the pore
should be large enough to fit also a glycerol molecule.
In GIP1;1, the four constriction region residues are
F49, V304, P313, and R319. The only difference with
GlaLlac is that the Tyr is replaced by Phe, resulting in
a more hydrophobic constriction region. Thus, it is
reasonable to expect GIP1;1 to have a relatively large
pore and to be permeable for glycerol in a similar
manner as has been shown for GlaLlac. It is more speculative to predict the water permeability of GIP1;1.
The hydroxyl group of Tyr in GlaLlac might be important for water transport by providing a site for hydrogen bond formation. This would suggest that GIP1;1 is
less permeable to water compared to GlaLlac. It has
been proposed that aquaglyceroporins facilitating both
water and glycerol have two polar residues in the
constriction region, whereas glycerol channels excluding water only have one single polar residue in the
constriction region (Thomas et al., 2002). In line with
this, we predicted that GIP1;1 is mainly a glycerol facilitator and less permeable to water.
Plant Physiology Preview
6 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Bacterial-Like Glycerol Channel in Mosses
Although the constriction region of GIP1;1 is quite
similar to the corresponding region of GlaLlac, the MIP
sequences closest related to GIP1;1 (see below) usually
have Trp, Gly, Tyr, and Arg at the constriction region.
This suggests that GIP1;1 has evolved from an aquaglyceroporin into a glycerol-specific channel.
Functional Analyses of Substrate Specificity
Consistent with the predictions, functional analyses
clearly show that GIP1;1 is a glycerol-specific channel
with no or very low water permeability. However,
the glycerol permeability is somewhat lower than expected. The C-loop connects the two direct repeats
that form all MIPs and is one of the more variable regions in MIPs, indicating that alterations can be accommodated here without interfering with the general
MIP structure. Among GLPs and GIP1;1 there is a
conserved element in the last part of the C-loop that
has been shown to be involved in binding of glycerol
in the extracellular vestibule of EcGlpF (Fu et al., 2000).
Mutations in an aquaglyceroporin from Plasmodium
falciparum show that this element can also modulate
the water permeability (Beitz et al., 2004). It is thus
possible that the extended C-loop, directly or indirectly, affects interactions with the substrate in the vestibule, resulting in lower glycerol permeability.
Horizontal Gene Transfer
In our alignments, the MIPs most similar to GIP1;1
are GlpFs from C. tetani and Bacillus halodurans with
41.5% and 41.0% identity, respectively. In contrast, the
identity of the best match in plants based on BLAST
searches, NIP1;1 from Zea mays, is only 30.0%. The
phylogenetic tree (Fig. 3) and the high identity to type
II glycerol transporters preferentially found in the
evolutionary distant gram-positive bacteria suggest
that the ancestral GIP gene was incorporated in a plant
genome via horizontal gene transfer.
In a recent investigation of the transcriptome of
P. patens (Nishiyama et al., 2003), it was found that
some transcripts in P. patens are more similar to bacterial transcripts than to higher plant transcripts. These
transcripts were suggested to originate from horizontally transferred genes or from genes that were lost
during evolution of vascular plants. We note that among
these transcripts the EST BQ826917, which encodes
a part of GIP1;1, also appeared. The peptide was defined as a glycerol uptake facilitator most similar to
a protein from O. iheyensis but was not investigated
more closely. Although, GLP from O. iheyensis (OihGLP)
is the best match in BLAST searches, OihGLP together
with GLPs from Bacillus subtilis, B. halodurans, and
Thermus aquaticus form a sister group to the clade containing GIP1;1 in the Bayesian analysis. This grouping
is also supported by the finding of a very interesting
and conserved extra transmembrane helix in the C
termini of GLPs in the former group but not in the
GIP1;1 clade (Figs. 2 and 3).
