Download Document 8318830

Chapter 1
General introduction
Karlijn C. Bastiaansen1,2 Wilbert Bitter2, and María A.
Llamas1
Department of Environmental Protection, Estación Experimental del
Zaidín-Consejo Superior de Investigaciones Científicas, C/Profesor
Albareda 1, 18008 Granada, Spain; and 2Section of Molecular
Microbiology, Department of Molecular Cell Biology, VU University
Amsterdam, De Boelelaan 1085, 1081HV Amsterdam, the
Netherlands.
1
Adapted from:
Bastiaansen, K.C., Bitter, W., and Llamas, M.A. (2012) ECF sigma
factors: from stress management to iron uptake. In Bacterial regulatory
networks. Filloux, A. (ed). Hethersett, Norwich: Caister Academic
Press, pp. 59-86.
8
|
Chapter 1
The σ70 family of sigma factors.
Regulation of gene expression is an essential mechanism that allows bacteria to rapidly
adapt to alterations in their environment. In this way, they preserve energy and profit by
expressing particular genes only when required. Gene expression in bacteria is regulated
primarily at the level of transcription initiation by modulating promoter recognition of
the RNA polymerase (RNAP) holoenzyme. The bacterial RNAP holoenzyme is composed
of a five-subunit core enzyme (RNAPc; subunit composition α2ββ’ω) and a dissociable
sigma (σ) subunit (Murakami and Darst, 2003). Whereas the core of the RNA polymerase
has enzyme activity, the sigma factor contains most promoter recognition determinants
and confers promoter specificity to the RNA polymerase. The sigma factor is also required
for transcription initiation, which is a stepwise process. First, the RNAP holoenzyme
forms a closed complex around the promoter in which the DNA is double-stranded.
After isomerisation (melting) of the DNA around the transcription start point, an open
complex is formed and transcription commences (Browning and Busby, 2004). Following
transcription initiation, the σ factor dissociates from the RNAP and is available to interact
with a next core enzyme complex, while the remaining RNAPc drives RNA transcription
until a transcription termination site is encountered (Mooney et al., 2005). All bacteria
contain a primary sigma factor that directs transcription of general housekeeping genes.
Specific activator or repressor proteins can affect binding of primary sigma factorcontaining RNAP complexes and as such regulate promoter activity. However, in addition
most bacteria also encode several alternative σ factors, which can redirect the RNAP to
initiate transcription from specific sets of alternative promoters (Helmann and Chamberlin,
1988). By alternating the use of standard and alternative sigma factors bacteria are able
to adequately regulate general cell functions as well as responses to specific signals.
Therefore, the versatility and adaptability of bacteria is partially reflected by the number
of alternative sigma factors they produce (Paget and Helmann, 2003).
Bacterial sigma factors have been classified into two main and unrelated families, namely
the σ70 and the σ54 families. σ54-type sigma factors are structurally different from σ70
family members and utilize a different mechanism of open complex formation (Merrick,
Figure 1. Structure of the σ70 family of sigma factors. The four conserved regions
with the subregions as well as the non-conserved region (NCR) of Group 1 sigma
factors are drawn. A typical promoter with -10, extended -10 and -35 boxes is shown
above the scheme and the interactions between the sigma subregions and the
different elements of the promoter are indicated by dotted lines. Furthermore, a
schematic representation of the Group 4 ECF sigma factors is shown. The structural
domains are displayed below the linear representation of the regions.
Introduction
|
1993). Differences between these families are also observed in promoter recognition.
σ54-dependent promoters contain highly conserved short sequences located at 24 and
12 basepairs upstream the transcription initiation site, while σ70 promoters are typically
located at -35 and -10 (Merrick, 1993). σ70-related sigma factors are present in all bacterial
genomes and are therefore probably the most archaic class of sigma-factors. For σ54 the
situation is different, since there are no known representatives of the σ54 family in any GCrich Gram-positive bacteria or cyanobacteria (Studholme and Buck, 2000). The σ70 family
is also the most diverse family and can be divided into four different groups based on
phylogenetic relationship and modular structure (Lonetto et al., 1992; Helmann, 2002).
The primary σ70 sigma factor of Escherichia coli is the model protein of this family. E. coli
σ70 is encoded by rpoD (RNA polymerase subunit) and its superscript reflects its molecular
weight (70 kDa). Originally Lonetto et al. (1992) proposed to divide this family into three
distinct groups or subfamilies based on sequence identity and functional characteristics:
the homologues of the E. coli primary sigma factor σ70 (Group 1), a group of highly related
but nonessential sigma factors (Group 2), and the more distantly related alternative sigma
factors (Group 3). Later this classification was refined and an additional group was added,
namely the extracytoplasmic function (ECF) sigma factors (Group 4) (Lonetto et al., 1994;
Helmann, 2002).
Members of group 1, the primary sigma factors, are highly conserved and recognize
similar target promoter sequences: TTGACA near the -35 and TATAAT near the -10
elements (Helmann, 2002). These sigma factors are usually between 40 and 70 kDa in
size and contain four conserved regions (Fig. 1) (Lonetto et al., 1992). These regions are
based on sequence alignments and can be divided into subregions, each of which plays a
different role in the interaction of the σ factor with the RNAPc and with the target DNA, i.e.
the promoter region. These subregions correspond to three distinct structural domains
(domain 2 to 4, Fig. 1) that are flexibly joined by linkers, which facilitate conformational
changes of the σ subunit (Gruber and Gross, 2003). Region 1.1 is considered to form an
N-terminal extension and is only present in primary σ70 factors (Helmann and Chamberlin,
1988; Lonetto et al., 1992; Murakami et al., 2002a). This domain prevents the interaction of
free sigma subunits with promoter DNA by inducing compact folding of the σ factor protein,
which is then unable to bind DNA (Dombroski et al., 1993; Schwartz et al., 2008). Structural
domain 2 (σ2) is formed by region 1.2, a stretch of non-conserved residues (NCR), and
regions 2.1 to 2.4. Region 1.2 plays an important role in transcription initiation. It makes
sequence-specific contact with the non-template strand downstream of the -10 promoter
element and supports the function of σ2 by stabilizing its conformation, thereby facilitating
processes like promoter binding and DNA melting (Haugen et al., 2008; Bochkareva and
Zenkin, 2013). The four subregions of region 2 are, amongst others, involved in promoter
recognition and interaction with the RNAPc. Region 2.4 is mainly implicated in binding
of the -10 box of the promoter (Fig. 1) (Lonetto et al., 1992). Region 2.3 is essential for
promoter melting and for interaction with the -10 promoter element (Young et al., 2004).
