FEMS MicrobiologyReviews75 (1990) 429-446
Published by Elsevier
429
FEMSRE~I~
Bacterial periplasrnic permeases belong to a family
of transport proteins operating from Escherichia coli to human:
Traffic ATPases
G i o w m n a F e r r o - L u z z i A m e s , C a r o l S. M i m u r a and V e n k a t a k r i s h n a S h y a m a l a
Division of Biochemistry and Molecular Biology. Barker Hall University of California. Berkeley, CA. U.S.A.
Received16 March 1990
Accepted 10 April 1990
Key words: Periplasmic permeases; Traffic ATPases; ATP binding proteins; Energy coupling; Conserved
amino acid homology; Multidrug resistance
1. S U M M A R Y
Bacterial periplasmic transport systems are
complex permeases composed of a soluble substrate-binding receptor and a membrane-bound
complex containing 2 - 4 proteins. Recent developments have clearly demonstrated that these perincases are energized by the hydrolysis of A T E
Several in vitro systems have allowed a detailed
study of the essential parameters functioning in
these permeases. Several of the component proteins have been shown to interact with each other
and the actual substrate for the transport process
has been shown to be the liganded soluble receptor. The affinity of this substrate for the membrane complex is approximately 10 /~M. The involvement of ATP in energy coupling is mediated
by one of the proteins in the membrane complex.
For each specific permease, this protein is a member of a family of conserved proteins which bind
Correspondence to: G.F.-L Ames. Division ef Bi ~henu,;try
and Molecular Biology,Barker Hall, University of C',~ifornJ~
Berkeley, CA 94720, U.S.A.
ATP. The similarity between the members of this
family is high and extends itself beyond the consensus motifs for ATP binding.
interestingly, over the last few years, several
eukaryotic membrane-bound proteins have been
discovered which bear a high level of homology to
the family of the conserved components of
bacterial periplasmic permeases, Most of these
proteins are known to, or can be inferred to participate in a transport process, such as in the case
of the multidrug resistance protein (MDR), the
S T E 6 gen¢ product of yeast, and possibly the
cystic fibrosis protein. This homology suggests a
similarity in the mechanism of action and possibly
a common evolutionary origin. This exciting development will stimulate progress in both the prokaryotic and eukaryotic areas of research by the
use of overlapping procedures and model buildiqg.
We propose that this universal class of permeases
be called 'Traffic ATPases" to distinguish them
from other types of transport systems, and to
b;,gblio_J~t their involvement in the transport of a
vas, '..-qety of substrates in either d~rection relative to the cell interior and their use of A T P as
energy source.
0168-6445/90/$03.50 © 1990 Federation of European MicrobiologicalSocieties
430
2. I N T R O D U C T I O N
Gram-negative bacteria have a complex cell
surface, consisting of an outer membrane, ~he cell
wall or peptidoglycan, and a cytoplasmic membrane [1]. The compartment between the outer
and the inner cell membranes is known as the
periplasm. Small solutes cross the outer membrane
by way of water-filled channels and diffuse readily
through the peptidoglycan, a rigid layer permeable
to most molecules. The cytoplasmic membrane, on
the other hand, is impermeable to most watersoluble compounds, which therefore require specialized transport systems to cross it.
The transport systems (permeases) can be
broadly divided into two classes depending on
whether they use the electrochemical ion gradient
or substrate level phosphorylation for energization. The distribution of the perraeases according
to this criterion reveals that they share also other
fundamentally different characteristics. Electrochemical ion gradient-energized permeases are, as
typified by the fl-galactoslde (lacY) permease [2],
shock-resistant systems, usually composed of a
single, very hydrophobic membrane protein that
acts as a symport or an antiport, utilizing an ion
or a proton gradient. In contrast, periplasmic permeases, which are energized by substrate leve!
phosphorylation, have a complex composition.
These transport systems are also referred to as
shock-sensitive permeases because they are inactivated during osmotic shock, which releases
proteins located in the periplasm amounting to
about 15~ of the total cell protein. In all periplasmic permeases studied, osmotic shock releases
an essential component, i.e. a soluble protein that
binds the transported solute with high affinity.
Periplasmic permeases, acting on extremely disparate substrates (sugars, amino acids, phosphate
esters, phosphonates, peptides, ions, and vitamins)
have been characterized and their properties reviewed recently [3]. Their overall composition is
invariably the same, irrespective of the nature of
the substrate being transported. Clearly, the uniformly similar structural design of these permeases
can serve to transport vastly different substrates.
Periplasmic permeases are typically composed
of one periplasmic substrate-binding protein and
hlltidine
Fig. 1. Schematic representation of a periplasmic permease as
detailed for the histidinc permcase. The three membrane components (HisQ, HisM, and HisP) are represented as forming a
complex within the cytoplasmic membrane. The periplasmie
binding protein (His.l) changes conformation upon binding
histidine and then interacts with the membrane complex. Since
the architecture of this complex has not been fully elucidated
yet, the interaction is shown as occurrin8 with the complexas a
whole; a direct contact between HisJ and HisQ has been
proven biocbemica~ly(see text). The squibble indicates the
involvementof ATP in energy coupfing.The histidine molecule
can penetrate the outer membrane through non-specifichydrophilic pores. Larger molecules may require specific pores (as
the case of maitodextrins), in some periplasmlc permca.u~ the
known membrane components are only two or as many as
four.
three membrane-bound components. This overall
organization is schematically represented in Fig. 1,
using the histidine permease as a model system.
The outer membrane is represented as containing
pores which allow entrance of the substrate into
the periplasm, where it is bound by the binding
protein (His J). The membrane-bound components
(HisQ, HisM, and HisP) form a complex within
the cytoplasmic membrane. Genetic evidence suggests a direct interaction between the periplasmic
binding protein and the membrane-bound proteins; biochemical data to be discussed later support the genetic evidence. The genetic structure of
the most extensively characterized permeases is
very similar [3]. In all cases a single operon (or
two divergent operons) contains all the genes cod-
431
ing for the known transport components. In some
cases the operon contains additional genes whose
function is either unknown at present (e.g. livL in
the branched-chain amino acid permease, [4]) or is
involved in further catabolism of the transported
substrates (e.g. rbsK codes for a ribose kinase,
[51).
It seems then, that complexity is the norm for
this particular mode of transport. These permeases
typically concentrate substrates inside the cell
against a very large gradient (e.g. 10S-fold in the
case of maltose; [6]); perhaps achieving and maintaining such large concentration gradients requires
a complex mechanism, possibly in relation to energy coupling. Alternatively, the high efficiency of
transport which these systems usually display requires a complex structure. They are in fact able
to scavenge solutes from very low concentrations:
the apparent KmS for uptake range from 0.01 #M
to 1 /tM [6-8]. By comparison, the apparent K m
for the transport of lactose through the shock-resistant, monocomponent, /~-galactoside permease
(lacY) is 190 /tM [9]. Such high efficiency may
constitute an evolutionary advantage for these
bacteria, at least where amino-acid transport is
concerned, since biosynthetically produced amino
acids can be lost from the cell and these high-affinity permeases are able to recapture the lost
amino acid [10].
An interesting aspect of some periplasmic systems is the fact that the membrane-bound complex can be multifunctional, i.e., it can be used for
transport of several unrelated substrates and for
that purpose it is utilized by more than one binding protein. For example, in the case of the histidine permease, the membrane-bound proteins are
also essential for transport of arginine via the
lysine arginine ornithine-binding protein (LAO
protein, coded for by the argT gene [3,11]). This
means that whatever mechanism is deduced from
the available data, it will have to include for each
of the different periplasmic components their alternating interaction with and removal from the
same set of membrane-bound components. The
existence of alternative periplasmic binding proteins utilizing the same set of membrane-bound
components has also been shos"n for the
branched-chain amino-acid permease [4,12].
3. GENERAL STRUCTURE
The peripiasmlc binding proteins are the most
thoroughly analyzed of these transport components. Figure 1 depicts the histidine-binding protein, HisJ. The properties periplasmie binding proteins share as a group are summarized here (reviewed in [3,13], see references for individual cases
therein). They are monomeric with molecular
weights varying between about 25 000 and 56 000;
several are stable to heat; they have high affinity
for their substrates; they undergo a conformational change upon binding of substrate; and they
have two functionally and genetically separable
active domains. The latter three properties, discussed below, are the most enlightening with respect to the mechanism of action of these proteins.
The binding affinity (KD) is between 0.i and 1
#M for sugar substrates and around 0.1 #M for
amino acids. Presumably because of this high affinity, some of these proteins have been purified
with tightly bound substrates, which can be removed by reversible denaturation with guanidineHCI [14,15]. Binding proteins undergo a conformational change upon binding of substrate, as has
been measured in the case of the histidine-,
maltose-, arabinose-, ribose-, galactose-, glutamine- and leucine-isoleucine-valine-binding proteins by a variety of methods (see [3,13] for references). Since a substrate-induced conformatinnal
change occurs in all binding proteins analyzed, it
is likely to be an essential aspect of their mechanism of action.
X-ray crystallographic studies of several binding proteins, have shown that they are organized
into two globular domains (lobes) forming a cleft
and connected by a flexible hinge. The binding
site for the substrate is located in the concave
region between the two lobes. The molecule is
flexible, the cleft becoming somewhat narrower
upon binding of substrate [16,17], thus 'trapping"
the substrate deep within the protein (hence the
name of "Venus's flytrap' for this type of binding
mechanism) [18-22]. Thus, these X-ray results
corroborate the evidence that binding proteins
undergo a conformational change upon interaction with substrate. Once the substrate is trapped
in the liganded form of the protein, it is likely that
432
tl'ds more stable form requires an external stimulus to release the substrate, presumably supplied
by the interaction with the membrane-bound
transport components.
