Structure of the Cystic Fibrosis transmembrane conductance regulator, what does this mean for future Cystic Fibrosis research?

By Bryony Ackroyd

Twitter: @BryonyAckroyd

In a previous blog post the implications of a mutated Cystic Fibrosis transmembrane conductance regulator (CFTR) receptor in Cystic Fibrosis (CF) was discussed, along with the pros and cons of the break through drug, Ivacaftor. Following on from this, in December 2016, the structure of the CFTR from zebrafish was determined via electron cryo-microscopy, how will this implicate future CF research?

CF is a genetic disease that affects 70,000 people worldwide and is characterised by an overly viscous mucus lining of the airways, resulting in difficulty in clearing the airways by coughing, and an increase in infections from opportunistic pathogens.  CF is caused by different mutations in the CFTR, an ABC transporter ion channel, which results in an imbalance in ion concentration and the observed phenotype of highly viscous mucus. The CFTR conducts chloride and as well as regulating other ion channels, such as chloride channels and glutathione transport. There are approximately 1900 known mutations within the CFTR, which is primarily expressed within the airway submucosal glands in the lungs.

Although the structure is of the zebrafish CFTR, the human and zebrafish CFTR share 55% sequence identity and 42 of the 46 mutations that cause CF are identical, making the zebrafish CFTR structure a useful tool for studying human CF.

The structure of the zebrafish CFTR is in the inwards facing conformation, i.e. open to the cytoplasm and closed to the outside of the cell. The electron microscopy (EM) density for the 12 transmembrane helices of the CFTR was good enough to unambiguously assign the amino acids. However, the density for the nucleotide binding domains (NBDs) was not as sharply resolved, therefore the crystal structures of the human and mouse NBDs were used as a way to guide model building of the zebrafish NBDs.

Structure of the zebrafish CFTR, determined via electron cryo-microscopy. The R domain and related density is shown in yellow, the Lasso motif is shown in red, transmembrane domin 1 in blue and transmembrane domain 2 in green. The lasso domain is shown to be partially integrated into the membrane and in close proximity to the R domain.


When determining the structure of the CFTR it was found to contain an “N-terminal interfacial structure” which has never previously been seen in an ABC transporter, it is referred to as the lasso motif.  The first 40 resides of the lasso motif are within the membrane and pack against one of the transmembrane helices. The part of the lasso motif extending outside the membrane forms a helix and tucks under helix one of the CFTR. Many of the mutations causing CF are found within the lasso motif region, highlighting its importance in the disease. Some hypotheses have suggested that the lasso motif regulates channel gating through interactions with the R domain, which fits well with the symptoms of CF. The R domain of the CFTR appears to inhibit the channel in the dephosphorylated state, this inhibition is reversed when the R domain is phosphorylated.

The missense CF-causing mutations were then mapped onto the structure of the CFTR, making it possible to categorise the mutations into 4 groups, pore construction mutations, folding mutations, ATPase site mutations and NBD/Transmembrane domain interface mutations. Pore construction mutations include mutations expected to alter the structure or electrostatics of the pore. Mutations that destabilised the CFTR and therefore caused folding mutations were classified as folding mutations.  ATPase site mutations comprised of mutations within the NBDs that are thought to interfere with ATP binding and the formation of the closed NBD dimer. NBD/transmembrane domain mutations cause defects in folding and gating and therefore impact on the transmission of conformational changes from the NBDs to the transmembrane domains.

The determination of the structure of the zebrafish CFTR has been a much needed breakthrough within the CF research field. For the first time researchers have been able to accurately pinpoint mutations involved in CF, giving a much greater insight into how these mutations cause the observed symptoms and allowing rational drug design to target these problem points.  This advancement can only be a positive thing for the future CF research.


Source: Zhang, Zhe et al., (2016). Atomic Structure of the Cystic Fibrosis Transmembrane Conductance Regulator. Cell, Volume 167 , Issue 6 , 1586 – 1597.e9.

