Structure
of the
sucrose-specific
porin
ScrY
from
Salmonella
typhimurium
and
its
complex
with
sucrose
Doris
Forst',
Wolfram
Welte
2
,
Thomas
Wacker'
and
Kay
Diederichs
2
The
X-ray structure of a
sucrose-specific
porin
(Scry)
from
Salmonella typhimurium
has
been
determined
by
multiple
isomorphous
replacement at 2.4 A resolution both
in
its
uncomplexed form
and
with
bound
sucrose.
ScrY
is
a noncrystallographic trimer of identical
subunits,
each
with
413
structurally well-defined amino
acids.
A
monomer
is
built
up
of
18
anti-parallel
~-strands
surrounding a hydrophilic pore, with a topology
closely
similar
to that of maltoporin.
Two
non-overlapping
sucrose-binding
sites
were identified
in
difference
Fourier
maps.
The
higher permeability for
sucrose
of
ScrY
as
compared
to maltoporin
is
mainly
accounted
for
by
differences
in
their
pore-lining
residues.
Gram-negative bacteria protect their vulnerable cytoplasmic
membrane
by a peptidoglycan layer
and
an
outer
membrane
(OM)I.2.
It
has been estimated
that
-10%
of
all genes
in
E.
coli
are involved
in
transport
of
small
nutrient
molecules across the
OM,
the
periplasmic space with
the
peptidoglycan
and
the cyto-
plasmic
membrane
(CM)3. Sugar-specific, active
transport
sys-
tems in
the
CM with
Km
values
in
the
micromolar range
maintain
a low concentration
of
monosaccharides, disaccha-
rides
and
unbranched
oligosaccharides
in
the
periplasm
4
•
5
.
In
contrast, sugar
transport
across
the
OM
is purely passive, driven
by
diffusion. According
to
Fick's equation
the
flux
of
a species
of
molecules accross
the
OM
is
proportional
to
the
concentration
difference between the external space
and
the
periplasmic space.
The
high value
of
the proportionality constant,
the
permeabili-
ty,
is
due
to
the porins, the
major
protein
component
of
the
OM. So-called 'general diffusion porins'6.7
or
'nonspecific
porins'
form water-filled channels
through
which ions
and
polar molecules, smaller
than
the exclusion limit
of
the channel,
may
pass. Nonspecific
porins
are between 30,000-50,000 Mr
and
form trimers.
The
ionic permeability
of
general diffusion
porins
can
be
measured as
current
flowing
through
porins
reconstituted
into
black lipid
membranes
6
or
through
small
OM
fragments
in
patch-clamp experiments
8
. Plots
of
current
against
ion
concentration are linear, showing
that
ions pass
through
the
OM
in
a diffusion-like process.
The
linear relation
of
current
and
ion
concentration
and
the
differences
in
the
permeability
of
anions
and
cations can
be
rationalized qualitatively by theoreti-
cal
models
9
.
The structures
of
five
general diffusion porins have been deter-
mined
by
X-ray
crystallographyl()",13.
These structures are varia-
tions
of
a 16-stranded antiparallel
~-barrel
with nearest-neighbor
connections between strands. The connections form loops (L)
of
varying length
on
the
external side
and
mostly hair-pin
turns
(T)
on
the periplasmic side
of
the
OM.
The
third
external loop (L3)
folds along the
inner
barrel wall
and
constricts its
lumen
in the
center
of
the
membrane, giving
the
channel
an
hourglass-like
cross section
that
allows for a high,
but
charge-dependent per-
meabilityl4.
The permeation
of
solute molecules through the channel can be
treated with reaction-rate theories, either transition state theoryl5
or
Kramers theoryl6. The latter appears to be more appropriate to
aqueous protein-substrate systems. The rate coefficient for perme-
ation contains a Boltzmann factor exp[
-~G
'
/RTl
with the activa-
tion free-energy
~G'
=
~H'-T~S'
of
the entry process (formula
(14)
in
ref. 16). The entropy term
-T~S'
will represent a conside-
rable fraction
of
the free-energy barrier,
as
the permeating mole-
cules are confined to the narrow cross section
of
the channel.