Thus, the first step in the action of these permeases involving the liganding of the substrate to
the periplasmic protein in the periplasm, results in
the formation of the actual transport substrate,
the liganded binding protein [23]. The practical
result of this first step is that the sol ~te to be
transported is presented in a concentrated form to
the membrane-bound complex, as a consequence
of the high binding affinity and very high concentration (in the order of raM) of the binding
protein in the periplasm. The affinity of the
iiganded binding protein for the membrane complex has been measured for the histidine permease
and shown to be around 10 /~M. This value is
considerably higher than that measured for the
transport process as a whole (apparent K,~:6
riM). This disparity is accounted for by the high
periplasmic concentration of the binding protein
and the relative amounts of the binding protein
versus the membrane complex [23].
The biochemical study of the membrane-bound
components has lagged far behind that of the
binding proteins, since they are more difficult to
study. Their number, between two and four, has
been determined for several permeases by genetic
or recombinant DNA techniques [3]. Two of the
components (when the total number is three or
four) are usually very hydrophobic [24-29]. In the
case of the histidine permease they are named
HisQ and HisM (Fig. 1). The remaining one(s)
(named HisP in Fig. 1) has an amino acid sequence that is not recognizably hydrophobic
[26,27,30], despite the fact that it is clearly, and
sometimes tightly, membrane-bound [31-33].
More will be said later about this component
(Section 5). The hydrophobic HisQ and HisM
proteins are accessible to protease digestion from
both the outer end the inner surface of the cytoplasmic membrane, thus indicating that they span
the membrane, as expected for integral membrane
proteins. Preliminary evidence suggests that both
proteins cross the membrane several times, with
the amino and carboxy termini located outside
and inside the membrane respectively, as estab-
lished by antibody and protease accessibility and
by Tn5-phoA fusion studies. Preliminary results by
crosslinking and antibody coprecipitation indicate
that the b_istidine permease membrane complex is
composed of one molecule each of HisQ and
HisM and two molecules of HisP (Kerppola and
Ames, in preparation). In cases where only 1 hydrophobic membrane component is known, the
complex may contain a homodimer of this protein. This view is supported by the finding that in
several cases the two hydrophobic components are
similar to each other, indicating a common evolutionary origin by duplication. This is the case for
example, for HisQ and HisM, whose predicted
topology is so similar that they have been postulated to form a heterodimer [34]. Along the
same lines, in cases where two hydrophilic membrane components are known (e.g. in the oligopeptide permease, [29]) rather than one, they may
exist as a heterodimer in the membrane complex,
thus maintaining the overall proportions in the
complex of two hydrophobic protein molecules for
each pair of hydrophilic membrane protein molecules.
4. P R O T E I N - P R O T E I N I N T E R A C T I O N S
WITHIN THE PERMEASE COMPLEX
Genetic data have been obtained which indicate the existence of direct interaction between the
binding protein and the membrane-bound complex, for both the histidine and the maltose permeases. In the case of the histidine permease a
mutant histidine-binding protein was isolated
which car,not function in transport despite the
fact that ~t has an intact histidine-bindingsite [35].
This suggested the existence of a region of the
protein which is essential for transport but not
necessary for binding histidine; this region was
postulated to be involved in interacting with the
membrane complex. Several mutations have been
characterized which suppress the mutation in the
binding protein, all located in the membranebound HisP protein ([36,37]; Speiser and Ames, L-I
preparation). This can be interpreted most easily,
but not exclusively, as a result of a direct interaction between the histidine-binding protein HisJ
and HisP. A comparison of the sequence of HisJ
433
with that of the closely related lysine-argirfineornithine-binding protein (LAd, the argT gene
product) lends support to the hypothesis that a
domain of the binding protein is specifically involved in interacting with the membrane complex.
Since both of these proteins require the same
membrane complex for function, they must be
very similar in their respective domains involved
in this interaction. Indeed, regions of these two
proteins are better than 90% homologous, which is
significantly higher than their overall level of homology, thus suggesting that these regions are
involved in forming the interaction domain. The
hisJ mutation mentioned above, which alters the
interaction domain of His.l, is indeed located in
one of these regions of very high homology [37,38].
Genetic evidence has also been obtained for the
existence of an interaction between the maltosebinding protein and the MalF and MalG proteins
of the maltose permease. Mutations were characterized in these genes which allow transport of
maltose in the absence of the binding protein; the
introduction into these mutants of a wild-type
malE gene (encoding the binding protein) inhibits
maltose uptake [39,40]. These r"esults have been
explained by postulating an interaction between
the maltose-binding protein, MalE, and MalF or
MalG. The binding protein-independent function
of the mutated membrane components has been
hypothesized to be inhibited by the presence of
wild-type binding protein, because it would result
in a non-productive and interfering interaction
with the altered membrane components. Evidence
that the maltose-binding protein has separate domains, for binding maltose and for interacting
with MalF and MalG, has not yet been obtained.
The direct interaction between binding proteins
and the membrane complex has been demonstrated bioehemically in the case of the histidii~e
permease, both in vivo and in vitro by two entirely
different erosslinklng methods [41]. The in vivo
method utilizes formaldehyde and has been appried to whole cells, in order to investigate protein-prot.ein interactions as they occur in vivo,
thus decreasing the possibility of artifactual architeetural alterations. A crosslinked product was
formed and shown by immunological methods to
contain both HisJ and HisQ. An entirely different
method was utilized in vitro to provide ad~tional
evidence of protein-protein interactions. A photolyzable crosslinking reagent (SUlfo-SANPAH) was
covalently reacted with pure HisJ. Addition of tl~s
derivatized protein to an in vitro reconstituted
vesicle system under conditions known to allow
active transport [23], followed by UV irradiat/on
resulted in a covalent bond between the His.l and
HisQ proteins ([41]; Prossrfitz and Ames, ira preparation). It was also confirmed by erosslink/ng
experiments that the mutant His3 postulated on
the basis of genetic data to be defective in its
interaction site indeed was erosslinked very poorly
to HisQ. Additional mutations have been introduced into HisJ by site-directed mutagenesis of
residues postulated to be located in the domain
that interacts with the membrane complex; the
decreased ability of the mutated proteins to interact with the membrane complex has been assessed, thus forming the basis for the initial characterization of the interaction domain in HAs.I
(Prossnitz and Ames, in preparation). Interestingly, unliganded H/sJ could also be erosslinked to
HisQ, though at least ten times worse than the
liganded form, indicating that the unl/ganded prorein has a conformation that still allows it to
interact with HisQ. Thus, with the avaflab/1/ty of
this biocherrfcal evidence supporting the genetic
implications, it is possible to conclude firmly that
HisJ and HisQ interact directly.
It is entirely possible that H/sJ makes dh'ect
contact also with HisM and HisP, but direct evidence is not yet available. The lack of a measurable cross!inked product between HJsJ and I-I/sP is
somewhat puzzling, in view of the genetic data
presented above, which strongly suggested such an
interaction. If the interaction occurred in a hydrophobic environment it might not be accessible to
our crosslinking reagents; alternatively, the genetic
evidence may reflect an indirect interaction between I-IisJ and HisP.
5. A FAMILY OF CONSERVED TRANSPORT
PROTEINS INVOLVED IN ENERGY COUPLING
Besides the hydrophobic components, all per/plasrnie permeases contain a protein which is usu-
434
ally thought to be peripheral. These peripheral
membrane components form a family of related
proteins that share extensive homology. We will
refer to these proteins as the 'conserved' components. An alignment of several of the sequences of
these conserved proteins has been presented in a
review [3]. Additional sequences which appeared
since are: oppF [29], ugpC [42], araG [43], fhuC
[44,45], chlD [,~5], glnQ [47], sfuC [481, proV
[49,50], btuD [51,120]. The similarity amongst
these sequences is extensive, including several long
stretches of amino acid residues, presumably involved in the performance of functions which are
common to the entire family. In particular, two
regions of extensive similarity share homology with
several other proteins known to bind ATP, such as
the a- and fl-subunits of the proton-translocating
ATPase, myosin, adenylate klnase, RecA protein
and others [52,53]. This homology strongly suggested that the transport proteins also bind ATP.
Indeed, both I-IisP and MalK have been shown to
carry a nucleofide-binding site by using the photoaffinity label 8-azido-ATP [54]. Similar results were
obtained with the RbsA protein (Hermodson, personal communication). Competition of 8-azidoATP labelling with a variety of nucieotide-containing compounds suggested that ATP a n d / o r
GTP are the natural substrates of HisP [54]. The
ATP analog 5'-p-fluorosulfonylbenzoyladenosine
was also shown to react with the HisP protein
(Hobson and Ames, unpublished results) and with
a chimeric lac fusion derivative of OppD [55].
Since nucleotide-binding activity has been demonstrated for several of the permease proteins, it was
reasonable to generalize that all conserved membrane components bind a nucleotide and that such
function is essential for all periplasmic systems,
probably involving the energy-coupling mechanism [54,55]. Indeed, this finding revived the interest and stimulated research in the mode of energy
coupling of these permeases, as reported below.
Besides the ATP-binding motifs, several additional regions of homology are also present. No
information is yet available regarding their function. Several individual residues, in particular
gly~:ine, are highly conserved. A secondary structure prediction of I-lisP [56] indicates that the
conserved glycines occur at join points between
a-helices and l-sheets or at predicted turn points.
A consensus sequence obtained from an alignment
of the known conserved components (16 sequences) was used to predict secondary structure.