The plant-specific NIP subfamily has also been suggested to have evolved through a horizontal gene
transfer event (Zardoya et al., 2002). In this case,
a bacterial aqpZ gene is believed to have been incorporated in the genome of a common ancestor to
land plants and later modified to allow the transport of
glycerol. The pairwise identities of EcAQPZ and the
closest NIP, NOD26 from soybean (Glycine max), and
GIP1;1 and the closest GLP, OihGLP, identified in
BLAST searches are 43.1% and 40.4%, respectively. The
high divergences could indicate that both types of
MIPs are relatively old. Using the nonparametric ratesmoothing method, it is estimated that the horizontal
gene transfer of NIP and GIP ancestors into plants
occurred 1,135 and 1,038 million years ago, respectively. The former estimate is in agreement with the
earlier published date for the horizontal gene transfer
of the NIP ancestor 1,188 million years ago (Zardoya
et al., 2002). However, one has to be cautious inferring
dates from sequence divergence because substitution
rates vary widely between different proteins and also
because both GIP1;1 and NIPs might have gone
through phases of positive selection/adaptive evolution that have resulted in a modified substrate specificity. Nevertheless, it is interesting to note that P.
patens harbors two different types of MIPs that both
might have been acquired independently via horizontal gene transfer from bacteria and selected for as
glycerol transporters. This suggests that it is a selective
advantage for plants to be able to facilitate transport of
glycerol across membranes. In a recent study, evidence
was presented indicating that glycerol is an important
carbon source transferred from a host plant to a fungal
pathogen (Wei et al., 2004). We speculate that exported
plant glycerol might also be important for symbiotic
fungi and bacteria and that this is a possible reason
why the glycerol transporters have been fixed in plant
genomes.
Adaptations of GIP1;1
If GIP1;1 was recruited by plants from a grampositive bacterium by horizontal gene transfer, there
were probably no introns in the original GIP gene.
Although it is well established that intron recruitment
occurs, the mechanisms are still unclear (Brady and
Danforth, 2004). One possibility is that intron insertion
into GIP1;1 did occur by homologous recombination
with another MIP. In our alignment, the first GIP1;1
intron is situated exactly at the same site as the conserved first intron site of PIPs. This region immediately before the first NPA-box is conserved among
MIPs, thus providing a possible site for homologous
recombination. However, the sequence flanking the
intron site in GIP1;1 deviates from the general MIP
sequence, and there are no traces of the PIP-like sequence that would be expected to be incorporated
along with the intron. Thus, there is no evidence
indicating that the intron was inserted via homologous recombination with another plant MIP gene. The
Plant Physiology Preview
7 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Gustavsson et al.
divergence between GIP1;1 and the closest bacterial
homolog suggests that GIPs have had time to adapt
a gene structure compatible with expression in plants.
In line with this, we fail to detect any significant
difference in codon usage or GC content compared to
other genes in P. patens (data not shown). The fact that
many functional important amino acid residues still
are conserved after these adaptations and the higher
conservation of coding versus noncoding regions in
GIPs from P. patens and F. hygrometrica implicate that
the gene is under negative selection (purifying selection) and, thus, fully functional in P. patens and
F. hygrometrica.
(Arabidopsis thaliana) MIPs, and the total length was 808 characters. N- and
C-terminal regions and loop regions, except loops B and E, were excluded from
the analyses. In total, 198 characters of the original alignment were included in
the analyses (characters 345–372, 396–471, 611–637, 655–693, and 710–737).
Phylogenetic analysis was performed with Mr Bayes V 3.0 Mac (Ronquist and
Huelsenbeck, 2003) for 1,000,000 generations, every 100th tree was collected
and sampled, and the 1,500 first trees were excluded from the consensus tree.
The amino acid model was fixed to Jones. Neighbor joining and parsimony
analysis were preformed on the same data set using PAUP* 4.0b10 (Swofford,
2002). For the neighbor joining tree, bootstrap values of 1,000 replicates were
calculated. Identities were calculated from pairwise alignments adjusted for
gaps. AQP0, 1, 3, and 7 (accession numbers JN0557, P50501, CAA10517, and
NP_001015726) from amphibians were included in the neighbor joining tree
used in estimations of divergence times, applying the nonparametric ratesmoothing method (Sanderson, 2003). The split of human and amphibian
orthologs was set to 360 million years, and all four pairs of orthologs were
used as calibration points (Kumar and Hedges, 1998).
