UvA-DARE (Digital Academic Repository)
Signal transduction in bacteria: phospho-neural network(s) in Escherichia coli?
Hellingwerf, K.J.; Postma, P.W.; Tommassen, J.; Westerhoff, H.V.
Published in:
FEMS Microbiology Reviews
Link to publication
Citation for published version (APA):
Hellingwerf, K. J., Postma, P. W., Tommassen, J., & Westerhoff, H. V. (1995). Signal transduction in bacteria:
phospho-neural network(s) in Escherichia coli? FEMS Microbiology Reviews, 16, 309-321.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),
other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating
your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask
the Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,
The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
Download date: 17 Jun 2017
ELSEVIER
MICROBIOLOGY
REVIEWS
FEMS MicrobiologyReviews 16 (1995) 309-321
Signal transduction in bacteria: phospho-neural network(s) in
Escherichia coli?
Klaas J. Hellingwerf a, *, Pieter W. Postma b, Jan Tommassen c, Hans V. Westerhoff d
a Vakgroep Microbiologie, E.C. Slater Instituut, BioCentrum Amsterdam, University of Amsterdam, Nieuwe Achtergracht 127, 1018 WS
Amsterdam, Netherlands
b Vakgroep Biocheraie/FS, E.C. Slater Instituut, BioCentrum Amsterdam, University of Amsterdam, Amsterdam, Netherlands
e Vakgroep Moleculaire Celbiologie, University of Utrecht, Utrecht, Netherlands
d Microbial Physiology, Free University, Amsterdam, Netherlands
Received 26 April 1995; accepted 26 April 1995
Abstract
The molecular basis of many forms of signal transfer in living organisms is provided via the transient phosphorylation of
regulatory proteins by transfer of phosphoryl groups between these proteins. The dominant form of signal transduction in
prokaryotic microorganisms proceeds via so-called two-component regulatory systems. These systems constitute phosphoryl
transfer pathways, consisting of two or more components. Most of these pathways are linear, but some converge and some
are divergent. The molecular properties of some of the well-characterised representatives of two-component systems comply
with the requirements to be put upon the elements of a neural network: they function as logical operators and show the
phenomenon of autoamplification. Because there are many phosphoryl transfer pathways in parallel and because there also
appears to be cross-talk between these pathways, the total of all two-component regulatory systems in a single prokaryotic
cell may show the typical characteristics of a 'phospho-neural network'. This may well lead to signal amplification,
associative responses and memory effects, characteristics which are typical for neural networks. One of the main challenges
in molecular microbial physiology is to determine the extent of the connectivity of the constituting elements of this
presumed 'phospho-neural network', and to outline the extent of intelligence-like behaviour this network can generate.
Escherichia coli is the organism of choice for this characterization.
Keywords: Escherichia coli; Two-component regulation system; Phosphoryl-transfer pathway; Neural network characteristics; Signal
transfer; Cross-talk;Auto-amplification
Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
309
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
310
* Correspondingauthor. Tel.: + 31 (20) 525 7055; Fax: + 31 (20) 525 7056; e-mail: [email protected]
0168-6445/95/$29.00 © 1995 Federationof European MicrobiologicalSocieties. All rights reserved
SSDI 0168-6445(95)00019-4
310
K.J. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
2. Two-component regulatory systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
311
3. Monomers versus dimers and the mechanism of transmembrane signalling . . . . . . . . . . . . . . . . . . . . . . . . .
313
4. Dephosphorylation of response regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
315
5. Auto-amplification in signal transduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
315
6. Cross-talk between two-component systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
316
7. Neural networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
316
8. Multiple phospho-neural networks? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
317
9. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
318
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
318
I. Introduction
To cope with the ever fluctuating environment,
microbial cells have to adjust their cellular make-up
to the extracellular conditions. This is particularly
true for unicellular organisms, which have a very
limited capacity to apply strategies in which the
immediate environment around the cell is subject to
a homeostatic regime. The molecular basis of many
of the specific responses to changing environmental
conditions have been described in the literature. Responses to stressful conditions, such as extremes of
temperature and nutrient deprivation [1], have received most attention. To adjust smoothly to these
stress conditions, many types of cells have developed
signal transduction systems that report to the cytoplasm aspects of the changes in the physico-chemical
conditions of the external world. In this manner,
entrance of the sometimes distressing signal-carrying
molecules into the cytoplasm is not necessary to
evoke a response (for review see [2]).
Biological signal transduction systems were discovered via the characterization, at the molecular
level, of the mechanism of action of hormones in
higher organisms. Up until the sixties, it was generally assumed (see e.g. [3] and [4]) that signal transduction across biological membranes would require
at least two proteins, each embedded - - from opposite sides - - into the hydrophobic interior of the
membrane. Through protein-protein interactions, the
two partner proteins were assumed to mediate signal
transfer. Since that time, our views have changed
drastically. It was argued on thermodynamic grounds
that such a topology for membrane proteins is unlikely to exist [5]. Simultaneously, it was demonstrated that hormone-induced signal transfer across
the cytoplasmic membrane in eukaryotic cells is
brought about by single intrinsic membrane proteins.
These proteins may have as few as only one transmembrane a-helix (e.g. the receptor tyrosine kinases) or many more, like seven in the best known
class of the eukaryotic signal-transducing proteins
from the cytoplasmic membrane, the so-called fladrenergic receptor family (or: the 7 a-helical bundle
family), of which eye-rhodopsin is the best characterized example. However, it has to be noted that
receptor oligomerization in the plane of the membrane is nowadays considered to be an integral part
of signal transduction through sensors with a single
transmembrane a-helix ([6]; see also below).
Systems with a similar molecular design were
subsequently described in unicellular eukaryotic microorganisms. Initially, the properties of these systems were considered to differ fundamentally from
those operative in prokaryotic microorganisms, in
their tactic responses to light and chemicals (i.e. in
phototaxis and chemotaxis, see e.g. discussion in
[7]). Again, this view has changed drastically. Signal
311
K.I. Hellingwerf et aL / FEMS Microbiology Reviews 16 (1995) 309-321
transfer in bacterial chemotaxis has turned out to be
one special manifestation of a general and abundantly present mechanism of signal transfer in bacteria [2,8], catalysed by so-called 'two-component regulatory systems'. These bacterial systems are composed of pairs of proteins (i.e. sensors and regulators), each with a characteristic domain structure and
with a very high degree of homology between corresponding domains. Representatives of these twocomponent systems play a role in signal transfer in
such complex and important, yet ill-understood processes as spornlation, competence development, virulence, response to nitrogen, phosphate or carbon
starvation, production of secondary metabolites, cell
division, etc., in a large variety of prokaryotic organisms.