Using Single Molecule FRET to Understand Substrate Binding Domains

By Bryony Ackroyd

Twitter: @BryonyAckroyd

ABC transporters can be either import or export systems for cells. They consist of two transmembrane domains (TMDs) and two cytoplasmic nucleotide binding domains (NBDs). ABC importers also use substrate binding domains (SBDs) or substrate binding proteins (SBPs). SBPs are separate proteins present in the periplasm, however SBDs are fused to the TMDs. Some ABC transporters even have two or three SBDs fused together in tandem. Although this is a known phenomenon, very little is understood about the system and how ABC transporters are able to interact with multiple and structurally distinct SBDs. In the work carried out in this particular paper the group focusses on GlnPQ from L. lactis, a Gram-positive bacterium, that imports asapargine, glutamine and glutamate via two different SBDs.

Although there are crystal structures and NMR data available for SBDs, not much is known about the mechanism of ligand binding e.g. induced fit or conformational selection. Bearing this is mind Poolman et al., used a unique combination of techniques to probe the conformational dynamics of the SBDs, single-molecule Forster resonance energy transfer (smFRET) coupled with isothermal titration calorimetry (ITC). Using this strategy the group was able to provide mechanistic insight into the transport mechanisms of ABC importers, showing that the SBDs of GlnPQ bind ligands via an induced-fit mechanism.

The SBDs can be in one of four states, closed-ligand bound (CL), open (O), partially closed (PC) or closed (C). The induced-fit mechanism of binding triggers the CL state from the O state, however in the conformation-selection model the SBD can be in the PC or C states without a ligand bound. Ligand binding stabilises the PC state and therefore pushes the SBD to the CL conformation. These differing conformational states were examined via smFRET and the changes between states was observed via FRET efficiency. The experiment was designed so that the O conformation of the SBD gave a low FRET efficiency and the closed conformations gave higher FRET efficiency. Fluorophores were designed on the SBDs to be between 3-6 nm apart in both the closed and open states.

The SBD1 of GlnPQ binds asparagine with high affinity and glutamine with low affinity whereas SBD2 solely binds glutamine with a high affinity.

Single molecule dynamics of SBDs probed with smFRET. (a) Schematic showing immobilisation of histidine tagged SBDs to a PEG-biotin coated surface in a flow cell. The surface scan on the right is shown in flase colour, orange indicates double-labelled SBDs, green is SBDs with only donor fluorophore and red is SBDs with only acceptor fluorophores. (b-d) Representative fluorescence time traces, blue is donor signal, red acceptor signal, grey FRET signal and orange is the fit. These graphs show that the FRET efficiency of SBD1 and SBD2 increased as the concentration of substrate increased. This indicates that closing of SBD1 and SBD2 increased as substrate concentration increased, in keeping with the induced-fit model.

By measuring the fluorescence emitted from an SBD immobilised on a surface when varying concentrations of substrate were added, it was possible to determine the conformational state of the SBD and therefore whether the induced-fit or conformational-selection model was being employed by the SBD. Poolman et al., showed that in the absence of ligand the SBDs of GlnPQ were continuously in the O conformation and not in the PC or C conformations, therefore demonstrating the induced-fit model is used by the SBDs of GlnPQ.

This clever and unique technique was able to beautifully show the different conformations of the SBDs and conformational changes that occur within SBDs during ligand binding. Hopefully this technique will be employed more widely in the future to elucidate ligands, binding mechanisms and conformations of other SBDs and SBPs.


Source: Poolman et al., (2015). Conformational dynamics in substrate-binding domains influences transport in the ABC importer GlnPQ. Nature Structural and Molecular Biology 22, 57–64.

Using a periplasmic binding protein as a biosensor for thiamine

By Sophie Rugg

Twitter: @sophiejrugg

Thiamine, also known as vitamin B1, is an essential micronutrient with an important role in metabolism for all life forms. Thiamine can’t be synthesised by animals, and so has to be obtained from their diet. Until recently, detecting thiamine was limited to either the use of expensive high-performance liquid chromatography (HPLC), or a slow assay involving microbial growth. This is because there is no antibody available that is specific for thiamine, making commonly used high throughput detection techniques such as enzyme-linked immunosorbent assay (ELISA) impossible.

Periplasmic binding proteins are components of the ABC transporters of Gram-negative bacteria, and bind their substrates with high affinity and specificity to enable them to be transported into the bacterial cell. These properties of high affinity binding and specificity make periplasmic binding proteins ideal for use as the recognition element in a biosensor. As bacterial ABC transporters are used for the import of nutrients, including a transporter for thiamine, many of these binding proteins have evolved to recognise small molecules which it may be difficult to raise an antibody against.

