Abstract
Prolonging the clinical effectiveness of β-lactams, which remain first-line antibiotics for many infections, is an important part of efforts to address antimicrobial resistance. We report here that inactivation of the predicted D-cycloserine (DCS) transporter gene cycA gene re-sensitizes MRSA to β-lactam antibiotics. The cycA mutation also resulted in hyper-susceptibility to DCS, an alanine analogue antibiotic that inhibits the alanine racemase and D-Alanine ligase enzymes required for D-Alanine incorporation into peptidoglycan. Amino acid transport studies showed that CycA functions as an alanine permease, and the cycA mutation was accompanied by increased accumulation of muropeptides with tripeptide stems lacking the terminal D-Ala-D-Ala, and reduced peptidoglycan cross linking. Exposure of MRSA to DCS was also associated with a dose-dependent accumulation of muropeptides with tripeptide stems and reduced PG crosslinking. Because impaired alanine transport or DCS-treatment have similar effects on PG structure, synergism between β-lactams and DCS was investigated. Therapeutically achievable concentrations of DCS re-sensitised MRSA to β-lactam antibiotics in vitro and significantly enhanced MRSA eradication in a mouse bacteraemia model. The susceptibility of several other ESKAPE group pathogens to β-lactams was also increased by DCS. These data reveal the potential of DCS, or new drugs targeting CycA, to facilitate the reintroduction of β-lactam antibiotics for the treatment of infections caused by MRSA.
Author Summary Treatment options for infections caused by ESKAPE group pathogens continue to dwindle with increasing antimicrobial resistance. Particularly important are the β-lactams, which remain first line antibiotics for many infections, and finding new ways to maintain their clinical effectiveness is a priority. Here we report that mutation of cycA in MRSA, which was found to encode an alanine permease, was associated with hyper-susceptibility to β-lactam antibiotics and D-cycloserine (DCS), used in the treatment of multi-drug resistant tuberculosis infections. DCS re-sensitised MRSA to oxacillin and other β-lactams in vitro and significantly enhanced oxacillin activity in a mouse model of bacteraemia. DCS also sensitised other ESKAPE pathogens to β-lactams revealing the therapeutic potential of DCS as an adjunct drug for these important antibiotics.
Introduction
Whilst many bacteria can exhibit resistance to select antimicrobials, isolates of the human pathogen Staphylococcus aureus can express resistance to all licensed anti-staphylococcal drugs. This results in significant morbidity and mortality, with up to 20% of patients with systemic methicillin resistant S. aureus (MRSA) infections dying, despite receiving treatment with anti-staphylococcal drugs [1]. Although aggressive hospital infection prevention and control initiatives appear to be having a positive impact on hospital-acquired MRSA rates in developed countries, S. aureus infections caused by community-associated MRSA strains and strains that are currently methicillin susceptible remain stubbornly high [2],[3].
As part of our efforts to identify improved therapeutic approaches for MRSA infections, we recently described the novel use of β-lactam antibiotics to attenuate the virulence of MRSA-induced invasive pneumonia and sepsis [4]. We demonstrated that oxacillin-induced repression of the Agr quorum-sensing system and altered cell wall architecture resulted in downregulated toxin production and increased MRSA killing by phagocytic cells, respectively [4]. Supporting this in vitro data, a randomised controlled trial involving 60 patients showed that the β-lactam antibiotic flucloxacillin in combination with vancomycin shortened the duration of MRSA bacteraemia from 3 days to 1.9 days [5, 6].
Because expression of methicillin resistance in S. aureus impacts fitness and virulence and is a regulated phenotype, further therapeutic interventions may also be possible. The complexity of the methicillin resistance phenotype is evident among clinical isolates of MRSA, which express either low-level, heterogeneous (HeR) or homogeneous, high-level methicillin resistance (HoR) [7–9]. Exposure of HeR isolates to β-lactam antibiotics induces expression of mecA, which encodes the alternative penicillin binding protein 2a (PBP2a) and can select for accessory mutations resulting in a HoR phenotype, including mutations that affect the stringent response and c-di-AMP signalling [10–14]. Because accessory genes can influence the expression of methicillin resistance in MRSA, targeting the pathways associated with the accessory genes may identify new ways to increase the susceptibility of MRSA to β-lactams. To pursue this, we performed a forward genetic screen to identify loci that impact the expression of resistance to β-lactam antibiotics in MRSA. Using the Nebraska Transposon Mutant Library, which comprises 1,952 sequence-defined transposon insertion mutants [15], inactivation of an amino acid permease gene was found to reduce resistance to cefoxitin, the β-lactam drug recommended by the Clinical and Laboratory Standards Institute for measuring mecA-mediated methicillin resistance in MRSA isolates. Disruption of this permease also resulted in hyper-susceptibility to D-cycloserine (DCS), an alanine analogue antibiotic that interferes with two enzymatic steps in the D-Alanine pathway of peptidoglycan biosynthesis. At therapeutically achievable concentrations, DCS was shown to increase the susceptibility of MRSA and other ESKAPE pathogens to β-lactam antibiotics. Peptidoglycan analysis revealed that the cycA mutation or exposure to DCS results in loss of the terminal D-Ala-D-Ala from the stem peptide and reduced cross-linking of the cell wall. Given that β-lactam antibiotics remain among the most effective, safe and affordable antibiotics, these data identify CycA as a new drug target to enhance the activity of β-lactam antibiotics.
