Identification of residues controlling transport through the yeast
aquaglyceroporin Fps1 using a genetic screen
Sara Karlgren
1
, Caroline Filipsson
2
, Jonathan G. L. Mullins
3
, Roslyn M. Bill
1,4
, Markus J. Tama
´
s
1
and Stefan Hohmann
1
1
Department of Cell and Molecular Biology/Microbiology, Go
¨
teborg University, Sweden;
2
Department of Biochemistry and
Biophysics, Go
¨
teborg University, Sweden;
3
Swansea Clinical School, University of Wales Swansea, UK;
4
School of Life
and Health Sciences, Aston University, Birmingham, UK
Aquaporins and aquaglyceroporins mediate the transport of

porins facilitate the diffusion of water across biological
membranes while the closely related aquaglyceroporins
mediate transport of water and solutes such as glycerol and
urea. These proteins are present in membranes where rapid
and controlled water or solute fluxes occur, for example, in
the mammalian kidney [2,4,5] and plant roots [6,7]. The
yeast Saccharomyces cerevisiae has four such MIP channels:
the aquaporins Aqy1 and Aqy2 and the aquaglyceroporins
Fps1 and Yfl054 [8]. Aqy1 is a strictly spore-specific
aquaporin while Aqy2 may play a role in osmoregulation
during cell growth (F. Sidoux-Walter & S. Hohmann,
unpublished observation). Possible roles in freeze tolerance
have been claimed for Aqy1 and Aqy2 [9]. The physiological
role of Yfl054 has not yet been established [8,10].
Yeast cells accumulate glycerol as a compatible solute in
osmoregulation [11]. The plasma membrane channel Fps1
mediates glycerol export and is required for survival of
a hypo-osmotic shock when glycerol has to be rapidly
exported from cells in order to prevent bursting [12,13]. On
the other hand, hyperactive Fps1 causes an inability to grow
at high external osmolarity because cells lose the glycerol
they produce [12,13]. Moreover, it has been shown that
Fps1 is required to control turgor and prevent cell lysis
during cell fusion of mating yeast cells [14]. Together, these
observations illustrate that Fps1 plays a central role in yeast
osmoregulation.
The transport function of Fps1 is controlled by osmotic
changes in order to prevent glycerol loss at high osmolarity
and to allow rapid export at low external osmolarity. The
capacity for glycerol transmembrane flux through the

transport and at the same time high selectivity of water or
glycerol transport [17–22].
Fps1 is an atypical aquaglyceroporin as the highly
conserved NPA motifs in the B- and the E-loop are NPS
(Asn-Pro-Ser) and NLA (Asn-Leu-Ala), respectively,
sequences that are also found in the Plasmodium glycerol
facilitator, although in the opposite loops [23]. While Fps1
can tolerate NPA in both positions, the Escherichia coli
homologue GlpF is inactive when its NPA motifs are
converted to NPS and NLA, suggesting a somewhat
different and more flexible arrangement of the Fps1 channel
[24]. In addition, Fps1 has unique long N- and C-terminal
domains only found in orthologues from other yeasts [15].
While large parts of these extensions can be removed
without apparent consequence, short domains close to the
first and the last TMD seem to be required for channel
control: deletions or mutations in these regions render the
channel hyperactive (K. Hedfalk, R. M. Bill, J. G. Mullins,
S. Karlgren, C. Filipsson, C. Bergstrom, M. J. Tama
´
s,
J. Rydstro
¨
m & S. Hohmann, unpublished observation)
[13,15]. The N-terminal regulatory domain may fold in a
similar way as the channel forming B- and E-loops, hence
dipping into the membrane. We suggested that this domain,
dubbed the N-loop, might directly interact with the channel
forming B-loop to control transport function [15].
As a novel approach to study the control of Fps1, we

