Post on 01-Jun-2018
transcript
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
1/19
EBS 1988
Review
Permeation of hydrophilic molecu les through the outer membrane
of
gram-negative bacteria
Review
on
bacterial porins
Roland
BENZ
a nd K a tha r ina BA U ER
Lehrstuhl fur Biotechnologie, Universitat Wiirzburg
(Received February 8 /Ma y 10, 1988) EJB 88 0165
The cell envelope of gram-negative bacteria such as
Escherichia coti,
Salmonella
typhimurium
and Pseudomonas
aeruginosa consists of three distinct layers : the outer mem-
brane, the peptidoglycan (murein) layer, and the inner mem-
brane. The inner membrane acts as a real diffusion barrier
and contains, in addition to the respiratory chain and the
H+-
ATPase, a large number
of
uptake systems for hydrophilic
substrates
[l] .
All these hydrophilic solutes have to pass the
outer membrane on their way into the cell. This means that
this membrane must have very special permeability properties
because these substrates have various structures and can all
pass through the same transport pathways.
On the other hand, the outer membrane makes the gram-
negative bacteria resistant to host-defense hctors such as
lysozyme, fl-lysin, and different leukocyte proteins
[2 -41.
Furthermore, in enteric bacteria living in the intestinal tract
of animals, the outer membrane represents a very effective
barrier, which protects the bacterial cells from the action of
bile acid detergents and degradation by digestive enzymes
[5]
At the same time the outer membrane of enteric and some
other gram-negative bacteria represents a strong permeability
barrier to many antibiotics that are effective against gram-
positive bacteria
[5 ,
61.
The outer membrane of gram-negative bacteria acts as a
molecular filter for hydrophilic compounds. The active
components of these molecular sieving properties are a major
class
of
proteins called porins
[7].
The porins have molecular
masses between 30 kDa and 50 kDa and normally form
oligomers in the outer membrane that are, in many but not
all cases, stable
in
sodium dodecyl sulfate. Dependent on their
permeability properties for hydrophobic solutes, the porins
can be divided into two classes. Most porins form, in the
outer membrane and reconstituted systems, large water-filled
channels and are generally diffusion porins because they sort
solutes mostly according to their molecular mass. Other
porins have a certain solute specificity and contain binding
sites for sugar
[8, 91,
phosphate [lo,
1 ],
and nucleosides
I121
inside the pores. These specific porins represent a considerable
advantage for the diffusion of defined substrates through the
outer membrane but they do not act as a general diffusion
pore.
Correspondence
tu
R . Benz, Lehrstuhl fiir Biotechnologie der
Universitat Wiirzburg, Rontgenring 11, D-8700 Wiirzburg, Federal
Republic
of
G e rma ny
T H E O U T E R M E M B R A N E O F G R A M - N E G AT I V E B AC T ER IA
The cell envelope of gram-negative bacteria consists
of
two membranes separated by the peptidoglycan (murein) layer
and the periplasmic space. This additional cellular compart-
ment contains a variety of soluble proteins, some of which
function as processing enzymes and convert nontransportdble
metabolites to transport substrates
[l].
Furthermore, it con-
tains a variety of binding proteins essential for the uptake of
nutrients into the cells and the chemotaxis of the organisms
[l].
The periplasmic space appears to be iso-osmolar to the
cell interior which means that the high osmotic pressure across
the cell envelope in very dilute solutions is maintained across
the outer membrane and the attached murein and not across
the inner membrane [13, 141. The peptidoglycan layer is re-
sponsible for the maintenance of cell shape and for the ability
of the cell to withstand the very high internal osmotic pressure
in dilute environments. It consists of a network of amino
sugars and amino acids (see
[I51
for a review). The amino
sugars N-acetylglucosaminyl-N-acetylmuramoyl dimers)
form long linear strands which are covalently linked together
between two muramoyl residues by short tetrapeptides
[I
51
Components of the outer membrane such as the lipoprotein
or the porins are either covalently bound to the murein or
interact with this macromolecule via ion bridges
[16, 171.
Whereas the basic structure
of
inner membrane is formed
by a normal phospholipid bilayer similar to most biological
membranes and contains a variety of different proteins, the
outer membrane has an unusual lipid composition and a
relatively simple protein composition. The outer monolayer
contains lipopolysaccharide as its major or exclusive (in
enteric bacteria) lipid, while the inner leaflet contains
phospholipids (mostly
phosphatidylethanolamine
and small
amounts of phosphatidylglycerol and cardiolipin
[5 ,
16,
181.
Interestingly, this asymmetry is dependent on the presence
oi
an intact peptidoglycan layer in the periplasm, since degra-
dation of peptidoglycan with lysozyme results in a redistri-
bution of lipopolysaccharide within both leaflets
[39 ] .
Lipopolysaccharide (see [20] for a review) is an amphi-
philic molecule containing a hydrophobic region (lipid A, also
known as endotoxin) that has five or six fatty acids linked to
diglucosamine phosphate. Covalently attached to this is the
rough oligosaccharide core containing in its proximal portion
an unusual sugar, 3-deoxy-~-rnanno-octulos0nic cid (dOclA,
previously known as
KDO),
as well as a variety of heptose
and hexose residues
[16,
201. The rough oligosaccharide core
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
2/19
2
may be substituted with a variable number of repeated tri- to
pentasaccharide units called the 0-antigen [20
-
21.
Lipopolysaccharides carry net negative charge resulting in a
strong negative surface potential of gram-negative bacteria
[20]. They have attracted special interest because of their role
in assembly and maintenance of the outer membrane per-
meability barrier [16,22], their importance for biological func-
tion of most outer membrane proteins [23]. Their function as
phage receptors [24] and their endotoxic effects [25] are also
important.
Lipopolysaccharides appear to be anchored in the outer
membrane by binding to outer membrane proteins [26, 271,
possibly through hydrophobic interactions with lipid A [27],
and by noncovalent cross-bridging of adjacent lipopolysac-
charide molecules with divalent cations [28- 01. Thus treat-
ment of gram-negative cells with EDTA generally results in
removal, by chelation, of divalent cations and consequent
disruption of the outer membrane [29]. In the absence of
such chelators, however, the ion bridges formed by negatively
charged groups and divalent cations are responsible for many
of the properties of the outer membrane of gram-negative
bacteria, including resistance to hydrophobic antibiotics, bile
salts, detergents, proteases, lipases, and lysozyme
[ 5 ,
16, 30,
311.
The outer membrane also contains, besides the porins, a
small number of so-called ‘major’ proteins present in high
copy number (OmpA and lipoprotein,
lo5
and 7 x
lo5
copies
per cell, respectively
[5 ,
14, 161). These proteins have definitely
no channel-forming activity but are important for the struc-
ture and the stability of the outer membrane. One third of the
lipoprotein (Braun’s lipoprotein,
M ,
7200 in E .
coli
[32]) is
covalently bound to the peptidoglycan, whereas the rest and
OmpA ( M , 35000, ‘heat-modifiable protein’ [33, 341) interact
with the murein via ion bridges. Little is known about the
function of many minor outer membrane proteins with the
exception of proteins important for the uptake of iron [35]
and vitamin
B 1 2
[36].
EXPRESSION OF PORINS
Major outer membrane proteins
Besides the most abundant outer membrane protein, the
lipoprotein, the outer membrane of wild-type cells of E . coli
K 1 2
contains several other major proteins (Omp proteins)
which are expressed in large amounts under standard labora-
tory conditions. The nomenclature for these proteins has been
different in different laboratories mostly because of their un-
usual behavior in
SDS
gel electrophoresis [37
-
01. At pre-
sent, it is agreed that the proteins are named according to
their structural genes (i.e. OmpF is the gene product of the
ompF gene [41]). The relative amounts of the OmpF and
OmpC proteins (both forming general diffusion pores) are
dependent on the growth conditions of the organisms, es-
pecially the osmolarity of the growth media and their tempera-
ture [42- 71. Three genetic loci are known to be involved in
their expression. The loci ompC and ompF mapping at
47.1 min and 20.7 min, respectively, on the E . coli
chromosome [48] are structural genes [49- 11. Mutations of
ompC and ompF result in the absence of OmpC and OmpF,
respectively in the outer membrane. A third genetic locus
mapping at 74 min, ompB, plays a central role in the regulation
of OmpC and OmpF. ompB contains two genes, envZ (coding
for a ‘sensor’ and located in the cell envelope) and ompR
(coding for a cytoplasmic ‘regulator’). The ompB locus affects
transcriptional expression of the ompF and ompC genes, both
absolutely and relatively [52, 531.
Mutations in ompB result in many different combinations
of porin phenotypes: OmpC- OmpF’, OmpC’ OmpF- or
OmpC- OmpF- [54,
551.
Several factors affect the expression
of the porin proteins. The relative level at which the two
proteins are expressed is determined by the growth medium.
