THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2002 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 277, No. 34, Issue of August 23, pp. 30991–30997, 2002
Printed in U.S.A.
Changes in Glucose Transport and Water Permeability Resulting
from the T310I Pathogenic Mutation in Glut1 Are Consistent with
Two Transport Channels per Monomer*
Received for publication, March 22, 2002, and in revised form, May 16, 2002
Published, JBC Papers in Press, May 24, 2002, DOI 10.1074/jbc.M202763200
Pavel Iserovich‡§, Dong Wang§¶, Li Ma‡储, Hong Yang¶, Felipe A. Zuniga‡, Juan M. Pascual¶,
Kunyan Kuang‡, Darryl C. De Vivo¶, and Jorge Fischbarg‡**‡‡
From the Departments of ‡Ophthalmology, **Physiology and Cellular Biophysics, and ¶Neurology, Columbia University,
New York, New York, 10032
We studied glucose and water passage across wild
type (WT) glucose transporter Glut1 and its T310I pathogenic mutant, expressing them in Xenopus laevis oocytes. We found that the T310I mutation produced a
8-fold decrease in glucose transport (zero-trans influx,
13 ⴞ 2% compared with WT), accompanied by a 2.8-fold
increase in the osmotic water permeability (Pf 280 ⴞ 40%
compared with WT), and no change in the diffusional
water permeability (Pd). The dependence of glucose and
water transports on the amounts of mutant cRNA injected was identical exponential buildups (k ⴝ 19.7 ng),
suggesting that they depend similarly on the quaternary
structure. The Ea values for Pf were 16 ⴞ 0.4 (WT) and
11 ⴞ 1 kcal molⴚ1 (T310I). We report for the first time
that 10 mM D-glucose and L-glucose inhibit Pf by ⬃45% in
the WT but not in the T310I mutant. In addition, 10 mM
maltose reduces Pf (15–20%) in both cases. However, 5
mM L-glucose increased the Pf of T310I, consistent with a
cooperative effect. These experimental observations
and an analysis of our three-dimensional model strongly
suggest the presence of two channels per Glut1 monomer, one of which can be blocked by the mutation T310I.
Glucose transporters (Gluts)1 are members of the major facilitator superfamily of membrane proteins (1) and facilitate
the transport of several monosaccharides. Among them, Glut1
is probably the one most extensively studied partly because of
its relative abundance within the erythrocyte cell membrane.
An extensive kinetic analysis of Glut1-mediated transport in
human erythrocytes is consistent with a simple alternating
conformational model for facilitative transport (2), although
alternate mechanistic models have been proposed previously
(3, 4). In addition, it has been suggested that the transport of
* This work was supported in part by National Institutes of Health,
U. S. Public Health Service Grants EY08918 (to J. F.), NS01698,
NS37949, and RR00645 (to D. C. D.), the Will and the Colleen Giblin
Foundations (to D. C. D.), and Research to Prevent Blindness, Inc. (to
J. F.). The costs of publication of this article were defrayed in part by
the payment of page charges. This article must therefore be hereby
marked “advertisement” in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
§ Both authors contributed equally to this work.
储 Recipient of a Fight for Sight Fellowship.
‡‡ To whom correspondence should be addressed: Dept. of Physiology,
College of Physicians & Surgeons, Columbia University, 630 W. 168th
St., New York, NY 10032. Tel.: 212-305-9092; Fax: 212-305-2461;
E-mail: [email protected].
1
The abbreviations used are: Glut, glucose transporter; DOG, deoxyglucose; WT, wild type; Pf, water osmotic permeability; Pd, water diffusional permeability; pM, plasmid; pCMBS, para-chloro-mercuribenzenesulfonate.
This paper is available on line at http://www.jbc.org
substrates would be accompanied by desolvation-solvation cycles that would store and then recover the associated energies
as Glut1 conformation changes (5). Based upon hydropathy
analysis of the primary sequence, a 12-transmembrane-spanning ␣-helical model for the structure of the Glut1 protein was
first proposed by Mueckler et al. (6). They also recognized that
several helices were amphipathic and therefore capable of
forming a water-accessible pore. This idea received experimental support from demonstrations of moderate water permeability through Gluts (7, 8). A later analysis led to the hypothesis
that perhaps five transmembrane helices can form a channel
that is able to admit hexoses but too small for disaccharides (9,
10). A water-filled transport channel is highlighted in a recent
review (11) in which it is asked of MF proteins: “has nature
hijacked a common structure for an aqueous pore?” An exploration of the hypothesis of a water-filled pore in Glut1 led to a
series of experimental papers (12–15) in which the Mueckler
group used cysteine-scanning mutagenesis to describe the possible roles of helices II, V, VII, and XI in solvent accessibility
and glucose transport. This information was used in suggesting
for Glut1 a scheme for helical packing (6, 13, 14) similar to that
of Lac permease (16, 17). Additional models have been proposed for Gluts. Based on the similarity with an ion channel
and general hints from aquaporin 1, Dwyer (18) has described
a three-dimensional spatial arrangement for Glut3, and we
have recently modeled a three-dimensional structure of Glut1
(19) based on the packing schemes above and have communicated its coordinates (Protein Data Bank code 1JA5 (19)).
Additional information on Glut1 structure can be deduced
from Glut1 pathogenic mutations found in patients with the
Glut1 deficiency syndrome discovered by Dr. D. C. De Vivo and
colleagues (20 –25). In this disease, a Glut1 deficiency disrupts
the glucose transport through the blood-brain barrier, which in
turn can cause serious clinical complications. The presence of
the mutated Glut1 results in reduced cerebrospinal fluid glucose concentrations (hypoglycorrhachia) and reduced erythrocyte glucose transporter activities in the patients (20). The
genetic abnormalities include hemizygosity, splice site mutations, deletions, insertions, and nonsense and missense mutations, as well as genetic transmission as the autosomal dominant trait (20 –25). There are eight known missense Glut1
mutations: S66F (21), G91D (25), R126L (21), R126H (23),
E146K (21), K256V (21), T310I (24), and R333W (21).