CONCLUSION
Water and Radiolabeled Glycerol Transport Assays in
Xenopus laevis Oocytes
A novel plant MIP with unique features has been
identified. The phylogenetic analyses and the high
identity to bacterial glycerol transporters support that
this gene has been acquired via horizontal gene transfer of a bacterial gene encoding a glycerol transporter.
Sequence-based predictions of substrate specificity
suggest that GIP1;1 is permeable for glycerol but not
for water. This is confirmed by functional analyses of
heterologously expressed GIP1;1 in X. laevis oocytes.
The high divergence of GIP1;1 from the proposed
gram-positive bacterial ancestors might suggest that
GIP1;1 is member of a relatively old and until now
unrecognized fifth subfamily of plant MIPs. Investigations of the presence of GIPs in other plant species
will bring more light to the origin and possible loss of
GIPs.
MATERIALS AND METHODS
Sequencing
EST clone PPAS20308 was achieved from the Leeds Institute for Plant
Biotechnology and Agriculture. The plasmid was sequenced with T3-, T7-, and
gene-specific primers. Genomic DNA was isolated from Physcomitrella patens
and Funaria hygrometrica kindly provided by Hans Ronne (Swedish University
of Agricultural Sciences, Uppsala, Sweden) and Nils Cronberg (Lund University,
Sweden), respectively. The genomic sequence from P. patens was obtained by
PCR amplification of overlapping segments from genomic DNA followed by
sequencing of PCR products with amplification primers and with internal
primers. Only a part of the GIP1;1 gene from F. hygrometrica was amplified and
sequenced. Primer sequences are available on request. The accession numbers
for cDNA, gene sequence from P. patens, and partial F. hygrometrica gene
sequence are AY611236, AY611237, and DQ092355, respectively.
Protein Predictions
The topology of the translated protein was predicted by the HMMTOP2.0
method (Tusnady and Simon, 1998, 2001). The sequence was scanned for
protein patterns in PROSITE (Bucher and Bairoch, 1994; Sigrist et al., 2002),
and subcellular localization was predicted by TargetP V1.0 and with PSORT
(Nakai and Kanehisa, 1991; Emanuelsson et al., 2000).
Multiple Alignment and Phylogenetic Analysis
A protein alignment of 67 taxa, including plant, human, fungi, bacterial,
and archaean sequences, was performed with ClustalW included in MacVector
7.2 (Accelrys) and refined by eye. The alignment included all 35 Arabidopsis
For in vitro transcription, the coding sequence of GIP1;1 was subcloned
into the BglII site of pXbG-ev1 by PCR. The construct was verified by
sequencing. In vitro transcription, X. laevis oocyte preparation, injection,
protein labeling, and the osmotic water permeability assay were performed
as previously described (Fetter et al., 2004). Glycerol uptake was performed 3 d
after injection. A group of five oocytes was incubated at room temperature in
1 mL of modified Barth’s solution (Fetter et al., 2004) containing 1 mM
unlabeled glycerol and 1 mCi/mL [14C]glycerol (138 mCi/mmol; Amersham
Pharmacia Biotech). After 0, 5, and 10 min, oocytes were washed rapidly four
times in ice-cold Barth’s solution, and individual oocytes were lysed for 2 to 3 h
in 1 mL of 5% (w/v) SDS for scintillation counting (Beckman LS 1701).
Sequence data from this article can be found in the GenBank/EMBL data
libraries under accession numbers AY611236, AY611237, and DQ092355.