Under growth-limitation conditions the cells must strive to
increase the permeability further to yield a sufficient flux
of
nutri-
ent
molecules at shallow concentration gradients. Increasing the
density
of
nonspecific porins in the
OM
and increasing the cross
section
of
the
porin
channel are inappropriate means because the
OM
is
already densely packed with porinsl7 and because an
increased exclusion limit would weaken the protective effect
of
the
OM. Another means to increase the permeablity
is
to decrease the
free energy barrier for entrance into the channel
so
that the rate
of
entrance
is
increased. This can be effected by introducing binding
sites specific for certain nutrient molecules, such as sugars
or
nude-
osides, since the negative contribution
of
the enthalpy
of
one
or
several binding sites will decrease
~G·.
Binding sites arranged along
the inner channel wall thus
will increase the permeability com-
pared to the case
of
a nonspecific channel, as long
as
the concen-
trations
of
the substrate molecules are well below their dissociation
constants (K
d
).
Indeed, porins specific for different small nutrient molecules are
expressed by Gram-negative bacteria under conditions
of
growth
limitation.
One
example
is
maltoporin (also called
lamB
due to its
role
as
a receptor for phage
A)
which
is
specific for maltose and
maltooligosaccharides. It
is
coded
on
the mal regulon together with
a maltose transport complex
of
the
ATP
binding cassette (ABC)
type and a periplasmic binding protein
l8
. Maltoporin
is
expressed
at disaccharide concentrations lower than
-100
JlMl9,
while the KdS
of
the binding sites for disaccharides
of
both glycoporins are in the
10
mM range. The
KdS
drop for oligosaccharides with
3,
4 and 5
glucosyl units
but
remain constant at a value
of
-300
JlM20
for
longer ones.
llnstitut
fOr
Biophysik
und
Strahlenbiologie,
Albert-Ludwigs-Universitat,
Albertstr.
23,
79104
Freiburg
im
Breisgau,
Germany.
l
Fakultat
fOr
Biologie,
Universitat
Konstanz,
Box
M656,
78457
Konstanz,
Germany
.
Correspondence
should
be
addressed
to
WW
email:
Wolfram.Welte@uni-konstanz.de
37
Ersch. in: Nature Structural Biology ; 5 (1998), 1. - S. 37-46
http://dx.doi.org/10.1038/nsb0198-37
Konstanzer Online-Publikations-System (KOPS)
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-210817
The
structure
of
maltoporin
with
and
without
bound
mal-
tooligosaccharides
and
also
with
sucrose has
been
determined
by
X-ray crystallograph
y
21-24.
Maltoporin
possesses
the
same overall
structure
as
the
general diffusion
porins,
except
that
its
barrel
contains 18 strands.
As
expected
from
binding
experiments
20
,
maltoporin
possesses a
chain
of
3-5
low affinity
binding
sites for
glucosyl
groups
extending along
the
wall
of
the
channel.
Schmidt
et
al.
25
have discovered a sucrose uptake system
on
plasmid
pUR400
in
Salmonella
typhimurium
and
Escherichia
coli
that
confers
to
the
cells,
under
growth
limitation
conditions,
the
ability
to
grow
on
sucrose as a sole
carbon
source.
The
scr
regu-
Ion codes for a sucrose-specific
phosphoenoltransferase-de-
pendent
transport
(PTS)
complex
in
the
cytoplasmic
Fig. 1
Ribbon
superposition of
ScrY
(yellow)
and maltoporin
(blue)
with
the periplasmic
side
and the external side at the bottom and top,
respec-
tively.
The
barrel
consists
of 18 antipara"el strands and
has
a
kidney-
shaped crossection of between
30
A and
50
A diameter.
The
height
varies
between
20
A near the trimer
axis
and
50
A at the trimer periphery.
The
tilt angles of the
~-strands
relative to the trimer
axis
is
between
35"
and
55",
and the barrel shear number
73
amounts to
+22.
This
figure
was
pre-
pared with
RIBBONS74.
membrane.
It
was
later
found
that
one
gene,
scrY,
codes for a
porin
that
is specific for sucrose
and
maltooligosaccharides
and
that
the
scr
regulon
is also
found
in
the
genome
of
Klebsiella
pneumoniae
26
,27.