This prediction indicates that the secondary structure has been well preserved. The structure features the nucleotide-binding fold characterized by
five fl---, turn-* a motifs found in many nucleotide-binding proteins (C.S. Mimura et al., in preparation).
The location of the conserved components
within the membrane is of particular interest. Their
ability to bind ATP strongly suggests that they
face the cytoplasm. Indeed HisP was shown to
present at least one domain to tb¢ cytoplasmic
side as shown by protease accessibility experiments (Kerppola and Ames, in preparation). On
the other hand, in the case of the histidine permease, the existence of several hisP mutations
that correct the defect in the interaction domain
of HisJ strongly suggests that HisP also has a
domain accessible to the periplasm. Preliminary
evidence that HisP may be accessible to proteas~
digestion from the exterior surface of the membrane supports this hypothesis (Kerppola and
Ames, in preparation). It is interesting that HisP is
relatively resistant to protease digestion from the
cytoplasmic side of the membrane, as compared to
HisQ and HisM. This may indicate that I-lisP is
physically protected from digestion by HisQ and
HisM. In the case of the corresponding conserved
protein, OppF, from the oligopeptide permease,
accessibility from the periplasmic face could not
be demonstrated [33]. Since the converse, i.e. the
demonstration of the accessibility of OppF to
protease digestion from the cytoplasmic side was
performed in the presence of detergent, its cytoplasmic orientation remains in question in this
case [33].
6. ENERGY COUPLING
6.1. Conflicting evidence
Until recently there has been considerable disagreement concerning this particular aspect of
transport (reviewed in [57,58]). It had been postulated that ATP or some form of phosphate-bond
435
energy is responsible for powering periplasrrfic
permeases [59,60]. This conclusion had been
achieved on the basis of experiments performed
with intact cells which had been starved to
eliminate endogenous energy sources and which,
in addition, were either treated with a variety of
metabolic poisons or were defective in the
proton-translocating ATPase. The various treatments, meant to affect differentially and specifically various metabolic steps, aiming at lowering
eitl~er the ATP level or the proton motive force,
gave conflicting results. The better awareness we
have today of the complicated interrelationships
between proton motive force and a variety of cell
functions, suggests that it is unlikely that a simple
relationship exists between addition of an inhibitor and its unique effect on proton motive force or
ATP level, and thus explains the conflicting resuits obtained in this early work.
More confusion was generated by results directly conflicting with the hypothesis that ATP is
the energy source, and rather implicating the proton motive force as the energy coupling mechanism. One set of experiments demonstrated that
under conditions where the ATP level of cells is
unchanged, but the proton motive force is decreased by the addition of valinomycin plus K +,
the activity of the glutamine permease (a periplasmic permease) is decreased. This was taken to
indicate (1) that ATP is not sufficient to power
transport, and (2) that the proton motive force
plays a role (directly or indirectly) in periplasmic
transport [61]. In agreement with Plate's results,
Singh and Bragg showed that periplasnfic permeases are functional only under conditions in which
a proton motive force is expected to be generated
[62]. However, some of these experiments should
be reinterpreted today taking into consideration
the fact that a proton motive force could have
been built up by way of the proton-conducting
ATPase activity [63]. On the other hand, against
an involvement of the proton motive force it was
shown that proton translocation during transport
by several periplasmic systems does not occur [64].
An interesting set of data implicated acetylphosphate [65] or a compound derived from it as
energy source for periplasmic permeases [66].
Tbesc data suffered from the same sources of
ambiguity as the above experiments, since the
correlation between transport and acetylphosphate
concentrations was indirect. Unfortunately, the
mutants used in these experiments were still capable of partial acetyl phosphate synthesis and
therefore capable of producing ATP. Later it was
shown that under conditions in which arsenate
inhibited gl~tamine transport very strongly, the
levels of both ATP and aeatylphosphate were normai or only slightly decreased. Therefore, it was
suggested that not acetylphosphate, but a compound derived from it, was the true energy-coupling factor [66].
Additional critical discussions concerning energy coupling in shock-sensitive systems are available [3,57,58,67-70].
6.2. Energization is by ATP hydrolysis
The notion that ATP is responsible for energy
coupling received some support from the finding
that several members of the conserved fan~ly bind
ATP and its analogs, as shown by affinity labelring of several of these proteins with ATP analogs.
These data, while not proving the involvement of
ATP in energy coupling, gave a strong impetus
into this aspect of research. Four different experimental systems, one in vivo and three in vitro,
were used in our laboratory to prove tiffs point.
(i) The fh'st one [71] used whole cells lacking
the entire proton-conducting ATPase (unc deletion mutants) and thus unable to interconvert the
two important pools of energy, ATP and proton
motive force, to manipulate separately the levels
of ATP and of the proton motive force without
the use of metabolic poisons. Using both the
histidine and maltose permeases, it was shown
that upon dissipation of the proton motive force
while maintaining a high ATP pool, transport was
unaffected. This result clearly demonstrates that
the proton motive force is unnecessary. Starved
cells with low ATP and high proton motive force
were unable to transport, thus showing that the
proton motive force, besides being unnecessary, is
also not sufficient and supporting the contention
that ATP is the energy source. Importantly, the
whole cells system was used to demonstrate that
the simultaneous inhibition of the proton motive
436
force and histidine transport by valinomycin plus
K + (i.e., the result obtained by Plate) was artifactual, since there was no such effect on maltose
transport. The artifact is probably due to specific
inhibitory effects on the histidine and glutamine
permeases resulting from an increase in the internal pH following the valinomycin treatment. Indeed, agents such as valinomycin/K+ and TPP +
are known to cause side effects [72]. Thus, dissipation of the membrane potential by these particular
methods cannot be used for studying the encrgetits mechanism of periplasmic permeases.
To demonstrate the involvement of ATP, its
level needs to be reduced. However, lowering the
ATP pool in whole cells with arsenate [73] unavoidably also lowers the pool of several other
energy-rich molecules. Arsenate-treated cells indeed lose the ability to transport histidine and
maltose in direct proportion to the drop in ATP
levels, while the proton motive force remains unaltered (thus supplying additional evidence for the
non-involvementof the proton motive force) [71].
Thus, these data are suggestive, but cannot be
considered satisfactory to draw a final conclusion,
since the arsenate treatment is likely to have additional effects, such as a direct inhibition of
permea~s. Indeed, an alternative method of
lowering the ATP level, gave interesting results.
The ATP pool was depleted drastically by creating
an idle cycle of ATP consumption. This is a very
mild treatment involving no addition of poisons
[74]: no proportionality could be seen between
ATP levels and transport and there was essentially
no change in transport levels, despite a decrease in
ATP to less than 10% of the normal level. These
results indicate that a very low concentration of
ATP is sufficient to energize transport, and that
arsenate has additional effects.
(ii) The second system consists of right-side-out
membrane vesicles reconstituted with added binding protein and energized with ascorbate and
phenazine methosulfate, or v-lactate [23]. This
system was initially developed using the glutamine
[67] and the galactose periplasmic [75] permeases.
We demonstrated that addition of ascorbate/
phenazine methosulfate or lactate allows the creation of a proton motive force which can be converted into ATP by the residua~ proton ATPase
activity. As a consequence: periplasmic transport
is energized. It was shown that unc mutants could
not transport and various treatments lowering or
raising the ATP pool correspondingly inhibited
and increased transport. The postulated involvement of acetylphosphate was excluded by the use
of mutants completely lacking acetylkinase and
phosphotransacetylase, which were shown to be
unaffected in their transport ability [23]. An interesting modification was introduced in the study of
maltose transport which used a mutant of the
maltose permease that could not fully secrete the
mahose-binding protein, thu~ producing vesicles
that carried the binding protein 'stuck' in the
membrane [76]. Despite the important information contributed by reconstituted right-side-out
vesicles, this system cannot be easily manipulated
to control accurately the level of ATP in order to
determine its hydrolysis concomitant with transport, and it is still metabolically too complicated
to be useful for detailed energetic studies. Thus,
the results obtained with the histidine permease
[23] and those with the maltose permease [76] can
be taken as strongly suggestive that ATP is the
energy source, but they cannot in any way be
considered conclusive.
(iii) Very strong support that ATP is indeed the
direct energy source and that its hydrolysis is
necessary was obtained for the histidine permease
with a novel inside-out vesicles system that corttained the histidine-bindingprotein trapped internally [77]. This system allows the controlled and
direct presentation of the normally impermeable
energy substrate to the epergy-coupling side of the
membrane. Addition of ATP caused histidine
translocation from inside out; this movement corresponds to inward transport in intact cells. A
non-hydrolyzable analog of ATP (AMPPNP) was
inactive, snggestJng the need for ATP hydrolysis.
Pretreatment of the vesicles with 8-azido-ATP and
UV fight (thus inactivating the nucleotide-binding
protein, HisP [54], eliminated transport, as expected. With this system it was also possible to
establish that the affinity of ).he permease for ATP
is about 200 pM. This value explains the ability of
whole cells to transport when the ATP level is
lowered 10-fold, which renders it about 300 pM
[711.
437
(iv) A reconstituted proteoliposome system finally resulted in the incontrovertible demonstration that ATP is the energy source. Proteoliposomes were formed utilizing a solubilized partially
purified histidine permease membrane complex
and E. coli phospholipids [78]. It should be
stressed that the success in developin8 this system
owes a great deal to the prior studies on the
reconstitution of the lac [79] and of anion-exchange permcases [80]. Also essential was the understanding of the important parameters of periplasmic permeases function, such as the nature of
the liganded binding protein as the true transport
substrate [23], the elimination of the proton motive force and acetyl phosphate as energy sources
[23,71], and the knowledge of the appropriate concentrations of liganded binding protein and of
ATP needed [23,77].