OmpC is preferentially expressed in media of high osmolarity
[50,
511, while OmpF
is
preferentially incorporated into the
outer membrane at low osmolarity. The expression of OmpC
is also favored in the presence of a fermentable carbon source.
Other cultural conditions lead also to different relative levels
in the expression of OmpC and Omp F [56]. The synthesis of
the porins and other outer membrane proteins (especially of
OmpA) is further modulated in such a way that the total
number of copies of these proteins remains constant [33, 44,
571. In addition, alterations of the lipopolysaccharide core
sugars [44] and of the lipid structure [58, 591 can affect ex-
pression of these proteins.
The construction of strains in which the lac operon is fused
to the
ompC
[49] and the ompF [52] genes, respectively, have
established that the ompB gene product has a dual role in
regulating both the absolute and the relative transcriptional
levels of expression of the ompF and ompC genes. Deletions
and base substitutions of the ompR gene and lacZ fusions
with this gene [60- 41 and subsequent mutant analysis re-
vealed that the OmpR protein activates the expression of
ompF and ompC genes most likely by binding to the upstream
regions of their promoters [60- 651. These studies, especially
those of the osmoregulation of such mutants also supported
the hypothesis that the OmpR protein has a two-domain
structure. Each domain plays a different role in the activation
of the
ompC
and
ompF
genes. Studies with the purified OmpR
showed that the interaction between this protein and the pro-
spective modulator EnvZ occurs at the N-terminal portion of
the protein, while the C-terminus is responsible for its binding
to the ompC and ompFpromotors [60-641. This is in accor-
dance with the fact that there is extensive conservation of
N-terminal regions between products of E .
coli
ompR and
other two-component regulatory systems like the phosphate-
starvation-dependent regulatory gene phoB and the nitrogen-
assimilation regulatory gene ntrC. There is also extensive con-
servation of C-terminal regions between the products of E .
coli envZandphoR, ntrB, and others [66]. It was proposed that
these regulatory genes comprise two-component regulatory
systems that evolved from a common ancestral system involv-
ing transduction of information about the status of the en-
vironment by one protein domain (C-terminal region
of
EnvZ) to a second one (N-terminal region of OmpR) [66].
Also, the fact that under certain conditions OmpR can acti-
vate the expression of OmpC or OmpF in the complete ab-
sence of envZ gives further evidence that EnvZ may function
as a transducer protein, which is not directely involved in
promotor recognition and DNA-binding [67]. Furthermore,
some exists evidence that the RNA polymerase is also involved
in the activation of porin genes [60- 4, 681.
The
micF
gene may serve as a third regulatory gene for
the expression of ompC and ompF. It codes for an RNA that
is complementary to the
5’
end of the ompF mRNA [69] and
seems to function as an antisense RNA in expression of the
ompF gene during osmoregulation. In this respect, tole,
coding for a minor outer membrane protein, may also play
an important role: to le mutants show reduced levels in OmpF
and overproduction of OmpC protein [70]. The properties of
micF mutants led to the hypothesis that the micF and the
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
3/19
3
+maltose
malA (74 ) malB(91 )
Fig. 1 . The maltose regulon qf E. coli. The lower part of the figure showes the genetic organization and regulation of the maltose regulon,
whereas the upper half of the figure indicates the molecular masses (given in kDa in circles) and cellular locations of the gene products. The
abbreviations I M, OM, PPS represent inner membrane, outer membrane and periplasmic space, respectively. Positive control is exerted by
the
malT
gene product and by the cyclic-AMP
atabolite-activator-protein
complex (CAMP-cap). The wavy line represents the direction of
transcription. This figure is adapted from
[256]
ompC
genes are probably coregulated [71]. It was proposed
furthermore, that
tolC
mutation leads to overexpression of
the coregulated micF and ompC genes. The increased level of
the micF product would then lead subsequently to a reduction
of OmpF.
Also the growth temperature seems to play an important
role in the expression of the major porins. Decreasing growth
temperatures result in increased amounts of OmpF and de-
creased quantities of OmpC. The regulatory gene responsible
for these effects was named
e n v y
[72]. It seems to map at
12.9 rnin on the
E . coli
linkage map
[48]
and codes for an
envelope protein which appears at low temperatures in
minicells [72]. How Envy functions and interacts with the
other porin regulatory components is not known.
Inducible major outer m embr ane proteins
Lam B prote in.
LamB is the product of the
lamB
gene
located at 91 rnin on the
E .
coli chromosome [48] and it is part
of the maltose regulon. This regulon consists of two regions,
malA
at 74 rnin
[73,
741 and
m a l B
at 90 rnin [75 -771. It con-
tains a number of genes coding for proteins involved in the
uptake of maltose and maltodextrins into the cell and their
degradation. LamB protein is involved in the permeation of
sugars across the outer membrane (see paragraph about
LamB under section on specific porins below). Maltose bind-
ing protein
( m a l E
gene product) is a periplasmic recognition
protein and is essential for chemotaxis. MalF, MalG, MalK
are inner membrane proteins responsible for the transport of
the sugars through the cytoplasmic membrane. Maltodextrin
phosphorylase
(malP
gene product) and amylomaltase male
gene product) are metabolic enzymes located in the cytoplasm
(see Fig. 1). The
ma1
system
is
positively regulated by the
malT
gene product [78], which is activated if maltose or
maltodextrins are added to the growth media. Furthermore,
it
is sensitive to catabolite repression: the complex of CAMP
and catabolite activator protein (cap) stimulates the ex-
pression of the regulator m al T gene and of the malE,F,G and
of the
malK-lam B
operons directly, whereas the stimulation
of the malP,Q expression may only be controlled via malT
V91.
PhoE protein.
The synthesis of the PhoE protein can be
induced in wild-type strains by growth under the conditions
of Pi starvation
[46, 801.
This process is coregulated with the
expression of serveral other Pi-starvation-inducible proteins
in a single regulon, the
pho
regulon [81]. The
p h o E
structural
gene is localized at 6 min of the chromosome of E. coli [81].
The other structural genes of the operon include
p h o A
(alka-
line phosphatase),
y h a S
(Pi binding protein), and
ugpB
(glycerol-3-phosphate binding protein). The gene products of
at least three regulatory genes are involved in this regulation:
p h o B ,p h o R ,
and
phoM
[82], together with a phosphate-specific
transport system for the uptake of Pi across the inner mem-
brane. The
p h o B
gene product is an activator required for
transcription of the structural genes, whereas transcription of
p h o B
is regulated by the
p h o R
and
p h o M
gene products (see
Fig. 2). Evidence for the model of Fig. 2 was obtained from
the cloning of the regulatory genes and the identification of
their products. The p h o R gene product represses the synthesis
of the PhoB protein in minicells
[83].
Furthermore, by using
p h o B - b e 2
[84] and
phoB-cat
[85]
gene fusions, it has been
shown that the synthesis of t h ep h o B product is induced under
Pi limitation. I t has to be noted, however, that Pi is probably
not the molecule that directly influences the state of the p h o R
gene product. In mutants with a defective phosphate-specific
transport system, a high intracellular Pi level can also be
maintained by another uptake system, the phosphate inor-
ganic transport system [86], while these mutants are still
derepressed for the
pho
regulon. It was recently shown that
the intracellular nucleotide pools are changed upon Pi limi-
tation and that the
phoU
gene is involved in these alterations
[87].These nucleotides are most likely the actual effectors of
thephoR gene product [88].
BIOSYNTHESIS OF THE PORINS
Common to all porins is their synthesis as a larger precur-
sor form with an N-terminal leader extension (see [89] for
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
4/19
4
I
Fig.
2.
The pho regulon
qf
E. coli.
B,
R, and M are the products of
the
genesphoB, p h o R , and phoM respectively.B is an activator which
is essential for transcription of the structural genes of thepho regulon
including phoE. The presence
of
sufficient Pi leads R to be in the
repressor form
RR,
which prevents transcription ofphoB. Pi starvation
converts R
to
the activator form RA. Both RA and M stimulate
transcription of phoB. The figure is adapted from
[88]
a recent review on export of protein). The leader sequence
contains between 21 and 22 amino acids in the case of the
three porins OmpF [90], OnipC [91], and PhoE [92] ofE.
oli,
whereas OmpA [93] and Braun’s lipoprotein [94] have leader
sequences of 21 and 20 amino acids, respectively. The existence
of leader sequences with approximately the same length and
a similar composition suggests that the major outer membrane
proteins (with the exception of the lipoprotein, see [95] for a
review) may share a common pathway for their biosynthesis
and their assembly. The genes most likely to be involved in
the ‘export machinery’ across the inner membrane are the sec
genes of the A, B,
D,
and
Y
(pr lA)
classes. The
secY ( p r lA )
gene product is an integral component of the cytoplasmic
membrane [96]and it is known to be essential for the translo-
cation of exported proteins across the cytoplasmic membrane
[97].