N-terminal
sequencing
28
showed
that
mature
ScrYand
malto-
porin
possess 483
and
421 residues respectively.
Upon
compa-
ring
sequences
it
was
found
that
a
C-terminal
ScrY
domain
of
411 residues
can
be
aligned
with
the
Maltoporin
sequence
with
an
overall
identity
of
around
20%26,27.
The
N-terminal
72
residues
of
ScrY have
no
equivalent
in
maltoporin.
We have crystallized
ScrY29
and
solved
by
multiple
isomor-
phous
replacement (m.i.r.)
the
structure
of
the
pore
domain
(residues
71-483)
at
2.4 A resolution
both
with
and
without
bound
sucrose.
Crystallographic problems and solutions
Three
presumably
interrelated
problems
rendered
the
struc-
ture
determination
of
ScrY difficult. First,
the
space
group
reported
for
the
low-resolution
data
29
of
the
preliminary
crys-
tallographic
investigation
proved
irreproducible.
At
resolution
of
4 A
and
beyond,
additional
reflections
(-h
+ k + I
"1=
3n)
were
found
between
the
reflections
of
the
obverse-indexed
hexago-
nal-rhombohedric
lattice
(-h
+ k + 1 =
3n),
turning
the
lattice
into
a
hexagonal-primitive
one.
These
reflections were
of
neg-
ligible
intensity
at
low
resolution
(6 A),
but
were
of
the
same
average
intensity
as
those
of
the
R3 lattice
at
resolution
better
than
3
A,
overall
increasing
the
proportion
of
weak reflections.
This
finding
can
be
explained
by
a
slight
distortion
of
the
R3
space
group,
leading
to
the
P3(l,2)
subgroup
with
a
trimer
in
the
asymmetric
unit,
whose
non-crystallographic
symmetry
axis is close to,
but
does
not
coincide
with,
the
rotation
axis
of
the
R3 system.
The
three-fold
rotational
symmetry
of
the
Fig. 2
View
of the membrane spanning surfaces of the glycoporin trimers. a,
ScrY;
b, maltoporin. Aromatic residues are added to the ribbon repre-
sentation of both porins.
The
two girdles of aromatic residues that bound the hydrophobic zone of
30
A width are discernible.
In
ScrY
they are
less
regular and some aromatic residues are within the hydrophobic zone.
Moreover,
maltoporin differs
from
ScrY
by
an additional clustering of aromat-
ic
residues
in
the external
loops,
This
figure
was
prepared with
Molscript
75
and
Raster3D'6.
3B
70
80
90
100
110
120 130 140 150 160
Fig. 3
Structural
alignment
between
ScrY
and
E.
coli
maltoporin.
Capital
letters denote
struc-
turally
equivalent
residues
based
on
a 3.8 A
cut-
off,
whereas
residues
without a
structural
counterpart
are
shown
in
lower
case
letters.
In
addition
to the
18
strands
of
the
main
/3-barrel,
short
pieces
of
regular
secondary
structure
(E
/3-
strand;
G,
3
w
helix)
as
classified
by
DSSp77
are
also
given.
SerY
sgFJilFHGYARSGVIMNDagastksgayITPAGetgGAIGRLGNQAOTYVEMNLEHKQtLDNGATTRFKVMVADGQTSYNDwrAstSDLNVRQA
11111111
II I
1111
II I I I II
III
I I I
Lamb - - VDFHGYARSGIGW'I'Gsggeqq- -
cfQ'l'TGA-
- - QSKYRLGNECEnAELKLGOEVWKEGDKSFYFDTNVAYSVAQQNDWEA- -TDPAFREA
10
20
30
40
50
60
70
80
~l__
00=
__
P'__
~3_
GOO
_p.