Using reconstituted proteoliposomes, it was
shown that internally trapped ATP allowed active
uptake upon addition of liganded binding protein.
The reconstituted system was entirely dependent
on all four permease proteins and ATP. Dissipators of the membrane potential had absolutely no
effect on transport. The demonstration that ATP
was hydrolyzed only concomitantly with histidine
transport finally confirmed incontrovertibly that
ATP drives solute transport in these permeases. It
should be, however, mentioned that GTP could
replace ATP reasonably well.
The proteoliposome reconstituted system functions with an efficiency that is comparable to that
obtained in reconstituted right-side-out vesicles:
0.55 mol of histidihe transported per min, per mol
of HisP, as compared to 1 tool per rain per tool of
HisP in vesicles. Both values are comparable to
those obtained for whole cells. Thus, the proteoliposome system must reflect reasonably well the
conditions in vivo.
At the time of writing, only the histidine permease has been reconstituted into proteoliposomes. The reconstitution of the maiiose permease
has been also achieved (quoted in [70]). It will be
useful to see whether the maltose permease behaves similarly to the histidine permease.
6.3. Stoichiometry of ATP hydrolysis
The stoiehiometry between ATP hydrolysis and
transport was preliminarily deterndned in the proteoliposomes system to be an average of 5 ATPs
per histidine. This value is most likely to be
artifactual and to be the result of damage infl/cted
upon the membrane complex during purification
and reconstitution resulting in sl/pping (uncoupling). A stoichiometry of one or two ATP molecules per histidine is more likely to be correct.
Several periplasmic permeases, or their eukaryotic
equivalents, have two ATP-bindingdomains, either
on the same protein or in the form of two separate
ATP-binding proteins [3,27,33], thus indicating
that two may be the correct value. Only from
careful studies in a weU-tuned in vitro system,
utilizing pure and undamaged proteins, and from
a variety of permeases, will an acceptable stoichiometric value be obtained. Data obtained in whole
cells inhibited with iodoacetate indicate a
stoichiometry of one to two for the maltose permease [81]. This observation is reassuring, but the
data should be interpreted with caution, since, as
discussed above, whole cells may have complicated metabolic routes for recycling ATP, despite and possibly because of the poisoning procedure. One possible problem is the known presence
of adenylate kinase [82], the activity of which can
only be eliminated by mutation.
7. TRANSPORT MODELS
From the overall information obtained with all
periplasmic permeases to date a general model can
be drawn using the histidine permease as an exam=
pie and building upon previous models [3] (Fig. 2).
The substrate (histidine) enters the periplasm and
encounters the binding protein (HisJ), that binds
it reversibly. The concentration of free substrate is
the same inside and outside the periplasm, but the
total concentration (bound plus free) is higher
inside and dependent on the binding protein concentration and on its affinity for the substrate.
Upon interaction with histidine, HisJ undergoes a
conformational change which increases its affinity
for the membrane-bound complex (composed of
proteins HisQ, HisM, and HisP). This interaction
(certainiy with HisQ, possibly also with HisM and
HisP) triggers conformational changes in the
membrane-bound apparatus. The substrate is re-
438
8. U N I V E R S A L I T Y O F T H E C O N S E R V E D
COMPONENT: RELATIONSHIP TO ENERGY
COUPLING
leased from the binding protein and a pore is
formed through the m e m b r a n e - b o u n d component(s), allowing passage of the substrate into the
cell. The A T P is hydrolyzed concomitantly, possibly as a consequence of the liganded binding
protein interaction with the m e m b r a n e complex.
The unliganded binding protein is released because of its decreased affinity for the membrane.
Alternatively, A T P hydrolysis may be needed to
cause the release of the binding protein a n d / o r to
close the pore. Since it has not been established
that substrates cross the m e m b r a n e through a
pore, an analogous model can be designed that
allows the formation or activation of substratebinding sites on the m e m b r a n e complex [83]. The
loaded site of the binding protein must be very
close (juxtaposed) to the pore or to the next binding site, thus not allowing release of the suhstrate
into the periplasm, which otherwise would negate
any special function of the binding protein itself.
/
Interestingly, it has emerged lately that numerous proteins, not necessarily involved with transport of small molecules, belong to the family of
the conserved components of periplasmic permeases [53,84-86]. A m o n g them, several eukaryotic
proteins are particularly interesting and will be
discussed individually.
8.1. Multidrug resistance
In 1986 the gene encoding the M D R protein
(P-glycoprotein) was sequenced (in two different
eukaryotic organisms: h u m a n and mouse) a n d
portions of it displayed a striking homology to the
sequence of the ATP-binding c o m p o n e n t of periplasmic permeases. W h a t business does a respec-
~
High affinity
LOWaffinity
Fig. 2. A speculative model for periplasmic transport. The components undergo a series of conformational changes initiated by the
binding of histidine to the periplasmic binding protein Hid. Tile liganded HisJ binds to the membrane complex. This interaction
occurs presumably with both. the hydrophobic membrane components (liisQ and HisM) and elicits a conformational change in the
ATP-binding membrane protein (HisP), thus causing ATP hydrolysis. Hydrolysis of ATP leads to the opening of a pore that allows
the unidirecUonal diffusion of the substrate to the interior. After the substrate has bccn released to the interior, an additional
conformational change (release of ADP?.)closes the pore. The unfiganded binding protein, having a poorer affinity for the membrane
complex, dissociatesfrom it. The membrane complex again binds ATP, ready to start the cycle. Alternatively, the substrate might be
transferred from the binding protein to specific binding site(s) on the membrane complex, from where it would then be finally
transferred to the interior, the rest of the cycle remaining the same.
439
table bacterial protein have being ha humans?
Since that finding several other eukaryotic proteins have been observed to display the same
homology. They all seem to have a transport function in common as summarized below. The development of resistance to cytotoxic agents in tumor
cells (Multi Drug Resistance: MDR; recent reviews cite the relevant literature: [87,88]) is one of
the causes of failure in cancer chemotherapy. The
problem is compounded by the fact that tumors
become resistant simultaneously to many of the
commonly used antineoplastic agents, most of
which are structurally unrelated. It is clear that at
least some aspects of this resistance have .t~.-,ir
basis in an altered ability to accumulate the drugs
intracellPtarly, specif~vally in the fact that resistant cells actively expel the internalized drug at
rates higher than the sensitive parent cells. The
biochemical basis of MDR lies in the overproduction of a large (170 kDa) phosphorylated membrane protein n;tmed P-glycoprotein (or gp 170)
which is presum,~l to be responsible for the active,
ATP-dependent excretion of the toxic agents. The
evidence for this model is strong, though still
somewhat indirect since the activity of the protein
in a purified in vitro system, such as reconstituted
proteoliposomes has not yet been demonstrated.
The MDR protein binds several of the toxic drugs
[89,901, it hydrolizes ATP [91], and membrane
vesicles containing elevated levels of P-glycoprorein transport the drugs in an ATP-dependent
manner [92]. Thus, amplification of the rode gene
and its genetic modification following exposure to
chemotherapeutic agents result in a lower internal
level of drugs and, therefore, in their rdat/ve ineffectiveness. The P-glycoprotein amino-acid sequence shows two homologous halves, each including a hydrophobic region and a hydrophific
region containing an ATP-binding motif [84,93,94].
The homology suggests that the two halves possibly arose by a gene duplication event followed by
fusion. Surprisingly, the hydrophilic region was
found to share strong homology with the bacterial
ATP-binding component of periplasmic permeases, including and extending beyond the consensus
ATP-binding motif, thus implying similar mechanisms of action. Interestingly, even the existence
of two ATP-binding moieties per transport com-
plex is shared with the bacterial systems. However, there is no homology between the prokaryofic hydrophobie transport componenL~ and
the hydrophobic moieties of P-glycoproteha.This
is not particularly surprising s/nee these components share rain/real homology even when b e
comparison is ILrnited to the prokaryotic co~aterparts. On the other hand, the protest,s may share
secondary structure homology as predicted by
computer modelling. Presumably these moieties
have extensively diverged during evolution towards special/zed functions, most 1/kely recogn/tion of different substrates a n d / o r substrate-binding proteins. It seems that P-glycoprotein might
have/ncorporated into one protein (as eukaryotes
often do) what are separate proteins in prokaryotes: the genes for the two hydrophobic components and for the ATP-binding protein of
bacterial perrneases have been fused into one gene,
which has then been duplicated and fused again.
However, an obvious problem is that of the relationship between the orientation of P-glycoprotein, the site of ATP hydrolysis, and the direction
of transport. The analogy with the periplasnfic
permeases does not appear to include a receptor
equivalent to the substrate-binding protein, nor
does it include the same directionality of lrnnsport. While a bacterial ~rmease transports in-'.,'~rdsutilizing the figanded binding protein as the
true substrate, the eukaryotic system transports
subs*rates outwards and there is no evidence for a
binding protein. Eukaryotic binding proteins nfight
exist [84]; an important function for such protehas
might be that of unparting specificity, as is the
case for the bacterial counterparts. In fact, particularly puzzling and pharmacologically troublesome is the fact that cells resistant to one drug to
which they have been exposed, are also res/stant
to seemingly unrelated drugs, thus resulting in the
multidrug aspect of the resistance phenomenon.
