Each bacterial species has a set of PBPs that are involved in the late stages of
peptidoglycan synthesis and are targets for β-lactam antibiotics. PBPs are acyl-serine transferases, the actions of which are mediated by an active-site serine moiety. PBPs
Acyl side chain
Thiazolidine ring
catalyse the final stages of murein biosynthesis and help to modify and shape of peptidoglycan exoskeleton by removing the terminal D-alanine residue from pentapeptide stems (Vollmer et al., 2008). The morphology of a bacterial cell is dependent on the composition and synthesis of the peptidoglycan layer. PBPs modify the exoskeleton and work with cytoskeleton proteins to manipulate the structure,
through specific protein localisations and interactions. PBPs are named for their affinity for penicillin, which mimics the peptide side chain structure and blocks cell wall
synthesis. However, they could be more appropriately named as ‘Peptidoglycan
Synthases’, as their role is mainly as the drivers behind bacterial cell wall biosynthesis.
1.5.1 Classification and Topology of PBPs
Penicillin-Binding Proteins (PBPs) are initially classified according to molecular weight and are then further classified according to function. PBPs are initially divided into high molecular weight (HMW, Mr >60,000) and low molecular weight (LMW, <60,000) PBPs. HMW PBPs are further divided in to Class A PBPs (bifunctional) and Class B PBPs (monofunctional transpeptidases) (Figure 1.9 a). Generally, two HMW PBPs are necessary for the bacterium to survive, one Class A and one Class B. LMW PBPs are shapeless mutants (Popham & Young, 2003) and not essential for survival of the bacterial cell. The role of LMW PBPs serves more to recycle peptidoglycan in preparation for the incorporation of nascent peptidoglycan by HMW PBPs.
Peptidoglycan synthesis and incorporation is mainly by HMW PBPs 1-3. HMW Class A bifunctional PBPs are the focus of this PhD (Figure 1.9 b), which carry out both transglycosylation and transpeptidation.
Figure 1.9 (a)The Classification of Gram-negative PBPs (using E. coli as the model). The topology of HMW Class A PBPs (blue) consist of a cytoplasmic tail, a transmembrane anchor, and two domains joined by a β-rich linker located on the outer surface of the cytoplasmic membrane where peptidoglycan synthesis takes place. The C-terminal of all HMW PBPs has a transpeptidase domain. LMW PBPs (red) are sometimes classified based on their migration through SDS-PAGE in to Classes A, B and C. Here they have been divided based on amino acid sequence alignment and knowledge of their structure and function (Massova & Mobashery, 1998). Type 4 and Type 7 PBPs are DD-endopeptidases, Type 5 PBPs are strict DD- carboxypeptidases and Type AmpH are of unknown function, except PBP4b has shown weak DD-carboxypeptidase activity (Vega & Ayala, 2006). (b) A Schematic representation of the two Class A bifunctional PBPs: E. coli PBP1A and PBP1B. NB: The ODD domain (Outer Docking Domain) of E. coli PBP1A is located between the transglycosylase and the transpeptidase domains, whereas the UB2H domain in E. coli PBP1B is after the TM helix, prior to the transglycosylase domain.
The E. coli genome has two major bifunctional Class A penicillin binding proteins, E. coli 1A and 1B. E. coli PBP1A serves to polymerise LII monomer units into polymers of a length of ~20 disaccharide units. All bacteria contain at least one PBP from each class. Gram-negative cocci tend to have one peptidoglycan-synthesising machinery i.e. the divisome at the septum, whereas Gram-negative rods have two: the elongation machinery in addition, so they have at least two Class A PBPs and two class B PBPs. Chlamydiae peptidoglycan is a particular case as it doesn’t have any class A PBPs.