YNB agar plates, resuspending them in YNB medium to an
D
600
of 0.4 and then performing 10-fold serial dilutions. Cell
suspensions (5 lL) were spotted onto agar plates supple-
mented with 1
M
xylitol, or with 0.8
M
NaCl as a negative
control, and on medium without osmoticum as a positive
control. Growth was monitored after 2–7 days at 30 °C.
For growth tests after osmotic shifts, transformants were
pregrown on YNB plates, then resuspended and spotted on
the same medium as the control. To invoke hyperosmotic
shock, cells were pregrown in medium without osmoticum
and shifted to medium with 0.8
M
NaCl. For a hypo-
osmotic shock, cells were pregrown in the presence of 0.8
M
NaCl and shifted to medium without salt. Growth was
monitored as above.
Mutagenesis and screening
Random mutations in FPS1 were introduced by transform-
ing YEpmyc-FPS1 into the E. coli strain XL1-Red from
Stratagene (La Jolla, CA, USA) following the manufac-
turer’s recommendations. Transformants were grown for
approximately 24 h yielding about 200 colonies per plate.
Colonies from each plate were pooled and grown in LB-

sequenced.
Western blot analysis
Cells were cultured in YNB supplemented with 2% glucose
to late log phase (typically D
600
is 0.8). The total membrane
fraction was isolated and visualized as described previously
(S.Karlgren,N.Pettersson,R.M.Bill&S.Hohmann,
unpublished observation).
Glycerol transport measurements
To determine glycerol influx following its concentration
gradient, cells were grown in liquid YNB medium to a D
600
of approximately 0.7. Cells were harvested, washed and
suspended in ice-cold Mes buffer (10 m
M
Mes, pH 6.0) to a
density of 40–60 mg cellsÆml
)1
. All subsequent steps were
performed at 4 °C. Glycerol influx in the presence or
772 S. Karlgren et al. (Eur. J. Biochem. 271) Ó FEBS 2004
absence of hyperosmotic stress was measured by adding
glycerol to a final concentration of 100 m
M
ÔcoldÕ glycerol
plus 40 l
M
[
14

glycerol and hence cannot grow at elevated osmolarity
caused by salt [26], or by various polyols including xylitol
and even glycerol (S. Karlgren, N. Pettersson, R. M. Bill &
S. Hohmann, unpublished observation). However, growth
of the gpd1D gpd2D mutant in the presence of polyols can be
rescued when transformed with a plasmid encoding hyper-
active Fps1 (FPS1-D1) that alleviates the osmotic dis-
equilibrium by permitting solute influx (S. Karlgren,
N. Pettersson, R. M. Bill & S. Hohmann, unpublished
observation) [12,13] (Fig. 1). Although hyperactive Fps1
can rescue growth of the gpd1D gpd2D mutant through
influx of various polyols (S. Karlgren, N. Pettersson, R. M.
Bill & S. Hohmann, unpublished observation) we chose
xylitol for the screen as it gave the clearest phenotype. We
note that we actually screen for hyperactive xylitol uptake
while the objective is to obtain mutants that fail to retain
internally produced glycerol under hyperosmotic stress, an
aspect that will be discussed when interpreting the muta-
tions obtained.
Transformants with mutagenized FPS1 in the gpd1D
gpd2D strain were grown to colonies on YNB and then
replica-plated onto plates supplemented with 1
M
xylitol.
Approximately 5000 colonies were screened and 31 grew on
xylitol plates. These were re-tested for growth on 1
M
xylitol.
Plasmids were isolated from these positive yeast colonies,
amplified in E. coli, checked by restriction analysis and

among yeast Fps1 orthologues, encompasses the stretch
from Met219 until about Ser248. A nine amino acid linker
follows this domain to the first TMD, which is predicted
to start at Leu257. Figure 2B provides an overview of the
relevant mutations from previous [15] and present analyses
within this region. One mutation was found in the approach
to the first TMD in a lysine (K250E) that is conserved
among yeast Fps1 orthologues [15]. Importantly, two
residues within the channel forming B-loop, which are both
highly conserved throughout the MIP family, were affected
by multiple exchanges. All three mutations occurring in the
C-terminal extension caused premature translation termin-
ation, either by generating a stop codon or a frame shift
leading to a stop some codons further downstream. As the
newly isolated mutations do not hit all residues that we
previously found to be critical for Fps1 control, while at the
same time we identified new relevant residues, the present
genetic screen is not saturated. A significant larger number
of mutations, probably more than 100, will be needed for a
fully comprehensive mutational map of channel control.
The genetic screen employed selected for mutated
versions of Fps1 that retain function, hence the proteins
Table 1. Summary of mutations obtained.
Mutation Nucleotide change Location
K223E AAGfiGAG In front of the N-terminal regulatory domain
Q227R CAGfiCGG Within the N-terminal regulatory domain
T231A ACAfiGCA Within the N-terminal regulatory domain
P232S CCTfiTCT Within the N-terminal regulatory domain
P236L (found five times) CCCfiCTC Within the N-terminal regulatory domain
S246P + Q592 stop TCTfiCCT + CAAfiTAA Between the N-terminal regulatory domain and TMD1