Other important genes are
secA
[98] and
secB
which are
also involved in protein secretion.
The biosynthesis and assembly of outer membrane pro-
teins have been studied using gene fusion products of outer
membrane protein genes with
lacZ [99
- 1021. So far it is not
clear
if
the results are meaningful because the gene fusion
products may be located in the cytoplasmic membrane and in
the periplasmic space and not in the outer membrane [103,
1041. The overproduction of OmpC using a multicopy plasmid
blocked the assembly of OmpA into the outer membrane but
not
of
a LacZ-LamB chimeric protein which is obviously
located in the periplasmic space [104].
The biosynthesis and assembly of OmpA was studied in
detail [105, 1061. As this protein probably shares the same
pathway of synthesis and assembly with the porins and LamB,
its description may serve as a model for porin biosynthesis.
OmpA protein is synthesized on free and membrane-bound
ribosomes
in
the cell as a water-soluble precursor which ac-
cumulates inside the cell
if
the membrane potential is partially
inhibited by the addition
of
an uncpupler of oxidative
phosphorylation [106]. At normal membrane potential, the
pro-OmpA protein is translocated through the cytoplasmic
membrane and processed by a membrane-bound leader
peptidase [107]. This leader peptidase also cleaves pro-mem-
brane proteins of the cytoplasmic membrane, proteins se-
creted into the periplasmic space, and most likely also the
porins, but it is not identical to the leader peptidase which
processes the lipoprotein [108].
So far, it is not clear in all aspects of processing and
assembly if the pro-porins and the pro-LamB share the ident-
ical mechanism to the pro-OmpA, i.e. if a membrane potential
across the cytoplasmic membrane is needed for the translo-
cation of the pro-porins across this membrane, although it
has been found that the presence of the FIF O s indispensable
for the export of LamB [109]. However, also in the case of the
OmpF porin, it is likely that the chain is almost, if not entirely,
complete before the processing star ts [110]. On the other hand,
some evidence also exists for a co-translational processing of
periplasmic proteins in
E .
coli, i.e. translocation of the nascent
chain after completion of approximately
80%
of its entire
length. When inner and outer membrane are separated, the
pro-OmpF is localized in the inner membranes, which indi-
cates its processing at the inner membrane by a membrane-
bound leader peptidase [110] (see above). The processing of
the pro-OmpF into the mature protein occurs very rapidly
with an apparent lifetime of the precursor of 30 s [110].
The mechanism of the assembly of the trimers in the outer
membrane from three mature monomers is still completely
unknown, although some indication exists for a somewhat
delayed formation of stable trimers from metastable trimers
in the case of LamB [ l l l ] . In any case, it is obvious from the
study of the mutant OmpC and the overexpression of OmpC
that OmpA, LamB and the porins share the same machinery
for the assembly into the outer membrane [104, 1121. This
could be shown by the use of an OmpC mutant of
E .
coli, in
which only the amino acids Val-300 and Gly-301 were re-
moved [112]. The expression of this protein caused a dramatic
decrease of the expression of OmpA, of LamB, and of the
other porins by blocking the assembly machinery into the
outer membrane but it did not effect the assembly of the
lipoprotein
[
1121.
ISOLATION AND PURIFICATION
OF
PORINS
The general diffusion pores of different gram-negative
bacteria were found to be tightly associated with the
peptidoglycan layer [113, 1141. Their isolation is relatively
simple because only a few outer membrane proteins are associ-
ated with the murein, while the others are lost during the
washing of the cell envelope fraction with SDS-containing
solutions. Other pore-forming proteins, such as the
Tsx
chan-
nel, are not murein-associated [12, 1151. In this case the outer
membrane has to be separated from the inner membrane by
procedures similar to that used first by Miura and Mizushima
[116].
Isolation of the outer membrane
The first step is always the disruption of the cells. This can
be done by French pressure cell treatment, by ultrasonication,
by shaking the cells in the presence of glass beads, or by
osmotic lysis of the spheroplasts. The disruption followed
by differential centrifugation leads to a crude cell envelope
fraction which is composed of outer membrane, cytoplasmatic
membrane, and peptidoglycan layer. The separation between
outer membrane and inner membrane can be done in the
following way [12,116]. The crude envelope fraction is layered
on a two-step sucrose gradient
(70%
sucrose and 54% su-
crose) and is centrifuged for 10 h at
80000 xg.
The outer
membrane accumulates at the interface between 54% and
70
sucrose, because of its higher density which may be
caused by the larger content of carbohydrates of the outer
membrane. Another method for the isolation of the outer
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
5/19
5
membrane uses its relatively low solubility in nonionic deter-
gents. The cytoplasmic membrane is removed in this case by
a washing procedure with Triton X-100 or sodium lauryl
sarcosinate in the presence of MgZf [40]. It has to be noted,
however, that in both cases a considerable fraction of
phospholipids and lipopolysaccharides may be lost from the
outer membrane during the washing procedure with deter-
gents.
Enzym atic degradation
o f
the porin-murein com plex
Porins, tightly associated with the peptidoglycan layer, can
easily be isolated by the method proposed first by Rosenbusch
11171. After breaking the cells, the cell envelope fraction is
solubilized in 2% SDS at 60°C. The insoluble material con-
tains the peptidoglycan with the covalently bound lipoprotein
and the associated porins. The porin can either be separated
from the peptidoglycan by digestion of the peptidoglycan
layer using lysozyme and trypsin [7, 1141 or in a more elegant
way by a salt extraction method 1118- 1201 (see below). Boil-
ing of the porin-peptidoglycan complex for 5 min in SDS
leads also to the dissociation but it denatures the protein.
Salt extraction method
A more recent method uses a salt extraction method for
the preparation of trimers first proposed by Nakamura and
Mizushima [118]. The method was extended by Tokunaga et
al. [119] and a recent excellent description has been given by
Nikaido [120]. A short scheme of the procedure is as follows.
The gram-negative bacteria are grown under normal culture
conditions at 37°C; the addition of 0.5% glucose represses
the expression of the maltoporin (LamB); after washing the
cells they are resuspended in a small volume and passed three
times through a French pressure cell; the cell envelope con-
taining the cytoplasmic membrane, the peptidoglycan, and
the outer membrane are pelleted by cenrifugation. The SDS-
soluble components
of
the outer and the inner membrane
are removed by washing the pellet with a buffer solution
containing 2% SDS at elevated temperature (up to 60°C).
After centrifugation the pellet should contain only peptido-
glycan and the associated protein. Considerable amounts of
non-porin proteins in the pellet can be removed by a repetition
of the SDS-wash step followed by centrifugation. The
peptidoglycan-associated protein preparation is resuspended
at 37°C in a buffer which contains as essential components,
besides 1 SDS, 0.4 M NaCl and 5 mM EDTA. The super-
natant of a subsequent centrifugation step is applied to a
Sepharose 4B or a Sephacryl S-200 superfine column. The
porin trimers are eluted from the column with the same buffer.
Ion-exchange columns (for example DEAE-Sephacel) can
also be used for the purification of porins because the pore-
forming complexes adsorb to the column material according
to their net surface charge [12]. For this, the large amount of
salts introduced by the salt extraction have first to be removed
by dialysis against low-ionic-strength buffer. Also, if SDS was
used for extraction, it is exchanged with non-ionic detergents
such as Triton X-100. The column is first washed with low-
ionic-strength detergent buffer. The porins are then eluted
from the column with linear salt gradients ranging over 50-
500 mM a t a given pH. The salt and detergent content of the
eluted protein can be decreased by another dialysis procedure
but this is not essential for the pore-forming activity which
remains constant for several months at 4°C in a refrigarator
or frozen in a freezer. The protein solutions can also be
lyophilized in many cases without any loss of the pore-forming
activity. In lypophilized form the protein remains active for
at least a year stored in a freezer at -20°C.
Saline extraction of whole cells
This method utilizes the susceptibility of the outer mem-
brane of photosynthetic bacteria towards 150 mM NaCl
solu-
tion [121, 1221. Freshly harvested cells are shaken for 2 h
in 150 mM NaCl at 37°C. The supernatant obtained after
centrifugation is dialysed against distilled water and lyo-
philized. Part of the material obtained is porin. The rest is
composed of lipid and lipopolysaccharide and occasionally of
other outer membrane proteins [121, 1221. The porin may be
purified as described above. It has the same pore-forming
properties as porin isolated by the salt extraction method
described above [123].