High Molecular Weight
(60-140 kDa) Low Molecular Weight (40-50 kDa)
Monofunctional Transpeptidases Class A Bifunctional PBP1A PBP1B PBP1C Class B Monofunctional Transpeptidases PBP2 PBP3 Type 4 PBP4 (DacB) Type 5 PBP5 (DacA) PBP6 (DacC) PBP6b Type 7 PBP7 Gram-negative PBPs (E. coli)
Type AmpH PBP 4b AmpH a" E. coli PBP1B E. coli PBP1A b"
There are even some bacteria without peptidoglycan and PBPs: Rickettsia, Mollicutes and Dehalococcoides.
1.5.2 Membrane Proteins as Drug Targets
Membrane proteins include receptors, ion channels, transporters, and enzymes, and constitute a significant fraction (20 %-30 %) of the proteome (Fagerberg et al., 2010). However they are highly underrepresented, and make up less than 1% of all protein structures solved. As of January 2017, there were 1194 published reports of membrane protein structures (671 unique) solved by NMR or crystallographic techniques.
1.5.3 Using Detergents to Study Membrane Proteins
Detergents are required for the solubility of membrane proteins and are needed to extract PBPs from their periplasmic associated environment. Despite Class A PBPs only having a single transmembrane (TM) domain, the transglycosylase domain is also embedded within the cytoplasmic membrane, to maintain a hydrophobic environment for its Lipid II substrate. Mainly non-ionic detergents are used to solubilise PBPs in this thesis. Non-ionic detergents are neutrally charged. The most common of the synthetic anionic surfactantsare based on the straight chain alkylbenzene sulfonates, which produce electrically negative colloidal ions in solution. Cationic detergents produce electrically positive ions in solution. Amphoteric detergents are capable of acting either as anionic or cationic detergents in solution depending on the pH of the solution. 1.5.4 E. coli PBP1A
E. coli PBP1A is a bifunctional PBP with both transglycosylase and transpeptidase functionality. Its structure has not been solved crystallographically yet, but the transglycosylase domain of Aquifex aeolicus PBP1A structure was solved in 2007 (Yuan et al., 2007). The A. aeolicus transglycosylase domain has also been crystallised in complex with Moenomycin (Fuse et al., 2010). E. coli PBP1A is likely to follow a similar structural pattern to H. influenzae PBP1A which was crystallised in December 2016 (Minasov et al., 2017). The structure of E. coli PBP1A is likely to consist of an N- terminal single α-helical TM domain, a transglycosylase domain, an Outer Docking
(ODD) Domain and a C-terminal transpeptidase domain. E. coli PBP1A interacts with the cell-elongation-specific Class B monofunctional transpeptidase PBP2 in vivo and localises mainly to the cylindrical wall of the cell, helping with cell elongation. It helps govern cell shape and inactivating PBP2 in E. coli causes cells to lose their rod shape and grow as enlarged spheres that eventually die unless compensatory mutations or conditions. PBP2 also binds to the outer-membrane lipoprotein LpoA, where the lipoprotein also helps in cell elongation. LpoA is attached to the outer-membrane and hangs down in to the periplasm to interact with PBP1A. The ODD domain of E. coli PBP1A and LpoA are restricted to γ-proteobacteria.
1.5.5 E. coli PBP1B
E. coli PBP1B is a major Class A bifunctional PBP that has been heavily studied over the past quarter century, as it is the workhorse of peptidoglycan synthesis, synthesising glycan chains de novo. E. coli PBP1B consists of five domains: (i) an N-terminal TM α- helix (residues 64-87) (ii) a UvrB domain 2 Homolog (109-200) (iii) a membrane- associated transglycosylase domain (203-367) (iv) a linker region connecting the transglycosylase and transpeptidase domains (391-443) (v) a C-terminal transpeptidase domain (444-736).