Edwards et al., (2016) developed a biosensor for thiamine based on the periplasmic binding protein for thiamine from Escherichia coli. This binding protein was incorporated into dye-encapsulating liposomes in order to amplify the signal. Immobilised to the surface of a streptavidin coated plate is biotin conjugated thiamine analogue. The thiamine analogue is connected to the biotin via a long polyethylene glycol (PEG)linker, so that the thiamine analogue doesn’t get in the way of the biotin binding the streptavidin coating. The immobilised thiamine analogue binds to the periplasmic binding protein with lower affinity than thiamine. After the thiamine containing sample has been added, any material not bound to the surface is removed. Any liposomes still stuck to the surface of the plate are lysed and the resulting dye concentration is inversely proportional to the thiamine concentration.

Overview of assay for thiamine detection taken from Edwards et al., (2016). Competitive assay with biotin conjugated thiamine analogue immobilized via streptavidin in microtiter plates and detected via periplasmic binding protein for thiamine conjugated to the lipid bilayer of dye encapsulating liposomes (left). After competition with sample thiamine, unbound materials are removed (middle) and liposomes remaining bound are lysed to release dye yielding a signal inversely proportional to thiamine concentration (right).

This work shows that periplasmic binding proteins can be used effectively in biosensors , particularly where there is no antibody available. With the wide range of periplasmic binding proteins evolved by bacteria to be able to transport nutrients into their cells, this technique is open to use across a wide range of applications provided that a suitable analogue of the substrate can be immobilised to the surface of a plate.

Source: Edwards et al., 2016. High-Throughput Detection of Thiamine Using Periplasmic Binding Protein-Based Biorecognition. Analytical  Chemisty88 (16), pp 8248–8256. DOI: 10.1021/acs.analchem.6b02092

ABC system for dietary oligosaccharides: a weapon of Bifidobacterium sp. in the ‘metabolic’ war of the gut

By: Constantinos Drousiotis

Twitter: @Ecolinnit

The human gut microbiota (HGM) is the community of microbes that thrive in the gastrointestinal tract. Lately, it has become evident that HGM has a profound impact on human health which has now attracted a great interest from the scientific community. Current research is aiming to understand the complicated metabolic interactions that exist in the microbiota which can reveal the type of imbalances in diet that could potentially lead to disease.

A rich diet in legumes and seeds leads to an increased population of Bifidobacteria in the HGM. This is ought to the fact that the latter can metabolise raffinose family oligosaccharides (RFO) as opposed to the human gut cells which are unable to. The specialised transport machinery enabling these bacteria to transport and utilise RFOs hasn’t been characterised previously.

The study carried out by Morten et al. aimed to characterise the substrate binding protein (SBP) of the ABC transport system that was expressed in response to growth on RFOs, ie. BlG16BP. The group solved the structure of the BlG16BP and showed that the binding pocket of the protein accommodates oligosaccharides with a glucosyl or galactosyl C4-OH at position 1(non-reducing glycosyl unit) and an α-glycosidic bond to a glucosyl moiety at position 2. Additionally, they showed that the fructosyl or glucosyl groups are tolerated well at position 3 because of the lack of direct polar contacts of the protein with the sugar at this position and additionally, the cleft’s open architecture. These are all features of the sugar structures of raffinose and panose.

Crystal structure of BlG16BP in complex with panose (A) and raffinose (B). The SBPs consist of an N-terminal domain (Domain 1, brown) and a larger C-terminal domain (Domain 2, green). The two domains are linked by hinge regions shown in light blue. Shown on right-hand side, is a close-up of the binding sites of BlG16BP in complex with panose (D) and raffinose (E). The non-reducing end glycosyl unit of both ligands (galactosyl in raffinose and glucosyl in panose) stacks onto Phe-392 (defined as position 1) and makes polar contacts to Asp-394, Asn-109, and His-395. Asp-394 is able to form hydrogen bonds to both the equatorial C4-OH of the non-reducing end glucosyl in panose and the axial C4-OH of the galactosyl in raffinose. The glucosyl moiety of raffinose and panose at position 2 stacks onto Tyr-291 and makes polar contacts to Lys-58, Glu-60, and Asp-326. The position 3 is tolerated well, as the glucosyl moiety of panose stacks onto Trp-216 with almost parallel planes of the sugar rings as opposed to the fructosyl group of raffinose which sits orthogonal against Trp216 due to a smaller area of Van der Walls contacts. As a point of reference, the oligosaccharide structures are provided on the bottom of the figure.