In one of the missense mutants, a polar amino acid threonine
at position 310 is replaced by a non-polar one, isoleucine (24),
which has also a larger side chain. The residue Thr-310 in helix
8 is conserved in Glut1 through Glut5 and also across species
(26), suggesting that Thr-310 may play an important role for
30991
30992
Glucose Transport and Water Permeability of T310I Glut1
glucose permeation. In this paper, we concentrate on the functional and structural changes that the mutation induces.
The effects of mutations of Glut1 on its Pf have not been
reported in the literature. We find that in the mutant T310I
glucose transport activity is decreased, and paradoxically, Pf is
increased. We also report that in the C-less mutant the Pf is
essentially unchanged. In addition, we find that D-glucose competes for water passage in the wild type but not the T310I
mutant. Analyzing the data and our Glut1 structure, we offer
an interpretation of these effects based on the existence of two
channels per monomer.
EXPERIMENTAL PROCEDURES
Mutagenesis and cRNA Preparation from cDNA Templates—A
1913-bp Eco47III and HindIII hGLUT1 cDNA fragment derived from
pcDNA3 (Invitrogen) was subcloned into a custom plasmid (pM)
containing fragments of 5⬘- and 3⬘-untranslated regions of Xenopus
␤-globin cDNA. This newly constructed plasmid was named pM-GLUT1
and was confirmed by direct DNA sequencing with the appropriate
primers. The mutation T310I was introduced by PCR. Mutagenic primers were: 5⬘-AGGGCGCAATTGCGGTACCCAGCTTGCTG-3⬘ (forward
1); 5⬘-CCTGTGTATGCCATCATTGGCTCAGGTATCGTCAACACGGCC-3⬘ (forward 2); 5⬘-GAAGGGCCGTGTTGACGATACCTGAGCCAATGATGGCATACAC-AGG-3⬘ (reverse 1); and 5⬘-GCCTGCAACGGCAATGGCAGCTGGACGTGG-3⬘ (reverse 2). PCR was conducted on the
pM-GLUT1 plasmid DNA using the above primers for 35 cycles with
denaturation at 95 C for 30s, annealing at 58 C for 1 min, and elongation at 72 C for 2 min. The right-sized PCR fragment was purified from
the gel and cut with StuI and MunI. The gel-purified StuI and MunI
fragment was then directionally ligated into the StuI and MunI sites of
pM-GLUT1. The presence of the mutation T310I was confirmed by DNA
sequence analysis. Capped, the runoff cRNA transcripts of both the wild
type and the mutant T310I GLUT1 were synthesized from the pMGLUT1 constructs after linearization at a unique NotI site using the
mMESSAGE mMACHINETM kit and T7 RNA polymerase (Ambion,
Austin, TX). The C-less mutant construct we used where all of the six
cysteines existing in the wild type were mutated (133, 207, 347, and 429
into serines and 201 and 421 into glycines (27)) was a gift from Dr. M.
Mueckler. This plasmid was linearized at the XbaI site, and the cRNA
was synthesized using the SP6 RNA polymerase.
Xenopus laevis Oocytes Injection—Oocytes were prepared as described previously (28). Oocytes were removed from the ovaries of X.
laevis (NASCO, Fort Atkinson, WI) and were defolliculated and separated (29). The largest undamaged oocytes (stages V and VI) were
transferred to Barth’s medium ((in mM) 88 NaCl, 2.4 NaHCO3, 1 KCl,
0.33 Ca(NO3)2, 0.41 CaCl2, 0.82 MgSO4, 10 Hepes; osmolarity: 178
mosM, pH 7.4). The oocytes were manually microinjected with 50 nl of
either water or water plus cRNA (1 ␮g/␮l) and were incubated at 18 °C
for 3 days.
Preparation of Purified Oocyte Membranes and Western Blot Analysis—Purified oocyte membranes were prepared according to the procedure of Garcia et al. (30) with minor modifications. After disruption of
oocytes with a Pipetman, the oocyte ghosts were washed free of yolk as
described previously (30). The ghosts were treated with 50 ␮l/oocyte
ice-cold homogenization buffer ((in mM) 83 NaCl, 1 MgCl2, 10 Hepes, 5
EDTA, 5 EGTA, 0.5 phenylmethylsulfonyl fluoride, 5 g/ml each of
aprotinin, pepstatin, and soybean trypsin inhibitor, pH 7.8) in a glass
homogenizer. The total cell homogenate was centrifuged three times at
3000 rpm for 10 min at 4 C in a microcentrifuge to pellet the yolk
granules and melanosomes. The final supernatant was loaded in 4 ml of
15% sucrose and was spun at 165,000 ⫻ g for 80 min to generate the
total membrane fraction. The membrane pellet was suspended in homogenate buffer at 5 ␮l/oocyte and stored at ⫺80 °C.
Ten ␮g of purified plasma membrane samples from oocytes injected
with wild type, T310I, and H2O were fractionated in 4 –20% SDSpolyacrylamide gradient gel (Bio-Rad) and transferred to a nitrocellulose membrane. Western blot analysis was performed using Glut1specific primary antibodies and horseradish peroxidase-conjugated
secondary antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) as
previously described (31). Glut1-specific signals were visualized and
quantified by electrochemiluminescence (ECL) reactions and then
transformed into digital values using a scanner and SigmaScan Pro
software (SPSS Inc., Chicago, IL).
Transporter Activity Assays, Uptake of 2-deoxy-D-[3H]Glucose (Deoxyglucose, DOG)—Groups of six oocytes were placed into 0.6 ml of Barth’s
solution containing 10 ␮Ci of [3H]DOG (5 Ci/mmol, PerkinElmer Life
Sciences) and 2 mM unlabeled DOG in the wells of a 24-well plate. After
incubation for 10 min, the oocytes were washed four times with ice-cold
Barth’s solution containing 0.2 mM phloretin. Individual oocytes were
then placed into vials, and 0.2 ml of 0.2 N NaOH plus 0.2% SDS was
added. After the oocytes were dissolved, 5 ml of scintillation fluid was
placed into each vial for counting. The results were expressed in picomoles of DOG uptake/10 min/oocyte.