The X-ray crystal structure of E. coli PBP1B was solved in 2009 by Sung et al. The single α-helix TM region functions as both a signal sequence for secretion and a stop transfer signal that serves as a membrane anchor. It stabilises the PBP-membrane interaction, but stable protein-membrane interaction is not critical for normal function of E. coli PBP1B. Removal of the TM domain does not affect the structure of the transglycosylase domain in the binding site (Sung et al., 2009). The function of transglycosylases is similar to that of lysozyme (Terrak et al., 1999), in that lysozyme cleaves the β1-4 glycosidic bond between GlcNAc and MurNAc peptidoglycan units, which are initially generated by transglycosylases. The 2009 crystal structure shows the transglycosylase domain to be located at a site near to where the cell membrane would be, in relation to the overall structure of the enzyme. If this structural model is accurate, it would support the theory that the proximity of the enzyme to the membrane would assist in its interaction with the membrane bound substrate (Sung et al., 2009). The recent crystal structures of transglycosylase domains in complex with moenomycin and
various β-lactams have highlighted essential interactions but their significance in structure based drug design efforts (King et al., 2016).
The transglycosylase domain is mostly α-helical and is responsible for polymerising LII in to linear glycan strands. It is part of a group of glycosyl transfer proteins ordered by sequence homology in to the CAZY (carbohydrate-active enzymes) classification of enzymes. Peptidoglycan glycosyl transfer proteins are in family 51 (GT51),
characterised by five conserved motifs and the use of LII. 80 % of the CAZY database are bifunctional PBPs and 20 % are monofunctional. The transglycosylase domain is comprised of two regions - a larger globular ‘head’ and a smaller, flexible ‘jaw’ region, separated by a cleft (Lovering et al., 2007). The globular head subunit has homology to lysozyme and the flexible jaw has high hydrophobicity and interacts with the lipid bilayer and the lipidic substrate. Positioning of the two subunits forms a cleft lined with residues that are conserved, shown in E. coli PBP1B (Terrak et al., 2008), S. aureus MGT and A. aeolicus PBP1A, with several similar structural features between members of the TG51 family (Lovering et al., 2008) (Figure 1.10a). These residues are clustered in to five motifs (Figure 1.10 b and c), with the first three motifs residing in the cleft. The cleft contains the catalytic residues, with motifs 1 and 3 harbouring the two conserved glutamic acid residues and motif 2 being involved in substrate recognition. Motifs 4 and 5 play a structural role in the cleft.
!
Figure 1.10 The Conserved Motifs in the TG51 Structural Family, with this E. coli PBP1B Domain Bound to Moenomycin A, Residing in the Catalytic Cleft.(a) The alignment of the five conserved TG51 motifs in Gram-negative species (b) The full transglycosylase domain and α-helix of E. coli PBP1B, the beginning of the linker region between the transglycosylase and transpeptidase domains is in purple TM α-helix = cyan, transglycosylase domain = red the five conserved motifs are shaded as follows: Motif 1 = yellow, motif 2 = orange, motif 3 = pink, motif 4 = blue, motif 5 = violet. (c) The active site cleft of the transglycosylase domain. Conserved residues include the catalytic glutamates E233 in motif 1 and E290 in motif 3 (Terrak et al., 2008). (d) The E. coli PBP1B transpeptidase domain (blue) containing the β- lactam ligand, ampicillin (yellow), bound to the catalytic active site Serine 510 (magenta). (King et al., 2016).
a" Motif 1
Motif 1 Motif 2 Motif 3
Motif 4
Motif 5
d
b TG Domain c TG Active Site
Glu 233 Glu 290 TP Active Site Ser 510
The amino acids tryptophan and tyrosine are found at higher frequency at a lipid-water interface in membrane proteins (Yau et al., 1998) and were found at the bottom of the transglycosylase domain, indicating that it is partially embedded in the lipid bilayer. The residues in the transglycosylase domain that interact with transglycosylase
inhibitors such as Moenomycin and are also conserved between other transglycosylases are critical in antibiotic drug design.
The UvrB Homology (UB2H) domain interacts with the MltA protein. It also interacts with the lipoprotein LpoB along with the transpeptidase domain, which stimulates its cross-linking ability. E. coli LpoB and the UB2H domain of E. coli PBP1B are further restricted within γ-proteobacteria to the enterobacteria (Typas et al., 2010) and so are not present in Pseudomonas spp species. PBP1B-LpoB may work in parallel with the Tol-Pal complex to aid membrane constriction.