By analogy with the bacterial counterpart, it is
possible that receptors (soluble binding proteins?)
for different substrates channel them through the
same membrane machinery, as is the case for
several perrneases in bacteria [3]. ExtensDe
searches by various laboratories have not yet
turned up possible candidates. The identification
of such hypothetical receptor proteins is ira-
440
portant since it might give a clue as to the physiological substrate of the eukaryotic transport systems. In fact, since the drug transport may belong
to the class known as 'illicit transport', i.e. it
utilizes a permease designed for another purpose
[95], an unresolved issue is the function of these
transport systems in normal, untreated cells. Two
obvious possibilities are the secretion of physiological substrates [88,96-98] and a defense mechanism against toxic compounds naturally found in
the environment. Needless to say, understanding
the mechanism of action of P-glycoprotein would
be extremely helpful in devising strategies to overcome the problem of chemotherapeutic failures
due to overexpression of this protein.
8.2. Chloroquine resistance in the malarial parasite
Importantly, a P-glycoprotein-related phenomenon is the one responsible for the appearance of
chloroqniue-resistant derivatives of the malarial
protozoan parasite, Plasmodium falciparum. This
is an area of mammoth concern since 200 million
cases of malaria occur every year, accompanied by
2 million deaths per year [99]. The evolvement of
resistant parasites has strongly interfered with efforts at controlling malaria. Recently it has been
shown that the phenotype of chloroquiue resistance shares features with that of MDR, by exhibiting a lower level of accumulation of chloroquine [100]. Indeed, the sequence of a gene called
pfmdr, the expression of which is amplified in
some resistant strains, is so similar to mdr, including the hydrophobic membrane-spanning regions,
that it has been classified as belonging to the mdr
family (as opposed to the larger family o f these
related transport proteins) [101,102]. Thus, the
family also includes protozoan proteins.
8.3. Cystic fibrosis
The surprise over the discovery of the MDR
homology to bacterial proteins had barely abated
when similar evidence emerged concerning other
eukaryotic proteins. The most dramatic of these is
undoubtedly the recent identification of the cystic
fibrosis gene [103-105]. Cystic fibrosis is the most
common recessive caucasian disease (one in 2500
live births) which results in early deatll due to
multiple problems as a consequence of the
abnormal composition of the secreted mucus,
especially in the lungs. Damaged lung function
leads to repeated infections that eventually cause
death. It has been postulated that impaired secretion of CI- across the apical membrane of epithelial cells results in the production of thick
mucus and is responsible for the symptoms of
cystic fibrosis [106]. The isolation and sequencing
of the eDNA for the gene carrying the defect has
now supplied important information in this respect, since the product of this gene is clearly
another eukaryotic counterpart of periplasmic permeases. The predicted protein structure (referred
to as CFTR: cystic fibrosis transmembrane conductanee regulator) has a mass of 169 kDa, with
the familiar motif of two repeated halves each of
which consists of a hydrophobic portion comprisi~g several membrane spanners, and a hydrophilic
portion carrying the ATP-binding consensus.
These, of course, are also the characteristics of the
P-glycoprotein. The cystic fibrosis defect was
identified as a deletion of 3 bp resulting in the
deletion of a phenylalanine residue in the first
hydrophilic domain of CFTR. This alteration occurs in 68~ of cystic fibrosis patients, thus constituting the major mutation group [103]. From
genetic analysis, at least 7 additional mutations
can be expected in the subgroup or" the pancreatic
insufficient patients, none of which has yet been
identified at the molecular level.
Since it is not yet known what is the precise
mechanism of action of CFTR, we do not understand why deleting the phenylalanine residue affects the activity of CFTR. If CFTR is directly
involved in the transport of CI- ( a n d / o r water) it
should be possible to study this phenomenon in a
simplified in vitro system such as reconstituted
proteoliposomes [78]. On the other hand, CFTR
may be involved in transport indirectly, by regulating ion channel activity of a transport protein(s),
since ion conductance is thought to be regulated
by a cyclic AMP-dependent protein kinase (prorein kinase A) or by protein kinase [107,108]. In
agreem¢~_,'~, with this latter possibility is the fact
that recently purified proteins functioning as chlo-
441
ride channels appear dissimilar from C F T R [109].
However, since the related P-glycoproteins and, in
particular, the prokaryotic periplasmic permeases
are clearly involved in transport, it seems most
reasonable that C F T R is also a transport protein.
The phenylalanine deletion may have impaired the
binding a n d / o r hydrolysis of ATP or have altered
the overall structure sufficiently to interfere with
its activation by the kinases, or both. The region
where the phenylalanine resides is not located
within either of the two ATP-binding consensus
sequences, and is in fact located in a region of
relatively poorer homology with comparable regions of other proteins from the same family (see
e.g., alignments in (Fig. 2; [101,110]). This could
be taken as an indication that it is a specialized
region that has evolved uniquely in C F T R for
dealing with ion transport; in this case, the phenylalanine deletion may affect exclusively this function. It should be mentioned that in the case of
periplasmic permcases it has been clearly shown
that these systems do not use or require a membrane potential or a K + or H + electrochemical
gradient [71,78] and therefore do not involve ion
movement [64]. Nothing is yet known of the
coinvolvement of ion movement in the activity of
P-glycoprotein or of the STE6 protein from yeast
(see below).
Independently from the usefulness of having
identified the most common defect in C F T R for
an understanding of its mechanism of action, its
usefulness for genetic diagnosis in cases without a
previous family history is also enormous, since
anyone can now be screened for being a carder of
the most common mutation, and hopefully in the
near future, tar the other mutations. Newer therapeutic approaches, once the biochemical nature
of the defect is understood, might be possible
through the design of specific drugs and may be
within sight.
8. 4. Insect proteins
Another eukatyotic protein, the product of "-he
white locus in Drosophila melanogaster (and the
closely related brown protein; [111]) also clearly
belongs to this family, having a hydrophobic and
a hydrophilic moiety, the latter clearly related to
the ATP-binding family of transport proteins [112].
The location and function of its product is not
known with certainty, but it is probably membrane-bound and involves the transport of pigments [113]. Thus, one more eukaryotic
transport-related group of proteins, this time found
in insects, can be assigned to this family.
8.5. Yeast mating pheromone
But wonders are not over: the yeast Saccharomyces cerevisiae produces at least one protein that belongs to the family and this time it
secretes a qpopeptide~ the mating pheromone, afactor. Two recent publications describe the sequence of the gene, STE6, encoding this transporter [110,114]. This is the first eukaryotic case in
which an essential physiological function for a
transporter is known. Thus, it is an exciting addition to the family. It is also particularly exciting
because it contrasts with the known mechanism of
secretion of its counterpart pheromone a-factor,
which is secreted through the classical secretory
pathway as a much larger precursor that is proteolyzed in transit to its final size, a tridecapeptide.
On the other hand, production of a-factor, which
is 12 amino acids long and carries a farnesyl
moiety and a methyl group, involves intracelhilar
processing from a 36 or 38 amino acid precursor,
and secretion independently of the classical protein secretory pathway (see; [115]). Yeast MATa
cells depend on a functional STE6 gene to secrete
the a-factor: ste6 mutants cannot mate ('sterile')
because they cannot secrete a-factor. The deduced
sequence of the STE6 protein has the unmistakable characteristics of the transport family: two
homologous halves, each half including a hydrophobic domain and a hydrophi|ic ATP-binding
domain. The homology with P-~ycoprotein is extensive, including portions of the hydrophobic domains, besides the usually extensively homologous
hydrophilic domain. Since a-factor appears to be
synthesized and process~'i in the cytosol [116], it is
most likely that the STE6 protein c~,-ries out the
actual translocation of a-factor. Th~ finding that
an MDR-like transporter translocates a poly-
442
peptide raises again the question of the physiological function of M D R and of the uniqueness of the
classic secretory pathway for proteins. As suggested by the sequencers of S T E 6 , the strong
possibility arises that members of the M D R family
may indeed be involved in the secretion of some
proteins or peptides, via a route that is independent of the classical secretory pathway, especially
in ~he case of those secreted proteins that lack
recognizable signal sequences or other obvious
hydrophobic domains (for example, interleukin 1).
A more detailed discussion of this proposal appears in [110].
Finally, it should be mentioned that bacteria!
transport systems responsible for polypeptide
secretion (hemolysin in E. coil, hlyB: [1!7];
cyclolysin in Bordetella pertussis, cya B: [118]; and
microcin B17 in E. coli, m c b E a n d mcbF: [119])
can be viewed as a link between the family of
prokaryotic permeases for small substrates and
that of the eukaryotic transporters for outward
transport of high molecular weight substrates.
9. C O N C L U S I O N S
It is certain that we will see more and more
examples of members of this family of transporters. Since all their structures have included an
ATP-binding domain, since it is likely that they all
hydrolyze ATP, and since they are involved in
extensive traffic through the membrane in either
direction and of a vast variety of substrates, we
propose that they be named 'Traffic ATPases'.
The general term 'traffic' helps distinguish them
from other transport, usually ion ATPases (such
as Ca, Mg ATPase or proton-conducting ATPase)
and eliminates the problem of making reference to
a specific substrate or specific accessory characteristics, such as periplasmic components, which may
be unique only to some members of the family.
Exciting developments are certainly in store as
this remarkable example of the unforeseen merger
of vastly different areas of basic research enters
the next step of cross-fertilization.
ACKNOWLEDGEMENTS
The work performed in the authors' laboratory
was supported by National Institutes of Health
Grant DK12121. Thanks are due to all past and
present members of the laboratory for their contributions to this research.
REFERENCES
[1] Nikaido, H. and Vaara, M. (1985) Molecular basis of
bacterial outer membranepermeability.Microb. Rev. 49,
1-32.