Recently, the relevance of two-component regulatory systems also for sensing and signal transfer in
eukaryotic organisms was demonstrated explicitly.
First, it was reported that proteins, homologous to
the two basic components of a two-component regulatory system, do occur in eukaryotes [9-11] and
subsequently, the phosphoryl transfer underlying signal transfer through such components was demonstrated to be (part of) the basis of osmosensing in
yeast [12].
The generality of the mechanism of signal transfer, as it is mediated through the sensors of two-component regulatory systems, has perhaps been demonstrated most convincingly with the chimeric protein,
containing the sensing domain of a prokaryotic aspartate sensor (see below) and the cytosolic portion
of the human insulin receptor. This chimeric receptor
activated the insulin pathway of eukaryotic cells in
response to aspartate [13]. Indeed, signal transfer in
the prokaryotic two-component regulatory systems
Table 1
Some canonical two-component systems
Abbreviation
S e n s o r Regulator
Ntr
NtrB
NtrC
Omp
EnvZ
OmpR
Pho
PhoR
PhoB
Che
CheAc
CheY,CheB
Signal
Gln/a KG a
Osmolarity
Pi b
Attractantsand repellants
signal
1
ADP~~2~ADP
response
Fig. 1. Schematic representation of a canonical two-component
regulatory system. S and R refer to the sensor and regulator
component, respectively. The numbers indicate the sequential
steps in the (linear) phosphorylation pathway, referred to in the
text.
and eukaryotic signal transfer show significant similarity [14].
2. Two-component regulatory systems
Initial results of sequence analyses of regulatory
proteins from bacteria revealed a surprising relation-
Elicited response
Expression of glutamine synthetase
Expression of outer membrane proteins
Expression of high-affinity uptake system
Swimming behaviour
Reference
[29]
[28]
[30]
[31]
Gln/aKG, ratio of the glutamine and a-ketoglutarate concentrations in the cytoplasm.
b An as yet unknown signal is generated by limitation in the availability of inorganic phosphate.
c Actually a hetero-trimeric complex between a methyl-accepting chemotaxis protein, CheW and CheA forms the regulated histidine kinase.
a
312
K.I. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
ship between proteins that had never been shown,
nor thought, to be related beforehand (c.f. refs.
[8,15]). A striking example is formed by sets of
proteins involved in the regulation of nitrogen
metabolism in enterobacteria, in chemotaxis in Escherichia coli and in sporulation in Bacillus subtilis.
These proteins can be classified into two protein
families: the sensors and the regulators; each corresponding pair then forms the basic unit of a twocomponent regulatory system (see also Table 1).
More than 70 members of each family have been
identified until now (for recent reviews see
[2,14,16,17]).
The overall structure of these two components is
as follows (see Fig. 1): most sensors are intrinsic
membrane proteins with two or more transmembrane
a-helices in their N-terminal part. Their C-terminal
part forms an independently folded domain that extends into the cytoplasm, and can bind ATP and
phosphorylate itself (i.e. it shows autokinase activity).
This C-terminal domain is approximately 250 amino
acid residues in size and shows significant similarity
between all members of the sensor family. In this
domain, two signature sequences for nucleotide-binding can be discerned, as well as a conserved histidine
residue, that becomes phosphorylated through autokinase activity. This autokinase activity of the sensor
is modulated by the presence of the cognate signal
(molecule), which is often sensed by its extracytoplasmic N-terminal domain. Sensors thus have histidine kinase activity; most likely, autophosphorylation occurs through phosphoryl transfer between the
two halves of a dimer of sensor molecules [18-20].
As a rule, the regulators are cytosolic proteins,
composed of two independently folded domains.
Their N-terminal domain (approx. 125 amino acid
residues) exhibits the mutual similarity among the
regulators, while the C-terminal domain only shows
significant homology amongst certain subclasses of
regulators (for instance, those that function together
with a specific minor sigma factor). The N-terminal
domain of a regulator can be phosphorylated by
phosphoryl transfer from the autokinase domain of
the corresponding sensor. However, the actual enzyme activity, required for this phosphoryl transfer
and determining its specificity, appears to be located
in the regulator itself [21], rather than in the sensor.
In turn, this phosphorylation activates (the C-termi-
nal domain of) the regulator, which then may result
in either (i) stimulation of transcription of specific
genes (for which there are many examples), (ii)
regulation of the direction of rotation of flagellae
(like CheY ~ P) or (iii) modulation of enzyme activity (like stimulated esterase activity of CheB ~ P).
Little is known about the conformational transitions that underlie these signalling events. Current
thinking about the structure of sensors is guided by
what is known of the methyl-accepting chemotaxis
protein(s) from E. coli. The structure of the periplasmic domain of Tar, the methyl-accepting chemotaxis
receptor protein, that functions in the detection of
aspartate in chemotaxis in E. coli, has been determined by X-ray crystallography [22]. On the basis of
subsequent molecular modelling, a hypothetical
structure of the intra-membrane and periplasmic part
of this sensor has been proposed [23]. The conformational transitions in this structure during transmembrane signalling will be discussed below.
Thinking about regulator structure is guided by
the information available for CbeY. The structure of
the unphosphorylated form of this protein was resolved through X-ray crystallography [24]. CheY,
which essentially corresponds with only the N-terminal domain of the average regulator, has an a/[3barrel structure with 5 a-helices surrounding a 5stranded parallel fl-sheet. In the other regulators,
phosphorylation of this N-terminal domain must activate the independently folded C-terminal domain.
Recently, the structure of CheY and CheY ~ P were
compared through the use of 1H.NMR spectroscopy.
These studies revealed that, upon phosphorylation,
nearly half of all the resonances in the spectrum of
CheY were shifted to a different position [25], indicating that a major conformational transition is induced by the phosphoryl transfer. This large conformational change was anticipated, based upon the
involvement of a phosphorylated aspartate residue in
the regulators [2]. The so-called y-turn loop of the
regulator domains may form a hinge for this conformational transition, which moves regulators into the
signalling state, as was already predicted by Volz
[26], on the basis of sequence comparisons. Recent
NMR work has shown that the overall structure of
SpoOF, a B. subtilis regulator involved in sporulation, is similar to the structure of CheY [27].
Sequence comparisons have been pivotal in the
K.J. Hellingwerfet al. / FEMS MicrobiologyReviews 16 (1995) 309-321
identification of many of the members of the sensor
and regulator families and the elucidation of their
role in signal transfer. Simon and colleagues have
discussed the limits of this approach [16]. Once
identified, straightforward predictions can be made
as to the critical amino acid residues for functioning
of a particular sensor or regulator, and as targets for
site-directed mutagenesis. The biochemistry and
physiology of many of these (linear) phosphoryltransfer pathways has been described in considerable
detail, in particular of the Omp [28], Ntr [29], Pho
[30] and Che system [31]. Only a single example (i.e.