Three-dimensional structures of the four proteins of interest in this thesis: E. coli and P. aeruginosa PBP1A and PBP1B have not been solved yet, apart from the full-length E. coli PBP1B protein (King et al., 2017, Sung et al., 2009) (Figure 1.11 b). The
transpeptidase domain of P. aeruginosa PBP1A has been crystallised with a
sideromimic conjugated compound in its transpeptidase serine active site (Starr et al., 2014). The transglycosylase domain is missing due to intrinsic flexibility and tryptic cleavage. No structures at all have been solved for E. coli PBP1A or P. aeruginosa PBP1B. There is a total of two Class A PBPs from species other than E. coli and P. aeruginosa, for which structures have been resolved: the transglycosylase domain only of the bifunctional PBP1A from the thermophilic species Aquifex aeolicus (Yuan et al., 2007) and the full-length bifunctional PBP1A from the Gram-negative Haemophilus influenzae (Minasov et al., 2017) (Figure 1.11 a).
Figure 1.11 The Crystal Structures of the Two Gram-Negative Class A Bifunctional PBPs (PBP1A and 1B). (a) H. influenzae PBP1A and (b) E. coli PBP1B. In E. coli PBP1B, the enzyme resides in the periplasmic space between the inner and outer membranes of E. coli cells, with the transmembrane α-helix anchoring the PBP to the membrane. The transglycosylase domain is at the N-terminal and the transpeptidase domain is at the C-terminal. The transglycosylase active site glutamic acid residue is at position 233 and the transpeptidase active site is residue 510. Cyan = TM domain, red = transglycosylase domain, orange = ODD domain in HiPBP1A, yellow = UB2H domain in EcPBP1B, blue = transpeptidase domain. H. influenzae PBP1A has been crystallised without its TM domain. PDBs: 3VMA & 5HL9 for EcPBP1B (Sung et al., 2009, King et al., 2017) and 5U2G for H. influenzae (Minasov et al., 2017) (The 5U2G structure was deposited on to the Protein Data Bank (www.rcsb.org) in December 2016, but the associated article has not yet been published, thus the domain colouring of HiPBP1A in (a) has been estimated at this time).
1.5.6 P. aeruginosa as a Clinically Important Pathogen
Pseudomonas aeruginosa was chosen as a species to study, as it is a Gram-negative of high clinical relevance with a major resistance problem. The P. aeruginosa enzymes earlier in the peptidoglycan synthesis pathway has been mostly characterised, but its PBPs have not. There is a need for high-through-put assays that could be used for screening, therefore part of the project was to investigate different assay methods for PBPs.
P. aeruginosa is one of the most common Gram-negative micro-organisms isolated from nosocomial isolates in recent years. It commonly causes hospital-acquired infections (HAIs) including pneumonia, bloodstream, urinary tract and surgical-site infections. More than 6,000 (13%) of the 51,000 health care-associated P. aeruginosa infections that occur in the U.S. each year are multi-drug-resistant (MDR) (Rossolini et
b E. coli PBP1B a H. influenzae PBP1A
al., 2014). The associated PBPs could play a crucial role in β-lactam resistance in MDR Gram-negative bacteria. P. aeruginosa colonises on the surface of medical equipment in surgical theatres and infects patients during surgery; this pathogen also infects the wounds of burns victims, and persistently occupies the lungs of Cystic Fibrosis patients. It is an opportunistic pathogen for which treatment options are limited, but include broad-spectrum β-lactams (carbapenems) and is of increasing concern in clinical situations. Some Gram-negative organisms including P. aeruginosa have mechanisms for evading the action of β-lactam antibiotics including the acquisition of β-lactamases, provoking the emergence of MDR strains.