Osmotic Water Permeability (Pf ) Measurement—To measure oocyte
volumes as a function of time, we used the method we first outlined in
1990 (7). The oocytes were placed in a 0.3-ml glass-bottom chamber
containing Barth’s medium at room temperature. The temperature of
the perfusing liquid could be set to a range of 5–37 C. The temperature
setting was considered the working temperature given the low thermal
capacity of the plastic chamber. The oocyte was loosely attached to the
chamber bottom using a small amount of Cell-Tak® glue (Collaborative
Research, Bedford, MA (29)). The oocyte equatorial cross-section was
viewed with a NikonTM inverted microscope equipped with a ⫻4 objective and a video camera (model NC-65, Dage-MTI, Michigan City, IN).
A computer with frame grabber digitized the oocyte image and calculated the oocyte area and volume every 6 s. For Pf measurements, the
oocytes were superfused with isotonic Barth’s solution (178 mosM) for a
period of 60 s and then with hypotonic solution (15 mosM, obtained by
omitting NaCl) for another 100 s. The Pf values were calculated from
the induced change in oocyte volume as shown in Equation 1
Pf ⫽ 共dV/dt)䡠[1/(A䡠Vw 䡠 ⌬C)]
(Eq. 1)
where A is the area (assumed spherical) of the oocyte, dV/dt is the rate
of volume change, ⌬C is the osmolarity gradient (all three at zero time),
and Vw is the water molar volume (29).
Diffusional Water Permeability (Pd ) Measurement—We calculated Pd
from the changes in oocyte volume resulting from H2O-D2O exchange.
Initially, the oocytes were equilibrated for 10 min in Barth’s solution
prepared with D2O or “heavy” water (Sigma) as a solvent (heavy waterbased Barth’s solution). After such incubation, the bathing solution was
replaced by Barth’s solution for which the solvent was H2O (waterbased Barth’s solution). In response, the oocytes swelled because of the
10% faster self-diffusion coefficient for H2O compared with D2O. In
other words, under these conditions, H2O diffuses into the oocyte faster
than D2O can leave it. Fig. 3 shows a family of curves obtained from
such experiments.
To determine oocyte membrane diffusional permeability to H2O (Pd),
we note that the extent of the cell volume change observed and the
intracellular H2O concentration are linearly related (32). Therefore,
monitoring the oocyte volume transient yields information on the oocyte
membrane diffusional exchange of D2O for H2O governed in turn by the
oocyte membrane Pd . In mathematical terms, because the oocyte volume is much smaller than that of the external solution, the intracellular concentration of water (Ci ) at time t during the D2O for H2O
exchange can be taken as Equation 2 (8, 33–35)
Ci共t兲 ⫽ Ce 䡠 关1 ⫺ exp(⫺t/␶兲]
(Eq. 2)
where Ce is the extracellular concentration.
If the initial oocyte volume and surface area are Vo and A from
Equation 1, the diffusional permeability Pd is as follows in Equation 3
Pd ⫽ Vo/(␶䡠A)
(Eq. 3)
We found the values of Vo and A as described above, and ␶ was
obtained by fitting the oocyte volume changes during H2O-D2O substitution (Fig. 3) to an exponential buildup curve (Equation 2) using
Origin™ software.
Pf , Pd Determination in the Same Oocyte—A sizable advantage of our
method is that, if desired, it allows one to obtain the values for both Pd and
Pf in the same experiment. This yields an accurate determination of the
Pf /Pd measurements, because both can be obtained from the same oocyte.
An oocyte was incubated in heavy water-based Barth’s solution for ⬃5
min and then was transferred to the chamber described above, which also
contained heavy water-based Barth’s solution. The oocyte was challenged
first with isotonic water-based Barth’s solution to determine the Pd and
after that with hypotonic (15 mosM) water-based Barth’s solution to determine the Pf . Typical curves are shown in Fig. 3.
Modeling—For molecular modeling, we used a Silicon Graphics octane work station with InsightII software (Accelrys, Inc.) as described
previously (19). To explore the cavities in Glut1, we had previously used
the program HOLE (36, 37), developed to explore the pore of ion channels. For the current purpose, as the internal structure of Glut1 ap-
Glucose Transport and Water Permeability of T310I Glut1
30993
FIG. 2. Western blot analysis of the expression of wild type and
T310I mutant Glut1 in X. laevis oocytes membranes. Groups of
three oocytes were incubated for 3 days after injection of 50 ng of cRNA
or water (as a control). Positive controls were purified Glut1 and red
blood cell membranes. The expression rate for the T310I mutant was
85% that of wild type (WT) and was used for normalization for all T310I
data. RBC, red blood cells.
FIG. 1. Effect of the pathogenic T310I and the C-less Glut1
mutants on glucose and water transports. The left ordinate corresponds to zero-trans influx uptake of 2 mM unlabeled DOG and 10 ␮Ci
of ]3H]DOG. The right ordinate corresponds to the Pf. The x axis legends
denote whether cRNA encoding wild type, T310I, and C-less mutant
Glut1 or only water (Control) were injected into oocytes. Values represent mean ⫾ S.E. Results for the T310I mutant were normalized using
the intensity of the Glut1 protein band.
peared more complex than anticipated, we decided to use a combination
of several algorithms, among them the program Swiss Pdb viewer (38).
RESULTS
It has been shown previously (22) that glucose uptake by
erythrocytes is only 35% of normal in a Glut1 deficiency syndrome patient affected with the T310I mutation in one of the
alleles. However, work with erythrocytes does not allow one to
quantify precisely the function of the mutant. We used instead
the X. laevis oocytes expression system.
Sugar Uptake by Wild Type, T310I Mutant, and C-less—Fig.