MODEL
M E M B R A N E
STUDIES W I T H P O R I N PORES
The permeability properties of porin pores for hydrophilic
molecules can either be studied
in vivo
or
in vitro. In vivo
experiments allow the study of the pores in their natural
environment which means that the substrates can interact
with the whole uptake system including periplasmic binding
proteins and periplasmic enzymes. The properties of the porin
pores were studied in vivo using the p-lactamase activity locat-
ed in the periplasmic space [124, 1251 and the uptake of
radioactively labeled solutes into the cells [126]. Both methods
only yield precise information on outer membrane per-
meability if the flow of the p-lactam antibiotics across the
outer membrane is rate-limiting [127], i.e. the p-lactamase
activity is
so
high that the enzyme is not saturated and the
gradient of the p-lactam antibiotics is established across the
outer membrane. Similar considerations apply to the uptake
studies. In these cases, it is necessary that the solutes are
actively taken up into the cell and that their gradient is again
established across the outer membrane which is
only
possible
at very low solute concentration because of the high affinity
of most inner membrane transport systems [127].
In
vitro
studies allow much better control of the experimental con-
ditions. On the other hand, the possibility of artifacts exists
(see below).
The vesicle permeability assay
This method was the first to be used for the identification
of the pore-forming proteins of the outer membrane of
E .
coli
and S. typhimurium 17, 113, 1141. The basic principle of the
vesicle permeability assay is as follows. Liposomes are formed
in a buffer solution containing two radiolabeled solutes (for
example [14C]sucrose and [3H]dextran) from bacterial or
other lipids in the presence of protein fractions or purified
porin. In the first investigations lipopolysaccharide was also
added at a molar ratio of 1 8 to the lipid. However, its pres-
ence is not essential for the pore-forming activity of the porins
and the formation of the liposomes [120]. Vesicles or lipo-
somes are formed by shaking the suspension, followed by mild
sonication in a water-bath-type sonicator. During liposome
formation both radiolabeled substances are entrapped. Sub-
sequently, the liposomes are passed through a Sepharose 4B
column and elute just after the void volume. The [3H]dextran
is retained in the liposomes, whereas the low-molecular-mass
['4C]sucrose may leave the liposomes through the pores dur-
ing the chromatography. In this case, it is retarded on the
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
6/19
6
column and the initial
1 4 C/3 H
atio is decreased and indicates
the permeability of the pores for the low-molecular-mass sol-
ute. Similar experiments can be performed with 14C-labeled
solutes of different molecular masses. The 14C /3H atio mea-
sured as a function of the molecular mass of the solutes allows
the evaluation of the exclusion molecular mass and of the
effective diameter of the pore [7, 113, 114, 1281.
Experiments of this type have been used to measure the
exclusion molecular mass of porins of
E .
coli, of
S .
t y p h i m u r i u m , of P . aeruginosa and of Haemophilus i n f l ue nza
[7, 113, 114, 128, 1291. The exclusion limits of the porins of
E. col i and of
S .
t y p h i m u r i u m were between 550-800 Da
which would be consistent with diameters around 1.1-
1.3 nm. The diameter of the protein F pore of P. aeruginosa
outer membrane appears to be much larger because solutes
with molecular masses of more than 5000 Da could permeate
the pore [130] (Fig. 3). All these diameters were found to be
consistent with in vivo studies, although the exclusion limit of
the outer membrane of P. aeruginosa was a matter of debate
[131, 1321. Furthermore, the role of protein F as a porin was
questioned [132].
The vesicle permeability assay can be termed an all-or-
none method. This means that it does ngt allow the measure-
ment of the kinetics of the permeation of different substrates
relative to one another. The time needed for the elution of the
column is so long that it is not possible to see any difference
between the rates of permeation of different solutes or any
specificity of the pore for one class of solutes. This all-or-none
assay may also lead to artificial results if more than one porin
is present in the protein preparation. In this case only the
larger pore can be detected even if it was only a small fraction
of the total number of pores. The presence of only one larger
pore per liposome would be enough to swamp the effects of
many small pores. This effect presumably played a role in the
first characterization of LamB incorporated into liposomes
[129]. Another possible artifact may arise if the permeability
of only one solute is measured by means of the assay described
above. Denaturated pore-forming complexes could in this
case cause a certain permeability of the liposomes for
hydrophilic solutes without the formation of a defined dif-
fusion pathway. However, the existence of a defined exclusion
limit may be questionable in this case.
The
liposome
swe l l i ng m e thod
This method was introduced in the study of the per-
meability properties of porin pores by Nikaido and coworkers
[120, 133- 361 following a method established by Bangham
et al. [137, 1381. Lipid and protein are dispersed in a buffer
first by shaking and then by sonication in a bath-type
sonicator. The buffer contains a large-molecular-mass dextran
or stachyose to maintain a certain osmolarity (around
40 mOsmol). The stachyose or the large-molecular-mass
dextran are not rendered permeable by the porin and are
entrapped inside the liposomes. The liposomes are added
under rapid mixing to an isotonic solution of a test solute. If
this solute can penetrate the pores, the tptal concentration of
solutes inside the liposomes increases because stachyose or
dextran are retained. Liposomes behave like an ideal
osmometer which means that water molecules permeate into
the liposomes according to the osmotic gradient. This swelling
process can be detected by a decrease of the average refractive
index of the liposome suspension, i.e. by a measurement of
the absorbance (see Fig. 4). The initial swelling rate can be
used as a measure of the penetration rate of the test solute
0.3 1.0 10
50
Molecular ma s s
k Da)
Fig.
3.
Exclusion limit fo r saccharides in vesicles reconstituted f ro m
outer membranes of P. aeruginosa
a)
and S . typhimurium LT2
M I
0 ) s obtained fr o m the vesicle permeability assay. The results imply
that porin of P. aeruginosa outer membrane has a larger exclusion
limit than porin from S . typhimurium. (From [258] with permission)
0.51
0 1 2 3 0 1 2 3 4
Time after mixing (min)
Fig.
4.
Absorbance tracings in a liposome-swelling experimen t recorded
at a wavelength
of 450
nm with liposomes reconstituted fr om lipid and
porin from E. coli
B.
The figure shows the results obtained after
dilution into L-arabinose (Ara), N-acetyl-D-glucosamine (GlcNAc),
sucrose (SUC),and buffer (5 m M Tris/HCl, pH
7.4) with
liposomes
reconstituted in the presence (right) and in the absence (left) of
porin.
(From [ I 331 by copyright permission of
the
Rockefeller
University
Press)
100
50
g 20
150
200
250 3 0 0
350
M o le cu la r m a ss
Fig. 5. Relative rates of permeation of different sugars into phospho-
lipid-porin liposomes. Different
sugars
(from top to bottom) of
150 kDa (1,-arabinose),
180
kDa
(D-galactose,
D-fructose,D-mannOSe,
D-glucose), 221 kDa (N-acetyl-D-glucosamine),
262
kDa
(2,3-
diacetamido-2,3-dideoxy-D-glucose)nd
342 kDa (sucrose,
meli-
biose, maltose, lactose). (From [133]
by
copyright permission of the
Rockefeller University Press)
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
7/19
7
Table 1. Physical properties and pare cha racteristics
of
general dif fusion po rin pores
The corresponding references are given after each result. The complete names of the organisms are: Aeromonas salmonicida, Anabaena
variabilis, Brucella a bortu s, Campylo bact er,fetu s, Cam pylob acter ejun i, Haemo philu s influenza, Klebsiella pneu mon iae, Legionella pneumophilia,
Neisseria gonorrhoae, Paracoccus denitrificans, Proteus mirabilis. Pseudomonas aeruginosa, Rhodobacter capsulatus (formerly
Rhodopseudomonas capsulata [255]), Rhodobacter sphaeroides (formerly Rhodopseudomonas sphaeroides [255]), Salmonella typhimurium,
Spirochaeata aurantia, Yersinia pestis
Porin Bacterial species Molecular mass Diameter derived from
exclusion liposome single-channel
limit swelling- conductance
General diffusion porins
OmpF
OmpC
PhoE
Lc
Lc (HK 253)
NmpC
K
OmpF
OmpC
OmpD
PhoE
37-kDa
40-kDa
PhoE
PhoE
I
F
E
MOMP
43-kDa
46-kDa
32-kDa
40-kDa
Group
2
33-kDa
43-kDa
47-kDa
Porin
42-kDa
40-kDa
Escherichia coli
Salmonella tyohimurium
typhimurium
Enterohacter cloacae
K. pneumoniae
N . gonorrhoae
P. aeruginosa
Y.
pestis
L. pneumophila
C.
ejuni
S. aurantia
A .
salmonicida
P. mirabilis
H . influenza
B. abortus
P. denitrgicans
R
.
capsula tus
R
.