Region 2.2, the most conserved region within the σ70 family of sigma factors, is a crucial
RNAP binding determinant and establishes the first interactions between sigma and RNAPc
(Sharp et al., 1999; Young et al., 2001). Structural domain 3 (σ3) is comprised of the regions
3.0 and 3.1 and forms three α-helices. The helix-turn-helix motif in region 3.1 is mainly
involved in the extensive interaction with the RNAPc (Murakami et al., 2002b; Murakami
et al., 2002a). The conserved region 3.0 interacts with the extended -10 promoter element
9
1
10
|
Chapter 1
(Fig. 1), which in general substitutes for poorly matched -35 and -10 promoter elements
(Sharp et al., 1999). Finally, domain 4 (σ4) is composed of subregions 4.1 and 4.2. Region
4.2 interacts with the -35 element of the promoter (Fig. 1) (Lonetto et al., 1992; Campbell
et al., 2002b). In addition, region 4 is also involved in RNAPc binding (Sharp et al., 1999)
although for some promoter sequences this domain seems dispensable for transcription
initiation (Campbell et al., 2002b).
Sigma factors belonging to the three other groups (Group 2 to 4) are smaller than primary
σ70 factors. Group 2 sigma factors lack most of region 1, but are still closely related to
Group 1 and direct RNAP to similar promoter sequences, although these proteins are not
essential for growth (Lonetto et al., 1992). Group 3 sigma factors diverge more in sequence
and are significantly smaller than the primary σ70 proteins (25-35 kDa). This group has
been designated as secondary sigma factors and generally coordinates activation of stress
responses or developmental processes (Lonetto et al., 1992; Helmann, 2002). Group 4
sigma factors are the smallest sigma factors and only contain structural domains 2 and
4 (Fig. 1), which therefore seem to form the minimal requirement for a functional sigma
factor. These sigma factors are the most recently identified group of σ70 family members
(Lonetto et al., 1994) and have been designated ECF sigma factors (σECF) because of their
role in the regulation of extracytoplasmic functions. Although they have been identified
relatively recently, the σECF factors represent the largest and most diverse subfamily of σ70
family members and are considered to be the third fundamental mechanism of bacterial
signal transduction (Staroń et al., 2009).
Extracytoplasmic function sigma factors (σECF).
σECF are mainly involved in the regulation of cell envelope related functions such as
secretion, iron transport or stress responses, usually in response to environmental stimuli
(Missiakas and Raina, 1998; Helmann, 2002; Mascher, 2013). Moreover, σECF are also known
to regulate virulence in pathogenic bacteria (Lamont et al., 2002; Kazmierczak et al., 2005;
Llamas et al., 2009). As mentioned before, σECF are significantly smaller than primary
sigma factors and only consist of structural domains 2 and 4 (Fig. 1) (Lonetto et al., 1994).
Region 4 is in structure and in sequence the most conserved region within this group
(Lonetto et al., 1994). In contrast, subregion 2.4, which is implicated in the recognition
of the -10 promoter element, shows high variation, which likely reflects differences in
promoter binding specificity (Lonetto et al., 1994). The almost complete absence of region
3 implies that extended -10 promoters do not play a role in σECF promoter recognition. This
rudimentary region 3 forms a short unstructured region that functions to link domain
2 and 4 (Campbell et al., 2003). Apart from their structure, there are several additional
features that characterize σECF. They generally regulate only a few genes as opposed to
the large regulons controlled by primary sigma factors. Most σECF, with the exception of
the FecI-like group (see below), autoregulate their own expression (Helmann, 2002). In
addition, σECF usually share a high degree of conservation in the binding specificity of
the -35 promoter element, implying that promoter specificity of these sigma factors is
predominantly determined by the less conserved -10 element sequence (Lane and Darst,
2006). Recently, σECF have been classified in 43 major phylogenetically distinct groups
(Staroń et al., 2009). Of those, the stress responsive RpoE-like sigma factors and the iron
starvation FecI-like sigma factors are the predominant classes. Although σECF represent the
most abundant type of sigma factors in bacteria, the distribution of σECF among bacterial
Introduction
|
genomes varies greatly. A minority of species do not contain any σECF, while other species
encode over 80 members (Butcher et al., 2008; Staroń et al., 2009). A high number of
σECF, which is often associated with high numbers of one- and two-component regulatory
systems, generally reflects the diversity of the bacterial living environment as well as the
bacterial genome size. Species that exist in a stable environment and have a small genome
typically do not contain many σECF. In contrast, species with large genomes living in
complex habitats encode up to dozens. For example, the versatile opportunistic pathogen
Pseudomonas aeruginosa thrives in diverse habitats, ranging from soil to the human
airways, and contains 19 σECF-encoding genes (Visca et al., 2002; Llamas et al., 2008). The
Gram-positive Streptomyces coelicolor that predominantly lives in soil and is known to
produce several antibiotics encodes over 60 σ70 family sigma factors, of which more than
50 are classified as σECF (Paget et al., 2002). The most impressive number of σECF up to date
is found in the Gram-negative Sorangium cellulosum. The genome of this myxobacterium
is approximately 13 million base pairs in size and encodes a striking 83 σECF (Staroń et al.,
2009).