12] K.~back, H.R. (1983) The lac carrier-protein in
Escherichia coil J. Membr. Biol. 76, 95-112.
[3] Ames,G.F.-L. (1986) Bacterialperiplasmictransport systems: structure, mechanism, and evolution. Annu. Rev.
Biochem. 55, 397-425.
[4] Landick, R., Oxender, D.L. and Ames,G.F.-L. (1985), in
The Enzymes of Biological Membranes, Voi. 3.
(Martonosi, A., Ed.), pp. 577-615. Plenum Press, New
York.
[51 Hope, J.N., Bell, A.W., Harmodson,M.A. and Groarke,
J.M. (1985) Ribokinase from Escherichia eoli K-12. J .
Biol. Chem. 261, 7663-7668.
[61 Szmelcman, S., Schwartz, M., Silhavy,T.J. and Boos, W.
(1976) Maltose transport in Escherichia coli K12. Eur. J.
Biochem. 65,13-19.
[7] Ames, G.F.-L. and Lever, J. (1970) Components of
histidine transport: histidine-bindingproteins and hisP
protein. Proc. Natl. Acad. Sci. U.S.A. 66,1096-1103.
[8] Rosenberg, H., Gerdes, R.G. and Chegwidden,K. (1977)
Two systems for the uptake of phosphate in Escherichia
coli. J. Bacteriol. 131,505-511.
191 Winlder, H.H. and Wilson, T.H. (1966) The role of
energy coupling in the transport of fl-galactosides by
Escherichia coli. J. Biol. Chem. 241, 2200-2211.
[IO1 Ames, G.F.-L. (1972), in Biological Membranes. Proceedingsof the 1972 ICN-UCLASymposiumin Molecu
lar Biology (Fox, C.F., Ed.), pp. 409-426. Academic
Press, New York.
[11] Kustu, S.G. and Ames,G.F.-L. (1973) The hisP protein,
a known histidine transport component in S.
typhimurium, is also an arginine transport component. J.
Bacteriol. 116,107-113.
1121 Landick, R. and Oxender, D.L. (1985) The comp!ete
nucleotide sequencesof the Escherichia coil LIV-BPand
LS-BP genes.J. Biol. Chem. 260, 8257-8261.
[131 Furlong, C.E. (1987) Osmotic-shock-sensitivetransport
systems, in Escherichia coli and Salmonella typhimurium
- Cellular and Molecular Biology (Neidhardt, F.C., Ingraham, J.L., Low, K.B. and Magasanik, B., Eds.), pp.
768-796. American Society for Microbiology,Washington.
443
[14] Miller, III, D.M., Olson, J.S. and Quiocho, F.A. (1980)
The tm~hanism of sugar binding to the periplasmic
ree~tor for galactose chemotaxis and transport in
F~cherichia coll. J. Biol. Chem. 255, 2465-2471.
[15] Miller, Ill, D.M., Olson, J.S., Pflugrath, J.W. and
Quiocho, F.A. (1983) Rates of ligand binding to periplasmic proteins involved in bacterial transport and
chernotaxis. J. Biol. Chem. 258,13665-13672.
[16] N©wcom~', M.E., Lc~s, B.A. and Quieeho, F.A. (1981)
The radius of gyration of L-arabinose-binding protein
d~'eases upon binding of ligand. J. Biol. Chem. 256,
13218-13222.
[17] Mao, B., Pear, M . P McCammon, J.A. and Quiocho,
F.A. (1982) Hinge-bending in L-arabinosc binding protein. J. Biol. Chem. 257,1131-1133.
[18] Gilliland, G.L. and Quiocho, F.A. (1981) Structure of the
L-arabino.~-binding protein from Escherichia co~7 at 2.4
,~ resolution. J. MOl. Biol. 146, 341-362.
[19] Saper, M.A. and Quiocho, F.A. (1983) l~ucine, isolencine, valine~binding protein from Escherichia coll. J.
Biol. Chem. 258,11057-11062.
[20] Vyas, N.K., Vyas, M.N. and Qulocho, F.A. (1983) The 3
resolution structure of a ~galactose-binding protein
for transport and cl~mok~is in Escherichia o~1i. Proc.
Natl. Acad. ,~i. U.S.A. 80,1792-1796.
[21] Mowbray, S.L and Petsko, G.A. (1983) The X-ray structure of the l~'~riplasmic galactose binding protein from
Salmonella tYlThimurium at 3.0-A resolution. J, Biol.
Chem. 258, 7991-7997.
[22] Pflugrath, J.W. and Quiocho, F.A. (1985) Sulphate
sequestered in the sulphate-binding protein of Salmonella
typhimurium is bound solely by hydrogen bonds. Nature
314, 257-260.
[23] Prosanitz, E., Gee, A. and Ames, G.F.-L. (1989) Reeonstitution of the histidine periplasmictransport system in
membrane vesicles. Energ3, coupling and interaction between the binding protein and the membrane complex. $.
Biol. Chem. 264, 5006-5014.
[24] Hisgins, C.F., Heag, P.D., Nikaldo, K., Ardeshir, F..
Gamin, O. and Ames, G.F.-L. (1982) Complete nucleotide sequence and identification of membrane components of the histidine transport operon of S. typhimurium.
Nature 298, 723-727.
[25] Frushaner, S. and Beekwith, J. (1984) The nuclentide
sequence of the gene for malF protein, an inner membrane component of the maltose transport system of
Escheriehia co//. 3. Biol. Chem. 259,10896-10903.
[26] Surin, B.P., Ruscoberg, H. and Cox, G.B. (1985) Phosphate-specific transport system of Escherichia coil:
nueleotide sequence and gane-polypoptide relationships.
J. Ba¢teriol. 161,189-198.
[27] Bell, A.W., Buckal, S.D., Groaske, J.M., Hope, 3.N.,
Kingsley, D.H. and Hermodson, M.A. (1986) The
nucleotide sequences of the rbsD, rbsA, and rbsC genes
of Escherichia coli K-12. J. Biol. Chem. 261, 7652-7658.
[28] Dassa, E. and Hcfnung, M. (1985) Sequence of gene
malG "m E. coli K-12: homologies between
integral
membrane components from binding protein-dependent
transport systems. EMBO J. 4, 2287-2293.
[29] Hiles, I.D., Gallagher, M.P., 3amieson, D.J. and Hi??~hx~.
C.F. (1987) Molecular charactsrLxation of the o~gopeptide permease of SalmoneUa ty/dffmurium. J. Mo|. Biol.
195,125-1,,2.
[30] Gilson, E., H i , i n s , C.F., Hofmmg, M., Ames, G.F.*L.
and Nikaldo, H. (1982) Extensive ~
between
membrane-associated components of histidine and
maltose transport systems of S. tyl~imurium and E. coll.
J. Biol. Chem. 257, 9915-9918.
[31] Ames, G.F.-L. and Nikakio, K. (1978) Identification of a
membrane protein as a histidine transport component in
S. typhimurium. Proc, Natl. Aead. Sci. U.S.A. 75, 54475451.
[32] Shuman, H.A. and Sillmvy, T3. (1981) Identification of
the malK gene product. $. Biol. Chem. 256, 560-562.
[33] Gallagber, M.P., Pearee, S.R. and Higom, C.F. (1989)
Identification and localization of the membrane-~sc~'iated, ATP-binding subunit of the oligopeptide permease
of Salmonella typhimurium. Eur. 3. Biocbem. 180, 133141.
[34] Ames, G.F.-L. (1985) The histidine transport system of
Salmonella typhimuri~, in Current Topics in Membranes and Transport, Volume 23, pp. 103-119.
Academic Press, New York.
[35] Kastu, S.G. and Ames, G.F.-L. (1974) The hisfidine-bincling protein J, a Idstidine transport component, has two
different functional sites. J. Biol. Chem. 249, 6976-6983.
[36] Ames, G.F.-L. and Spudieh, E.N. (1976) Protein-protein
interaction in transport: periplasr~dc histidine-blndin 8
protein 3 interacts with P protein. Proc. Natl. Acad. Sci.
U.S.A. 73,1877-1881.
[37] Shyamala, V. and Ames, G.F.-L. (1988) Amplification of
bacterial genomic DNA by the polyn~ras© chain reaction and direct sequencing after asymmmtric amplification: application to tim study of periplasmie pro'mens,.
J. Bacteriol. 171,1602-1608.
[38] Higgins, C.F. and Ames, G.F.-L. (1981) Two periplasmic
proteins which interact with a common membrane
tor show extensive homology: Complete nuclentid¢ sequenees. Proc. Natl. Aead. Sci. U.S.A. 78, 6038-6042.
[39] Tr~tow, N.A. and Shuman, H.A. (1985) Genetic offdenim for substrate and p¢.~riplasmie-binding-protein recognition by the MalF and MalG proteins, cytoplasmic
membrane components of the Eseherichia celi maltose
transport system. J. Bacteriol. 163, 654-660.
[40] Treptow, N.A. and Shumam H.A. (1988) Aflele-specific
malE mutations that restore interactions between
maltose-binding protein and the inner taembrane components of the maltose transport system. J. MUl. Biol. 202,
809-822.
[41] Prossnitz, E., Nikaido, K., Ulbrich, S.3. and Ames, G.F.L. (1988) Formaldehyde and photoactivatable crosslinking of the periplasmic binding protein to a membrane
444
component of the histidine transport system of
Salmonella typhimurium. J. Biol. Chem. 263, 1791717920.
[421 Overduin, P., Boos, W. and Tommassen, J. (1988)
Nucleotide sequence of the ugp, ,es of Escherichia coil
K-12: homology to the maltose system. J, Mol. Microbiol. 2, 767-775.