170 180 190
200 210 220
230
240
ScrY
FVELGNLpt
fagpFKGSTLWAGKRFDRDNFDIHWIDSDVVFLAGTGGGIYDVKWNclGLRSNFSLYGRNFGdid- - - - - - - -
-dssNSVQMYILTM
I II
111111111
I I II I I I I I I
Lamb
NVQGKNLiew-
- - LPGSTIWAGKRFY -QRHDVHMIDF'lYWDISGPGAGLENIDVG-
FGlU.SLAATRSSEAggsssfasnniydytNBTANDVPDV
90
100 110 120 130 140 150 160 170
GGO
_135_
BEE
EEBE
_~6
__
~7__
GGG_p,_
250
260
270
280 290
300
310 320
SerY
NHFa-
- - - - -gPLOMMVSGLRAKc1nderkdsngnlakgdAANTGVHALLGLHNDSfyglrdGSSKTALLYGHGLGABvkgigsdga- - - - - - -
--
Description
of
the
structure
I I I II I I I I I I I
Lamb
RLAqmeinpggTLELGVDYGRANlrdnyr-
- - - -
-1
vdgASKDGWLFTABHTQSV1- - - -
kGFNKFVVQYATDSMTsqgkglsqgsgvafdnekf
The model
of
the C-terminal413 residues
of
ScrY
represents the complete trans-
membrane domain.
As
in all porins whose
structures have been analyzed
so
far,
ScrY
is
basically a
~-barrel
with nearest neigh-
bor
connections between strands and with
large loops
(L)
of
irregular length exposed
to the external surface.
L3
is
attached by
polar and apolar interactions to the inner
barrel wall.
It
constricts the channel cross
section near the center to 8 x
11
A
so
that
the channel lumen possesses an hour-glass
180 190
200
210
220
230
240
250
_P9_
ClGG
_P10__
__Pll_GOO
330
340
350
3:60
370 380
390
400 410
420
SerY
• _ -lrpGADTWRIASYGTTPLSENWSVAPAMLAQRSKDryaOOPSYQWATFNLRLIQAINQNF.ALAYEGSYQYMDLKpegyndrqaVNGSFYKLT
II
I II I I I I I
II
II I
Lamb ayninnNGHMLRILDHGAISHODNWDMMYVOMYQDINWd- -NDNGTKWWTVGIRPMYKWTPIMSTVMEIGYDNVESQrt - - - - -
-gDKNNQYKIT
260
270
280 290
300
310
20
330
340
_Pl'__
__P13__
__P14__
__P15__
GOO
__
P16_
430
440 450 460
470
480
SerY
FAPTFKVGSIgDFFSRPEIRFYTSWMDWSkklnnyasdda-
- - - - - - - -
_.
- - - -lgsdgt'nsOOBWSFGVQMETWF
I
III
II I I I II I II II
III
I
Lamb
LAOQWQAGDS
- -
IWSRPAIRVFllTYA.KWDekwmrdytgnadnnanfgkavpadfnggsfgrgdSDBWTFGAQMBIWW
350
3:60
370 380
390
400 410 420
_P17_
GGGGG
_Pl'_
hexagonal primitive lattice was well fulfilled, and the absence
of
in ten sites
of
the
(0,0,1)
reflections with
1"*
3n confirmed the
space group
P3
0
,2)'
Second,
we
found the
ScrY
crystals to be highly anisomor-
phous, with crystals even from the same crystallization setup
exhibiting variations
of
unit cell axes in the order
of
2-3%,
leading to difficulties in interpretation
of
difference Patterson
and difference Fourier syntheses. R-factors between native
crystals ranged mostly between
20-30%. Therefore, complete
data sets had to be collected on single crystals.
Third,
ScrY
crystals were twinned, with twinning fraction a
evenly distributed in the range 0-0.5 . Procedures
as
described
in the Methods section were used to determine the twinning
fraction according to equation 1 and for detwinning the data
according to equation
2.
Despite these difficulties initial single isomorphous replace-
ment phases were obtained by solving the difference Patterson
map
of
the cocrystallized
KAu(
CN)z-derivative. Interpretation
of
other derivatives followed (Table 1). It was found that
detwinning provided cleaner difference Patterson functions
and was necessary for their interpretation. However, for heavy-
atom refinement and phasing the observed (and not the
detwinned) amplitudes were used,
as
detwinning produces
datasets
of
lower completeness, and
we
were able to identify
derivative datasets with low twinning fractions. M.i.r phasing
allowed us to obtain an interpretable density map at 2.9
A reso-
lution, into which residues 71-483
of
the mature sequence
could be built in several macrocycles
of
X-PLOR refinement
and manual improvements
of
the model, using
O.