An important question raised by the presence of such a large sigma factor repertoire is
how the RNAPc is able to distinguish between different sigma factors to ensure proper
gene regulation, especially when σECF are activated. The concentration of primary σ70
factor within the cell is known to remain stable, even under circumstances that increase
the concentration of one or more alternative sigma factors. Different sigma factors have
to compete for the limited amount of RNAPc present in the cell (Ishihama, 2000; Maeda
et al., 2000). The primary σ70 has the highest binding affinity for RNAPc and is present in
the highest amount, which is sufficient to fully saturate the RNAPc (Maeda et al., 2000;
Tiburzi et al., 2008). Then how do alternative sigma factors compete with σ70 for the RNAP
holoenzyme? A model has been proposed in which the σ70-containing RNAP is mainly bound
in the cytoplasm to strong promoters, leaving weak promoters, such as σECF -dependent
promoters that do not contain strong -10 and -35 consensus sequences, free (Grigorova
et al., 2006). When an alternative sigma factor reaches a threshold concentration, which
usually needs to exceed the total amount of RNAP in the cell (Grigorova et al., 2006;
Tiburzi et al., 2008), it will be able to compete for the RNAPc and direct transcription from
weak promoters. In addition, specific regulators can be present or induced that facilitate
sigma factor substitution of the RNAP (recently reviewed in (Österberg et al., 2011)). One
strategy to achieve this is to decrease the ability of σ70 to bind to the RNAPc or by inhibiting
RNAP-σ70 activity. A well-studied example of this includes ppGpp nucleotides and the cofactor DksA (Costanzo and Ades, 2006; Costanzo et al., 2008) or Rsd, a stationary phase
protein that sequesters σ70 (Jishage and Ishihama, 1998). An alternative approach is to
increase the binding affinity of the alternative sigma factor for RNAPc, as has been shown
for the regulatory protein Crl of Salmonella enterica. This protein binds the alternative
sigma factor σS and increases its association rate to the RNAP holoenzyme (England et al.,
2008; Monteil et al., 2010).
Post-translational control of σECF - the anti-sigma factor.
Both synthesis and activation of σECF are tightly regulated processes that usually occur
in response to environmental signals. The post-translational control of these proteins is
carried out by anti-sigma factors that bind to and sequester the σECF, which is only released
and activated in the presence of a specific inducing signal. The functional unit of the
11
1
12
|
Chapter 1
σECF-dependent signalling is formed by the σECF and its cognate anti-sigma factor, and the
genes encoding these two proteins are normally in a single operon and co-transcribed.
Anti-sigma factors typically have a small N-terminal cytoplasmic domain, which is often
linked to a C-terminal periplasmic domain via one or sometimes multiple transmembrane
segments. The structures of four σECF in complex with their anti-sigma factor have shown
that, despite a low degree of sequence similarity, structural homology exists between the
N-terminal domains of most of these anti-sigma factors (Campbell et al., 2003; Campbell
et al., 2007; Maillard et al., 2014; Shukla et al., 2014). This domain, which contains about
85-90 amino acids is sufficient to bind the σECF and has therefore been designated ASD (for
anti-sigma domain). Binding of the anti-sigma factor to its cognate σECF occludes the RNAP
binding determinants and keeps the sigma subunit in an inactive conformation (Campbell
et al., 2002a; Campbell et al., 2003; Sorenson et al., 2004). As a result, binding of the σECF to
the anti-sigma factor and the RNAPc is mutually exclusive. So far, two different classes of
anti-sigma factors have been identified. The first class (‘mere’ anti-sigma factors) includes
proteins that function solely as anti-sigma factors and inhibit the σECF in the absence of
the inducing signal. The second class includes proteins that do not only inhibit sigma
factor activity, but are also required for full σECF activity in presence of the signal (Koster
et al., 1994; Ochs et al., 1995; Stiefel et al., 2001; Mettrick and Lamont, 2009). Anti-sigma
factors that belong to this second class are also referred to as sigma factor regulators
(SFRs) (Mettrick and Lamont, 2009; Llamas and Bitter, 2010). Expression of the cytosolic
Figure 2. Activation of the E. coli ECF sigma factor σE (RpoE) by cell envelope stress. In unstressed
cells, σE is inactive through its interaction with the RseA anti-sigma factor. The DegS and RseP proteases
are inactive by inhibitory interactions involving their respective PDZ domains, the RseB protein, and a
glutamine-rich region of RseA. In stressed cells, unfolded proteins accumulate in the periplasm and
activate the DegS protease by binding to its PDZ domain. RseB is displaced from the RseA protein by
mislocalized LPS species in the periplasm allowing DegS to cleave the periplasmic domain of the anti-sigma
factor. This site-1 cleavage removes the inhibitory interaction between RseP and RseA, and enables RseP to
cleave in the transmembrane region of RseA. The remainder of the RseA protein is degraded in the cytosol
by ClpXP proteases, upon which σE is free to interact with the RNAPc and direct transcription of its target
genes. The σE regulon mainly encodes proteins that are involved in the restoration of the proper folding
of outer membrane proteins. The σE, RseA, RseB, DegS and RseP homologues of P. aeruginosa (σAlgT, MucA,
MucB, AlgW and MucP, respectively) are indicated between brackets. CW, cell wall; OM, outer membrane;
P, periplasm; CM, cytoplasmic membrane; C, cytoplasm. Figure adapted from (Bastiaansen et al., 2014).
Introduction
|
tail of SFRs induces σECF activity independent of the presence of the signal (Mettrick and
Lamont, 2009). It has been proposed that this fragment protects the σECF from degradation
and increases its affinity for the RNA polymerase (Mahren and Braun, 2003; Mettrick and
Lamont, 2009), although its exact function is still unknown.