Growth assays with a mixture of RFOs as the carbon source revealed that raffinose and melibiose were utilised first in order throughout the course of growth assays of B. animalis subsp. lactis Bl-04, indicating that are preferentially recognised over tetra- and pentasaccharides. The group suggests that the preferential binding by this transporter could potentially reflect the levels of the respective sugars in the gut or reveal the ability of bifidobacteria to further process the tetra- and pentasaccharides extracellularly. Nonetheless, BlG16BP SBP has a lower affinity to the only two previously characterised oligosaccharide binding proteins from bifidobacteria; the lower affinity could indicate that RFOs are found in higher concentrations than the ligands of the other two oligosaccharide binding proteins.  Also, it could point out to the low level of competition for RFOs which would be agreeable with the fact that this ABC system is not phylogenetically diverse.

Notably, phylogenetic analysis showed, that as far as Lactobacillus species are concerned, this α-galactoside transporter is only found in the human-gut adapted clade of thereof. The lack of this system in Lactobacillus which thrive in other ecological niches suggests that the transporter was acquired by horizontal gene transfer as a survival strategy in response to the fierce microbial competition in the gut. This is in accordance with previous studies which point towards horizontal gene transfer viewed as an adaptation strategy to gut niche.

The study reports the first biochemical and structural insight into an ABC-associated glycoside transport protein and provides evidence that ABC-mediated uptake may confer a competitive edge in the fierce competition for metabolic resources in the human gut niche. Altogether, the findings improved our understanding of the impact of oligosaccharide uptake in preferential glycan utilization.


Source: Morten et. al., (2016) , An ATP Binding Cassette Transporter Mediates the Uptake of α-(1,6)-Linked Dietary Oligosaccharides in Bifidobacterium and Correlates with Competitive Growth on These Substrates. The Journal of Biological Chemistry, 291(38), 20220-231.

Eating the Poison

By Bryony Ackroyd

Twitter: @BryonyAckroyd

The oligopeptide permease system (Opp) is an ABC transporter that commonly transports peptides into Gram-positive and Gram-negative bacterial cells. However, it has been demonstrated that Opp can also transport the antibiotic GE81112. This means the bacteria are effectively “eating the poison” that will eventually kill them.

GE81112 belongs to a structurally novel class of antibiotics and is key in the fight against antibiotic resistance and “super-bugs”. The tetrapeptide antibiotic GE81112 binds the 30S ribosomal subunit and interferes with the binding of initiator fMet-tRNA to the 30S subunit therefore inhibiting protein synthesis.

When Maio et al., began testing the microbiological activity of GE81112 on a series of microorganisms they obtained a number of unusual results. For example, the same bacteria (S. aureus, B. subtilis and E. coli) that in complete media are insensitive to GE81112 were sensitive to GE81112 in minimal or chemically defined rich media. One explanation for these results could be that GE81112, once in the cytoplasm, was disrupting 30S subunit with a different efficiency, however in vitro studies disproved this theory.

It was then hypothesised that a possible inhibitory or inactivating molecule was present in the rich media, causing the discrepancies in antibiotic sensitivity between rich media and minimal media. It was also noted that in chemically defined complete medium the activity of GE81112 is only slightly reduced compared to minimal media, indicating that the ineffectiveness of GE81112 in complete medium is not due to the concentration of nutrients.

To test the above hypothesis a series of experiments were conducted. The activity of GE81112 was measured by the change in the minimum inhibitory concentration in different growth medias. Whereas addition of individual amino acids to the growth media did not have any influence on GE81112 activity, the addition of casamino acids resulted in an increase in the minimum inhibitory concentration of GE81112. The difference between these two results was put down to the presence of di-, tri- and oligopeptides in casamino acids that may compete with GE81112 for an import system.

Due to GE81112 being a tetrapeptide the dipeptide and tripeptide transport systems were ruled out and instead the oligopeptide transport system, Opp, was investigated. An E. coli opp- mutant and wild-type were grown on minimal medium agar plates with the addition of the GE81112 antibiotic. The opp- mutants were not inhibited by GE81112, whereas the wild-type cells produced a large halo of inhibition indicating that Opp is the means of import for GE81112. Further experiments were carried out showing that presence of the whole Opp transporter was necessary for transport and sensitivity to GE81112.