1.5.7 P. aeruginosa PBP1A
Pseudomonas aeruginosa have a full compliment of PBPs as E. coli, including the Class A PBPs 1A and 1B, three Class B enzymes (PBP2, PBP3, and PBP3a:
orthologues of the corresponding enzymes in E. coli) and the Class C PBPs 4 and 5. PaPBP3 shares 42% sequence identity with the corresponding protein in E. coli and has been identified as the primary target of a number of β-lactams used to treat
pseudomonal infections, including the cephalosporin analogues cefsulodin (Gotoh et al., 1990) and ceftazidime (O’Callaghan et al., 1980),piperacillin (Godfrey et al., 1981), and the parenteral carbapenem, doripenem (Davies et al., 2008). PBP1A follows the same structural format as E. coli PBP1A: The TM helix followed by a domain of unknown function, known in E. coli PBP1A as the ‘ODD’ domain. This is followed by a transglycosylase domain, linked by a β-rich linker to a transpeptidase domain. The interface between the transglycosylase and transpeptidase domains comprises of 6 sequential β-strands, with two serving as a linker between the two domains. The transpeptidase domain has a similar fold to other transpeptidases and serine β-
lactamases. Within the transpeptidase domain, there is an oligosaccharide/ nucleotide- binding (OB) fold, as in A. baumanii (Han et al., 2011). The crystal structure of the transpeptidase domain of PaPBP1A was solved in 2014 (PDB: 4OON) with a sideromimic-conjugated compound bound to its transpeptidase domain (Starr et al., 2014), achieved after proteolysis with trypsin. The construct crystallised was the full- length PBP minus the N-terminal TM domain: residues 36-822, to a resolution of 3.20 Å. The transglycosylase domain cannot be seen in the crystal structure due to intrinsic flexibility and tryptic cleavage.
1.5.8 P. aeruginosa PBP1B
Very little is published on Pseudomonas aeruginosa PBP1B, including any structural or functional information. It has been assumed thus-far that this enzyme behaves similarly to E. coli PBP1B and is likely to have a similar structure, with the main difference being that there is no LpoB protein present in Pseudomonas species and the LpoB protein is thought to have evolved along with the UB2H domain, as in E. coli PBP1B. Therefore it is also unlikely that PaPBP1B has a UB2H like that in E. coli.
1.5.9 P. aeruginosa Resistance to β-Lactams
Many β-lactam antibiotics are used in combination with β-lactamase inhibitors, as there is such widespread β-lactam resistance. Pseudomonas aeruginosa cells exhibit
resistance to β-lactams through over-expression of AmpC and OprD, as well as efflux pumps. Six P. aeruginosa nosocomial isolates were studied for resistance mechanisms, all of which lacked OprD and over-expressed AmpC, and five of them showed modified PBP profiles such as diminished expression of PBP1A and modified binding affinities to β-lactams (Moya, et al., 2012). Ceftaroline (a cephalosporin B-lactam) has a broad- spectrum activity against Gram-positive pathogens and several Gram-negative
pathogens, with the notable exceptions of P. aeruginosa (and extended spectrum β- lactamase (ESBL)-producing Enterobacteriaceae). Current antibiotics active and in use against Pseudomonas aeruginosa infections include the carbenicillin, the
cephalosporins cefepime and ceftazidime and the fluoroquinolone ciprofloxacin.! 1.5.10 Proteins that Interact with PBPs
PBP1B in particular makes a series of important protein-protein interactions (PPIs) that are important in cell metabolism providing a network of interactions. PBP1A is
involved in adding new sugar units to the peptidoglycan sacculus during cell elongation. PBP1B localises to the septum of a bacterial cell during cell division, forming a
complex machinery of enzymes as part of the Z-ring, the driving force behind constriction of the cell wall. Both of these activities require a plethora of enzymes to work together to coordinate these fundamental processes (Figure 1.12). Cell division proteins such as FtsN interact with PBP1B, building a network of protein interactions
that work together to polymerise peptidoglycan including during enlarging of the sacculus during cell division. FtsN interacts with E. coli PBP1B (Müller et al., 2007) as well as PBP3 (FtsI) (Bertsche et al., 2006) and FtsW (Fraipont et al., 2011; Derouaux et al., 2013, Derouaux et al., 2008). FtsN is thought to initiate cell division and divisome assembly, being the final component in the divisome complex (van der Ploeg et al.,