Different mechanisms involved in the activation of σECF have been identified. The best
studied is the mechanism that activates RpoE-like sigma factors, which occurs via
regulated intramembrane proteolysis (RIP), a widely used signalling pathway by which
a transmembrane anti-sigma factor is subjected to sequential proteolytic degradation
resulting in the release and activation of the σECF (Brown et al., 2000; Heinrich and Wiegert,
2009). In Gram-negative bacteria, the most common and important mechanism for the
control of σECF activity is Cell-Surface Signalling (CSS), which usually controls activity of
FecI-like sigma factors. In response to an environmental signal the sigma factor is activated
via a signalling pathway that involves not only the σECF and anti-sigma factor protein, but
also a cell-surface exposed outer membrane receptor (Braun et al., 2006; Llamas and
Bitter, 2010; Llamas et al., 2014). For their relevance to the contents of this thesis, both the
RIP and CSS mechanisms are described in detail below. Furthermore, activity of some σECF
is regulated by anti-anti-sigma factors that are phosphorylated in response to a stimulus,
upon which they bind to the anti-sigma factor allowing the release of the σECF (Staroń and
Mascher, 2010). Another mechanism has been described in which oxidation of cysteine
residues within the ASD domain of specific zinc-dependent anti-sigma factors leads to
the formation of intramolecular disulphide bonds and the dissociation of the Zn2+ ion and
the σECF (Campbell et al., 2008). Finally, σECF analysis has identified anti-sigma factors that
carry conserved domains of still unknown function, suggesting the existence of alternative
mechanisms for σECF activation (Staroń et al., 2009). However, not all σECF are linked to
potential anti-sigma factors, which suggests the existence of alternative pathways in
the control of sigma factor activity. In some cases the σECF might be regulated only at the
transcriptional level (Hong et al., 2002). Some σECF not associated with anti-sigma factors
have long C-terminal extensions that might play a role in the signalling pathway and
activation. Another group of σECF is associated with completely unrelated proteins, such as
putative serine/threonine kinases or other enzymes (Staroń et al., 2009). Functional links
between these protein pairs remain to be discovered.
Activation of stress responsive (RpoE-like) σECF by Regulated Intramembrane
Proteolysis (RIP).
All organisms have stress response systems that allow them to sense and respond to
damaging conditions by altering gene expression. In response to cytoplasmic stress, both
the stationary phase sigma factor RpoS (σS) (Group 2) and the heat-shock sigma factor
RpoH (σH) (Group 3) are considered to be master regulators in several bacteria (HenggeAronis, 2002; Battesti et al., 2011). However, an additional level of complexity is produced
when the stress signal is generated outside the cytoplasm and the information must be
communicated across the cytoplasmic membrane. This is known as cell-envelope or
periplasmic stress. Envelope stress responses play an important role in many processes,
including protein folding, cell wall biosynthesis, and pathogenesis. σECF, together with
two-component systems, are known to be essential in this process (Raivio and Silhavy,
2001). The best characterized σECF controlling cell-envelope stress are the RpoE-like
sigma factors, which form a widely distributed group that is found in most bacterial phyla
13
1
14
|
Chapter 1
(Staroń et al., 2009). This group is represented by E. coli σE, which plays a central role in
maintaining cell envelope integrity in this bacterium (Raivio and Silhavy, 2001; Hayden
and Ades, 2008). σE is not only important under damaging conditions that generate stress,
but also during basic cellular physiology. In fact, the σE encoding rpoE gene is essential
for bacterial viability (de las Peñas et al., 1997). σE is activated by stresses that disrupt
protein folding in the cell envelope, such as heat shock, overproduction of outer membrane
proteins and mutations that inactivate periplasmic chaperones. The molecular mechanism
that controls σE activation has been studied in great detail (reviewed in (Ades, 2008)).
In unstressed cells, σE is sequestered at the cytoplasmic side of the inner membrane by
the anti-sigma factor RseA (Fig. 2). RseA is a transmembrane protein that binds σE with
high affinity, thereby preventing the interaction with the RNA polymerase (Campbell et al.,
2003). Release of σE from RseA is regulated by a proteolytic cascade that is activated upon
accumulation of unfolded peptides in the periplasm. This proteolytic cascade involves two
cytoplasmic membrane anchored proteases, DegS and RseP (also known as YaeL), that act
sequentially (Fig. 2) (Kanehara et al., 2002). The signal that induces DegS activity is the
presence of unfolded outer membrane proteins in the periplasm. The C-terminal peptides
of the porins OmpC or OmpF can bind the PDZ domain of DegS, thereby allowing this
protease to cleave in the periplasmic region of RseA (Fig. 2). However, the periplasmic
protein RseB binds tightly to RseA and protects this anti-sigma factor from proteolytic
degradation (Cezairliyan and Sauer, 2007). Therefore, a second signal is needed to activate
the σE response. Recent research has shown that RseB acts as a LPS sensor and detects
mislocalized LPS species accumulating in the periplasm (Lima et al., 2013). Binding
of these LPS molecules releases RseB from RseA, making this protein susceptible to
cleavage by DegS. Following this first cleavage (site-1), the RseP protease cleaves RseA
in the transmembrane segment (site-2) and the cytoplasmic domain of RseA bound to σE
is released into the cytoplasm. Subsequently, the RseA cytoplasmic domain is degraded
by cytoplasmic proteases such as the ClpXP complex and then the free σE can bind the
RNAPc and direct the transcription of the σE regulon genes (Flynn et al., 2003; Flynn
et al., 2004). Although the σE regulon includes genes that affect many different aspects
of the cell, a significant fraction is coding for chaperones required for the delivery and
assembly of outer membrane proteins and LPS, and periplasmic proteases such as DegP
to degrade misfolded proteins (Rhodius et al., 2006). σE also positively autoregulates its
own expression, thereby ensuring a rapid amplification of the signal required to maintain
cell envelope homeostasis. Simultaneously, it also induces the expression of the rseA antisigma factor, required for the negative feedback loop (Ruiz et al., 2006).