[431 Scripture, J.B., Voelker, C., Miller, S., O'Donnell, R.T.,
Polgar, L., Rade, J., Horazdorsky, B.F. and Hogg, R.W.H.
(1987) High-affinity L-arabinnse transport operon.
Nucleotide sequence and analysis of gene products. J.
Moh Biol. 197, 37-46.
[44] Burkhardt, R. and Braun, V. (1987) Nuclentide sequence
of the fhuC and .[had genes involved in iron(Ill) hydroxamate transport: domains in FhuC homologous to
ATP-bindin~ proteins. Moh Gen. Genet. 209, 49-55.
[45] Coulton, J.W., Mason, P. and Allatt, D.D. (1987) fhuC
and fhuD genes for iron(lll)-ferrichrome transport into
Escherichia coil K12. J. Bacterioh 169, 3844-3849.
[46] Johann, S. and Hinton. S.M. (1987) Cloning and nucleotide sequence of the chlD locus. J. Bacterial. 169, 19111916.
[47] Nohno, T., Saito, T. and Hang, J.-S. (1986) Cloning and
complete nuclcotide sequence of the Escherichia coil
giutamine pennear, e operon (ginHPQ). Moh Gen. Genet.
205, 260-269.
[48] Angerer, A., Gaisser, S. and Braun, V. (1990) Nucleotide
sequences of the SfuA, SfuB and SfuC genes of Serratia
marcescens suggest a periplasmic-binding-protein-dependent iron transport mechanism. J. Bacteriol. 172, 572578.
[49] Overdier, D.G., Olson, E.R., Erickson, B.D., Ederer,
M.M. and Csonka, L.N. (1989) Nucleotide sequence of
the transcriptional control region of the osmotically regulatea ProU operon of Salmonella typhimurium and identification of the 5' end point of the ProU mRNA. J.
Bacterial. 171, 4694-4706.
[50] Gowrishankar, J. (1989) Nucleotide sequence of the
osmoregulatory ProU operon of Escherichia earl. J.
Bacterial. 171,1923-1931.
[51] Friedrieh, M.J., DeVeaux, L.C. and Kadner, R.J. (1986)
Nucleotide sequence of the btuCED genes involved in
vitamin BI2 transport in Escherichia coil and homology
with components of periplasmic binding protein-dependent transport systems. J. Bacterioh 167, 928-934.
[52] Walker, J.E., Saraste, M., Runswick, M.J. and Gay, N.J.
(1982) Distantly related sequences in the a- and ~8-subunits of ATP synthase, myosin, kinases and other ATPrequiring enzymes and a common nucleotide binding
fold. EMBO J. 1, 945-951.
[53] Higgins, C.F., Hiies, I.D., Salmond, G.P.C., Gill, D.R.,
Dowuie, J.A., Evans, I.J, Holland, I.B., Gray, L., Buekel,
S.D., Bell, A.W. and Hermodson, M.A. (1986) A family
of related ATP-binding subunits coupled to many distinct biological processes in bacteria. Nature 323, 448450.
[54] Hobson, A.C., Weatherwax, R. and Ames, G.F.-L. (1984)
ATP-binding sites in the membrane components of the
histidine permease, a periplasmic transport system. Proc.
Natl. Acad. Sei. U.S.A. 81, 7333-7337.
[55] Higgins, C.F., Hiles, I.D., Whalley, K. and Jamieson,
D.K. (1985) Nuclcotide binding by membrane components of bacterial periplasmic binding protein-dependent
transport systems. EMBO J. 4,1033--1040.
[56] Novotny,J. and Auffray, C. (1984) A program for prediction of protein secondary structure from nucleotide sequence data; application to hlstocompatibility antigens.
Nucleic Acids Res. 12, 243-255.
[57] Ames, G.F.-L. (1990) Energeties of peripla~mic transport
systems, in Bacterial Energetics, (Krulwich, T.A., Ed.),
The Bacteria VoL 12, pp. 225-245.
[58] Ames, G.F.-L. (1990) Energy coupling in periplasmic
permeases: the histidine permease as a model system, in
Research in Microbiology, (Saier, Jr., M.H., Ed.), in
press.
[59] Berger, E.A. (1973) Different mechanisms of energy coupling for the active transport of praline and glutamine in
Escherichia coll. Proc. Natl. Acad. Sci. U.S.A. 70, 15141518.
[601 Berger, E.A. and Heppel, L,A. (1974) Different mechauisms of energy coupling for the shock-sensitive and
shock-resistant amino acid permeases of Escherichia carl.
J. Bioh Chem. 249, 7747-7755.
[611 Plate, C.A. (1979) Requirement for membrane potential
in active trznsport of giutamine by Escheriehia coll. J.
Bacterial. 137, 221-225.
[62] Sing,h, A.P. and Bragg, P.D. (1977) Energetics of galactose, praline, and giutamine transport in a cytochromedeficient mutant of Salmonella typhimurium. J. Suprareal. Struct. 6, 389-398.
[631 Singh, A.P. and Bragg, P.D. (1976) Anaerobic transport
of amino acids coupled to the glycerol-3-phosphatefumarate oxidoreductase system in a cytochrome-deficient mutant of Escherlchia coll. Biochim. Biophys. Acta
438, 450-461.
[64] Daruwalle, K.R., Paxton, A.T. and Henderson, P.J.F.
(1981) Energization of the transport system for arabinose
and comparison with galactose transport in Escherichia
eoli. Biochem. J. 200, 61!-627.
[651 Hang, J.-S., Hunt, A.G., Masters, P.S. and Lieberman,
M.A. (1979) Requirement of acetyl phosphate for the
binding protein-dependent transport systems in
Escherichia coll. Proc. Natl. Acad. Sci. U.S.A. "76, 12131217.
[66] Hang, J.-S. and Hunt, A.G. (1980) The role of
acetylphosphate in active transport. J. Supramol. Struct.
4, 77.
[67] Hunt, A.C. and Hang, J.-S. (1981) The reconstitution of
binding protein-dependent active transport of gintasuine
in isolated membrane vesicles from Escherichia coll. J.
Biol. Chem. 25~:, 11988-11990.
[681 Hunt, A.G. and Hang, J.-S. (1981) in Membranes and
Transport, Vol. II (Mastonosi, A.N., Fed.), pp. 9-13.
Plenum Press, New York.
445
[69] Hengge, R. and Boos, W. (1983) Maltose and lactose
transport in Escherichia coli: examples of two different
types of concentrative transport systems. Biechim. Bidphys. Acta 737, 443-478.
[70] Dean, D.A. and Nikaido, H. (1990) in Research in
Microbiology, (Saicr, Jr., M.H., Ed.), in press.
[71] Joshi, A., Ahmed, S. and Ames, G.F.-L. (1989) Energy
coupling in bacterial periplasmic transport systems: studies in intact Escherichia colt cells. J. Biol. Chem. 264,
2126-2133.
[72] Pooiman, B., HellingwcrL KJ. and Konings, W.N. (1987)
Regulation of the glutamate-glutamine transport system
by intracellular pH in Streptococcus lactis. J. Bactcriol.
16% 2272-2276.
[73] Klein, W.L. and Boycr, P.D. (1972) Energization of
active transport by Escherichia coil J. Biol. Chem. 247,
7257-7265.
[74] Galloway, RJ. and Taylor, B.L. (1980) Histidine starvation and adenosine 5'-triphosphate depletion in chemotaxis of Salmonella typhimurium. J. Bacterial. 144, 10681075.
[75] Rotman, B. and Guzman, R. 0984) Galactose-binding
protein-dependent transport in reconstituted Eseheriehia
co// membrane vesicles, in Microbiology 1984, (Leive, L.
and Schlessinger, D., Eds.), pp. 57-60. American Society
for Microbiology, Washington, DC.
[76] Dean, D.A., Fikes, J.D., Gehring, K., Bassford, Jr., PJ.
and Nikaido, H. (1989) Active transport of maltose in
membrane vesicles obtained from Escherichia coli cells
producing tethered maltose-binding protein. J. Bacteriol.
171, 503-510.
[77] Ames/G.F.-L., Nikaido, K., Groarke, J. and Petithory, J.
(1989) Reconstitution of periplasmic transport in insideout membrane vesicles: energization by ATP. J. Biol.
Chem. 264, 3998-4002.
[78] Bishop, L., Agbayarfi, Jr., R., Ambudkar, S.V., Maloney,
P.C. and Ames, G.F.-L. (1989) Reconstitution of a
bacterial periplasmic permease in proteoliposomes and
demonstration of ATP hydrolysis concomitant with
transport, pro¢. Natl. Acid. Sci. U.S.A. 86, 6953-6957.
[79] Newman, M.J., Foster, D., Wilson, T.H. and Kaback,
H.R. (1981) Purification and reconstitution of functional
lactose carrier from Escher,'hia coil J. biol. Chem. 256,
11804-11808.
[80] Ambudkar, S.V., Larson, T.J. and Maloney, P.C. (1986)
Reconstitution of sugar phosphate transport systems of
Escherichia col/, J. Biol. Chem. 261, 9083-9086.
[81] Mimmack, M.L., Gallaghcr, M.P., Pearce, S.R., Hyde,
S.C., Booth, I.P,. and Hisgins, C.F. (1989) Energy couplk~g to periplasmic binding protein-dependent transport
systems: stoichiometry of ATP hydrolysis during transport in vivo. Pro¢. Natl. Acid. Sci. U.S.A. 86, 8257-8261.