1 shows the results of zero-trans DOG influx by oocytes expressing one of the proteins above. The unlabeled DOG concentration in the medium was 2 mM. As can be seen, the T310I
mutant transports DOG at a rate of only ⬃13 ⫾ 2% of that of
wild type Glut1. As mentioned below, the expression of the
proteins referenced was assessed with the control experiments
shown in Fig. 2.
We also utilized the C-less Glut1 mutant. In accordance with
the data from Mueckler and Makepeace (27), the uptake of
DOG by the C-less Glut1 mutant (Fig. 1) was 80 ⫾ 9% of that
by wild-type Glut1. The uptake of DOG by water-injected oocytes was ⬍1% than in oocytes expressing wild type Glut1.
Water Permeability of Wild Type and T310I and C-less Mutants—Fig. 1 shows the Pf values determined for oocytes expressing the proteins described. Interestingly, the Pf for the
pathogenic T310I mutant was approximately three times
(280 ⫾ 40%) larger than that for the wild type (after subtraction of the background Pf for water-injected oocytes). As for the
C-less, the Pf determined (Fig. 1) was 69 ⫾ 13% of that of the
wild type (also after the subtraction of the background Pf for
water-injected oocytes). Clearly this mutant, even if all six of
the original cysteines in wild type Glut1 have been replaced,
can transport glucose and water at rates near those obtained
with the wild type (Fig. 3).
Fig. 4 shows the Pf and Pd data we obtained. The Pf values
we determined here are comparable with those we reported
earlier (39) for the Pf of oocytes injected with GLUT1 cRNA
(28 ⫾ 2 here versus 23 ⫾ 5 ␮m s⫺1) and water (control: 15.8 ⫾
1 here versus 13 ⫾ 1 ␮m s⫺1). As for Pd values in oocytes, from
what we know this is the first such report. We found that in
contrast to the increase seen in Pf for the T310I mutant (Fig.
4B), the Pd for this mutant was quite similar to that of the wild
type (Fig. 4B). Consequently, the Pf /Pd ratio (Fig. 4C) was
FIG. 3. Typical curves for determinations of diffusional and
osmotic water permeabilities on same given oocytes. For these
experiments, oocytes expressing Glut1 were preincubated for 10 min in
isotonic Barth’s medium prepared with D2O. We calculated the Pd from
the kinetics of the fast increase in oocyte volume after challenging
oocytes with isotonic Barth’s medium prepared with H2O. After steady
state, Pf was determined from the oocyte volume in response to osmotic
challenge (15 mosM).
FIG. 4. Water passage across Glut1 and the T310I mutant. A and
B, values of Pf and Pd obtained from oocytes expressing either wild type
Glut1 or mutant T310I or just injected with water as a control. To
obtain the Pf /Pd ratios for wild type and T310I mutant shown in C, we
subtracted the corresponding Pf and Pd of water-injected (control) oocytes. The data in A are the same as those in Fig. 1. Values represent
mean ⫾ S.E. Results were normalized using the intensity of the Glut1
protein band.
approximately three times higher for the T310I mutant than
for Glut1. The significance of these findings is considered under
“Discussion.” The Pf /Pd ratio for water-injected oocytes presumably means that endogenous transporters or channels contribute to water passage. There is in fact evidence for the
30994
Glucose Transport and Water Permeability of T310I Glut1
FIG. 5. Dependence of DOG uptake and osmotic water permeability on the amount of T310I Glut1 mutant cRNA injected. A,
the actual experimental values plus curves fitted to the data using
exponential buildup functions: y ⫽ y0 ⫹ A [1⫺exp(x/xc)]. Values represent mean ⫾ S.E. in five oocytes. B, the values were normalized by
subtracting y0 and dividing by the asymptotic maximum for each set.
The exponential constant xc was 19.7 ng.
presence of some endogenous membrane proteins in oocytes
(28). From our data, they could have some water permeability.
Assessments of Expression—We also determined the expression rate of the wild type and mutant Glut1 in oocyte membranes using Western blot analysis with Glut1-specific antibodies. We found that the expressions of wild type Glut1 and
T310I mutant in X. laevis oocytes are very close (Fig. 2); the
calculated expression ratio was used to normalize the Pf , Pd ,
and glucose uptake data obtained with oocytes injected with
T310I mutant RNA. It has also been reported that the C-less
expression by oocytes is comparable with that of the wild type
(40).
It has been shown previously (28) that in oocytes expressing
wild type Glut1, the rate of DOG uptake characteristically
depends on the amount of cRNA injected. The dependence of Pf
on the amount of Glut1 cRNA injected has been also demonstrated (8). In our present case, the relatively high Pf of the
mutant appeared useful for us to try to determine the dependencies of glucose transport and water permeability on the
amount of T310I cRNA injected. Fig. 5A shows that they
change in concert as the amount of the cRNA injected increases
and that their dependence can be approximated to a single
exponential buildup y ⫽ y0 ⫹ A [1 ⫺ exp(x/xc)]. In Fig. 5B, the
values were normalized by subtracting y0 and dividing by the
asymptotic maximal A for each set. The exponential constant xc
was 19.7 ng. After normalization, the same Fig. 5B shows that
DOG uptake and Pf data practically overlap with each other,
displaying the same exponential buildup dependence.
Activation Energy—Fig. 6 shows the temperature dependence of the Pf for the T310I mutant. The data were fit to the
standard expression: Pf ⫽ A exp(⫺⌬E/RT), where ⌬E is the
activation energy, R is the gas constant, and T ⫽ 310 K. The
activation energies for water transport through the wild type
and the T310I mutant were 17.0 ⫾ 2.0 and 13.0 ⫾ 0.5 kcal
mol⫺1, respectively.