Activation of iron starvation (FecI-like) σECF by Cell-Surface Signalling (CSS).
The FecI-like σECF constitute one of the predominant groups of σECF and have been classified
as iron starvation sigma factors, based on a common role in the regulation of iron uptake
(Leoni et al., 2000). Iron is essential for bacterial growth and survival, especially because
of its ideal redox potential for biological reactions, but in most aerobic environments
iron acquisition is greatly complicated due to its highly insoluble nature. To fulfil their
requirements for iron bacteria produce and secrete high affinity iron-chelating compounds
called siderophores, or use siderophores produced by other organisms (referred to as xenoor heterologous siderophores). In addition, pathogens can also use host-iron complexes as
iron sources, such as transferrin, lactoferrin, haem or haemoglobin (Ratledge and Dover,
2000; Wandersman and Delepelaire, 2004). However, although iron is crucial for bacterial
Introduction
|
15
1
Figure 3. General model of Cell-Surface Signalling (CSS) regulation. In the uninduced state (i.e.
absence of the iron carrier), the σECF is sequestered through an inhibitory interaction with the transmembrane anti-sigma factor. Upon binding of the inducing signal (i.e iron-siderophore, ferric-citrate)
to the CSS receptor, the TonB protein interacts with the receptor providing the energy required for
the transport across the outer membrane. In this situation, an interaction between the periplasmic
signalling domain of the receptor and the anti-sigma factor also occurs, which results in the release
and activation of the σECF. Upon activation, the σECF mediates expression of target genes, which usually
includes the gene coding for the cognate CSS receptor. Depending on the role of the CSS anti-sigma
factor as a mere anti-sigma or as a sigma factor regulator (see text for details), two models for FecIlike σECF-driven transcription have been proposed (A and B, respectively). In A, a completely free
sigma subunit associates with the RNAPc, whereas in B a complex formed by the cytosolic tail of the
anti-sigma factor and the σECF binds to the RNAPc. CW, cell wall; OM, outer membrane; P, periplasm;
CM, cytoplasmic membrane; C, cytoplasm.
growth, bacteria need to strictly regulate iron acquisition to prevent the accumulation of
iron-induced toxic radicals inside the cell. Iron homeostasis is mainly controlled by the
general iron regulator Fur that represses gene expression in iron-sufficient conditions
(Escolar et al., 1999; Hantke, 2001). Furthermore, the production of iron uptake systems
for heterologous siderophores and host molecules only makes sense if these molecules
are present in the environment. In contrast to other σECF (i.e. σE), the expression of iron
starvation sigma factors and of their cognate anti-sigma factors is not autoregulated, but
induced by low iron concentrations through Fur. Once produced, these sigma factors need
to be activated and will mainly mediate expression of genes required for iron uptake.
In Gram-negative bacteria the activity of iron-starvation σECF is controlled by a signal
transduction mechanism involving both proteins in the outer membrane and in the
cytoplasmic membrane, referred to as Cell-Surface Signalling (CSS). In these systems,
the functional unit of σECF-dependent signalling is formed not only by the σECF and its
cognate anti-sigma factor, but also by an outer membrane receptor (Fig. 3). The CSS outer
membrane receptor belongs to the family of TonB-dependent receptors. These special outer
membrane receptors are usually involved in the transport of iron-siderophore complexes
across the outer membrane, although some members transport other macromolecules
such as vitamin B12, nickel complexes and carbohydrates (Noinaj et al., 2010). In order
to facilitate substrate passage, these receptors are energized by a cytoplasmic membrane
protein complex composed of the TonB-ExbB-ExbD proteins. Of those, the TonB protein is
the one that physically interacts with the outer membrane receptor, hence the name TonBdependent receptor (Braun and Endriss, 2007; Krewulak and Vogel, 2008). Coupling with
the cytoplasmic membrane is necessary as the substrate (i.e. iron-siderophore) has to be
16
|
Chapter 1
actively transported across the outer membrane, where there
is no other source of energy available.
Crystal structures of several TonB-dependent receptors
have revealed a common domain architecture, despite low
sequence similarity (reviewed in (Noinaj et al., 2010)). The
C-terminal part of these proteins forms a large β-barrel in
the outer membrane, consisting of 22 antiparallel β-strands
that are connected via short periplasmic turns and larger
surface exposed loops (Fig. 4; grey). This β-barrel produces
a big pore of approximately 35-40 Å. The pore is occluded
by a globular plug domain, also called cork or hatch, in
absence of the substrate, which prevents the aspecific
diffusion of large molecules across the outer membrane (Fig.
4; black). In order for the substrate to pass, the plug must
undergo conformational changes that temporarily allows
Figure 4. Structure of the
P. aeruginosa FpvA CSS
opening of the pore. Whether the plug domain opens the
receptor. The mature apopore by complete retraction or by creating an opening upon
FpvA receptor (without pyoverdin bound) is shown. The Cconformational rearrangement is still a matter of debate
terminal
β-barrel
(grey)
(Noinaj et al., 2010). TonB-dependent receptors also contain
consists of 22 anti-parallel
β-strands and forms a pore in
a so-called TonB box, a β-strand of only 6 semi-conserved
the outer membrane which
amino acids situated near the N-terminus of the protein.
is occluded by a plug domain
(black).
The
periplasmic
This region is located in the periplasm and is the domain
signalling
domain
which
where the interaction with the TonB protein occurs (Noinaj
determines the specificity of
the CSS signal transduction
et al., 2010). How TonB exactly energizes the receptor is
pathway and is responsible for
unknown, but simulation experiments have demonstrated
the interaction with the antisigma factor is coloured blue.
that, despite the small interface, the interaction between
Figure adapted from (Brillet et
TonB and the receptor is relatively stable, which would
al., 2007).
allow a strong mechanical coupling (Gumbart et al., 2007).