After completing the final X-PLOR refinements,
SHELXL
refinements
of
the native and sucrose-complexed models con-
firmed the twinning fraction as determined by equation 1 and
showed that inclusion
of
ex
as
a refinable parameter (as opposed
to no twinning treatment) decreased
both
R
free
and
Rcryst
by only
0.12% and 0.20% respectively. However, results
of
these refine-
ments were discarded since - in addition to requiring almost
prohibitive computing resources - it
was
not possible to
restrain the positions
of
NCS-related water molecules in a sim-
ple
way.
Therefore, final statistics (Table 2) except for twinning
fractions are based
on
X-PLOR refinement.
shaped longitudinal cross section with a large external and a nar-
rower periplasmic vestibule.
Like
other porins,
ScrY
forms stable trimers in the outer
membrane. In contrast to the structures
of
the nonspecific
porins, which all consist
of
16
~-strands,
the two glycoporins
ScrY
and maltoporin21 possess
18
strands. The observed barrel
structure has a kidney-shaped cross section with dimensions
of
roughly
50
A x
28
A and a height
of
between
25
A near the
trimer center and
50
A at the trimer periphery.
When the MIR structure analysis
of
ScrY
was begun, the low
sequence similarity with maltoporin led us to expect consider-
able structural differences between the two porins. However, the
two independently solved structures show a surprisingly large
degree
of
structural conservation (Fig. 1). The gross differences
between both glycoporins are in the external loops which tend to
tilt over the pore vestibule
in
both porins like petals and thereby
form a constriction at the entry to the common vestibule
of
all
three monomers. The residues
of
the phage-A binding site
of
maltoporin on
L9, L4,
L6
are not conserved in
ScrY.
L9
contains a
short
3
w
helix and thereafter a Ca
2
+-binding loop segment. In
the latter, the ion
is
bound in a distorted tetragonal-bipyramidal
geometry. The longest loop
11, which consists
of29
amino acids
and contains a short 3
1O
-helix, folds down towards the constric-
tion site and
is
involved in hydrogen bonds with
L3.
These
hydrogen bonds stabilize the structure
of
both loops. Generally,
ScrY
is
less 'closed' than maltoporin
as
L4, L6,
L8
and
L9
are
shorter. The similarity
of
the backbones
of
both glycoporins
is
quite high in view
of
the modest sequence homology: using a
3.8
A cutoff,
315
of
the
413
Ca
atoms ofScrY can be superimposed
with those
of
maJtoporin with a root-mean-square (r.m.s.) devi-
ation
of
only
1.3
A.
Among the periplasmic turns (T),
T5
is
larger in
ScrY
and pro-
trudes in a radial direction from the barrel, whereas
T4
is
markedly shorter in
ScrY.
T5
is
involved in the crystal-contacts
and
T8
folds into the channel mouth.
The monomer-monomer interface
Trimer formation
is
stabilized through many polar interactions
between monomers at the top and bottom
of
the
~-barrel.
In
between, there exists an apolar region formed by small
39
b
hydrophobic amino acids. The low average B-factor
of
around
18
A
2
indicates the rigidity
of
this region.
On
the periplasmic side
of
the trimer center, near the
Nand
C termini, a cluster
of
six
phenylalanines, formed by residues
73
and
75
of
each monomer,
is
found. Here, the aromatic rings from adjacent monomer inter-
faces
are mutually perpendicular to each other, optimizing
quadrupole interactions
30
and connecting adjacent monomers.
In
ScrY
and all
E.
coli
porins, loop
L2
fills
a gap above
L3
of
a
neighboring monomer!!,2!. This loop contributes considerably
to the stability
of
the trimer because it connects adjacent porins
in a manner similar to an outward-bent hook.