NifL of Azotobacter vinelandii) has been reported
until now in which such a predicted phosphorylation
site could not be confirmed in subsequent biochemical experiments [32].
For most of the two-component systems, we know
very little about the actual signal molecule (like for
EnvZ, which senses medium osmolarity somehow;
see also Table 1) that is/are perceived by the sensors. Of course, in this respect, too, there are exceptions, like the methyl-accepting chemotaxis proteins,
UhpBC (which detects glucose 6-phosphate [33]) and
FixL (which detects molecular oxygen [34]).
Variations on the basic theme outlined in Fig. 1
do occur: the sensor may be a soluble cytoplasmic
protein (like in the Ntr system [35]), and even when
the sensor is an intrinsic membrane protein, the
signal sensing domain may be located on the cytoplasmic rather than on the periplasmic side of the
membrane [36]. Also, the regulator may essentially
be a single domain protein (like CheY and SpoOF;
[2,37]) and a sensor and a regulator domain may be
combined into a single protein (like in FrzE [38]).
Particularly this latter class of sensors has gained
considerable importance.
During the last 2 years, it has been demonstrated
that components with homology to sensors and regulators also occur in eukaryotic cells, like in yeast [9]
and Arabidopsis [10] and maybe even in mammals
[16] (see also above). Amongst the two-component
sensors from eukaryotes, particularly the type with a
combined sensor and a regulator domain, within a
single protein, abounds.
The covalent coupling of a sensor and a regulator
domain may play an important role in regulating the
rate of phosphoryl transfer from a sensor to the
corresponding, but separate, regulator. Phosphoryla-
313
tion of the sensor-linked regulator domain may open
up the site containing the histidine-linked phosphoryl
group for access by the regulator. Also, ArcB from
E. coli conforms to this type of multiple-domain
protein [39]. For at least one prokaryotic representative of this atypical type of sensor (FrzE), a separate
regulator appears to be absent, which implies that for
this system, regulation of gene expression by FrzE
must take place at the membrane surface (just like
for instance for ToxR; see [40]).
In two-component systems like Spo, Pho and
Deg, the situation is even more complex. Some of
these phosphoryl transfer pathways diverge, because
more than one regulator is phosphorylated by a
single kinase (e.g. the Spo system in B. subtilis [41],
which is involved in the regulation of sporulation
and competence development). Others converge,
when more than one kinase phosphorylates a single
response regulator (like for instance PhoB [42], NarL
[43], RegA [44], LuxN from Vibrio [45], etc.). In
Myxococcus, the latest variant of a part of a twocomponent system has been discovered: a protein
containing only the transmitter domain of a sensor
and, at its N-terminus, a receiver domain of a regulator [46]. This protein may play a role in the proposed
phospho-relay system that leads to differentiation of
this organism [46].
In many organisms additional phosphoryl transfer
pathways play an essential role. The phosphoenolpyruvate-dependent carbohydrate phosphotransferase system (PTS) is a well known example (for a
review, see [47]). In this four-step phosphoryl transfer pathway a phosphoryl group from phosphoenolpyruvate is invested in the uptake and concomitant phosphorylation of carbohydrates. Some of the
PTS components show a weak similarity with sensors and regulators [15].
3. Monomers versus dimers and the mechanism of
transmembrane signalling
An important aspect of the functioning of sensors
is the mechanism of signal transfer across the membrane. The dominant secondary structure element of
sensors, intrinsic to the cytoplasmic membrane, is
the a-helix. Because of the rigidity of this latter
structure, it is very unlikely that signals can be
314
K.L Hellingwerfet al. / FEMSMicrobiology Reviews 16 (1995) 309-321
transferred through a single transmembrane a-helix.
In agreement with this notion, in eukaryotic signaltransducing receptors, which contain only a single
transmembrane helix (like the epidermal growth factor receptor), association/dissociation equilibria, laterally in the plane of the membrane, are at the basis
of signal transfer through the membrane [6]. In the
prokaryotic sensors, monomer-multimer transitions
do not appear to be involved in signal transfer
[49,50], be it that sensors do appear to function as
dimers, as it has been demonstrated for instance for
the methyl-accepting chemotaxis proteins (which
function together with pairs of CheA, mutually linked
through a pair of CheW; see e.g. [51]), as well as for
VirA [19]. Surprisingly, a dimeric structure does not
appear to be required for transmembrane signal
transfer through the sensors [52,53]. As the periplasmic and cytoplasmic domain of individual sensors
are often linked by two trans-membrane helices, it is
generally assumed that the molecular basis of signal
transfer across the membrane must be contained in
some form of mutual movement of these helices,
with respect to each other (for a recent discussion
see [48]). This can for instance be a piston-like, a
scissor-like or a rotational movement.
Binding of a signal molecule (e.g. aspartate) to
the Tar sensor induces a small rearrangement in its
periplasmic domain, as has been revealed through
crystallization of the periplasmic domain of Tar, with
and without aspartate, by D.E. Koshland's group
[23,54,55]. Binding of aspartate to Tar is negatively
cooperative, in E. coli even to such an extent that
crystals can be analysed with only a single aspartate
bound per dimer of the periplasmic domain of Tar
[55]. The extent of the transition, initially blurred by
binding of a sulfate anion to the aspartate-binding
site in an unoccupied periplasmic domain of Tar
[23], is very small. In Salmonella Tar, the average
displacement of amino acids around the two binding
sites in a dimer is about 1-2 A, which also explains
the negative cooperativity. Nevertheless, this transition induces a rotational movement of the two
periplasmic domains, relative to each other [48].
Since signals can also be relayed through sensor
monomers [53], most likely also the two helices,
mutually, of a single sensor within the membrane
rotate (see Fig. 2) [48,55]. Based on these results, it
is speculated that a rearrangement of the autokinase
o
®
®
Fig. 2. Conceptual models for the mechanismof transmembrane
signal transferwithin sensors. PartsA and B representa piston-like
and a scissor-likemovementof the two trans-membrane a-helices
of a single membrane-boundsensor, respectively.
domain results from this rotational transition, with
subsequent modulation of its activity. Cross-linking
experiments, too, have made it unlikely that a piston
model can explain the activation of the sensors [56]
and evidence has been presented (through helix-swop
experiments between methyl-accepting chemotaxis
proteins [57]) that the transmembrane helices indeed
may function as quite non-specific and inert rods.
One should keep in mind, however, that many of the
inferences about conformational transitions in transmembrane signalling actually are based only on measurements of (Tar) methylation, instead of on measurements of the rate of phosphoryl transfer to a
histidine kinase domain of a true sensor.