Effects of Saccharides on Glut1 Pf —Fig. 7, A and B, shows
the effects of external saccharides on the osmotic water flow
across membranes of oocytes expressing either wild type Glut1
or the T310I mutant. The compounds used were (a) the normal
substrate of Gluts, D-glucose, (b) the non-transported enantiomer, L-glucose, and (c) a disaccharide that inhibits glucose
transport, maltose. Fig. 7A exemplifies the original data ob-
FIG. 6. Temperature dependence of the osmotic permeability.
Background oocyte Pf values were subtracted in each case. Values
represent mean ⫾ S.E. (five oocytes).
FIG. 7. Effect of saccharides on the Pf of oocytes expressing
either wild type or the pathogenic mutant T310I. Oocytes were
preincubated for 10 min with the saccharides indicated. A, effects of
D-glucose. Oocytes were injected with cRNA encoding wild type Glut1 or
T310I mutant and with water only as controls. The numbers inside the
bars denote the D-glucose concentration (mM) used in each case. B,
permeabilities for both wild type and T310I mutant. Base-line permeabilities (for water-injected controls) were subtracted, and data were
normalized with respect to the corresponding permeabilities in the
absence of saccharides. Asterisks, averages significantly different from
controls at a level of 0.05 (one-way ANOVA).
tained. Importantly, after taking into account the base-line Pf
of water-injected oocytes, 10 mM D-glucose inhibits osmotic flow
by ⬃45% in oocytes expressing Glut1. On the other hand,
D-glucose (up to 30 mM) does not significantly affect the Pf of
oocytes expressing the T310I mutant. Fig. 7B further shows
that, interestingly, the inhibitory effects (40 – 45%) on water
permeability in oocytes expressing wild type Glut1 were also
observed with 10 mM (but not with 5 mM) L-glucose and maltose. Crucially, for oocytes expressing the mutant T310I, 10 mM
maltose decreased Pf by 20%, 5 mM L-glucose increased Pf by
Glucose Transport and Water Permeability of T310I Glut1
35%, and 10 mM D- and L-glucose did not significantly affect
water permeation.
DISCUSSION
Movements of Glucose and Water Across Wild Type and Pathogenic T310I Mutant—Our finding of reduced glucose transport (down to 13%, Fig. 1) by the pathogenic mutant was in line
with our expectations. Still, the mutation reduces but does not
abolish glucose transport. Moreover, the Pf of the mutant was
paradoxically increased (Fig. 1) over that of the wild type.
These results lead us to consider what might be the shape of
the putative water pathway(s) in Glut1; whether glucose and
water may share the same pathway(s), and in which manner
could these have changed because of the T310I mutation.
There is already evidence for moderate water passage
through oocyte membranes in connection with Glut expression
as first shown by our group (7) and quickly confirmed by
Verkman’s laboratory (8), and that the rate of water passage in
Glut1 rises in proportion to the amount of Glut1 cRNA injected
in oocytes (8). Here we have ascertained the simultaneous
behavior of glucose and water passage upon the expression of
T310I. As shown in Fig. 5, we determined that the rates of
glucose and water passage rose together exactly in parallel
with the amount of copies of T310I expected to be expressed
(Fig. 5B). This reaffirms the idea that both glucose and water
traverse the same transporter. Furthermore, the dependence
shown in Fig. 5B suggests that both glucose and water transports depend similarly on the oligomeric structure of the functional transporter.
The Pf of the C-less and the Function of the Monomer—DOG
glucose uptake and osmotic water permeability of the C-less
are 80 and 69% of those of the wild type, respectively. In other
words, the function of this mutant is similar to that of the wild
type. From a previous report (41), the formation of Glut1 oligomers would depend on the formation of disulfide bonds between residues 347 and 421. Accordingly, when Glut1 and/or
its mutants are expressed in the X. laevis system, C-less Glut1
mutants would be expressed by oocytes as functional monomers capable of transporting both glucose and water. In addition, the co-expression of Glut isoforms differing in kinetic
properties suggested a functional monomeric form and do not
support oligomerization as being a prerequisite for functional
activity (42). However, to be noted from our Glut1 structure,
cysteines 347 and 421 are in the periphery of the monomer and
face outwards (data not shown), so clearly S–S-bonded dimers
could still form in the wild type and T310I mutant. In addition,
the ability of Glut1 to be functionally active as a monomer does
not exclude the significant cooperativity among subunits as
proposed by Carruthers and colleagues (41).
Number of Channels per Monomer—We consider first the
one-channel alternative. Is a channel wide enough to be described by classical hydrodynamics, or should we use a single
file paradigm? In the first case, an increase in Pf /Pd can be
explained by an increase in the channel radius (43). However,
that would also have led to an increase in Pd, which did not
happen in our case (Fig. 4B). As for single-file water channels,
the Pf /Pd ratio allows one to draw inferences on the number of
water molecules occupying the pore and, hence, on the length of
the pore (35, 44). In this context, a larger Pf /Pd ratio for the
T310I mutant would entail higher occupancy, a longer pore,
and lower Pf . However, this possibility contradicts the experimental finding of a higher Pf in the T310I mutant (Fig. 4A). In
other words, assuming the presence of only one channel leads
to difficulties in explaining the present evidence.
Therefore, we turn to the two-channel alternative. Although
it is not known with certainty how many transport channels
exist across Glut1, there is evidence consistent with the pres-
30995
ence of two channels. A channel for glucose was theorized (6)
and was located between helices 7 and 5 using mutagenesis
data from Mueckler’s laboratory (12, 40, 45). As we detail
below, the locations of the mutagenized residues fit very well in
the boundaries of what we term “main channel” in our Glut1
structure (Ref. 19 and this study). As for a second channel
through Glut1, Brasseur and colleagues (46) presented theoretical arguments in favor of this idea, and Keller and colleagues (15) detected a solvent-accessible exofacial cleft between helices 2 and 7 and suggested that it could lead to a
possible alternative pathway for transported substrates. This
finding is consistent with the presence of the auxiliary channel
we have described in our structure (19) located in that part of
the protein.