sphaero ides
A. variabilis
C . e t u s
kDa
32.7 [90, 2421
36.0 [91]
36.8
[9]
36.5 [21 I]
40 [242]
40 [243
39.3 [149, 1191
39.8 [149, 1191
38.0 [149, 1191
36 [245]
37 [246]
40 [246]
39.5 [211]
37 [212]
37 [212]
34 [141 , 2241
39 [215]
33 [247]
-25 [258]
43 [248]
46 [248]
32 [249]
42 [250]
40 [251]
40 [128]
40 [I471
33 [252]
43 [143]
47 [I221
nm
600
[I141
-
-
-
-
-
800
[113]
800 [I 131
800 [I131
-
-
-
-
-
5000 [130]
-
-
-
-
-
1400 [128]
700 [I471
-
-
-
1.2 [134]
1.1 [I341
1.1 [I341
-
-
-
-
-
[257]
- [257]
-
1.2 [241]
1.6 [241]
-
1.0 [141]
2.0 [144]
-
-
-
-
-
1.4 [128]
1.2 [I471
1.7 [252]
1.6 [I431
1.2 [I221
-
1.2 [I581
1.1 [158]
1.1 [I581
1.3 [242]
1.0 [158], 1.4 [I501
1.1 1243, 2441
1.2 [I481
1.3 [I481
1.3 [148]
1.1 [240]
-
1.1 [170]
1.1 [I701
1.1 [156]
2.0 [155]
1.1 [I581
1.0 [258]
0.9 [248]
0.9 [248]
2.3 [249]
I O
[250]
1 O”
1.0 [I571
1.7h
1.6 [I231
1.6 [253]
-
Specific porins Specificity
LamB
E. coli
Tsx
E. coli
D1 P. aeruginosa
P P. aeruginosa
LamB
S. typhimurium
48.0 [I721
31.4‘
specific for nucleosides
[12]
46
[115, 1961 specific for sugarsd [I151
4 8 [ l l ]
specific for anions
[lo , 11, 1591
44 [254] specific for sugars (?)
specific for sugars [9, 135, 145, 154, 1611
a Schmid,
A, ,
Bauer,
K.,
Benz, R., unpublished results.
‘
Bremer,
E.,
Martinussen,
J.,
Valentin-Hansen, P., unpublished results.
Benz,
R. ,
Schmid, A. , Rosenberg, E. Y., Nikaido,
H.,
unpublished results.
Nikaido, H., personal communication.
through the porin pore. Using the same liposome preparation,
the pentration rate of different solutes can be compared
striking of the pore edge and by the friction of the pore
interior) is given by
-
(Fig: 5). For a general diffusion pore, i.e. a water-filledcyl in-
der, the logarithm of the relative rate of permeation is a linear
A /
=
[l
(a /r)]’[1-2 .104(u /r)
+
2.09
-
.95 (u/r)’]
,”,
function of the molecular mass of the solutes, which allows
1)
where A is the effective area of the pore with the total area
A,, r and a are the radii of the pore and of the solute respec-
tively. The use of the liposome swelling assay allowed a mean-
an estimation of the effective diameter of the pore according
to the theory of Renkin [139].According to this theory the
total hindrance to diffusion of the solutes (caused by the
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
8/19
8
ingful comparison to be made between different porin pores
of
E .
coli outer membrane [I341. Furthermore, this method
was used to demonstrate the specificity of the LamB channel
for maltose and maltodextrins compared to other saccharides
[9, 1351 and to characterize the propert ies of different LamB
mutants [140]. The properties of a variety of porin pores
of
different gram-negative bacteria were also investigated using
this method [128, 141 -1441. The pore diameter varied be-
tween 1 I - .2 nm (see Table 1).
Although the liposome swelling assay is far more precise
than the vesicle permeability method, i t may possibly have a
number of complications. The theoretical evaluation
of
the
Renkin equation as used for the calculation of the pore diame-
ter is based on a number
of
simplifications [139]. In the case
of the experiments with porin pores, these may not be justified.
A second problem is the poor time resolution which is limited
by the stirring that is necessary after the addition of the
liposomes to the test solute [133]. Information may be lost
because the early decay of the absorbance cannot be recorded.
This may be the reason for the non-linear dependence of the
swelling rate on porin concentration which was observed for
OmpC of E . coli K12 [134]. A final difficulty is estimation of
the effective diameter of the test solutes. Disaccharides of
identical molecular mass caused a substantial variability of
the swelling rates (3 -7-fold) [134]. This variability may be
caused by differences in the hydration shell of the solutes. The
large-molecular-mass solute (dextran or stachyose) may also
create problems. The osmolarity of the large-molecular-mass
dextran was difficult to adjust in the experiments with LamB
[9].
Stachyose, on the other hand, which was also used in these
experiments [140], binds to the binding site inside the LamB
channel [145]. This means that the relative rate of permeation
of different sugars through LamB mutants may be influenced
by the binding of stachyose.
Lipid hiluyer
experiments
The swelling experiments with reconstituted liposomes
provide excellent information about the size of porin pores.
Less detailed information is available on the selectivity of
the pores. This is because membrane potentials occur in the
liposome swelling assay. These are created by the use of
charged test solutes which may greatly complicate the analysis
of
the experimental data [134].
More detailed informat ion about the pore interior and the
pore selectivity can be obtained from lipid bilayer exper-
iments. Three different method have been used successfully
to reconstitute porin pores into lipid bilayer membranes. In
the first
of
these, detergent-solubilized porin was added di-
rectly to the aqueous-phase-bathing solvent containing
(‘painted’) lipid bilayer membranes [146]. In the second
method, solvent-free (‘folded’) membranes were formed from
vesicles reconstituted from lipid and protein according to the
Montal-Mueller method [147]. The third method inserted the
porin pores into lipid bilayer via fusion of reconstituted ves-
icles with membranes [148]. This methodhas only occasionally
been used. It will not be further described here.
The simplest method consists of the addit ion of detergent-
solubilized porin, in very small concentration
(1
- 00
nglml),
to the aqueous phase bathing a black lipid bilayer membrane.
After an initial lag of some minutes, presumably caused by
the diffusion of the protein through unstirred layers, the con-
ductance (i.e. membrane current per unit voltage) of the mem-
brane increased by many orders of magnitude within about
20-30 min [14, 149,
1501.
Only a slight additional increase
400
p s
A
1 *in
Fig.
6 .
Single-channel recording
of
a
diphytanoylglycerophospho-
cholineln-decane
in
the presence of
20
nglmlprote in
P ojP.
aeruginosa
a n d 3 0
mM
KCI
in the ayueousphase.
A
voltage
of 50 m V
was applied
through calomel electrodes with salt bridges; temperature
= 25‘ C
(as compared with the initial one) occurred after that time.
The conductance increase was found to be dependent on the
type of porin used for the reconstitution and on the
membrane-forming material. In membranes from oxidized
cholesterol, a completely unphysiological lipid, porins usually
showed about 100-fold more pore-forming activity than in
membranes f rom normal phospholipids [14,148]. The reason
for the dependence of the reconstitution rate on the
membrane-forming lipid is not clear. It may arise from small
structural differences between the membranes. Using this
method, the surface area of the membrane is in principle not
limited. Multi-channel experiments (i.e. experiments with a
large number
of
channels) were usually performed with sur-
face areas between 1 and 2 mm2, whereas the membrane areas
were below 0.1 mmz in the case of the single-channel record-
ings [146, 1511. A maximum of between l o 6 - lo8 pores/cm’
could be incorporated into lipid bilayer membranes using this
method. This means that the reconstitution of porin pores is
not a rare event.
On
the other hand, it is obvious that this
number is far below the pore density in the outer membrane
of gram-negative bacteria (1012 pores/cm2 [149]).
A n
alternative method for the reconstitution of porin
pores was used by Schindler and Rosenbusch [147, 1521. Ves-
icles from lipid and protein were spread on the surface of the
aqueous phase on both sides of thin teflon foil which has a
small circular hole (about 100 pm diameter) above the initial
water levels. The surfaces of both aqueous compartments were
covered by monolayers. The water levels on both sides of the
membrane are now raised and a folded lipid bilayer membrane
is formed across the small hole.
So
far it is not clear how the
porin pores are incorporated into the lipid bilayers. In the
early investigations the pores were always activated as mul-
tiples of the single conductance unit at large voltages, presum-
ably because of the fusion of porin-containing vesicles with
the membrane [147]. More recently, especially in the case of
PhoE [153] and LamB [154], the pores were activated as single
conductive units. The diameter of folded membranes is limited
to about 100-200 pm (area about 0.07 mm’).
Both experimental approaches allow the resolution of
single channels [14, 1521. This is one of the main advantages
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
9/19
9
Table 2. Average single-channel conductance of Rhodobacter capsu-
latus porin in differen t salt solutions
The solution contained
10
ng/ml porin and less than
0.1
Fg/ml SDS
or Triton
X-100.
The pH of the aqueous salt solutions was around 6
unless otherwise indicated. The membranes were formed from
diphytanoyl glycerophosphocholine/n-decane. Temperature
=
25°C;
V , =
25 mV. (Taken from [123])
Salt c A A / .