Although the antimicrobial activity of GE81112 is not very efficient on bacteria growing in rich media, due to the competition for the Opp transport systems by other oligopeptides, it is important for antimicrobial resistance as it has been shown to be effective against methicillin resistant bacteria. Evidence suggests that mutations altering the cytoplasmic antibiotic target of GE81112 are few and far between, indicating that bacterial resistance to GE81112 could be slowed if entry into the bacterial cell is not blocked by oligopeptides. Could it then be possible to modify GE81112 to enter the bacterial cell without the aid of Opp to improve GE81112 efficiency and reduce resistance?


Source: Gualerzi et al., (2016). The Oligopeptide Permease Opp Mediates Illicit Transport of the Bacterial P-site Decoding Inhibitor GE81112. Antibiotics, 5(2): 17.

A food poisoning bacterium could aid in the fight against multidrug resistant cancers

By Caroline Pearson

Twitter: @CarolineRosePea

Salmonella enterica serovar Typhimurium is a food borne bacterial pathogen that commonly causes gastroenteritis in humans. However, it has been found that this pathogen can selectively grow inside tumours and modulate many biochemical pathways. This resulted in its recognition as a possible tool in the treatment of cancer to deliver therapeutic agents directly to the source of the cancer following systemic infection. Although many applications for this surprisingly therapeutic pathogen have been suggested, translating them into clinical use has been a stumbling point due to the possibility of systemic infections or immune mediated toxic responses to the invading bacteria.

An alternative approach to delivering the live salmonella bacteria to a cancer patient is to identify the therapeutic agents produced by S.Typhimurium which allow it to modulate biochemical pathways and administer these directly to the patient without the risk of systemic Salmonella infection. This approach has been taken by Mercado-Lubo et al., who have identified the molecule responsible for reducing the levels of multidrug resistance (MDR) transporter P-glycoprotein (P-gp) in tumour cells which increases their susceptibility to chemotherapeutic drugs.

Upregulated P-gp expression is associated with poor prognosis in several types of cancer.  The P-gp protein is encoded by MDR1, and is a MDR ABC transporter responsible for one aspect of the MDR phenotype in cancer cells. Recent studies have found that S. Typhimurium was able to reduce levels of P-gp in cancer cells and that the Salmonella type III secretory system was essential for this modulation. Therefore, S. Typhimurium type III secreted effector proteins were screened for their ability to modulate P-gp resulting in the identification of SipA.

SipA is able to modulate P-gp by activation of caspase 3 which then cleaves the P-gp protein so that it can no longer be presented at the cell surface to function as a drug efflux pump.

caroline blog post
Working model of SipA downregulation of P-gp taken from Mercado-Lubo et la., 2016. (a) Cancer cells express different types of ABC transporters, especially P-gp, to gain multidrug resistance. This allows tumour cells to extrude cytotoxic drugs from the intracellular space. (b) The SipA-AuNP may act extracellularly, by interacting with a transmembrane receptor to induce a CASP3 dependent cleavage of P-gp. The activation of caspase-3 also results in apoptosis; a cell death process. (c) Cleavage of P-gp results in the appearance of two protein fragments of about 90 and 60 kDa. Such cleavage destroys the P-gp scaffold essentially removing this transporter from the plasma membrane thereby preventing the active efflux of doxorubicin and enhancing its cytotoxic activity.

To harness the therapeutic potential of this effector protein without having to infect patients with potentially pathogenic S. Typhimurium, Mercado-Lubo et al., built a Salmonella nanoparticle mimic by fusing an inert gold nanoparticle with multiple copies of the SipA protein.  In vitro and in vivo studies both showed that the SipA nanoparticle possessed the ability to reduce P-gp levels in multiple cancer cell lines and increase their susceptibility to treatment with doxorubicin (a chemotherapeutic drug). The nanoparticle structure also enhanced SipA functionality in comparison to free SipA, presumably due to the nanoparticle complex stabilising SipA and preventing its degradation before reaching its target.

The writers suggest that this semi-synthetic Salmonella nanoparticle mimic could be applied to various chemotherapeutic drugs to overcome MDR in tumours and that the findings represent an important step forward in demonstrating the potential of this strategy as a ‘stand alone’ approach to increase cancer cell sensitivity to conventional chemotherapeutics.