Evidence for Two Channels from Water Permeation Across
Wild Type and Pathogenic T310I Mutant Inhibitory Effects of
Saccharides—Interestingly, externally added 10 mM (but not 5
mM) D-glucose, L-glucose, and maltose cause a 40 – 45% inhibition in the Pf of the wild type (Fig. 7). This observation is
apparently the first documented instance of inhibition of water
passage through Glut1 by its substrate, D-glucose and suggests
that glucose and water share the same path(s) through Glut1.
The fact that both a transported sugar (D-glucose) and its
non-transported enantiomer (L-glucose) inhibit water passage
suggests that such inhibition takes place in a channel externally to a selectivity filter.
On the other hand, in the case of the mutant, D-glucose (up to
30 mM) or L-glucose (10 mM) do not affect the Pf value. This
finding is consistent with the existence of a second channel that
would become a preferential pathway for water passage. The
access of the sugars to the main channel might be greatly
reduced or blocked at the 310 site (Fig. 8, right panel), which
would partially account for the lack of sugar-induced inhibition
of Pf . Furthermore, the mutant transports only ⬃10% glucose
that the wild type does. This would be consistent with reduced
sugar penetration into the auxiliary channel and the virtual
lack of water flow inhibition observed.
The effects of 10 mM maltose are revealing. The decrease in
Pf for the case of the wild type is in line with its known effect of
glucose transport blocker and seems to implicate the main
channel. It also decreases the Pf of the mutant by a similar
amount. Maltose would not be expected to enter the transport
channel(s) and would inhibit Pf at its own binding sites at the
openings of both channels. In this connection, there is strong
recent kinetic evidence from Carruthers laboratory for the existence of two sugar import sites (and two maltose binding
sites) for the human erythrocyte Glut1 (47). Remarkably, the
study (47) argues convincingly for two channels per functional
glucose transport unit. Our evidence here leads us to very
similar conclusions with the exception that they are inclined to
equate the functional unit to a dimer, whereas from previous
(42, 48) and present studies given experimental evidence and a
structural analysis of our three-dimensional model, we think
that a monomer can be a functional transport unit and have the
two channels in question when using the X. laevis expression
system.
Lastly, there is a trend toward the stimulatory effects on Pf
by 5 mM saccharides, significant for L-glucose. This finding is
consistent with cooperative effects that might develop given the
presence of two channels.
Activation Energy—As mentioned under “Results”, the activation energies for water transport through wild type and
T310I mutant were 17.0 ⫾ 2.0 and 13.0 ⫾ 0.5 kcal mol⫺1,
respectively. These numbers are of the order of the activation
energies reported for glucose transport into human erythrocytes (14.1 kcal mol⫺1 (49)), for water flow into large unilamel-
30996
Glucose Transport and Water Permeability of T310I Glut1
FIG. 8. Effects of the pathogenic
mutation T310I on the Glut1 structure. The main channel is colored in
green, and the auxiliary channel in blue,
both in space-filling representation. The
yellow structure consists of segments of
both channels plus a narrow connection.
The left panel represents wild type Glut1,
and the right panel represents the pathogenic T310I Glut1 mutant. The channel
coordinates were generated with the program HOLE (36, 37). Channels are cut at
both ends to depict only the interhelical
segments. The residue 310 is in surface
rendering. Thr-310 is in light green, and
the mutated residue Ile-310 in blue. Only
helix 7 is shown (white ribbon).
lar vesicles with reconstituted purified human erythrocyte
Glut1 (10 –13 kcal mol⫺1 (50)), and for the passage of water into
oocytes expressing Glut1 (13 kcal mol⫺1 (8)). Such relatively
large activation energies contrast with the low values associated with water channels (3– 4 kcal mol⫺1 (51, 52)) and indicate
that water passage across Glut1 depends on the conformational
changes of the protein.
As shown in Fig. 6, the activation energy for water transport
in T310I is approximately 5 kcal mol⫺1 smaller than that for
wild type Glut1. This may correspond to an increase in the
rigidity of the mutant with consequent increase in the number
of open states.
Evidence from Structural Analysis: Update on the Glut1
Channels—As mentioned above, we have recently modeled a
three-dimensional structure for Glut1 (19), which accounts for
existing data. We have also drawn from it new insights on
transports pathways through the protein and possible effects of
pathogenic mutations. Our work on the structure continues as
we incorporate new experimental evidence. In this regard, in a
recent paper by Mueckler and colleagues (54), the authors
mention that two residues in helix 5 (Gln-161 and Val-165) of
our structure do not seem to fit well with the experimental data
observed (40) as pCMBS blocked the single Cys mutants at
those sites, suggesting that they lie next to the transport pathway. We have examined this issue. It would seem that the
program HOLE (36, 37) that we had been using to determine
the coordinates of cavities in the Glut1 structure is optimal for
gramicidin-type simple pores, but the more complex cavities in
Glut1 would require additional procedures. Therefore, we are
now using the Swiss Pdb viewer (38) to provide the initial
coordinates to be explored further with HOLE. We were
pleased to ascertain that the main transport channel that we
had located near the exofacial end of helix 7 actually extended
all the way through the protein with an opening on the endofacial side. With this extension, helix 5 is in the vicinity of the
main channel (data not shown), which immediately gives a
structural basis for the inhibitory effects on Q161C and V165C
crucially pointed out (54). We also reexamined which side of
helix 5 faces the main channel using the mutational data
available for this amphipathic helix (40). We verified that with
the angular position we chose for helix 5 (19), it has most of its
pCMBS-accessible residues facing that channel (data not
shown) from which our coordinates seem to require no change.