An as yet unexplored area of research is the
putative modulation of the intensity of signal transfer
through sensors by the size of the electrochemical
potential gradient of protons across the cytoplasmic
membrane. Observations on the functioning of sensory rhodopsin I from Halobacterium salinarium
[58] make it likely that indeed such modulation can
exist. This photoactive signal receptor protein is an
K.J. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
intrinsic membrane protein, but it does not translocate charge across the membrane. Nevertheless, its
turnover is modulated by the electrical potential gradient across the membrane, most likely because the
conformational changes during signalling are accompanied by rearrangements of charged or dipolar
groups in the electric field across the membrane.
Similar rearrangements may underly the transmembrane signal transfer through sensors.
Dimer formation is of utmost importance for the
functioning of regulators, too. This is true for transcriptional activators in general, which often are
functional in dimeric form (see e.g. [59,60]). Dimer
formation of activators is reflected in their target
promoter structures. For example, the pho box,
which is the binding-site for PhoB of phosphate
limitation inducible promoters, contains a two-fold
repeat of a 7-bp element [30]. Recently, it was
observed that a mutant form of OmpR, when forced
to form dimers through a mutation that introduces a
cysteine and which leads to disulfide bridge formation, is simultaneously activated for transcription of
ompC [61].
4. Dephosphorylation of response regulators
The pathway of phosphoryl-transfer within a
two-component system ends with a phosphorylated
response regulator. The in vitro stability of this
component, though quite variable amongst the regulators of two-component systems (ranging from seconds to hours), is typical for acyl-phosphates [62].
Also quite variable is the nature of the components
that are involved in the release of the phosphate
group from regulators. In the first place, response
regulators themselves may be involved in the catalysis of phosphate release. However, in many twocomponent systems, the sensor, too, stimulates phosphate release from the corresponding response regulator (like in Ntr, Omp, Uhp, etc.), particularly when
the former is in the unstimulated form [14]. This type
of nonlinear regulation allows an exquisite fine-tuning of the intensity of signalling, detected by the
sensor, to the degree of phosphorylation of the response regulator. In the Nar system, in which two
sensors govern a single regulator, both sensors also
have phosphatase activity in addition to kinase activ-
315
ity. Interestingly, the ratio of kinase over phosphatase activity of these two sensors is characteristically different [43]. In a limited number of systems,
specifically dedicated proteins function directly or
indirectly in the de-activation of the phosphorylated
response regulator. Examples are the P-II protein,
which stimulates the phosphatase activity of the sensor NtrB, and CheZ in the Che system, both in
enterobacteria.
Through these mechanisms, the in vivo stability
of phosphorylated response regulators may vary from
a few seconds (for CheB and CheY [63,64]) to
several hours, i.e. as long as several generation times
of the organism (for OmpR and VirG [65,66]; for
review see [14]). In combination with the amplification of the signalling machinery, which is observable
in some signal transfer pathways (see below), the
extremely long lifetime of a number of the phosphorylated response regulators provides the cell with
some sort of memory. As a consequence, the response of a cell upon excitation with a particular
stimulus will depend on the history of that particular
cell for that type of stimulus.
5. Auto-amplification in signal transduction
In some of the signal transfer systems, signalling
itself gives rise to an amplification of the amount of
the two components of that particular signalling system in the cell, via transcriptional activation. The
best characterized example in this respect is NtrB/C
[67]. Phosphorylation of NtrC, upon nitrogen deprivation, induces the synthesis, not only of glutamine
synthetase, but also of the signal-transduction components themselves, i.e. NtrB and NtrC. Consequently, upon maximal stimulation of signal transfer,
the amount of the signal transduction components in
the cell may increase up to ten-fold. A similar mechanism of amplification of the amount of signalling
components is present in the VirA/G system [68]
and in PhoR/B of E. coli [69,70]. As will be
discussed below, it is of utmost importance to determine in future work how widespread this phenomenon of autoamplification is amongst the ensemble of two-component systems and to determine for
each the dynamic range of this autoamplification.
316
KJ. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
6. Cross-talk between two-component systems
In view of the large degree of homology among
sensors and among regulators, it is not very surprising that phosphoryl-transfer pathways converge and
diverge. In in vitro experiments, it was observed that
a sensor from one particular system rapidly phosphorylated several different response regulators [71].
This has led to the notion that in vivo there may be
significant cross-talk in the signal transfer between
various systems that otherwise operate in parallel
(for review see [42]; note, however, that the term
'cross-talk' is used in the literature also to refer to
completely unrelated processes [72]). The notion of
the occurrence of cross-talk has recently been
strengthened by the observation that some sensors
(like ArcB) contain a second domain with a histidine
residue that can be phosphorylated [73]. In particular
this second histidine residue is active in non-cognate
phosphoryl transfer. Stock et al. [74] estimated that
on average any sensor-linked histidine kinase can
phosphorylate any regulator with a relative rate of at
least 1% of the rate of phosphorylation of its cognate
sensor. The total flux of phosphoryl groups will of
course also strongly be affected by the relative concentration of the components involved.
One may question the extrapolation of the in vitro
data on cross-talk to the in vivo situation. In many
cases, a putative low background rate of phosphorylation of a given regulator, by an unrelated sensor,
may be counteracted by the phosphatase activity of
the cognate sensor. For instance, cross-regnlation of
the Pho regulon by the sensor CreC was only observed in a mutant strain, lacking PhoR [42]. However, it is conceivable that under the appropriate
growth conditions such cross-talk does have a physiological meaning. In this respect, it is relevant to
note that additional evidence has been obtained that
supports this assumption. Several observations (e.g.
in the Ntr, Omp and Nar systems [75-77]) indicate
that the remaining activation of a particular response
regulator, after elimination of the cognate sensor
through mutation, is still meaningful for the cell
within the physiological context of the regulator's
signalling system (i.e. nitrogen starvation for Ntr and
osmotic stress for Omp). It may be more than just
coincidence that these examples of cross-talk were
reported for systems like Ntr, Nar and Pho, which
function in the center of intermediary metabolism. It
is therefore particularly relevant to further investigate the cross-talk between systems regulating the
response to carbon, nitrogen and phosphorus limitation. Each of these three limitations is known to lead
to the induction of not only a set of specific but also
of general stress proteins [1].
In addition to this cross-talk sensu strictu, sensor
and/or regulator proteins can also be phosphorylated
by metabolites from intermediary metabolism
(acetyl-phosphate, carbamoyl-phosphate, etc. [78] and
possibly y-glutamyl phosphate [79]). But in the rate
of phosphorylation by intermediary metabolites, the
different regulators show specificity, too. For instance, in contrast to most regulators, CheB is hardly
phosphorylated by acetyl-phosphate [74]. As a third
source, regulators might be phosphorylated by proteins involved in other phosphoryl transfer networks
in prokaryotes, such as the phosphoenolpyruvate
phosphotransferase system [80]. Such a transfer has
already been suggested to occur between the PTS
and chemotaxis (Che) proteins [81].