The
membrane
exposed
surface
The peripheral transmembrane surface
of
the
ScrY
porin trimer
can be divided into three zones that encircle the trimer: a
hydrophobic zone with a width
of
up to
33
A composed
of
aliphatic residues (such
as
Ala, Leu,
He
and
Val)
and two
hydrophilic zones at the periplasmic and external surfaces.
Two
aromatic girdles form borderlines between the hydrophobic
zone and the two hydrophilic zones (Fig. 2a). A neutron diffrac-
tion analysis
3
! showed that the hydrocarbon core
of
detergent
micelles just covers the hydrophobic zone
of
OmpF porin, which
is
thus identified
as
the contact interface with the lipid chains in
the outer membrane. The accumulation
of
aromatic amino acids
at the polar/apolar border
is
a common feature
of
all known
structures
of
porins and
is
also seen,
but
is
less obvious
in
other
membrane
proteins
32
•
These girdles are thought to seal the
porins into the outer membrane lipid bilayer
as
the 'amphiphilic'
aromatic residues are well suited for partitioning into the inter-
phase between the hydrocarbon and polar membrane
layers
lO
•
12
•
The comparison with maltoporin (Fig.
2b)
and with nonspecific
porins shows that the girdles in
ScrY
are much less regular, with
more aromatic residues in the hydrophobic zone.
We
speculate
that the irregularity
of
the girdles in
ScrY
may contribute to the
specific toxicity
of
ScrY
to
E.
coli
cells at high expression levels
26
since this would likely impair the sealing
of
the porins with the
membrane lipid bilayer, possibly resulting in membrane instabil-
ity
or
porin accumulation somewhere before the outer mem-
brane.
The
'greasy slide'
and
its
'ionizable track'
Schirmer et
al.
2!
identified six aromatic residues on the inner sur-
face
wall
of
the maltoporin from the external vestibule across the
constriction site toward the periplasmic vestibule.
As
sugar bind-
ing by proteins often involves stacking
of
pyranosyl rings on aro-
matic ri
ngs
33 these authors proposed that these aromatics are a
40
Fig. 4 Images
7
•
comparing
the
hourglass-
shaped
inner
channel surfaces
of
a.
ScrY
com-
plexed
with
two
sucrose molecules and
b.
maltoporin. The
cutting
plane
is
spanned by
the
trimer
axis and a radial beam connecting
the
axis and
the
constriction site center. The molec-
ular
surface near Asp. Glu. Arg.
Lys
is
colored
red whereas near
Phe.
Tyr.
Trp
it
is
colored
green.
chain
of
binding sites
or
a binding slide for
maltose and maltoligosaccharides.
In
structures
of
maltoporin complexed with
maltooligosaccharides, most
of
the gluco-
syl
groups are in van der Waals contact
with some
of
the aromatic residues.
In
ScrY,
the four external
of
the six aro-
matic residues found
in
maltoporin are
absolutely conserved (Trp 151, Tyr 118, Tyr 78, Trp 482; Fig. 3),
even up to the level
of
a similar side chain conformation. The
fifth, Trp 358,
is
conservatively replaced by Phe 435 whereas the
sixth near the periplasmic mouth
is
missing. Following Wang et
aU>,
we
use the terms
Sl-S5
for the glucosyl binding sites near
Trp
lSI,
Tyr 118, Tyr 78, Trp 482 and Phe 435.
The structure
of
the residues lining the constriction site
(78-80, 110, 114, 118, 120-121, 140, 142, 161, 187-188,
194-201, 204, 207, 322-323)
is
rather similar
in
both
glyco-
porins.
Of
these residues, all except three are conserved
or
con-
servatively replaced and most adopt a similar side
chain
conformation in
ScrY
and maltoporin.
In
spite
of
the very simi-
lar architecture
of
both
glycoporins around the constriction site,
the cross section
of
the channel in
ScrY
is
larger (8.5 A x
11
A)
than in maltoporin (7 A x
10
A).