We originally described two channels termed main and auxiliary (19). However, as anticipated above, we have found that
the main channel extends in the endofacial direction among
helices 7, 5, 8, and 10 (Fig. 9, green color) all through the
FIG. 9. Transport channels in the Glut1 structure and the positions of seven known pathogenic mutants. For clarity, we only
show the mutated residues in red surface rendering. The channels are
shown in space-filling representation as in Fig. 8. Only helix 7 is shown
(white ribbon).
protein, including the endofacial loop region. This finding is
consistent with this channel serving to transport glucose. As
for the second or auxiliary channel, we have found a prolongation that traverses the region of endofacial loops. Together with
a short connecting piece described previously, these two channels are now seen forming a H-like structure (Fig. 9). They
have separate ends on both the exofacial and endofacial sides
and are connected together in the middle around helix 7 (Fig. 9,
yellow color). In the protein conformational state represented
by our model, the main channel is open on both sides, whereas
the auxiliary channel is open endofacially but closed exofacially
(Fig. 9, blue color). Importantly, as explained below, it seems
that the mutations in the vicinity of the auxiliary channel
disable glucose transport (Fig. 9).
Glucose Transport and Water Permeability of T310I Glut1
Evidence from Structural Analysis: Changes Attributed to the
T310I Pathogenic Mutation—In our model, the threonine residue at position 310 lines the main channel (Fig. 8, left). Hence,
the replacement of Thr (hydrophilic) by Ile (hydrophobic) plus
the fact that the Ile side chain is ⬃2.3 Å longer than that of Thr
could provide a straightforward explanation for the drastic
decrease in glucose transport function shown experimentally.
To investigate these possible effects, we substituted Ile for
Thr-310 in our structure and ran the molecular dynamics for 50
ps. We examined the main and auxiliary channels of the mutant at the end of the simulation. The main channel was clearly
blocked by the isoleucine residue, which protrudes to the middle of the putative glucose pathway (Fig. 9, right), consistent
with the deficit in glucose transport observed. In addition, the
auxiliary channel happened to be open on its exofacial end (Fig.
9), which supports its possible role for water passage, and is
consistent with the elevated water permeation through the
mutant.
This information on the arrangement of the channels after
molecular dynamic simulation gives insight into the consequences of the mutation. It seems that the simple introduction
of Ile-310 results in the mutant reaching a stable conformation
accompanied by a lower mobility (data not shown) plus an
important steric blockage of a glucose pathway. The mobility
loss in itself would make it difficult for the mutant to undergo
the conformation changes necessary for glucose transport, and
the steric blockage would make the transport deficit even more
pronounced. In fact, it may be surprising that any transport at
all does remain under the circumstances.
Structural Analysis: the Locations of Pathogenic Mutants—
Interestingly, these mutants are not distributed randomly in
the structure but are very close to well defined regions (Fig. 9).
Three of them are on the exofacial side: S66F (in the loop
between helices 1 and 2), T310I (in helix 8), and R126L or
R126H (in helix 4). Of these, R126L or R126H is on the auxiliary channel, S66F is close to it, and T310I is on the main
channel. The other pathogenic mutants are located on the
endofacial side: G91D (loop between helices 2 and 3), E146K
(loop between helices 4 and 5), K256V (loop between helices 6
and 7), and R333W (loop between helices 8 and 9). They seem
to be distributed in pairs. G91D and E146K are close to each
other and to the auxiliary channel near the endofacial side of
the protein. Interestingly, this is close to the cytochalasin B
binding site (55). It is possible that these mutations could act
on Glut1 flexibility, perhaps mimicking in some way the inhibitory effect that cytochalasin B has. Finally, K256V and R333W
are also located close to each other and to the putative main
channel exit where they could obviously affect glucose
transport.
CONCLUSION
It seems possible to offer consistent explanations for the
complex effects that the mutation T310I causes in the functional properties of Glut1 with the help of our three-dimensional model for Glut1. An examination of water passage gives
useful clues for the study of transport of substrates through
this protein. Both our experimental evidence and an analysis of
our Glut1 structure agree in suggesting the presence of two
channels per monomer. They also indicate that the obstruction
of glucose flux and increased rigidity of the T310I mutant are
probably responsible for its pathogenicity. The changes that
this mutation causes in the dynamic behavior of Glut1 will be
detailed separately.
Acknowledgment—We thank Dr. M. Mueckler for the generous gifts
of purified Glut1 and C-less plasmid.