T231A
P323S
P236L
S246P +
Q592stop
G348D
G348R
G348S
H350L
H350Y
L451W
Q592stop + S246P
B-loop
E-loop
N-terminus
C-terminus
W541stop
+
H
3
N
I531FIRVMNLQSTG
S537QLVFTSL
-
OOC
A
Fig. 2. Summary of mutations obtained. (A)
Topology map of Fps1 indicating mutations
found in the genetic screen reported here. For
clarity the N- and the C-termini are not shown

M
xylitol to the
gpd1D gpd2D mutant, albeit clearly to different extents
(Fig. 1). The three most upstream mutations as well as the
C-terminal truncations conferred the weakest growth on
xylitol. The double mutant, S246P plus stop at 591,
conferred particularly robust growth on xylitol and in
contrast to all other mutations even allowed growth in the
presence of 1
M
of the C6 polyol
D
-sorbitol. As all other
mutations in the C-terminal extension occur much further
upstream and caused much weaker growth on xylitol and
because we did not observe any effects conferred by
truncations that far downstream in the C-terminus in a
different study (K. Hedfalk, R. M. Bill, J. G. Mullins,
S. Karlgren, C. Filipsson, C. Bergstrom, M. J. Tama
´
s,
J. Rydstro
¨
m & S. Hohmann, unpublished observation), we
believe that the effect in the double mutant is mainly due to
the S246P mutation, although we can not fully exclude a
synergistic effect of both mutations.
As we had isolated the mutants based on their ability to
mediate xylitol influx we wished to test if the novel
Fps1 derivates were still fully able to exert their normal

previously that this ability is well reflected by growth
characteristics in the presence of high external osmolarity
[12,13,15].
The different mutants showed varying degrees of osmo-
sensitivity (Fig. 1). The strongest effects were observed for
mutations within the N-terminal regulatory domain and in
particular for mutations of the two prolines P232 and P236,
the double mutant (S246P plus stop at Q591), K250E in
the linker to the first TMD and mutations in G348 in the
B-loop. Interestingly, different exchanges of the highly con-
served G348 caused different effects, with G348D causing
strongest osmosensitivity. The two different exchanges of
the conserved His350 caused similar effects.
Two residues in the N-terminal regulatory domain had
previously been studied by alanine-scanning mutagenesis:
Gln227 and Pro236 [13]. In order to compare the effects of
the different exchanges directly they were tested side-by-side
(Fig. 4). Q227A and Q227R caused similar strong osmo-
sensitivity indicating that the two exchanges, although
chemically very different, affected channel control in a
similar way. While exchange of Pro236 with alanine only
Fig. 3. Western blot analysis of the whole membrane fraction probed
with an anti c-myc Ig against the C-terminal c-myc-tag of Fps1.
Amounts of protein loaded from left to right: 50, 20, 30, 50, 50, 30, 50,
50, 30, 50, 50, 100, 50 and 50 lg. The lower double band is probably a
degradation product.
Fig. 4. Growth on plates. Cells were pregrown on YNB and dropped in
a 1 : 10 dilution series on the same medium with or without NaCl.
Ó FEBS 2004 Control of aquaglyceroporin Fps1 (Eur. J. Biochem. 271) 775
caused moderate osmosensitivity, exchange with the some-

mutant proteins
We have previously shown that the ability of Fps1
derivatives to mediate glycerol transport and to down-
regulate glycerol transport upon a hyperosmotic shock
can be monitored by measuring the influx of radiolabelled
glycerol following its concentration gradient in unstressed
cells and in cells exposed to 0.8
M
NaCl. We selected some
mutations from the new collection that represented different
characteristics including P236L, G348D, H350L and
L451W. As observed previously, glycerol transport through
hyperactive Fps1 is higher than that through wild type
under all conditions and is down-regulated by hyperosmotic
shock. This is also the case for the mutants studied here
(Fig. 5). All of them, however, maintained a much higher
glycerol transport capacity then wild type even after
hyperosmotic shock. The apparent rate of glycerol transport
is lower for H350L than for the other mutants, which is
consistent with the fact that it only conferred moderate
osmosensitivity, a measure for glycerol loss (Fig. 1). The
three other mutants conferred similar glycerol influx while
they caused different degrees of osmosensitivity. Glycerol
uptake rates correlated better with the ability to grow on
xylitol, suggesting that in some of the mutants isolated
influx may be enhanced to a higher degree than efflux.
Discussion
The transmembrane transport of glycerol in yeast is rapidly
controlled by osmotic changes to ensure glycerol accumu-
lation under hyperosmotic stress and fast glycerol release