Source: Mercado-Lubo, R., Zhang, Y., Zhao, L., Rossi, K., Wu, X., Zou, Y., Castillo, A., Leonard, J., Bortell, R. & other authors. (2016). A Salmonella nanoparticle mimic overcomes multidrug resistance in tumours. Nat Commun 7, 12225. Nature Research.

ABC Transporter is a Key Component in Bacitracin Resistance

In the past few years a number of bacterial transport proteins have been shown to act as co-sensors for signal transduction pathways. This process generally occurs via a protein-protein interaction between the membrane bound sensor domain, which binds specific substrates, and the signalling domain, which transfers the signal information into the cytoplasm of the cell.

In this paper by Dintner et al., as well as in previously published studies, it has been shown that in the absence of the transporter component these signal transduction pathways are rendered inactive. This is due to signalling activation being entirely dependent on a sensory transporter sensing its specific substrate. All currently known examples of these systems are involved in resistance to antibiotics and the role of a transporter in signalling is conserved.

The system used in this paper to investigate this phenomenon in greater detail was the BceRS-BceAB system from the Bacillus subtilis bacterium, which confers resistance against the antibiotic bacitracin. The BceRS component of the system is a two-component regulatory system (TCS) and the signal transduction domain, whereas BceAB is an ABC transporter and the sensor domain. It is not known exactly how BceRS-BceAB confers resistance to bacitracin, however it is possible that it’s sequestered into the cytoplasm via the BceAB ABC transporter. Bacitracin is known to inhibit both cell wall and peptidoglycan synthesis in bacteria.

Schematic model diagram showing BceAB and BceRS. BceAB constitutes the ABC transporter sensory domain, whilst BceRS constitutes the TCS signal tansduction domain. Double headed arrows indicate direct interactions between domains. Dotted arrows indicate transcription events. BceAB and BceS interact within the membrane. ATP hydrolysis by BceAB causes activation of BceS which allows phosphorylation of BceR. BceR then triggers increased production of BceAB. Taken from Dintner et al., 2014.

It has previously been shown that BceS, the histidine kinase component, is unable to detect the presence of bacitracin without BceAB, the ABC transporter component. This therefore lead to the assumption that BceAB is the sensory part of the system.

Initial experiments showed clear interactions between BceS and BceB or BceAB, however BceA was not observed to interact with any components of the TCS (BceS, BceR or BceRS). Dintner et al., also showed that BceR production in the absence of BceS resulted in a lack of interaction with the transporter (BceA, BceB or BceAB). This lead to the conclusion that BceS and BceAB form a scaffold that allows BceR to interact with the complex. Addition of the bacitracin antibiotic did not appear to have an effect on complex formation.

Following on from these discoveries the group wanted to identify whether BceAB, the ABC transporter, interacted directly with the substrate bacitracin or not. They investigated this via surface plasmon resonance (SPR) spectroscopy. This technique uses light diffracted off the underside of a surface containing the molecule of interest to create a spectrum. The change in this spectrum as a substrate is added to the surface, and possibly binds the molecule of interest, can be measured accurately along with the association and dissociation rates.

Unfortunately the BceAB complex was unstable under the SPR conditions and so BceB alone was used in the studies. Zn2+-bacitracin, the active form of the antibiotic, was used as the substrate along with the peptide nisin as a nonsubstrate control. The KD of Zn2+-bacitracin under steady state was calculated to be 60nM, whilst nisin showed no binding to the BceB. Interestingly the absence of Zn2+ prevented bacitracin binding BceB, giving further evidence of the specificity of BceB to the active peptide, Zn2+-bacitracin. The data obtained from these experiments show that the transporter, BceAB, binds free Zn2+-bacitracin specifically and with high affinity.

Dintner et al., conclude by stating that they have proposed a “working model for the mechanism of signal transduction within Bce-like models”. Bce-like systems “represent widely spread resistance determinants against peptide antibiotics in Firmicutes bacteria” and therefore make this study important in the war against antibiotic resistance.


Source: Dintner et al., (2014). A sensory complex consisting of an ATP-binding cassette transporter and a two-component regulatory system controls bacitracin resistance in Bacillus subtilis. The Journal of Biological Chemistry, 289(40)27899-910.

Bryony Ackroyd

Twitter: @BryonyAckroyd