7. Neural networks
A network classifies as a neural network when it
conforms to a number of criteria [82]. These criteria
imply the presence of (i) more than one signal
transduction route, (ii) cross-talk between those
routes, (iii) logical operators at the cross-talk points
(i.e. AND, OR, NOR, etc., ports that give an output
that depends on the value of more than a single
signal), and (iv) amplification of signal transduction
elements in response to signal transduction through
those elements. The molecular basis of all these
properties differs between the various materializations of neural networks, of which the most accurately described examples are: (i) certain computer
programmes, (ii) the immune system of higher eukaryotes and (iii) the human brain.
Neural networks [83] are characterized by their
ability to generate a pattern of output signals (a
'response') as a consequence (and function) of a
pattern of input signals (a 'signal'), such that the
relation between the former and the latter pattern
depends on the signal transduction history of the
network. The latter dependence is such that a re-
K.J. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
sponse corresponding to a signal is achieved more
readily or extensively when similar signals have
been received in the past; i.e. the network is able to
'learn' (note the difference with 'memory'). This
learning may even take the form of association: a
network can be provided with a training set of
signals. After processing of these signals, the network subsequently will associate a similar response
to both these initial signals and to incomplete variants of the signals it was trained with. For instance,
if A and B are 'proper' responses, corresponding to
signals a and b, respectively, one may train a network by always giving the signals a and b in
combination. After such training, a neural network
will produce both A and B, even if only signal a is
offered.
Signal transduction in a single living cell has not
yet been identified as an example of a neural network, in part perhaps because it falls far short of the
highly sophisticated behaviour of the human brain.
Yet, essentially all the fundamental mechanistic requirements which classify a system as a neural network are present in prokaryotic signal transduction:
(i) a significant number of transduction routes operate in parallel, (ii) these routes are interconnected
through branching and cross-talk, and (iii) some of
their elements are present at concentrations that depend on previous signalling, i.e. they show autoamplification. All these properties are present in the
phosphoryl transfer through two-component regulatory systems. Future work on the physiological levels
of cross-talk and auto-amplification may reveal the
extent to which these properties bestow prokaryotic
signal transduction with (perhaps primitive) aspects
of neural network behaviour, such as memory and
learning.
8. Multiple phospho-neural networks?
A single E. coli cell may contain as many as 50
different two-component systems [2]. Through a specific regulation of expression, and because of their
high biochemical specificity, these systems primarily
form linear phosphoryl-transfer routes. Nevertheless,
additionally, these pathways mutually interact and
they interact with other phosphoryl transfer cascades,
such as the PTS (c.f. [47]) and phosphorylated inter-
s
317
hV
. . . .
I
,,
/
[,
OUTPUT (RESPONSE)'~'~KN
Fig. 3. Diagrammatic representation of the essential reactions that
lead to tile hypothesis that signal transfer in a prokaryotic organism shows neural network characteristics. S, signal; Z~p, proton
gradient; Ac-P, acetyl-phosphate; Carb-P, carbamoyl-phosphate;
and AGp, phosphorylation potential (of adenine nucleotides).
mediary metabolites. Furthermore, because the sensor/regulator combinations of individual phosphoryl
transfer pathways comply with all the prerequisites
for neural network elements, the picture emerges of a
'phospho-neural network' in each individual bacterial cell. Upon signalling from the environment,
input of phosphoryl groups into this network, primarily via sensors, is followed by transfer to various
output systems, involved in the regulation of gene
expression, of enzyme activity, of flagellar rotation
a n d / o r of cell division (see Fig. 3). This network
has neural network characteristics since e.g. the concentration of its components is linked to the intensity
of signal transfer through the network. Qualitative
(i.e. mechanistically) and quantitative resolution of
such a 'phospho-neural network' is of utmost importance for understanding the very essence of the
performance of a prokaryotic cell, and it is extremely
valuable as a model system to study regulation and
signal transfer in more complex systems.
An additional challenge is the determination of
the extent of connectivity of the constituting elements of this presumed 'phospho-neural network'.
This should reveal whether or not it is appropriate to
describe the sum of the phosphoryl transfer pathways
318
K.I. Hellingwerf et a l . / FEMS Microbiology Reviews 16 (1995) 309-321
in an individual prokaryotic cell as a single network,
or whether it is more relevant to distinguish more or
less 'isolated' subdomains in this network. E. coli is
the organism of choice for this characterization, because it has been well-characterized, both genetically
and physiologically [84], and because a large array of
mutants of this organism is available for detailed
molecular biological studies.
via chemical reduction. Independently, the phosphorylation level of a specific two-component system
may be 'read out' by a reporter enzyme, under the
control of the regulator of the system under study.
Through such studies, in the end, also our understanding of signal transduction in eukaryotic cells
may strongly benefit. In those cells, too, the problem
of a quantitative understanding of multiple, parallel,
overlapping, and interacting signal transduction pathways has to be solved [89,90].
9. Outlook
A major challenge for molecular microbial physiology in the coming years will be to derive equations
that quantitatively characterize the response of an
intact phospho-neural network to extracellular stimuli. Signalling via these phosphorylation cascades
often leads to changes in the level of gene expression. Such a description is therefore necessarily of a
hierarchical nature, which generates additional problems in modelling [85]. Expertise to derive such a
hierarchical model is emerging [86]. The quantitative
description of this network will strongly benefit from
these recent developments in the theory of control
analysis of regulatory networks. Of particular interest
in the phospho-neural network of a prokaryotic cell
is the low copy number of some of the components
involved, so that stochastic phenomena may come
into play [17]. Most studies performed up to now on
the biochemical and genetic aspects of signal transduction in prokaryotes have mainly concentrated on
signal transfer within individual two-component systems. The same applies to modelling studies of
signal transduction (see e.g. [87,88]). An approach in
which the interactions between signalling pathways
are characterized may reveal unexpected properties
[73].
For a quantitative basis of these modelling studies, it is of utmost importance to develop assays,
with proper accuracy and time resolution, for the
determination of the in vivo phosphorylation level of
a particular sensor or regulator, under physiological
conditions. This problem can be tackled via attempts
to separate the phosphorylated and non-phosphorylated forms of a sensor and/or regulator on a nondenaturing gel, in combination with antibody detection. Another approach may be to stabilize the linkage between the phosphate group and the regulator
References
[1] Matin, A. (1991) The molecular basis of carbon-starvationinduced general resistance in Escherichia coli. Mol. Microbiol. 5, 3-10.
[2] Stock, J.B., Ninfa, A.J. and Stock, A.M. (1989) Protein
phosphorylation and regulation of adaptive responses in bacteria. Microbiol. Rev. 53, 450-490.