In addition to their transport function, many TonB-dependent receptors also have a role
in CSS regulation. These bifunctional receptors are referred to as CSS receptors or TonBdependent transducers (Koebnik, 2005; Llamas et al., 2014). They induce a signalling
cascade in response to the presence of their cognate substrate (i.e. iron-siderophore)
that results in the activation of a FecI-like σECF. The transport and signalling functions are
independent and substrate passage across the outer membrane is not required for σECFmediated transcription induction to occur and vice versa (Härle et al., 1995; Kim et al.,
1997). CSS receptors are easily distinguished from ordinary TonB-dependent receptors
only involved in transport by the presence of an N-terminal extension of approximately
70-80 amino acids (Koster et al., 1993; Braun et al., 2003; Schalk et al., 2004; Llamas and
Bitter, 2010) (Fig. 4; blue). This extension contains the so-called signalling domain, which
is the domain that interacts with the anti-sigma factor in the cytoplasmic membrane and
transduces the presence of the signal. The signalling domain determines the specificity of
the transduction pathway but has no effect on the transport function of the protein (Koster
et al., 1994; Schalk et al., 2009).
The signal transduction pathway of CSS starts by binding of an environmental signal (i.e.
an iron-siderophore) to the cognate CSS receptor. In the current model this induces an
Introduction
|
interaction between the signalling domain and the anti-sigma factor in the periplasm,
which ultimately leads to the activation of the σECF and the expression of target genes,
which usually includes the gene encoding the cognate CSS receptor. The best studied
CSS system is the E. coli Fec system, which responds to and regulates the uptake of the
iron carrier iron-citrate (Braun et al., 2006). The E. coli iron-citrate transport system is
encoded by the fecABCDE operon and is regulated by the fecIR operon located upstream
of fecABCDE (Braun, 1997). The fecA gene codes for a CSS receptor and the fecBCDE genes
for a cytoplasmic membrane ABC transport system that carries iron from the periplasm
into the cytoplasm (Braun, 1997). Transcription of the fecIR operon, which encodes σFecI
and the FecR anti-sigma factor, is iron-regulated through the Fur repressor protein, while
transcription of fecABCDE is both negatively regulated by Fur and positively by σFecI (Enz et
al., 1995; Härle et al., 1995). The FecR anti-sigma factor is, as most CSS anti-sigma factors,
a three domain protein that contains an N-terminal cytoplasmic tail (residues 1-85), a
transmembrane segment (residues 86-100) and a large periplasmic C-terminal domain
(residues 101-317) (Braun et al., 2006). The interaction of FecR with the CSS receptor
FecA has been demonstrated in vitro by binding assays, which showed that the N-terminal
domain of FecA interacts with residues 237-317 of the FecR periplasmic domain (Enz
et al., 2003a). The FecR N-terminal domain binds σFecI and mainly domain 4 of the sigma
factor is involved in this interaction (Mahren et al., 2002). Although full-length FecR is not
able to activate σFecI in the absence of iron-citrate, the cytoplasmic tail of FecR (FecR1-85) is
required for σFecI activity (Ochs et al., 1995; Stiefel et al., 2001; Mahren and Braun, 2003),
showing that FecR is not a mere anti-sigma but a sigma factor regulator (SFR). In fact,
it was demonstrated that FecR1-85 increases the affinity of σFecI for the RNA polymerase
and protects σFecI from proteolytic degradation (Mahren and Braun, 2003). Interestingly,
FecR1-85 was co-purified with σFecI bound to the RNAP and it is therefore possible that the
N-terminal domain of FecR is included in the σFecI-RNAP complex (Mahren and Braun,
2003). A model was proposed in which, upon iron-citrate induction, the cytoplasmic tail
of FecR is cleaved, possibly by the transmembrane protease RseP, and together with σFecI
released into the cytoplasm where the complex will bind to the RNAPc initiating fecABCDE
transcription (Braun et al., 2006) (Fig. 3B). However, experimental data supporting this
model is lacking.
σECF and CSS in Pseudomonas.
The Pseudomonas genus constitutes a group of saprophytic bacteria with a remarkable
genetic and metabolic versatility, which is reflected by the number of niches they inhabit.
Although these bacteria are mainly isolated from soil and water ecosystems, they are also
associated with plant, animal and human environments (Spiers et al., 2000; Palleroni,
2010). P. putida is a persistent colonizer of the plant rhizosphere and can suppress growth
of fungi and bacteria pathogenic for the plant, underscoring its potential as a biocontrol
agent (Haas and Defago, 2005). In contrast, P. syringae is one of the most important bacterial
plant pathogens (Mansfield et al., 2012). This bacterium is able to cause disease in a wide
range of different plant species, thereby seriously affecting the commercial production of
fruits and vegetables. Another species is P. aeruginosa, an opportunistic human pathogen
of clinical relevance and is a main cause of severe hospital-acquired infections, especially in
immunocompromised patients with cancer, cystic fibrosis or burn wounds. Unfortunately,
these infections are often difficult to treat due to a high intrinsic resistance of P. aeruginosa
against antibiotics (Lyczak et al., 2000; Lister et al., 2009; Breidenstein et al., 2011).
17
1
18
|
Chapter 1
Pseudomonas species are known to contain a high proportion of regulatory genes in their
genomes, which reflects the importance of controlling gene expression in response to
the diverse environmental conditions that these bacteria encounter. Together with twocomponent and quorum sensing systems, σECF and CSS are among the most important
and common regulatory mechanisms of Pseudomonas. While E. coli only encodes two
σECF (σE and σFecI), P. aeruginosa and P. putida both contain 19 σECF (Martínez-Bueno et al.,
2002; Visca et al., 2002; Oguiza et al., 2005; Llamas et al., 2008). Activity of most of these
Pseudomonas σECF is controlled by CSS, as indicated by their genomic association with antisigma factors and CSS receptors (Llamas and Bitter, 2010; Llamas et al., 2014), although
only a few of those have been characterized. For example the PupIRB system of P. putida
was, together with the E. coli FecIRA system, the first CSS pathway reported (Koster et al.,
1994). PupIRB regulates the uptake of the heterologous siderophores pseudobactin BN7
and BN8 produced by other Pseudomonas strains.