xylitol. Hence
we screen for enhanced uptakeof glycerol, while the
physiological role is glycerol export. Most, although not
all, mutants we obtained in this way also conferred
osmosensitivity (and hence enhanced glycerol loss under
these conditions). Moreover, we obtained several mutations
in residues that we previously identified as important by
targeted mutagenesis. These observations confirm the
validity of the approach.
Although we screened for enhanced uptake, all mutations
faced the inside of the cell. Recently we screened for
suppressors of truncated, hyperactive Fps1 and obtained
mutations that reduced transport. The four mutants char-
acterized faced the outside of the cell [15]. Structural analysis
of AQP1 and GlpF suggested that these channels are largely
symmetric (except for the tails facing the inside) [17,32].
While our mutational analysis may not yet be representa-
tive, the distribution of mutations may suggest that the
Fig. 5. Uptake profile of 100 m
M
radiolabelled glycerol by the strains
indicated, before and after an osmotic shift to 0.8
M
NaCl.
776 S. Karlgren et al. (Eur. J. Biochem. 271) Ó FEBS 2004
outside face is mainly important for transport and the inside
face for control, at least in the case of the somewhat unusual
Fps1.
Some mutations, such as L451W and C-terminal trun-
cations but also alterations in His350 allow solid growth in

important residues were identified on the basis of mutations
that block transport. Hence, the approach used here, which
is novel as it screens for gain of function, leads to completely
novel insight into the structure-function relationship of MIP
channels.
Based on the structure of GlpF [17] and previous
modelling [15] we have attempted to rationalize the
mutations obtained in this study. We have shown previously
that the N-terminal regulatory domain, dubbed the N-loop,
has sequence and predicted structural similarity with the
channel forming B- and E-loop. We suggested that B- and
N-loops may interact.
In the models shown (Fig. 6), the 226–236 region of the
N-loop and 347–356 of the B-loop are in close proximity.
Residues affected by mutation are then located for the most
part in the functionally critical pore region. These include
Lys223 (violet), Gln227 (white), Thr231 (brown), Pro232
and Pro236 (grey), and on the B-loop they include Gly348
(orange) and His350 (blue). Lys451 (black) is the only
mutation to be clearly located away from the pore. Ser246,
Lys250, Ile531 and Ser537 lie in regions of the protein that
are not currently structurally modelled.
K223E (violet) and K250E (not modelled) result in a
charge reversal, which is likely to introduce electrostatic
interactions with other nearby lysine residues. This, in turn,
is likely to reduce the flexibility of the section linking the
N-terminal regulatory domain with TMD1, thereby holding
the pore more permanently open.
Q227R (white) lies directly adjacent to the NPQ (Asn-
Pro-Gln) region of the regulatory motif (which is similar to

arrangement with nearby residues, notably Q227 (white) on
the N-loop and His350 (blue) on the B-loop. The mutation of
this glycine residue could also disrupt the capacity of the
B-loop for membrane insertion, as it is clearly located in
the interfacial environment between the membrane face and
the cytosolic compartment (the orange residue, best viewed
in the side-on view of the model). Indeed, the mutations to
charged or polar residues have the most capacity for disrupt-
ing membrane insertion. A striking finding is the greater
effect observed with G348D than with G348R. This suggests
that the region surrounding Gly348 requires some flexibility
and distancing from His350, and perhaps to lie more
intimately with Gln227 on the N-loop to regulate normally.
In the model, where the beginning and end of the N-loop
region are aligned so that they face TMD1 as much as
possible, and with the NPQ turn of the B-loop maximally
immersed in the pore cavity, there is also a notably close
arrangement between Gln227 of the N-loop (shown in
white) and His350 on the B-loop (coloured blue). Both these
residues are highly polar, and have complementary charges
for interaction with each other.
The mutations of His350 (blue) caused marked effects
despite changing to similarly large residues. Hence, the
positive charge of His350 appears to be critical for normal
function. In this regard, its proximity in the model to the
negative dipole of Gln227 is of significant interest. His350 is
also conserved across the MIP family. Indeed, the random
mutations reported here regarding Gly348 and His350 are
likely to be of more general relevance for the understanding
of the structure-function relationship of the MIP family of