30997
REFERENCES
1. Saier, M. H., Beatty, J. T., Goffeau, A., Harley, K. T., Heijne, W. H., Huang,
S. C., Jack, D. L., Jahn, P. S., Lew, K., Liu, J., Pao, S. S., Paulsen, I. T.,
Tseng, T. T., and Virk, P. S. (1999) J. Mol. Microbiol. Biotechnol. 1, 257–279
2. Carruthers, A. (1990) Physiol. Rev. 70, 1135–1175
3. Carruthers, A. (1991) Biochemistry 30, 3898 –3906
4. Baldwin, S. A. (1993) Biochim. Biophys. Acta 1154, 17– 49
5. Widdas, W. F., and Baker, G. F. (1991) Cytobios 66, 179 –204
6. Mueckler, M., Caruso, C., Baldwin, S. A., Panico, M., Blench, I., Morris, H. R.,
Allard, W. J., Lienhard, G. E., and Lodish, H. F. (1985) Science 229,
941–945
7. Fischbarg, J., Kuang, K. Y., Vera, J. C., Arant, S., Silverstein, S. C., Loike, J.,
and Rosen, O. M. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 3244 –3247
8. Zhang, R., Alper, S. L., Thorens, B., and Verkman, A. S. (1991) J. Clin. Invest.
88, 1553–1558
9. Zeng, H., Parthasarathy, R., Rampal, A. L., and Jung, C. Y. (1996) Biophys. J.
70, 14 –21
10. Deleted in proof
11. Barrett, M. P., Walmsley, A. R., and Gould, G. W. (1999) Curr. Opin. Cell Biol.
11, 496 –502
12. Hruz, P. W., and Mueckler, M. M. (1999) J. Biol. Chem. 274, 36176 –36180
13. Hruz, P. W., and Mueckler, M. M. (2000) Biochemistry 39, 9367–9372
14. Olsowski, A., Monden, I., and Keller, K. (1998) Biochemistry 37, 10738 –10745
15. Olsowski, A., Monden, I., Krause, G., and Keller, K. (2000) Biochemistry 39,
2469 –2474
16. Frillingos, S., Sahin-Toth, M., Wu, J., and Kaback, H. R. (1998) FASEB J. 12,
1281–1299
17. Venkatesan, P., Liu, Z., Hu, Y., and Kaback, H. R. (2000) Biochemistry 39,
10649 –10655
18. Dwyer, D. S. (2001) Proteins 42, 531–541
19. Zuniga, F. A., Shi, G., Haller, J. F., Rubashkin, A., Flynn, D. R., Iserovich, P.,
and Fischbarg, J. (2001) J. Biol. Chem. 276, 44970 – 44975
20. De Vivo, D. C., Trifiletti, R. R., Jacobson, R. I., Ronen, G. M., Behmand, R. A.,
and Harik, S. I. (1991) N. Engl. J. Med. 325, 703–709
21. Wang, D., Kranz-Eble, P., and De Vivo, D. C. (2000) Hum. Mutat. 16, 224 –231
22. Klepper, J., Fischbarg, J., Vera, J. C., Wang, D., and De Vivo, D. C. (1999)
Pediatr. Res. 46, 677– 683
23. Brockmann, K., Wang, D., Korenke, C. G., von Moers, A., Ho, Y. Y., Pascual,
J. M., Kuang, K., Yang, H., Ma, L., Kranz-Eble, P., Fischbarg, J., Hanefeld,
F., and De Vivo, D. C. (2001) Ann. Neurol. 50, 476 – 485
24. Klepper, J., Wang, D., Fischbarg, J., Vera, J. C., Jarjour, I. T., O’Driscoll, K. R.,
and De Vivo, D. C. (1999) Neurochem. Res. 24, 587–594
25. Klepper, J., Willemsen, M., Verrips, A., Guertsen, E., Herrmann, R., Kutzick,
C., Florcken, A., and Voit, T. (2001) Hum. Mol. Genet. 10, 63– 68
26. Bell, G. I., Kayano, T., Buse, J. B., Burant, C. F., Takeda, J., Lin, D.,
Fukumoto, H., and Seino, S. (1990) Diabetes Care 13, 198 –208
27. Mueckler, M., and Makepeace, C. (1997) J. Biol. Chem. 272, 30141–30146
28. Vera, J. C., and Rosen, O. M. (1989) Mol. Cell. Biol. 9, 4187– 4195
29. Echevarria, M., Kuang, K., Iserovich, P., Li, J., Preston, G. M., Agre, P., and
Fischbarg, J. (1993) Am. J. Physiol. 265, C1349 –C1355
30. Garcia, J. C., Strube, M., Leingang, K., Keller, K., and Mueckler, M. M. (1992)
J. Biol. Chem. 267, 7770 –7776
31. Seidner, G., Alvarez, M. G., Yeh, J. I., O’Driscoll, K. R., Klepper, J., Stump,
T. S., Wang, D., Spinner, N. B., Birnbaum, M. J., and De Vivo, D. C. (1998)
Nat. Genet. 18, 188 –191
32. Kitching, J. P., and J. E. (1959) Exp. Biol. 37, 73– 82
33. Lawaczeck, R. (1984) Biophys. J. 45, 491– 494
34. Macey, R. I. (1984) Am. J. Physiol. 246, C195–C203
35. Ye, R. G., and Verkman, A. S. (1989) Biochemistry 28, 824 – 829
36. Smart, O. S., Goodfellow, J. M., and Wallace, B. A. (1993) Biophys. J. 65,
2455–2460
37. Smart, O. S., Neduvelil, J. G., Wang, X., Wallace, B. A., and Sansom, M. S.
(1996) J. Mol. Graph. 14, 354 –360, 376
38. Guex, N., and Peitsch, M. C. (1997) Electrophoresis 18, 2714 –2723
39. Loike, J. D., Cao, L., Kuang, K., Vera, J. C., Silverstein, S. C., and Fischbarg,
J. (1993) J. Gen. Physiol. 102, 897–906
40. Mueckler, M., and Makepeace, C. (1999) J. Biol. Chem. 274, 10923–10926
41. Zottola, R. J., Cloherty, E. K., Coderre, P. E., Hansen, A., Hebert, D. N., and
Carruthers, A. (1995) Biochemistry 34, 9734 –9747
42. Burant, C. F., and Bell, G. I. (1992) Biochemistry 31, 10414 –10420
43. Renkin, E. M. (1954) J. Gen. Physiol. 38, 225–243
44. Finkelstein, A. (1987) in Distinguished Lecture series of the Society of General
Physiologists, Vol. 4, pp. 32– 41, John Wiley & Sons, Inc., New York
45. Seatter, M. J., De la Rue, S. A., Porter, L. M., and Gould, G. W. (1998)
Biochemistry 37, 1322–1326
46. Ducarme, P., Rahman, M., Lins, L., and Brasseur, R. (1996) J. Mol. Mod.
(Online) 2, 27– 45
47. Hamill, S., Cloherty, E. K., and Carruthers, A. (1999) Biochemistry 38,
16974 –16983
48. Wellner, M., Monden, I., and Keller, K. (1994) Biochem. J. 299, 813– 817
49. Hu, X. J., Peng, F., Zhou, H. Q., Zhang, Z. H., Cheng, W. Y., and Feng, H. F.
(2000) Biochim. Biophys. Acta 1466, 306 –314
50. Zeidel, M. L., Albalak, A., Grossman, E., and Carruthers, A. (1992) Biochemistry 31, 589 –596
51. Eggena, P. (1983) Am. J. Physiol. 244, C44 –C49
52. Zeidel, M. L., Ambudkar, S. V., Smith, B. L., and Agre, P. (1992) Biochemistry
31, 7436 –7440
53. Deleted in proof
54. Mueckler, M., and Makepeace, C. (2002) J. Biol. Chem. 277, 3498 –3503
55. Holman, G. D., and Rees, W. D. (1987) Biochim. Biophys. Acta 897, 395– 405