ACKNOWLEDGMENTS
We thank Dr. Stavros Bashiardes as part of the P. patens EST Program at the
University of Leeds (Leeds, UK) and Washington University (St. Louis) for sharing the cDNA clone PPAS20308. Prof. Hans Ronne (Swedish University of
Agricultural Sciences, Uppsala, Sweden) and Nils Cronberg (Lund University)
are acknowledged for providing P. patens and F. hygrometrica, respectively.
Received March 22, 2005; revised June 12, 2005; accepted June 16, 2005;
published August 19, 2005.
LITERATURE CITED
Agre P, Bonhivers M, Borgnia MJ (1998) The aquaporins, blueprints for
cellular plumbing systems. J Biol Chem 273: 14659–14662
Beitz E, Pavlovic-Djuranovic S, Yasui M, Agre P, Schultz JE (2004)
Molecular dissection of water and glycerol permeability of the aquaglyceroporin from Plasmodium falciparum by mutational analysis. Proc
Natl Acad Sci USA 101: 1153–1158
Borgnia MJ, Agre P (2001) Reconstitution and functional comparison of
purified GlpF and AqpZ, the glycerol and water channels from Escherichia coli. Proc Natl Acad Sci USA 98: 2888–2893
Borstlap AC (2002) Early diversification of plant aquaporins. Trends Plant
Sci 7: 529–530
Brady SG, Danforth BN (2004) Recent intron gain in elongation factor-1a
of colletid bees (Hymenoptera: Colletidae). Mol Biol Evol 21: 691–696
Bucher P, Bairoch A (1994) A generalized profile syntax for biomolecular
sequence motifs and its function in automatic sequence interpretation.
Proc Int Conf Intell Syst Mol Biol 2: 53–61
de Groot BL, Engel A, Grubmuller H (2001) A refined structure of human
aquaporin-1. FEBS Lett 504: 206–211
de Groot BL, Grubmuller H (2001) Water permeation across biological membranes: mechanism and dynamics of aquaporin-1 and GlpF.
Science 294: 2353–2357
Plant Physiology Preview
8 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.
Bacterial-Like Glycerol Channel in Mosses
Emanuelsson O, Nielsen H, Brunak S, von Heijne G (2000) Predicting
subcellular localization of proteins based on their N-terminal amino
acid sequence. J Mol Biol 300: 1005–1016
Fetter K, Van Wilder V, Moshelion M, Chaumont F (2004) Interactions
between plasma membrane aquaporins modulate their water channel
activity. Plant Cell 16: 215–228
Froger A, Rolland JP, Bron P, Lagree V, Le Caherec F, Deschamps S,
Hubert JF, Pellerin I, Thomas D, Delamarche C (2001) Functional
characterization of a microbial aquaglyceroporin. Microbiol 147:
1129–1135
Fu D, Libson A, Miercke LJ, Weitzman C, Nollert P, Krucinski J, Stroud
RM (2000) Structure of a glycerol-conducting channel and the basis for
its selectivity. Science 290: 481–486
Gonen T, Sliz P, Kistler J, Cheng Y, Walz T (2004) Aquaporin-0 membrane
junctions reveal the structure of a closed water pore. Nature 429:
193–197
Harries WE, Akhavan D, Miercke LJ, Khademi S, Stroud RM (2004) The
channel architecture of aquaporin 0 at a 2.2-A resolution. Proc Natl Acad
Sci USA 101: 14045–14050
Heller KB, Lin EC, Wilson TH (1980) Substrate specificity and transport
properties of the glycerol facilitator of Escherichia coli. J Bacteriol 144:
274–278
Heymann JB, Engel A (1999) Aquaporins: phylogeny, structure, and
physiology of water channels. News Physiol Sci 14: 187–193
Hohmann I, Bill RM, Kayingo I, Prior BA (2000) Microbial MIP channels.