This
is
due to three major ex-
changes
of
L3-residues protruding into the lumen: Asn
192
in
ScrY
becomes Arg
109
in maltoporin,
but
with a reversed orien-
tation so that the side chain
is
oriented into the lumen
in
malto-
porin;
Asp
201
and Phe 204 in
ScrY
are replaced by Tyr
118
and
Asp
121
of
maItoporin, the latter two residues protruding more
into the constriction site lumen than
the
former. The larger cross
section
of
the constriction site and the wider external channel
mouth qualitatively explain the much higher single channel con-
ductance
ofScrY in 1 M
KCI
(1,4 nS)
as
compared to maltoporin
(0,15
nS)28.
In the neighborhood
of
the greasy slide, the degree
of
conser-
vation
of
ionizable residues - the 'ionic track'
of
Schirmer
23
-
is
high and above the average conservation found between
ScrY
and maltoporin. Many
of
these are conserved even
in
their side
chain conformation between
ScrY
and maltoporin. Some
of
them were found to be involved in hydrogen bonding
of
two
maltose molecules, a maltotriose molecule, a maltohexaose mol-
ecule and a sucrose molecule
bound
to maltoporin
22
,23.
Sucrose
binding
sites
in
ScrY
The structure
of
ScrY
cocrystallized with sucrose reveals two
sucrose molecules,
'Sud'
and
'Suc2~
bound
to the channel sur-
face
(Fig. 4a) with occupancies ofO.? and 0.8 respectively. Their
glucosyl groups are in van der Waals contact with Tyr
78
(S3),
located in the constriction site, and with Trp 482 (S4), adjacent
to the periplasmic end
of
the constriction site (Figs 5, 6b), the
same sites as for the glucosyl groups
(g3
and g4)
of
a maltose
found in
maltoporin
22
• The sucrose molecules form hydrogen
bonds with the ionizable track in the neighborhood
of
the greasy
slide and
of
L3,
as shown schematically in Fig.
7.
Sud
and Suc2,
as
well as a single sucrose molecule
in
malto-
Fig. 5 Constriction site viewed (stereo)
from
the
periplasmic
space
with
superimposed electron
density map.
5uc2
is
seen
in
the
front
bound
near Trp 482
(54
binding site),
5uc1
is
seen
in
the
background near
Tyr
78
(53
binding site).
Green:
2F
ob
, - F"" density
at
2.0
cr.
Cyan:
Fob,
-
F""
'omit'
density
at
1.5
cr
of
the
5crY-sucrose
complex
with
phases calculated
after
refine-
ment
ofthe
model
without
sucrose.
porin
23
, are all similarly positioned with
the glycosyl group nestled against the
greasy slide, the fructosyl group bending
away
from it and the nonreducing gluco-
syl-C4 pointing towards the periplasm.
The greasy slide seems to bind the sugars
in a defined direction, that
is
with the
nonreducing end towards the periplasm.
This orientation has been found for all bound sugars in glyco-
porins22-24.
Sucrose binding to
S3
is
sterically blocked in maltoporin due
to the presence
of
Tyr
118
and Arg 109,
both
of
which protrude
into the lumen
of
the constriction site (Fig. 6a). These malto-
porin residues
as
well
as
Asp
121
are represented in
Fig.
6a
by
their transparent van der Waals surface. In addition to rendering
the binding
of
sucrose to
S3
sterically impossible, these residues
of
maltoporin represent a 'bolt' in the constriction site blocking
the path between
S3
and
S4.
The remainder
of
this region
is
very
similar in both glycoporins. Indeed, the glucosyl group
of
the
single sucrose molecule found by Wang
et
alP
in maltoporin
binds to a position between
S2
and
S3,
just before the constric-
tion site.
From the very low permeation rate
of
sucrose through malto-
porin, one can infer that an extended conformation
of
sucrose,
which should permeate through maltoporin efficiently,
is
very
unfavorable.
Our
results demonstrate how
ScrY,
in contrast to
maltoporin,
is
constructed to allow the bulky sucrose to slide
easily through the channel. The observation that, in the absence
ofScrY,
E.
coli
cannot grow on low sucrose concentrations
as
sole
carbon source
26
shows that
ScrY
achieves a high permeability for
this disaccharide.