KCI
RbCl
NaCl
LiCl
K z S 0 4
MgClz
CaCI,
KCH3C02 (PH
7)
Tris/HCI
K/Hepes (pH
8)
Tris/Hepes (pH 8)
M
0.003
0.01
0.03
0.1
0.3
1
o
3.0
1
o
1
o
1 o
0.5
0.5
0.5
1 o
0.5
0.5
0.5
nS
0.012
0.038
0.12
0.35
1.1
3.3
9.5
3.4
1.8
1.4
2.1
1.3
I
.8
2.3
0.52
0.83
0.12
nm
0.29
0.29
0.29
0.29
0.29
0.30
0.35
0.30
0.21
0.20
0.28
0.20
0.23
0.33
0.17
0.28
0.18
of the lipid bilayer assay. The experiments were performed in
the following way. Small amounts of porin
(1
-
0
ng/ml) were
added to the aqueous phase bathing a black membrane of
small surface. Subsequently, a step-wise increase of the current
through the membrane was observed. Fig.
6
shows an exper-
iment of this type with protein P of P.
aeruginosa [159].
All
conductance steps were directed upward. Closing events were
in
general only rarely observed. This means that the lifetime
of porin pores from gram-negative bacteria usually exceeded
1
min. The only pores with a much shorter lifetime (around
50
-
00
ns) were those observed in the presence of total outer
membrane of
P.
aeruginosa [155],
although the absolute level
of the pore conductance (i.e. pore current per unit voltage)
was the same as with purified protein F.
The single-channel conductance, A , of many but not all
porin pores was a linear function of the specific conductance
ci
of
the bulk aqueous phase. This means that, despite a large
variation of the average single-channel conductance, the ratio
A / a
varied
on ly a
little (see Table
2).
The single-channel con-
ductance
of
these general difusion pores (see the next section)
can be used to calculate the effective diameter of the porin
pores. Assuming that the pores are filled with a solution of
the same specific conductivity as the external solution and
assuming a cylindrical pore with a length
1
of
6
nm, the effec-
tive pore diameter
d (
=
2r)
and the cross-section can be calcu-
lated according to the equation
[146]:
A
=
m r 2 / 1 .
The diameter of the porin pore of Rhodob'acter
capsulatus
(compare Table
1)
may be calculated to be about
1.6
nm from
the single channel conductance
(0.40
nS) in 0.1
M KC1
and
the corresponding specific conductance of the aqueous phase
(ci
= 14
mS/cm)
[123].
As already pointed out, the liposome swelling method
cannot be used to obtain detailed information on the selec-
tivity properties of porin pores. Experiments with lipid bilayer
Table 3. Zero-current membrane potentials
V,
for diflerent porins
in
the presence oJ u tenfold KCl gradient
Species Porin
V P , / P a
Reference
mV
E.
coli OmpF (B)
27
3.9
OmpF (K12) 28 3.8
OmpC
50
26
K
46
1.6
PhoE -24 0.30
NmpC -26 0.21
ti 581
OmpC 54 41 11581
OmpD
0
4 23 .54
~ 5 8 1
245]
S . ~ ~ p ~ i ~ u r j u ~mpF 46 14
PhoE
P . aeruginosa P -59
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
10/19
10
The basic advantage of the lipid bilayer method is the
measurement of molecular events. Every step of Fig. 6 reflects
the conductance due to insertion of one conductive unit, i.e.
of one protein
P
trimer of
P.
aeruginosa
into the membrane.
This allows an easy check
of
the purity of the porin prep-
aration and of the integrity of the pore-forming units after
the isolation and purification of the porin [161]. The interpre-
tation of single-channel studies requires a statistically signifi-
cant number of measurements. This means that the measure-
ment of only ten channels is not sufficient to characterize a
porin pore. Instead, hundreds of pores must be studied in
order to avoid identifying random fluctuations, which may
occur in a limited number of events in single-channel measure-
ments, as characteristic properties of the pore [162].
GENERAL DIFFUSION PORINS
Most porin pores of gram-negative bactrial outer mem-
brane that have been studied in vivo or in vitro are general
diffusion pores. Table 1 summarises the properties of all the
general diffusion pores that have been characterised to date
using one of the three methods described in the previous
section. The diameter of the different porin pores varies f rom
about
1 0
nm to 2.2 nm. This means that the diameter of the
porin pores is much larger than that of the ‘ion’ channels in
nerve and muscle membranes. Reconstitution experiments
have shown that the porins of
E.
coli outer membrane are
permeable to trisaccharides and tetrapeptides, i.e. the pores
allow the penetration of nutrients and ions through the outer
membrane
[ 5 ,
14, 17, 1601. Lipid bilayer experiments have
demonstrated that the ions move inside the porin pores similar
to the way in which they move in the aqueous phase. This
finding allowed the calculation of an effective radius on the
basis of Eqn (2). It is interesting to note that the radii as
calculated from the different methods described in the preced-
ing section show satisfactory agreement (compare Table 1).
Only in the case of the major outer membrane protein of
Huemophilus influenza did the radii estimated f rom the differ-
ent methods not coincide [157]. The reason for this is not
completely clear. It has to mentioned, however, that the
unheated porin did not run as oligomers on SDS/
polyacrylamide electrophoretograms and that different lipids
were used for the reconstitution experiments. On the other
hand, it cannot be excluded that the explanation proposed by
the authors (i.e. that the branching of the channel is respon-
sible for the difference) may be valid in this case [157, 1601.
I t
has already been mentioned above that in vivo and
certain in vitro studies give only vague information about the
selectivity of the porin pores [134, 1631. The swelling rate of
PhoE-containing liposomes was only a factor of two larger,
in
the case of negatively charged solutes than that of a neutral
solute with a similar molecular mass [134]. The lipid bilayer
assay allows good access to both sides of the membrane and
it is easy to establish a salt gradient across the membranes
and to perform zero-current potential. measurements [151,
1581. The use of salts with equally mobile ions like K + and
CI-
(limiting molar conductivities 73.5 mS/M and 76.4 mS/
M [164]) yields precise information about the selectivity of the
porin pores. Other salts such as LiCl and KC H3 CO0 can be
used to test whether the pore is a general diffusion pore
or not. Li’ and acetate- have similar aqueous mobilities
(38.7 mS/M and 40.9 mS/M [164]) but smaller mobilities than
K + and C1- (see above). This means that a general diffusion
pore must have a larger cation selectivity in KCI than in LiCl
and a smaller cation selectivity in KCI than in KCH3CO0.
Table 3 shows that these considerations are correct for the
general diffusion pores but not for the anion-specific porin
protein
P
from
P.
aeruginosa
outer membrane.
Is
PhoE
a specific porin?
PhoE has been shown to be effective for the uptake of
phosphate, polyphosphate and other negatively charged
sol-
utes [126, 163, 165-1681, The general diffusion pores listed
in Tables 1 and 3 have been accepted as such, except in the
case of PhoE pore of E. coli outer membrane. In this case some
authors have proposed the existence of specific phosphate or
polyphosphate binding site, because in in vivo and in in vitro
experiments, polyphosphate could block the pore [126, 153,
1691. However, more recent investigation of this ‘specific’
inhibition effect showed that it only occurred if the divalent
cation Mg2 and polyphosphate (or other polyvalent anions
like citrate) were present in the aqueous phase [170, 1711. The
inhibition
of
the pore function in the presence
of
both ions
can be explained by complex formation inside the pore, which
reduces the flow
of
solutes. Polyphosphate alone is not able
to reduce the permeability of the PhoE pore in the in vivo or
in vitro
experiments but changed its selectivity from anion to
cation selectivity, because the excess of positively charged
groups inside the channel
is
converted into an excess of nega-
tively charged groups by the presence of the polyphosphates
[171].
SPECIFIC PORINS
L am B o f E. coli
Transport of maltose and maltodextrins in E . coli is me-
diated by a transport system that is composed of several
components including LamB (see above). The
lumB
gene
codes for the receptor of phage /z in the outer membrane
(molecular mass 48 kDa [172]) which was found to be
important for the maltose uptake because
lamB
mutants were
found to be impaired in maltose transport when the concen-
tration of this sugar in the growth media was less than 0.1 mM
[173]. This result indicated an important role of the LamB
protein in the maltose uptake machinery. Early reconstitution
experiments with purified LamB suggested that
it
formed a
general diffusion pathway similar to the general diffusion
pores but with a larger diameter [129,174]. More recent inves-
tigations have shown that LamB (also known as ‘maltoporin’)
has a defined substrate specificity [9, 135, 145, 154, 161, 1751
and that i t is able
to
discriminate between disaccharides of
identical molecular mass, for example between sucrose and
maltose [9, 1351 (compare Table 4). Furthermoe, the per-
meation of glucose through LamB could be blocked by
maltodextrins [I 351, and LamB was identified as a cell-surface-
binding site for sugars in vivo [8, 1761. LamB is obviously not
a general diffusion pore which means that the early reconsti-
tution data and the finding that LamB acts as a general dif-
fusion pore in vivo may be caused by an unidentified general
diffusion pore. However, the permeation of small solutes with
molecular masses less than 300 Da through LamB seems to
be possible.