Within the genus Pseudomonas, σECF have been mainly analysed in the pathogen P.
aeruginosa. The best characterised σECF of this bacterium is the RpoE-like sigma factor
σAlgT, which controls the production of the capsular exopolysaccharide alginate. Alginate
is responsible for the mucoid colony morphology found in most P. aeruginosa clinical
isolates from chronic lung infections (i.e. from cystic fibrosis patients). Isolation of
mucoid P. aeruginosa is normally an indicator of clinical deterioration and decreased life
expectancy (Govan and Deretic, 1996). The accumulation of mucus in the airways produces
limitations in pulmonary functions and alginate enables P. aeruginosa to grow as compact
microcolonies or biofilms, thereby increasing its resistance to host defences and antibiotic
treatment. The key to mucoid conversion is the expression of the alginate biosynthetic
operon, of which transcription is regulated by σAlgT (Hershberger et al., 1995). σAlgT is
controlled by the MucA anti-sigma factor and its activation follows a pathway similar to
that of E. coli σE (Qiu et al., 2007; Qiu et al., 2008; Wood and Ohman, 2009) (Fig. 2). Most
mucoid P. aeruginosa isolates contain missense or nonsense mutations in mucA (Anthony
et al., 2002), which bypass the requirement for site-1 cleavage and lead to the release of
active σAlgT and concomitant overproduction of alginate. σAlgT/σE homologues are present
in other species belonging to the Pseudomonas genus and, although they are less studied
than in P. aeruginosa, these RpoE-like sigma factors also seem to control exopolysaccharide
production and tolerance towards stress (Schnider-Keel et al., 2001; Li et al., 2010).
Besides σAlgT, P. aeruginosa contains multiple FecI-like σECF that are involved in CSS
regulation. The Fpv CSS system is the best studied in this bacterium and responds to the
presence of the siderophore pyoverdine, which is produced by the bacterium itself. It
regulates pyoverdine synthesis and uptake as well as virulence factor production (Lamont
et al., 2002; Beare et al., 2003). The CSS receptor FpvA (Fig. 4) can bind both iron-free and
iron-loaded pyoverdine, but only upon interaction with iron-loaded pyoverdine a number
of conformational changes are induced (Schalk et al., 2001). Especially the signalling
domain of FpvA is subjected to major changes upon iron-pyoverdine binding (Brillet et al.,
2007). In the apo-FpvA form (without pyoverdine), the signalling domain of the receptor
interacts with the TonB box. Binding of Fe-pyoverdine to FpvA destabilizes this interaction
upon which these domains are free to interact with their partners in the cytoplasmic
membrane, the FpvR anti-sigma factor and the TonB protein, respectively (Fig. 3). The P.
aeruginosa Fpv system shares common features with the E. coli Fec system but also contains
Introduction
|
important differences that make it a unique CSS system. The E. coli Fec system activates
only genes involved in the transport of the inducing signal (i.e. iron-citrate). However, the
pyoverdine CSS system regulates not only pyoverdine uptake, but also genes involved in
pyoverdine biosynthesis (Shen et al., 2002) and, most importantly, also the synthesis of the
two virulence factors exotoxin A and the endoprotease PrpL (Lamont et al., 2002; Beare
et al., 2003). Furthermore, the anti-sigma factor FpvR does not interact with one, but with
two σECF, i.e. σPvdS and σFpvI (Lamont et al., 2002; Beare et al., 2003). In contrast to E. coli
FecR, FpvR functions solely as an anti-sigma factor (Mettrick and Lamont, 2009). The first
67 residues of FpvR are sufficient to bind both σPvdS and σFpvI (Rédly and Poole, 2005), and in
absence of pyoverdine, FpvR promotes proteolytic degradation of σPvdS thereby reinforcing
its role as anti-sigma factor (Spencer et al., 2008). In agreement with a strict anti-sigma
function, deletion of fpvR results in the activation of σPvdS- and σFpvI-dependent transcription
in absence of pyoverdine. Furthermore, fpvR overexpression inhibits the activity of both
sigma factors (Lamont et al., 2002; Beare et al., 2003). σPvdS and σFpvI are quite different
in sequence and apparently they bind FpvR in a similar but not identical manner (Rédly
and Poole, 2005), which could explain this difference. The fpvR gene is located adjacent
to fpvI but in the opposite orientation, which means that, in contrast to most σECF/antisigma pairs, they are not co-transcribed. The pvdS gene is not located near fpvR, but next
to the pyoverdine biosynthetic locus as a single gene. The expression of the three genes
is controlled by iron via Fur (Ochsner and Vasil, 1996). σFpvI only controls the expression
of the pyoverdine outer membrane receptor FpvA, whereas σPvdS regulates the expression
of more than 20 genes. The σPvdS regulon mainly includes genes involved in pyoverdine
biosynthesis, but also genes encoding exotoxin A and PrpL, as well as the regAB and ptxR
genes that positively regulate exotoxin A production (Vasil et al., 1998; Leoni et al., 2000;
Wilderman et al., 2001; Wilson et al., 2001; Gaines et al., 2007).