[82] Maloney, P.C., Ambt~&~ar, S.V. (1989) Functional reconstitution of prokaryote and eukaryote membrane proteins. Arch. Biocbem. Biophys. 269,1-10.
[83] Ames, G.F.-L. and Hisgins, C.F. (t983) The organization, mechanism of action, and evolution of periplasmic
transport systems. Trends Biochenl. Sci. 8, 97-100.
[84] Ames, G.F.-L. (1986) The basis of malt/drug resistance
in mammalian cells: homology with bacterial transport.
Cell 47, 323-324.
[85] Doolittle, R.F., Johnson, M.S., Husaim. L, Van Houten,
B., Thomas, D.S. and Sanea, A. (1986) Danmin~ evolution of a prokaryotic DNA repair protein and its relationship to active-transport proteins. Nature 323. 451453.
[86] Mount, S.M. (1987) Sequence similarity. Nature 325.
487.
[87] Endicott, J.F_ and Ling, V. (1989) 'l'V,e biochenfistry of
P-glycoprotein-mediated multidtxig ic,,~is~anc¢. Armu.
Rev. Biochem. 58, 137-171.
[88] Gottesman. M.M. and Pastan, 1. (1988) The mulddrng
transporter, a double-edged sword. £ Biol. Chem. 263.
12163-12166.
[89] Cornwen, M.M., GoResman, M.M., Pastan, I.H. (1986)
Increased vinblastine binding to membrane vesicles from
muhldrug-resistant KB cells. J. Biol. Chem. 261, 79217928.
[90] Comwell, M.M., Sara, A.R., Felsted, R.L., Gottesman,
M.M., and Pastan, |. (1986) Membrane vesic|es from
multidrug-resistant human cancer ~:Hs contain a speeific
150- to 170-kDa protein detected by photoaffmity labeling. Prec. Nad. Acid. Sci. U.S.A. 83. 3847-3850.
[91] Hamada, H. and Tsunm, T. (1988) Purification of the
170- to 180-kilodaiton membrane glycoprntein assc~ated
with maltidrug resistance. J. Biol. Chem. 263,1454-1458.
[92] Horio, M., Gottesman, M. and Pastan, L (1988) ATP-dependent transport of vinblasfine in vesicles from human
mulfidrug-resistant ceils. Proc. Natl. Acid. Sci. U.S.A.
85, 3580-3584.
[93] Chen, C.-J., Chin, J.E.., Ueda, K., Clark, D.P., Pastan, L,
Gottesman, M.M. and Roainson, L (1986) Internal duplication and homology with bacterial transport proteins
in the MDR1 (P-glycoprotein) gane from mulfidrug-resistant human cells. Cell 47, 381-389.
[94] Gros, P., Croop, J. and Housman, D. (1986) Mammalian
multidrng resistance gene: complete eDNA sequence indicates strong homology to bacterial transport proteins.
Cell 47, 371-380.
[95] Ames, B.N., Ames, G.F.-L., Young, .I.D., Tsuch/ya, D.
and Lecocq, 3. (1973) Illicit transport: the oligopeptide
permease, prec. Natl. Acid. Set. U.S.A. 70, 456-458.
[96] Fojo, A., Ueda, K., Slamon, D..L, Poplack, D.G., Gottesman, M.M. and Pastam L (1987) Expression of a muitidrug-resistance gene in human tumors, prec. Natl. Acid.
Set. U.S.A. 84, 265-269.
[97] Areeci, R.J., Croop, J.M, Horwitz, S.B. and Housman,
D. (1988) The gene encoding multidrug resistance is
induced and expressed at high levels during pregnancy in
the secretory epithelium of the uterus. Pro¢. Natl. Acid.
Sci. U.S.A. 85. 4350-4354.
[98] ThiebanlL F., Tsurno, T., Hamada, H., Gottesmnno M.M.,
Pastan, I. and wininglmm, M.C. (1987) Cellular localization of the multidrug-resistrmce gene product P-glycoprotein in normal human tissues, pro:. Natl. Acid. Sei.
U.S.A. 84, 7735-7738.
446
[99] Wyler, D.J. (1983) Malaria-resurgence, resistan.ce, and
research. New Engi. J. Med. 308, .~75-878.
[100] Krogstad, D.J., Gluzman, I.Y., Kyle, D.E., Oduola,
A.MJ., Martin, S.K., Milhoas, W.K. and Schlesinger,
P.H. (1987) Efflux of chlornqulne from Plasmodium
falciparum: mechanism of chloroquine resistance. Science
238,1283-1285.
[101] Foote, SJ., Thompson, J.K., Cowman, A.F. and Kemp,
DJ. (1989) Ampfification of the mnltidrug resistance
gene in some chloroquine-resistant isolates of P.
falciparum. Cell 57, 921-930.
[102] Wilson, C.M., Scrrano, A.F., Wasley, A., Bogenschutz,
M.P., Shankar, A.H. and Wirth, D.F. (1989) Amplification of a gene related to mammalia,-~ mdr genes in
drug-resistant Plasmodium falciparum. Science 244,
1184-1186.
[103] Kerem, B.-S., Rommens, J.M., Buchanan, J.A., Markiewicz, D., Cox, T.K., Chakravarti, A., Buchwald, M. and
Tsui, L.-C. (1989) Identification of the cystic fibrosis
gene: genetic analysis. Science 245,1073-1080.
[104] Riordan, J.R., Rommens, J.M., Kerem, B.-S., /don, N.,
Rw.mahel, R., Grzelczak, Z., Zielenski, J., Lok, S.,
Plavsic, N., Chou, J.-L., Drnmm, M.L., |armuzzi, M.C.,
Collins, F.S. and Tsoi, L.-C. (1989) Identification of the
cystic fibrosis gene: cloning and characterization of complementary DNA. Science 245: 1066-1073.
[105] Rommens, £M., lannuzzi, M.C., Kerem, B.-S., Drumm,
M.L., Meimer, G., Dean, M., Rozmahel, R., Cole, J.L.,
Kennedy, D., Hidaka, N., Zsiga, M., Buchwald, M.,
Riordan, J.R., Tsui, L.-C. and Collins, F.S. (1989) Identification of the cystic fibrosis gene: chromosome walking
and jumping. Science 245,1059-1065.
[106] Welsh, M3. (1987) Electrolyte transport by airway epithelia. Physiol. Rev. 67.1143-1184.
[107] Shoumacher, R.A.. Shoemaker, R.L., Halm, D.R., Tallant, E.A., Wallace, R.W. and Frizzeli, R.A. (1987) Phusphorylation fails to activate chloride channels from cystic
fibrosis airway cells. Nature 330, 752-754.
[108] Li, M., McCann, J.D., Liedtke, C., Nail-n, A.C., Greenpard, P., and Welsh, M.J. (1988) Cyclic AMP-dependent
protein kinase opens chloride channels in normal but not
eye;tic fibrosis airway epithelium. Nature 331, 358-360.
[109] Lan0~'y,D.W., Akabas, M.H., Redhead, C., Edelman, A.,
Cragoe, Jr. EJ. and AI-Awgati, Q. (1989) Purification
and reconstitution of chloride channels from kidney and
trachea. Science 244,1469-1472.
[110] Kuchler, K., Sterne, R.E. and Thorner, J. (1989) Saccharomyes cerevisiae STE6 gene product: a novel pathway for protein export in eukaryotic cells. EMBO $. 8,
3973-3984.
[111] Dreesen, T.D., Johnson, DJ. and Henikoff, S. (1988)
The Brown protein of Drosophila melanogaster is similar
to the White protein and to components of active transport complexes. MoL Cell. Biol. 8, 5206-5215.
[112] O'Hare, K., Murphy, C., Levis, R. and Rubin, G.M.
(1984) DNA sequence of the white locus of Drosophila
melanogaster. J. Mol. Biol. 180, 437-455.
[113] Suifivan, D.T. and Sullivan, M.C. (1975) Transport defects as the physiological basis for eye colour mutants of
Drosophila melanogaster. Biochem. Genet. 13, 603-613.
[114] McGrath, J.P. and Varshavsky, A. (1989) The yeast
STE6 gene encodes a homologue of the mammalian
multid~'ug resistance P-glycoprotein. Nature 340, 400404.
[115] Schekman, R. (1985) Protein localization and membrane
traffic in yeast. Annu. Rev. Cell Biol. 1,115-143.
[116] Sterne, R. (1989) A Hovel Pathway for Peptide Hormone
Biogenesis: Processing and Secretion of the Mating Phenomone, a-factor, by Saecharomyces eerevisiae. Ph.D.
Thesis, University of California, Berkeley.
[117] Felmlce, T., Pellett, S. and Welch, R.A. (1985) Nucleotide sequence of an Escherichia coil chromosomal hemolysin. J. Bacteriol. 163, 94-105.
[118] Glaser, P., Sakamoto, H., Bellalou, J., Ullmann, A. and
Danchin, A. (1988) Secretion of eyclolysin, the calmoduiin-sensitive adenylate cyclase-haemniysin bifunctional
protein of Bordetella pertussis. EMBO J. 7, 3997-4004.
[119] Del Carmen Garrido, M., Herrero, M., Kolter, R. and
Moreno, F. (1988) The export of the DNA replication
inhibitor, microcin B17, provides immunity for the host
cell. EMBO J. 7,1853-1863.
[120] Sirko, A., Hryniewicz, M., Hulanicka, D. and BtSck, A.
(1990) Sulfate and thiosulfate transport in Eseherichia
coli K-12: Nucleotide sequence and expression of the
CYSTWAM gene cluster. J. Bacteriol. 172, 3351-3357.