The exact function of the MalE protein (the maltose bind-
ing protein) is not completely clear at present although it is
essential for maltose uptake [177, 1781. It has been reported
that MalE binds to LamB [179] and that this is important for
voltage-dependent pore function [1621. However, more recent
investigations
of
the interaction between MalE and LamB in
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
11/19
11
Table
4. Stability constants.
K ,
(half-saturation constant K,
= l / K )
, fbr
fh e binding of different sugars to the La m B channel, and relative
ra te q fpermeation , P,
of
the sugars through L am B relative to that
01
maltose
The data for the relative rate of permeation (adjusted to
100
for
maltose) were taken from
[9].
The L amB-containing liposomes were
added
to
buffer solutions containing
40
m M of the corresponding test
sugars. The da ta for the sugar binding were taken fro m [145]. K (=
1
K m )was calculated from titration experiments similar to tha t shown
in Fig. 8 for the
Tsx
channel
Sugar K S P
M- ' mmol / l
Maltose 100 10 I00
Maltotriose
2 500
0.40 66
Maltotetraose 10000
0.10 19
M
altopentaose
17000 0.059 -
Maltohexaose 15000 0.067 -
M altoheptaose
5000 0.067 2.5
Trehalose
Lactose
Sucrose
Gentibiose
Melibiose
Cello biose
D-Glucose
L-Glucose
D-Galactose
D-Fructose
D-Mannose
46 22 76
I 8 56 9
67 15 2.5
250
4.0
42
180 5.5
33
6.7 150 13
9.5 110 290
22 46
24 42 225
1.7 600 135
6.3 160 160
-
Stachyose
20 50 < I
Raffinose
46 22
vivo
[180] and
in
vitro [154, 1611 have shown that this modu-
lation of the pore function is likely to be an artifact.
Lipid bilayer experiments in the presence of purified LamB
resulted in a strong conductance increase caused by the forma-
tion of small ion-permeable pores with a 10-times-smaller
single-channel conductance (160 pS) than OmpF under other-
wise identical conditions [145, 154, 1611. The ion permeation
through the pores could be blocked completely by the addition
of increasing concentrations of sugars, i.e. by titration exper-
iments [145, 154, 1611. Because of the absence of any interac-
tion between binding site and ions, the half-saturation con-
stant could be calculated from the t itration experiments. The
structure of the binding site was studied by the investigation
of
the binding of large variety of sugars (included in Table 4).
The data for the binding of the sugars allowed the deduction
of a rough steric model of the binding site [145]. The half-
saturation of the binding of the maltodextrins decreased with
their chain length, but was saturated after five residues
(Table 4, see also Fig. 7). This result makes it very likely that
the binding site has a length of approximately five glucose
residues, corresponding to a maximum length of 3 nm. Mu-
tations of LamB [181] can change the half-saturation constant
of the binding between the sugars and the site in the channel.
This is shown for a mutant (BW 1333,Trp74 Arg; R. Benz,
A .
Schmid, T. Ferenci and T. Nakase, unpublished results) in
Fig.
7.
The effect of the mutation may be explained by a
reduction of the number of binding sites inside the channel.
The maltodextrins apparently move through the channel in
single file and block the channel for the passage
of
ions. From
this it can be concluded that the channel is only 0.7 nm wide,
0
2
4
6
8
glucose
res idues
Fig. 7. Binding constants of glucose and diffe rent maltooligosaccharides
to the binding site inside the La m B channel
[I451
and inside La mB
m u ta nt ( B W 1333 Trp74 Arg ) as derived fr o m titration of ion
conductance with increasing concentrations of sugars.
wt, wild-type
La mB; BW
1333,
mu tant (Benz, R., Schmid,
A
Ferenci, T., Nakae,
T., unpublished results). Note that the half-saturation constant
of
Lam B for maltohexaose is identical to th at derived by a completely
different m ethod
[I761
whereas the diameter of the general diffusion pores of E .
coli outer membrane is of the order of 1.1 nm (see Table 1).
Further studies on mutant LamB and of the investigation of
the influence of the mutations on the sugar binding may give
further insight into the three-dimensional structure of the
binding site and may help to decide the question of which
amino acids are involved in the binding of the sugars [140,
181
-
841. The cation selectivity of LamB makes it very likely
that carbonyls are present inside the channel [145]. This means
that hydrogen bonds are presumably responsible for the bind-
ing of the sugars. It is interesting to note that the data of
Table 4, i.e. the combination of the relative swelling rate and
the half-saturation constant, can be used for the evaluation
of mechanism of sugar permeation through the LamB channel
on the basis of a one-site, two-barrier model [145]. The pro-
cedure is not discussed here and the interested reader is refered
to the original literature [145]. The result of the analysis is
similar to that shown in the next paragraph for Tsx protein
from
E.
coli
outer membrane. The important advantage
of
the binding site is the high turnover of molecules at very low
substrate concentration, although the pore saturates at high
substrate concentration. In contrast to this, the
flow
of mol-
ecules through a general diffusion pore is always a linear
function of the substrate concentration.
T sx
o E. coli
The pore function of another outer membrane protein,
the Tsx protein, was recently established [12]. This protein
was shown to serve as the receptor of colicin K and bacterio-
phage T6 [186]. It is known that Tsx protein
is
part of the
nucleoside uptake system because mutants lacking this protein
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
12/19
12
1.0
I
1
-
[Adenos ine] mM)
10
rnin
Fig. 8. Titrution o the Tsx-induced membrane conductance with increasing concentrations
of
adenosine. The membrane was formed from
diphytanoyl
glycoerophosphocholine/n-decane
in 1
M
KCI
and 200 ng/ml Tsx of
E. coli .
V
=
50 mV;
temperature
=
25°C
are impaired in the uptake of several nucleosides [187-1891.
The specificity of the nucleoside-specific pathway in the outer
membrane seems to be very high because the rate of uptake
of adenosine and thymidine was strongly reduced in these
tsx-
mutants whereas the uptake of cytidine was similar in
tsx’ and t s x - strains. Similar to the situation described above
for LamB and maltose uptake, transport by Tsx protein be-
comes rate-limiting at nucleoside concentrations below about
Tsx protein is not murein-associated like the ‘normal’
porins OmpF and OmpC and
i t
is only a minor outer mem-
brane protein under normal growth conditions. Furthermore,
it was found that the presence of SDS is deleterious for the
pore-forming activity in reconstitution experiments and for
the phage inactivation [12]. Tsx was isolated and purified from
outer membranes of an overproducing E.
coli
strain using the
detergent Triton X-100. It is interesting to note that, no matter
what the solubilization temperature, Tsx has the same molec-
ular mass of 28 kDa on SDS/polyacrylamide electropho-
retograms [12] whereas its correct molecular mass is
31.418 kDa
(E.
Bremer,
J .
Martinusson, P. Valentin-Hanson,
unpublished results). This observation is completely different
to the behavior of outer membrane porins of
E. coli
which
show temperature-dependent mobility on SDS gels as a conse-
quence of their SDS-resistant trimeric structure.
So
far it
remains unclear if the active Tsx channel is also an oligomer.
Reconstitution experiments with purified Tsx showed that it
is able to increase the conductance of lipid bilayer membranes
by many orders of magnitude [12]. This conductance could
be inhibited by the addition of nucleosides to the membranes
(Fig. 8) and a half-saturation constant could be calculated
for the binding of the nucleoside to the membrane from a
Lineweaver-Burke plot of the data of Fig.
8
(Fig. 9). This
result indicated that Tsx protein formed a channel which
contained a binding site for nucleosides. Surprisingly, the
single-channel conductance in 1 M KC1 was extremely small.
It was only 10 pS which can be compared with the 160 pS of
LamB and 1.9
nS
(1900 pS) of OmpF of E. coli K12 [158].
1 PM [187- 1891.
I
llbinding
I
I
llbinding
,
Adenos ine :
K =
0.5
rnM
’
-
5 -
1
3 -
2 -
I I
I
-l/fts 0 2
6
8
10
Fig.
9.
Lineweaver-Burke
plot
qf the data given in Fig.
5 .
The half-
saturation constant for the binding of adenosine to the binding site
inside the Tsx channel was calculated using the assumption that the
channel does not conduct ions when it is occupied by a nucleoside
The titration experiments could be used for the evaluation
of the half-saturation constant for the binding of different
nucleosides to the binding site inside the pore (Table
5)
[190].
As can be seen from Table 5, the stability constant for
adenosine binding was approximately the same as for
thymidine (corresponding to a similar half-saturation con-
stant). Surprisingly, the binding of the deoxynucleosides to
the channel interior was much stronger than that of the
nucleosides which may indicate that the binding of the mol-
ecules occurs via hydrophobic interaction between site and
nucleosides. Accordingly, a saturation of the flux of
nucleosides through the Tsx channel may be expected for
increasing concentrations of nucleosides [12, 1901.