In addition to the Fpv system, P. aeruginosa contains another twelve CSS systems, most of
them involved in the regulation of heterologous siderophore uptake (reviewed in (Llamas
and Bitter, 2010; Llamas et al., 2014)). This includes the Fox and Fiu CSS pathways, which
are induced by the heterologous siderophores ferrioxamine and ferrichrome, respectively
(Llamas et al., 2006). These systems have the normal CSS features represented in Fig. 3
and show mechanistically high similarities to the E. coli Fec system. The ECF sigma factors
σFoxI and σFiuR are co-transcribed with their respective anti-sigma factors FoxR and FiuR,
respectively, and expression of these genes is regulated by iron through Fur. The genes
encoding the FoxA and FiuA CSS receptors form separate transcriptional units that are
regulated by both Fur and σFoxI or σFiuR, respectively (Llamas et al., 2006). Similar to E.
coli FecR, P. aeruginosa FoxR and FiuR function as sigma factor regulators that do not
only inhibit but are also required for sigma factor activity (Mettrick and Lamont, 2009).
Overproduction of the cytosolic tail of these proteins causes σECF activity independent of the
presence of the CSS inducing signal (Mettrick and Lamont, 2009). CSS systems controlling
the response to the presence of iron-citrate, haem and the siderophore mycobactin have
been also described in P. aeruginosa (Ochsner et al., 2000; Banin et al., 2005; Llamas et al.,
2008).
Finally, P. aeruginosa contains the PUMA3 CSS system (also referred to as Vre) that seems
to be independent of iron and could be dedicated to the regulation of several potential
virulence factors (Llamas et al., 2009). The first interesting characteristic of the PUMA3
19
1
20
|
Chapter 1
system is the receptor component VreA, which is considerably smaller (~23 kDa) than
other CSS outer membrane receptors (75-85 kDa) and lacks the C-terminal b-barrel
domain, suggesting that this protein is only involved in signal transduction and not in
signal uptake (Llamas et al., 2009). Microarray analyses showed that the σECF regulated
by this CSS system, σVreI, targets the expression of genes encoding secreted proteins and
components of secretion systems (Llamas et al., 2009), including a putative two-partner
secretion system (TPS) and the type II Hxc secretion system involved in the secretion of
a low-molecular weight alkaline phosphatase (Ball et al., 2002). TPS systems mediate the
secretion of high-molecular-weight surface-exposed proteins that are often involved in
cell adhesion and pathogen dissemination (Jacob-Dubuisson et al., 2001). Although the
role of these σVreI-regulated genes in P. aeruginosa virulence has not been established yet,
it has been shown that overexpression of σVreI increases the virulence of P. aeruginosa in
zebrafish (Danio rerio) embryos (Llamas et al., 2009). Moreover, the sera of most patients
infected with P. aeruginosa contain antibodies directed against σVreI-regulated proteins
(Llamas et al., 2009), which indicates that the σVreI regulon is transcribed in vivo during
infection. Apparently, the Vre system is activated by a host signal, which is consistent with
previous reports showing that interaction of P. aeruginosa with human airway epithelial
cells induces the expression of many σVreI-regulated genes (Frisk et al., 2004; Chugani and
Greenberg, 2007).
Scope of the thesis.
CSS is an important regulatory mechanism that allows Gram-negative bacteria to sense
and respond to their environment, including animal and plant hosts. Since the discovery
of CSS (Koster et al., 1994; Härle et al., 1995), several of these systems have been analysed
and microarray experiments have been instrumental in identifying the CSS regulons
(Llamas et al., 2008; Llamas et al., 2009). However, many aspects of the molecular
mechanism behind CSS regulation are still poorly understood, especially how the signal is
transduced from the outer membrane to the cytoplasm is unclear. The key component in
this process is the CSS anti-sigma factor, which receives the signal from the CSS receptor
to activate the σECF. The aim of the research described in this thesis is to examine the role
of the anti-sigma factor in the CSS transduction pathway and to determine how CSS σECF
are activated in response to the inducing signal. In Chapter 2 we describe an unusual
CSS system of P. putida that contains a hybrid cytoplasmic membrane protein named IutY,
combining both a sigma and an anti-sigma domain in a single polypeptide. By studying
this unusual CSS system with a reduced complexity we hope to more easily find universal
CSS activation mechanisms. We show that the periplasmic domain of IutY needs to be
removed in order to liberate and activate the cytosolic σIutY domain, and demonstrate that
the Prc and RseP proteases are required for this sequential process. We also prove that
these proteases are indeed involved in the regulation of classical CSS systems, in which
the sigma and anti-sigma factors are produced as two separate proteins. This confirms a
previous hypothesis suggesting that CSS σECF activation requires a proteolytic cascade to
remove the CSS anti-sigma factor, similar to the regulation of RpoE-like σECF. In Chapter 3
we analyse a proteolytic event that occurs in several CSS anti-sigma factors prior to signal
recognition. This event, termed initial cleavage, separates the anti-sigma factor in two
domains already in absence of the CSS inducing signal. Using the P. aeruginosa FoxR antisigma factor as a model protein, we were able to demonstrate that this cleavage is due to
autoproteolytic activity of the protein and that the resulting domains interact and function
Introduction
|
together to transduce the CSS signal. We further investigated this event in Chapter 4 and
propose that the autoproteolytic cleavage of P. aeruginosa FoxR is not an enzyme-mediated
process, but occurs through a chemical sequence utilizing an N-O acyl rearrangement as
a first step. In Chapter 5 we isolated and examined constitutively active mutants of the
hybrid sigma/anti-sigma protein IutY by genetic screening and introducing single amino
acid substitutions to examine the mechanism of P. putida Iut activation in more detail.
Our results indicate that in response to the inducing signal the Prc protease trims the
IutY protein down to ~50 amino acids, which produces a substrate that can be cleaved
by RseP. Finally, in Chapter 6 we describe that the P. aeruginosa PUMA3 CSS system is
not expressed in response to iron limitation, but is regulated by phosphate concentration
through the PhoB regulator. Upon production this pathway is partially, but not completely,
active. Full activation of PUMA3 requires the removal of the VreR anti-sigma factor and
we show a role of the Prc and RseP proteases in this process. The results obtained in this
thesis are summarized and discussed in Chapter 7.
21
1