[3] Robison, G.A., Butcher, R.W. and Sutherland, E.W. (1967)
Adenyl cyclase as an adrenergic receptor. Ann. N.Y. Acad.
Sci. 139, 703-723.
[4] Birnbaumer, L. (1990) Transduction of receptor signal into
modulation of effector activity by G proteins: the first 20
years or so. FASEB J. 4, 3178-3188.
[5] Chabre, M. (1987) The G protein connection: is it in the
membrane or the cytoplasm? Trends Biochem. Sci. 12, 213215.
[6] Schlessinger, J., Schechter, Y., Willingham, M.C. and Pastan, I. (1978) Direct visualization of binding, aggregation,
and internalization of insulin and epidermal growth factor on
living fibroblastic ceils. Proc. Natl. Acad. Sci. USA 75,
2659-2663.
[7] Hellingwerf, K.J. (1988) Phylogenetic relations between unicellular organisms and the mechanism of vertebrate signal
transduction. Trends Biochem. Sci. 13, 128.
[8] Nixon, B.T., Ronson, C.W. and Ausubel, F.M. (1986) Two
component regulatory systems responsive to environmental
stimuli share strongly conserved domains with the nitrogen
assimilation regulatory genes ntrB and ntrC. Proc. Natl.
Acad. Sci. USA 83, 7850-7854.
[9] Ota, I.M. and Varshavsky, A. (1993) A yeast protein similar
to bacterial two-component regulators. Science 262, 566-569.
[10] Chang, C., Kwok, S.F., Bleecker, A.B. and Meyerowitz,
E.M. (1993) Arabidopsis ethylene-response gene err1: similarity of product to two-component regulators. Science 262,
539-544.
[11] Brown, J.L., North, S. and Bussey, H. (1993) SKN7, a yeast
multicopy suppressor of a mutation affecting cell wall /3glucan assembly, encodes a product with domains homologous to prokaryotic two-component regulators and to heat
shock transcription factors. J. Bacteriol. 175, 6908-6915.
K.I. Hellingwerf et al. / F E M S Microbiology Reviews 16 (1995) 309-321
[12] Maeda, T., Wurgler-Murphy, S.M. and Saito, H. (1994) A
two-component system that regulates an osmosensing MAP
kinase cascade in yeast. Nature 369, 242-245.
[13] Moe, G.R., Bollag, G.E. and Koshland, D.E. Jr. (1989)
Transmembrane signalling by a chimera of the Escherichia
coli aspartate receptor and the human insulin receptor. Proc.
Natl. Acad. Sci. USA 86, 5683-5687.
[14] Parkinson, J.S. and Kofoid, E.C. (1992) Communication
modules in bacterial signaling proteins. Annu. Rev. Genet.
26, 71-112.
[15] Kofoid, E.C. and Parkinson, J.S. (1988) Transmitter and
receiver modules in bacterial signalling proteins. Proc. Natl.
Acad. Sci. USA 85, 4981-4985.
[16] Swanson, R.V., Alex, L.A. and Simon, M.I. (1994) Histidine
and aspartate phosphorylation: two-component systems and
the limits of homology. Trends Biochem. Sci. 19, 485-490.
[17] Parkinson, J.S. (1993) Signal transduction schemes of bacteria. Cell 73, 857-871
[18] Yang, Y. and Inoue, M. (1991) Intermolecular complementation between two defective mutant signal-transducing receptors of Escherichia coli. Proc. Natl. Acad. Sci. USA 88,
11057-11061.
[19] Pan, S.Q., Charles, T., Jin, S., Wu, Z.-C. and Nester, E.W.
(1993) Preformed dimeric state of the sensor protein VirA is
involved in plant-Agrobacterium signal transduction. Proc.
Natl. Acad. Sci. USA 90, 9939-9943.
[20] Ninfa, E.G., Atkinson, M.R., Kamberov, E.S. and Ninfa, A.J.
(1993) Mechanism of autophosphorylation of Escherichia
coli nitrogen regulator II (NR u or NtrB): trans-phosphorylation between subunits. J. Bacteriol. 175, 7024-7032.
[21] Lukat, G.S., McCleary, W.R., Stock, A.M. and Stock, J.B.
(1992) Phosphorylation of bacterial response regulator proteins by low molecular weight phospho-donors. Proc. Natl.
Acad. Sci. USA 89, 718-722.
[22] Jancarik, J., Scott, W.G., Milligan, D.L., Koshland, D.E. Jr.
and Kim, S.-H. (1991) Crystallization and preliminary X-ray
diffraction study of the ligand-binding domain of the bacterial chemotaxis-mediating aspartate receptor of Salmonella
typhimurium. J. Mol. Biol. 221, 31-34.
[23] Milburn, M.V., Privr, G.G., Milligan, D.L., Scott, W.G.,
Yeh, J., Jancarik, J., Koshland, D.E. Jr. and Kim, S.-H.
(1991) Three-dimensional structures of the ligand-binding
domain of the bacterial aspartate receptor with and without a
iigand. Science 254, 1342-1347.
[24] Stock, A.M., Mottonen, J.M., Stock, J.B. and Schutt, C.E.
(1989) Three-dimensional structure of CheY, the response
regulator of bacterial chemotaxis. Nature 337, 745-749.
[25] Lowry, D.F., Roth, A.F., Rupert, P.B., Dahlquist, F.W.,
Moy, F.J., Domaille, P.J. and Matsumura, P. (1994) Signal
transduction in chemotaxis. A propagating conformation
change upon phosphorylation of CheY. J. Biol. Chem. 269,
26358-26362.
[26] Volz, K. (1993) Structural conservation in the CheY superfamily. Biochem. 32, 11741-11753.
[27] Cavanagh, J., Zapf, J., Hoch, J.A., Feher, V., Dahlquist,
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
319
F.W. and Whiteley, J.M. (1994) Aspartyl phosphates in the
regulatory control of bacterial response. Amino Acids 6,
131-140.
Igo, M.M., Ninfa, A.J. and Silhavy, T.J. (1989) A bacterial
environmental sensor that functions as a protein kinase and
stimulates transcriptional activation. Genes Dev. 3, 598-605.
Ninfa, A.J. and Magasanik, B. (1986) Covalent modification
of the glnG product, NRI, by the glnL product, NRu,
regulates the transcription of the glnALG operon in Escherichia coli. Proc. Natl. Acad. Sci. USA 83, 5909-5913.
Makino, K., Shinagawa, H., Amemura, M., Kimura, S.,
Nakata, A. and Ishihama, A. (1988) Regulation of the phosphate regulon of Escherichia coli. Activation of pstS transcription by PhoB in vitro. J. Mol. Biol. 203, 85-95.