If
we as-
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
13/19
13
Table 5. Stability constants K,, fo r the binding ofnucleosides to th e Tsx
channel (holf'saturation constant K,
=
1 j K )
The data were taken from [190]. K was calculated from titration
experiments similar to that given in Fig. 8
Nucleoside K K ,
M - ' m M
Adenine
500
2.0
Adenosine
2 000 0.50
Deoxyadenine
7100
0.14
Thymine
112
5.8
5 -Deoxyth ymidine 20000
0.050
Thymidine 5
000 0.20
l
0.01 0.1
1.0
10
100
=-.n./mM
Fig.
10. Flow
ofadenosine through Tsx and through a general diffusion
pore us
function qf
rhe
nucleoside concentration
on one
side
o j
the
pores.
The concentration on the othe r side of the pores is set to
zero.
The flow through Tsx,
4 ,
is given relative the maximum flux
For the flux of adenosine through the general diffusion pore, it was
assumed thai, at 0.1 m M , only
1%
of the flux was throu gh Tsx
sume that the concentration of adenosine on one side of the
Tsx channel is zero and that on the other side is c , the flux
4
of the substrate through Tsx as a function of
c
is given by the
following equation
:
4
=
4rnaxKc./(2+ Kc)
(4)
where is the maximum flux at very high substrate concen-
trations and
K
= l/ Km s the stability constant for the binding
of adenosine
to
the binding site (Table
5).
The flux saturates
at high substrate concentrations as shown in Fig.
10
for
adenosine. On the other hand, the flux of adenosine through
a general diffusion pore is linearly dependent on the concen-
tration (Fig. 10). This means that the relative throughputs of
the two types of pores is dependent on the substrate concen-
tration. Fig.
10
clearly shows that the flux of adenosine at
high concentrations through a general diffusion pore can ex-
ceed that through the specific Tsx channel. This is in spite of
the fact that in 0.1 mM adenosine, the flux through the general
diffusion pore is only 1 of that through a Tsx pore. This
result is consistent with the in
vivo
situation, in which the
general diffusion pores have an insignificant role at low
adenosine concentration. However, according to the data
given in Fig.
10,
we would expect the crossover of the two
throughputs curves to occur in response to an adenosine con-
centration between 10- 100pM, and not a below 0.1 pM [12].
It has been reported that
Tsx
also allows the flux of amino
acids [191]. Titration experiments with lipid bilayer mem-
branes showed that no binding site for amino acids or sugars
exists inside the channel [190]. This means that if Tsx acted as
pore for these substrates, it is a general diffusion pore for
them.
Porin
P
of Pseudomonas aeruginosa
The outer membrane of
P .
aeruginosa has very special
sieving properties which make this organism very resistant to
most antibiotics [192, 1931. Part of these sieving properties
could arise from the limited permeability of the large protein
F channel [136, 144, 1551. On the other hand, the role of
protein F as a porin has recently been questioned and the low
outer membrane permeability of P. aeruginosa was explained
by the presence of an as yet unidentified porin with a small
exclusion limit [131, 1321. Under conditions of phosphate
limitation, another porin (protein P) is induced together with
an alkaline phosphatase, a phosphate binding protein, and
some inner membrane proteins [lo, 11, 1941. The induced
system therefore has some similarities with the
p h o
system
of the Enterobacteriaceae and immunological cross-reactivity
exists between protein P and PhoE [195]. Protein P forms
highly anion-selective channels in lipid bilayer membranes
which a re the only
in vitro
system for the study of this channel.
Zero-current membrane potentials have a Nernstian slope
(corresponding to P,/Pa
<
I), and they indicated that the
protein P channel was at least
1000
times more permeable for
chloride than for potassium [lo,
11,
1591.
The conductance steps of the porin P channel were found
to be fairly homogeneous in size as compared with the steps
measured in the presence of the general diffusion pores [I591
(compare Fig. 6). Current-voltage curves linear up to 200 mV
were observed for the protein P channel. The cation present
in the aqueous salt solutions had no influence on the single-
channel conductance. In
100
mM chloride solutions this con-
ductance was always about 160 pS irrespective of the size and
charge of the cations. The change of the anion had a more
substantial influence on the single-channel conductance and
anion size could be used to probe the diameter of the protein
P channel [159].
Fig. 11 shows the dependence of the single-channel con-
ductance on the radii of different anions in
0.1
M salt solu-
tions. The diameter of the selectivity filter of the channel
was approximately between
0.5-0.7
nm, as estimated from
Fig. 11 and the observation that chlorate and bromate could
traverse the channel, whereas perchlorate and iodate could
not. The selectivity among the halides followed the Eisenmdn
sequence
AVI
which also indicated a closely spaced selectivity
filter [159]. The binding of anions to this filter accounts for
the concentration dependence of the single-channel conduc-
tance as shown in Fig. 12 for chloride and monovalent phos-
phate (pH 6) [lo] (and
R.
Benz, R. E.
W .
Hancock, unpub-
lished). The data
of
Fig. 12 could easily be fitted to the follow-
ing equation, which assumes single occupancy of the pore
[I591 (i.e. a single binding site inside the pore):
A c ) =
A,,,Kc/(I +KC).
(5)
The binding constant K (corresponding to the inverse half
saturation) and the maximum single-channel conduatance,
A,,, can be used to calculate the kinetics for the movement
of the different ions through the protein P channel. Fig. 12
8/9/2019 Trecerea moleculelor hidrofile prin membrana bacteriilor gram negative.pdf
14/19
14
200
100
A
PS
-
5 0 -
20
10
5 -
-
-
-
-
0.10
aa QZO
rlnm25
Fig. 11. Single-channel conductance o the protein P channel in 0.1 M
salt solutions as a func tion o the ionic radius o the anions. The
posi-
tively charged counterion was in all cases potassium
A l p s
10’
10
1
K I
I
I
I
1 10 10’ 10’ c l m M
Fig. 12.
Average single-channel conductance o protein P of
P.
aeruginosa given as a function
o
the KCI
(
x
)
and
o
the K H 2 P 0 4
concentration
0
in the aqueous phase. pH 6, temperature
=
25 “C
shows that this channel is not efficient at high phosphate
concentrations. However, under phosphate starvation (i.e.
phosphate concentrations below 1 niM) the channel conducts
phosphate better than chloride because of its smaller half-
saturation constant for phosphate (150 pM . This is another
clear demonstration of the advantage to be gained by the
possession of a binding site inside the channel for specific
substrates.
Protein
DI
of
P. aeruginosa
The addition of glucose to the grpwth medium of
P .
aeruginosa
resulted in the synthesis of an outer membrane
protein D1 with a molecular mass of 46 kDa [115, 1961. Pro-
tein D1 could be considered as a glucose-specific channel
because its presence resulted in a release of glucose from
liposomes [115]. The rate of penetration of different sugars
was measured using the liposome swelling assay (H. Nikaido,
personal communication). Comparison between protein F
and protein D1 clearly indicated that D1 is sugar-specific. The
diffusion of a variety of different sugars was facilitated by
protein D1 including that of L-glucose which binds also to
LamB [145] (and H. Nikaido, personal communication). This
indicated that sugar transport through protein D1 and
through LamB cannot be compared with that in epithelia
[197, 1981. This is because L-glucose is not transported in
epithelial cells but it binds to LamB [145] and is transported
through D1 (H. Nikaido, personal communication). The use
of a variety of different glucose analogs allowed the identifi-
cation of the groups responsible for the interaction with the
pore interior
(H.
Nikaido, personal communication).
As
in
the case of LamB, it seems that the interaction between the
sugars and the channel interior occurs with carbonyl groups
by means of hydrogen bonds.
ARE
T H E
PORIN PORES VOLTAGE-REGULATED?
Neither the vesicle permeability assay nor the liposome
swelling method allow the rapid application and measurement
of the effects of a transmembrane voltage. This is the unique
advantage of the lipid bilayer instrumentation. The first two
lipid bilayer studies of porin pores appeared as early as 1978
[146, 1471. These studies agreed on the size and selectivity of
the pores, but they disagreed about the voltage-dependence
of the pores because porin pores require a considerable voltage
for activation and inactivation when they are in folded mem-
branes [147, 1521. As a consequence, a voltage-control of the
outer membrane of gram-negative bacteria was postulated
which resulted in considerable confusion because many
microbiologists found the idea of voltage control of the outer
membrane permeability very attractive. However, in painted
membranes, porin pores were not found to be voltage-depen-
dent [146]. In addition, they become active in single steps and
not in large bursts as seen in folded membranes. It is generally
believed today that voltage control of outer membrane per-
meability in bacteria does not exist [17]. This is based on the
consideration of the physiology of the cells and on experimen-
tal results over the last ten years. Firstly,
in vivo
experiments
have demonstrated that the Donnan potential (up to 80-
100 mV [199, 2001) across the outer membrane has no influ-
ence on its permeability [17, 2001 and other potentials are