Trends Microbiol 8: 33–38
Johanson U, Gustavsson S (2002) A new subfamily of major intrinsic
proteins in plants. Mol Biol Evol 19: 456–461
Johanson U, Karlsson M, Johansson I, Gustavsson S, Sjovall S, Fraysse L,
Weig AR, Kjellbom P (2001) The complete set of genes encoding major
intrinsic proteins in Arabidopsis provides a framework for a new
nomenclature for major intrinsic proteins in plants. Plant Physiol 126:
1358–1369
Johansson I, Karlsson M, Johanson U, Larsson C, Kjellbom P (2000) The
role of aquaporins in cellular and whole plant water balance. Biochim
Biophys Acta 1465: 324–342
Kumar S, Hedges SB (1998) A molecular timescale for vertebrate evolution. Nature 392: 917–920
Maurel C, Reizer J, Schroeder JI, Chrispeels MJ, Saier MH Jr (1994)
Functional characterization of the Escherichia coli glycerol facilitator,
GlpF, in Xenopus oocytes. J Biol Chem 269: 11869–11872
Nakai K, Kanehisa M (1991) Expert system for predicting protein localization sites in gram-negative bacteria. Proteins 11: 95–110
Nishiyama T, Fujita T, Shin IT, Seki M, Nishide H, Uchiyama I, Kamiya
A, Carninci P, Hayashizaki Y, Shinozaki K, et al (2003) Comparative
genomics of Physcomitrella patens gametophytic transcriptome and
Arabidopsis thaliana: implication for land plant evolution. Proc Natl
Acad Sci USA 100: 8007–8012
Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic
inference under mixed models. Bioinformatics 19: 1572–1574
Sanderson MJ (2003) r8s: inferring absolute rates of molecular evolution
and divergence times in the absence of a molecular clock. Bioinformatics
19: 301–302
Savage DF, Egea PF, Robles-Colmenares Y, Iii JD, Stroud RM (2003)
Architecture and selectivity in aquaporins: 2.5 a x-ray structure of
aquaporin z. PLoS Biol 1: E72
Sigrist CJ, Cerutti L, Hulo N, Gattiker A, Falquet L, Pagni M, Bairoch A,
Bucher P (2002) PROSITE: a documented database using patterns and
profiles as motif descriptors. Brief Bioinform 3: 265–274
Stroud RM, Miercke LJ, O’Connell J, Khademi S, Lee JK, Remis J, Harries W,
Robles Y, Akhavan D (2003) Glycerol facilitator GlpF and the associated
aquaporin family of channels. Curr Opin Struct Biol 13: 424–431
Sui H, Han BG, Lee JK, Walian P, Jap BK (2001) Structural basis of waterspecific transport through the AQP1 water channel. Nature 414: 872–878
Swofford DL (2002) PAUP*. Phylogenetic Analysis Using Parsimony (*and
Other Methods). Version 4. Sinauer Associates, Sunderland, MA
Thomas D, Bron P, Ranchy G, Duchesne L, Cavalier A, Rolland JP,
Raguenes-Nicol C, Hubert JF, Haase W, Delamarche C (2002) Aquaglyceroporins, one channel for two molecules. Biochim Biophys Acta
1555: 181–186
Tusnady GE, Simon I (1998) Principles governing amino acid composition
of integral membrane proteins: application to topology prediction. J Mol
Biol 283: 489–506
Tusnady GE, Simon I (2001) The HMMTOP transmembrane topology
prediction server. Bioinformatics 17: 849–850
Wei Y, Shen W, Dauk M, Wang F, Selvaraj G, Zou J (2004) Targeted gene
disruption of glycerol-3-phosphate dehydrogenase in Colletotrichum
gloeosporioides reveals evidence that glycerol is a significant transferred
nutrient from host plant to fungal pathogen. J Biol Chem 279: 429–435
Zardoya R, Ding X, Kitagawa Y, Chrispeels MJ (2002) Origin of plant
glycerol transporters by horizontal gene transfer and functional recruitment. Proc Natl Acad Sci USA 99: 14893–14896
Plant Physiology Preview
9 of 9
Downloaded from on June 14, 2017 - Published by www.plantphysiol.org
Copyright © 2005 American Society of Plant Biologists. All rights reserved.