Hess, J.F., Oosawa, K., Kaplan, N. and Simon, M.I. (1988)
Phosphorylation of three proteins in the signaling pathway of
bacterial chemotaxis. Cell 53, 79-87.
Woodley, P. and Drummond, M. (1994) Redundancy of the
conserved His residue in Azotobacter vinelandii NifL, a
histidine autokinase homologue which regulates transcription
of nitrogen fixation genes. Mol. Microbiol. 13, 619-626.
Weston, L.A. and Kadner, R.J. (1988) Role of uhp genes in
expression of the Escherichia coli sugar-phosphate transport
system. J. Bacteriol. 170, 3375-3383.
Gilles-Gonzalez, M.A., Ditta, G.S. and Helinski, D.R. (1991)
A haemoprotein with kinase activity encoded by the oxygen
sensor of Rhizobium meliloti. Nature 350, 170-172.
Miranda-Rios, J., Sanches-Pescador, R., Urdea, M. and Covarrubias, A.A. (1987) The complete nucleotide sequence of
the glnALG operon of Escherichia coli K12. Nucleic Acids
Res. 15, 2757-2770.
Scholten, M. and Tommassen, J. (1993) Topology of the
PhoR protein of Escherichia coli and functional analysis of
internal deletion mutants. Mol. Microbiol. 8, 269-275
Trach, K.A., Chapman, J.W., Piggot, P.J. and Hoch, J.A.
(1985) Deduced product of the stage 0 sporulation gene
spoOF shares homology with the SPO0A, OmpR, and SfrA
proteins. Proc. Natl. Acad. Sci. USA 82, 7260-7264.
McCleary, W.R. and Zusman, D.R. (1990) FrzE of Myxococcus xanthus is homologous to both CheA and CheY of
Salmonella typhimurium. Proc. Natl. Acad. Sci. USA 87,
5898-5902.
Iuchi, S. and Lin, E.C.C. (1992) Purification and phosphorylation of the Arc regulatory components of Escherichia coli.
J. Bacteriol. 174, 5617-5623.
DiRita, V.J. (1992) Co-ordinate expression of virulence genes
by ToxR in Vibrio cholerae. Mol. Microbiol. 6, 451-458.
Burbulys, D., Trach, K.A. and Hoch, J.A. (1991) Initiation of
sporulation in Bacillus subtilis is controlled by a multicomponent phosphorelay. Cell 64, 545-552.
Wanner, B.L. (1992) ls cross regulation by phosphorylation
of Two-component response regulator proteins important in
bacteria? J. Bacteriol. 174, 2053-2058.
Schrrder, I., Wolin, C.D., Cavicchioli, R. and Gunsatus, R.P.
(1994) Phosphorylation and dephosphorylation of the NarQ,
320
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
K.J. Hellingwerf et al. / F E M S Microbiology Reviews 16 (1995) 309-321
NarX, and NarL proteins of the nitrate-dependent two-component regulatory system of Eseherichia coil J. Bacteriol.
176, 4985-4992.
Mosley, C.S., Suzuki, J.Y. and Bauer, C.E. (1994) Identification and molecular genetic characterization of a sensor kinase
responsible for coordinately regulating light harvesting and
reaction center gene expression in response to anaerobiosis.
J. Bacteriol. 176, 7566-7573.
Bassler, B.L., Wright, M. and Silverman, M.R. (1994) Multiple signalling systems controlling expression of luminescence in Vibrio harveyi, sequence and function of genes
encoding a second sensory pathway. Mol. Microbiol. 13,
273-286.
Plamann, L., Li, Y., Cantwell, B. and Mayor, J. (1995) The
Myxococcus xanthus asgA gene encodes a novel signal transduction protein required for multicellular development. J.
Bacteriol. 177, 2014-2020.
Postma, P.W., Lengeler, J.W. and Jacobson, G.R. (1993)
Phosphoenolpyruvate carbohydrate phosphotransferase systems of bacteria. Microbiol. Rev. 57, 543-594.
Kim, S.-H. (1994) 'Frozen' dynamic dimer model for transmembrane signaling in bacterial chemotaxis receptors. Protein Sci. 3, 159-165.
Falke, J.J. and Koshland, D.E. Jr. (1987) Global flexibility in
a sensory receptor: a site-directed cross linking approach.
Science 237~ 1596-1600.
MiHigan, D.L. and Koshland, D.E. Jr. (1991) Intra-subunit
signal transduction by the aspartate chemoreceptor. Science
254, 1651-1654.
Gegner, J.A., Graham, D.R., Roth, A.F. and Dahiquist, F.W.
(1992) Assembley of an MCP, CheW, and kinase CheA
complex in the bacterial chemotaxis signal transduction pathway. Cell 70, 975-982.
Nakashima, K., Kanamaru, K., Aiba, H. and Mizuno, T.
(1991) Signal transduction and osmoregulation in Escherichia coil A novel type of mutation in the phosphorylation domain of the activator protein, OmpR, results in a
defect in its phosphorylation-dependent DNA binding. J.
Biol. Chem. 266, 10775-10780.
Miliigan, D.L. and Koshland, D.E. Jr. (1988) Site-directed
cross linking. J. Biol. Chem. 263, 6268-627.
Scott, W.G., Milligan, D.L., Milbum, M.V., Priv6, G.G.,
Yeh, J., Koshland, D.E. Jr. and Kim, S.-H. (1993) Refined
structure of the ligand-binding domain of the aspartate receptor from Salmonella typhimurium. J. Mol. Biol. 232, 555573.
Biemann, H.-P. and Koshland, D.E. Jr. (1994) Aspartate
receptors of Escherichia coli and Salmonella typhimurium
bind ligand with negative and half-of-the-sites cooperativity.
Biochemistry 33, 629-634.
Falke, J.J., Demburg, A.F., Sternberg, D.A., Zalkin, N.,
Milligan, D.L. and Koshland, D.E. Jr. (1988) Structure of a
bacterial sensory receptor: a site-directed sulfhydryl study. J.
Biol. Chem. 263, 14850-14858.
Tatsuno, I., Lee, L., Kawagishi, I., Homma, M. and Imae, Y.
(1994) Transmembrane signalling by the chimeric chemosensory receptors of Escherichia coli Tsr and Tar with heterolo-
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
gous membrane-spanning regions. Mol. Microbiol. 14, 755762.
Manor, D., Hasselbacher, C.A. and Spudich, J.L. (1988)
Membrane potential modulates photocycling rates of bacterial rhodopsins. Biochemistry 27, 5843-5848.
Anderson, W., Schneider, A., Emmer, M., Perlman, R. and
Pastan, I. (1988) Purification and properties of the cyclic
adenosine 3',5'-monophosphate dependent gene transcription
in Escherichia coli. J. Biol. Chem. 246, 5929-5937.