Other
protein-sucrose complexes
In contrast to maltose and maltoligosaccharides, sucrose in crys-
talline form and
bound
to proteins has a globular conformation
with the glucosyl and the fructosyl moieties bent towards each
other. The torsion angles,
11>,
'1',
of
sucrose
34
are found to be sim-
ilar in crystallographical and molecular modelling studies to the
values
of
crystalline sucrose (108°/-45°), indicating a similar
overall conformation in solution and in the crystal
3
4-36.
In
ScrY,
the torsion angles about the two glycosidic bonds
of
sucrose,
11>,
'1',
arefound
to be 75°/-31 ° for
Sud
and 90°/-61 ° for Suc2.
In a complex with lentil-lectin
37
, sucrose exhibits torsion
angles
of
107°
and -58°, with a water intercalated between gluco-
syl-02 and fructosyl-03.
All
of
the glucosyl hydroxyls are hydro-
gen bonded to the protein
but
only two fructose hydroxyls form
indirect hydrogen bonds to the protein through waters.
Moreover, the glucosyl group
is
stacked on a phenylalanine ring.
The features
of
the bound sucrose in lentil-lectin are also found
in
ScrY
and maltoporin (Fig. 7). The two glucosyl groups are
more intimately attached to the channel wall than the fructosyl
groups, that
is
they are in van der Waals contact with aromatic
residues and form a larger number
of
hydrogen bonds to the pro-
tein.
As
the fructosyl group contributes
less
to the free energy
of
binding and since maltose can be considered
as
two glucosyl
groups the binding
of
maltose to
ScrY
should be stronger than
that
of
sucrose to
ScrY.
Indeed, experimental determination
of
the respective binding constants
28
yield
KD
values
of
6.25 mM
and
50
mM for maltose and sucrose.
Oligosaccharide binding sites involving chains
of
aromatic
residues flanked by potential hydrogen
bond
forming residues
have also been found in other proteins. Cellulose-binding
domains
of
cellulolytic enzyme complexes from fungi and bacte-
ria indicated binding sites for three adjacent crystalline cellulose
chains with aromatic residues interacting with roughly every sec-
ond
glucose unit
of
one chain
38
.
E.
coli
maltodextrin phosphory-
lase shows stacking
of
a glucosyl group on a
Tyr39.
The authors
discuss two factors contributing to the favorable stacking
of
glu-
cosyl groups on aromatic residues: hydrophobic interactions and
electrostatic interactions between partial positive charges on the
hydrogens bound to carbons
of
the sugar ring and the 1t electrons
of
the aromatic ring.
On
the other hand, chains
of
aromatic residues are not a gen-
eral feature
of
oligosaccharide binding,
as
seen from the struc-
ture
of
a bacterial endoglucanase
4o
, and lysozyme
41
. C-type
(Ca
2
+ -dependent) lectins show how the surroundings
of
a Ca
2
+
atom,
bound
to the surface
of
a protein, can create a low-affinity
binding site for monosaccharides
(K
D
between 6 and
22
mM)
which can bind two neighboring equatorial sugar hydroxyls to
the protein
42
,43.
Ca
2
+ has also been found to be involved in indi-
rect binding
of
glucose through water molecules
44
or
by appar-
ently optimizing the conformation
of
residues for hydrogen
bonding to sugar
hydroxyls45.
Glycoporins possess a series
of
low-affinity binding sites for
glucosyl groups. Binding occurs through hydrophobic contacts
with aromatic residues and hydrogen bonding to ionizable
residues. The binding sites are arranged
as
an extended slide
from the extracellular mouth to the periplasmic mouth
of
the
channel. The observed increase
of
binding constants for malto-
oligo saccharides with increasing number
of
glucosyl residues in
maltoporin
20
and
ScrY28
suggests a series
of
four
or
five
binding
sites with the correct spacing and geometry to bind such helical
maltooligosaccharide chains
21
,24.
As
binding sites increase the
sugar permeability
of
the outer membrane, it
is
expected that
glycoporins confer a growth advantage to bacteria whose growth
is
limited by low concentration in the culture medium
of
disac-
charides like maltose or sucrose
or
maltotriose and may be indis-
pensable for the uptake
of
maltooligosaccharides with more than
three glucosyl residues. Indeed in the presence
of
maltooligosac-
charides longer than three glucose units
as
the only available car-
41