8. Hohmann, S., Bill, R.M., Kayingo, I. & Prior, B.A. (2000)
Microbial MIP channels. Trends Microbiol. 8, 33–38.
9. Tanghe, A., Van Dijck, P., Dumortier, F., Teunissen, A., Hohmann,
S. & Thevelein, J.M. (2002) Aquaporin expression correlates with
freeze tolerance in baker’s yeast, and overexpression improves
freeze tolerance in industrial strains. Appl. Environ. Microbiol. 68,
5981–5989.
10. Oliveira, R., Lages, F., Silva-Graca, M. & Lucas, C. (2003) Fps1p
channel is the mediator of the major part of glycerol passive dif-
fusion in Saccharomyces cerevisiae: artefacts and re-definitions.
Biochim. Biophys. Acta. 1613, 57–71.
11. Hohmann, S. (2002) Osmotic stress signaling and osmoadaptation
in yeasts. Microbiol. Mol. Biol. Rev. 66, 300–372.
12. Luyten, K., Albertyn, J., Skibbe, F., Prior, B.A., Ramos, J.,
Thevelein, J.M. & Hohmann, S. (1995) Fps1, a yeast member of
the MIP-family of channel proteins, is a facilitator for glycerol
uptake and efflux and it is inactive under osmotic stress. EMBO J.
14, 1360–1371.
13. Tama
´
s, M.J., Luyten, K., Sutherland, F.C.W., Hernandez, A.,
Albertyn,J.,Valadi,H.,Li,H.,Prior,B.A.,Kilian,S.G.,Ramos,J.,
Gustafsson, L., Thevelein, J.M. & Hohmann, S. (1999) Fps1p
controls the accumulation and release of the compatible solute
glycerol in yeast osmoregulation. Mol. Microbiol. 31, 1087–1104.
14. Philips, J. & Herskowitz, I. (1997) Osmotic balance regulates cell
fusion during mating in Saccharomyces cerevisiae. J. Cell Biol. 138,
961–974.
15. Tama
´

24. Bill, R.M., Hedfalk, K., Karlgren, S., Mullins, J.G., Rydstrom, J.
& Hohmann, S. (2001) Analysis of the pore of the unusual major
intrinsic protein channel, yeast Fps1p. J. Biol. Chem. 276, 36543–
36549.
25. Reference withdrawn.
26. Ansell, R., Granath, K., Hohmann, S., Thevelein, J. & Adler, L.
(1997) The two isoenzymes for yeast NAD-dependent glycerol
3-phosphate dehydrogenase encoded by GPD1 and GPD2,have
distinct roles in osmoadaption and redox regulation. EMBO J. 16,
2179–2187.
27. Thomas, B.J. & Rothstein, R.J. (1989) Elevated recombination
rates in transcriptionally active DNA. Cell. 56, 619–630.
28. Sherman, F., Fink, G.R. & Hicks, J.B. (1983) Methods in Yeast
Genetics. Cold Spring Harbor Laboratory Press, Cold Spring
Harbor, NY.
29.Gietz,D.,StJean,A.,Woods,R.A.&Schiestl,R.H.(1992)
Improved method for high efficiency transformation of intact
yeast cells. Nucleic Acids Res. 20, 1425.
30. Reference withdrawn.
31. Koradi, R., Billeter, M. & Wuthrich, K. (1996)
MOLMOL
: a pro-
gram for display and analysis of macromolecular structures.
J. Mol. Graph. 14 (51–5), 29–32.
32. Murata,K.,Mitsuoka,K.,Hirai,T.,Walz,T.,Agre,P.,Heymann,
J.B., Engel, A. & Fujiyoshi, Y. (2000) Structural determinants of
water permeation through aquaporin-1. Nature 407, 599–605.
Ó FEBS 2004 Control of aquaglyceroporin Fps1 (Eur. J. Biochem. 271) 779