Kim, B. and Little, J.W. (1992) Dimerization of a specific
DNA-binding protein on the DNA. Science 255, 203-205.
Tsuzuki, M., Aiba, H. and Mizuno, T. (1994) Gene activation by the E. coli positive regulator, OmpR. Phosphorylation-independent mechanism of activation by an OmpR mutant. J. Mol. Biol. 242, 607-613.
Koshland, D.E. Jr. (1952) Effect of catalysts on the hydrolysis of acetyl phosphate. Nucleophilic displacement mechanisms in enzymatic reactions. J. Am. Chem. Soc. 74, 22862292.
Hess, J.F., Oosawa, K., Kaplan, N. and Simon, M.I. (1988)
Phosphorylation of three proteins in the signalling pathway
of bacterial chemotaxis. Cell 53, 79-87.
Wylie, D., Stock, A., Wong, C.Y. and Stock, J. (1988)
Sensory transduction in bacterial chemotaxis involves phosphotransfer between Che proteins. Biochem. Biophys. Res.
Commun. 151,891-896.
Igo, M.M., Ninfa, A.J., Stock, J.B. and Silhavy, T.J. (1989)
Phosphorylation and dephosphorylation of a bacterial transcriptional activator by a transmembrane receptor. Genes
Dev. 3, 1725-1734.
Jin, S.G., Prusti, R.K., Roitsch, T., Ankenbauer, R.G. and
Nester, E.W. (1990) Phosphorylation of the VirG protein of
Agrobacterium tumefaciens by the autophosphorylated VirA
protein: essential role in the biological activity of VirG. J.
Bacteriol. 172, 4945-4950.
Reitzer, L.J. and Magasanik, B. (1983) Isolation of the
nitrogen assimilation regulator, NRI, the product of the glnG
gene of Escherichia coli. Proc. Natl. Acad. Sci. USA 82,
1979-1983.
Clarke, H.R.G., Leigh, J.A. and Douglas, C.J. (1992) Molecular signals in the interaction between plants and microbes.
Cell 71, 191-199.
Guan, C.-D., Wanner, B. and Inouye, H. (1983) Analysis of
regulation of phoB expression using a phoB-cat fusion. J.
Bacteriol. 156, 710-717.
Maldno, K., Shinagawa, H. and Nakata, A. (1985) Regulation of the phosphate regulon of Escherichia coil K-12:
regulation and role of the regulatory gene phoR. J. Mol.
Biol. 184, 231-240.
Ninfa, A.J., Ninfa, E.G., Lupas, A.N., Stock, A., Magasanik,
B. and Stock, J. (1988) Cross-talk between bacterial chemotaxis signal transduction proteins and regulators of transcription of the Ntr reguion: evidence that nitrogen regulation and
chemotaxis are controlled by a common phosphotransfer
mechanism. Proc. Natl. Acad. Sci. USA 85, 5492-5496.
Kessler, B., Marqu6s, S., K/Shler, T., Ramos, J.L., Timmis,
K.N. and de Lorenzo, V. (1994) Cross talk between catabolic
K.J. Hellingwerf et al. / FEMS Microbiology Reviews 16 (1995) 309-321
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
pathways in Pseudomonas putida: XylS-dependent and -independent activation of the Tol meta operon requires the
same cis-acting sequences within the Pm promoter. J. Bacteriol. 176, 5578-5582.
lshige, K., Nagasawa, S., Tokishita, S. and Mizuno, T.
(1994) A novel device of bacterial signal transducers. EMBO
J. 13, 5195-5202.
Stock, J.B., Surette, M.G., Levitt, M. and Park, P. (1995)
Two-component signal transduction systems: structure-function relationships and mechanisms of catalysis. In: TwoComponent Signal Transduction (Hoch, J. and Silhavy, T.,
Eds.), ASM Press, Washington, D.C., in press.
Schneider, B.L., Shiau, S.-P. and Reitzer, L.J. (1991) Role of
multiple environmental stimuli in the control of transcription
from a nitrogen regulated promoter in Escherichia coli with
weak or no activator-binding sites. J. Bacteriol. 173, 63556363.
Russo, F.D. and Silhavy, T.J. (1991) EnvZ controls the
concentration of phosphorylated OmpR to mediate osmoregulation of the porin genes. J. Mol. Biol. 222, 567-580.
Egan, S.M. and Stewart, V. (1990) Nitrate regulation of
anaerobic respiratory gene expression in narX deletion mutants of Escherichia coli K-12. J. Bacteriol. 172, 5020-5029.
McCleary, W.R., Stock, J.B. and Ninfa, A.J. (1993) Is acetyl
phosphate a global signal in Escherichia coli? J. Bacteriol.
175, 2793-2798.
Tsuzuki, M., Aiba, H. and Mizuno, T. (1994) Phosphorylation-independent mechanism of activation by an OmpR mutant. J. Mol. Biol. 242, 607-613.
Reizer, J., Reizer, A., Saier, M.H. Jr. and Jacobson, G.R.
[81]
[82]
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
321
(1992) A proposed link between nitrogen and carbon
metabolism involving protein phosphorylation in bacteria.
Protein Sci. 1, 722-726.
Griibl, G., Vogler, A.P. and Lengeler, J.W. (1990) Involvement of the histidine protein (HPr) of the phosphotransferase
system in chemotactic signaling of Escherichia coli K-12. J.
Bacteriol. 172, 5871-5876.
Dayhoff, J.E. (1990) Neural Network Architectures. Van
Nostrand Reinhold, New York, NY.
Hopfield, J.J. and Tank, D.W. (1986) Computing with neural
circuits: a model. Science 233, 625-633.
Neidhardt, F.C., et al. (1987) Escherichia coli and
Salmonella typhimurium. Cellular and Molecular Biology.
American Society for Microbiology, Washington, D.C.
Kahn, D. and Westerhoff, H.V. (1991) Control theory of
regulatory cascades. J. Theor. Biol. 153, 255-285.
Van Dam, K., Van der Vlag, J., Kholodenko, B.N. and
Westerhoff, H.V. (1993) The sum of the control coefficients
of all enzymes on the flux through a group-transfer pathway
can be as high as two. Eur. J. Biochem. 212, 791-799.
Segel, L.A. (1976) Incorporation of receptor kinetics into a
model for bacterial chemotaxis. J. Theor. Biol. 57, 23-42.
Bray, D. and Lay, S. (1994) Computer simulated evolution of
a network of cell-signaling molecules. Biophys. J. 66, 972977.
Davis, R.J. (1994) MAPKs: new JNK expands the group.
Trends Biochem. Sci. 19, 470-473.
Cano, E. and Mahadevan, L.C. (1995) Parallel signal processing among mammalian MAPKs. Trends Biochem. Sci.
20, 117-122.