Results
Mutation of cycA increases the susceptibility of MRSA to β-lactam antibiotics
To identify new ways of controlling expression of methicillin resistance, we sought to identify novel mutations involved in this phenotype. To achieve this, an unbiased screen of the Nebraska Transposon Mutant Library (NTML) was performed using a disk diffusion assay to identify mutants with increased susceptibility to cefoxitin. The parent strain of the library (JE2, a derivative of USA300 LAC cured of plasmids p01 and p03 [15]) was used as a positive control. Phage 80α-mediated backcross of candidate transposon mutations into the parent strain JE2, together with genetic complementation were used to confirm the role of putative genes involved in cefoxitin resistance. Only mutants verified by both transduction and complementation were examined further. Among the mutants identified was NE810 (SAUSA300_1642) (Fig. 1A), which exhibited significantly increased susceptibility to cefoxitin as determined by a disk diffusion assay (Fig. S1A) and a >500-fold increase in susceptibility to oxacillin as determined by E-test (Fig. 1B). Consistent with this, NE810 was recently reported to be more susceptible to amoxicillin [16]. RT-qPCR analysis revealed that expression of mecA was not significantly affected in NE810 compared to JE2 (Fig. S1B), whilst whole genome sequence analysis of NE810 further revealed that the SCCmec element was fully intact and that no other mutations were present (data not shown). These findings implied that increased cefoxitin susceptibility was not attributable to a failure of this strain to express a functional mecA gene product. Transduction of the cycA transposon insertion into the USA300 strains JE2, FPR3757 and LAC-13C (all SSCmec type IV, CC8), as well as DAR173 (SSCmec type IV, CC5) [17], DAR22 (SSCmec type III, CC5) [17] and DAR169 (SSCmec type I, CC8)[17] was also accompanied by significant increases in cefoxitin and oxacillin susceptibility (Table 1).
Mutation of cycA increases susceptibility to D-cycloserine
SAUSA300_1642 is annotated as a D-serine/L- and D-Alanine/glycine transporter with homology to CycA in Mycobacterium tuberculosis [18, 19]. Nonsynonymous point mutations in cycA contribute, in part, to increased D-cycloserine (DCS) resistance in Mycobacteria [18, 19], prompting us to investigate the susceptibility of the S. aureus cycA mutant to DCS. In contrast to the observations in Mycobacteria, our data showed that NE810 was significantly more susceptible to DCS than the wild type JE2 and was successfully complemented by the wild-type cycA gene (Fig. 2A). By broth dilution, the DCS MIC decreased from 32 μg/ml in JE2 to 2 μg/ml in NE810 (Table 1).
The cycA mutants of the MRSA strains LAC-13C (data not shown), USA300, DAR173, DAR22 and DAR169 exhibited similar and significant increases in susceptibility to DCS (Table 1). Transduction of the cycA allele from NE810 into the MSSA strains 8325-4 and ATCC29213 was also associated with a reduction in the DCS MIC from 32 to 4 μg/ml. DCS (a cyclic analogue of alanine and serine) is a broad-spectrum antibiotic produced by several Streptomyces species used as a second line drug in the treatment of TB infections in humans. This drug inhibits the alanine racemase enzyme that converts L-Alanine to D-Alanine, as well as the ligase enzyme that links two D-Alanine amino acids together [20]. These are important enzymes in both peptidoglycan (cell wall cross-linking) and teichoic acid biosynthesis in Gram-positive bacteria. A mutant in the putative D-Alanyl:D-Alanine ligase (ddl, SAUSA300_2039) is not available in the NTML library, suggesting that this gene may be essential. However, inactivation of the alanine racemase (alr) gene in the NTML library mutant NE1713 was associated with significantly increased susceptibility to cefoxitin (Fig. S1A) and hyper-susceptibility to DCS (Fig. S1C), which was consistent with the important role for D-Alanine in S. aureus peptidoglycan biosynthesis and β-lactam resistance.
Screening a subset of the NTML library containing known/predicted membrane transporters, failed to identify any mutation(s) associated with markedly increased DCS resistance. Two mutants, NE923 (YnfA) and NE1025 (ThrE), which were identified only exhibited a DCS MIC of 64 μg/ml, up from 32 μg/ml (data not shown) suggesting that DCS may be transported inefficiently by multiple transport systems in S. aureus.
CycA is an alanine permease that is required for D-Ala-D-Ala incorporation into the peptidoglycan stem peptide
To investigate the role of CycA as a potential permease, JE2 and NE810 were grown for 8 h in chemically defined media containing 14mM glucose (CDMG) and amino acid consumption in spent media was measured. Although no growth rate or yield difference was noted between JE2 and NE810 in CDMG (Fig. 2B), alanine uptake by NE810 was significantly impaired compared to JE2 (Fig. 2C). Utilisation of other amino acids by NE810 and JE2, including serine and glycine, were similar (Figs. 2D and 2E and Fig. S2). Impaired alanine transport in the cycA mutant grown in CDMG correlated with increased susceptibility to oxacillin (1 mg/l) (Fig. 2F). These data demonstrate for the first time that CycA in S. aureus functions as an alanine permease.
Scanning and transmission electron microscopy revealed gross differences in whole cell morphology (Fig. 3A and B) and cell wall thickness (Fig. 3C and D) in NE810 compared to JE2. NE810 exhibited a highly irregular cell shape, and many cells appeared collapsed (Fig. 3B). TEM imaging revealed that NE810 cells had a wrinkled cell surface (Fig. 3D), while cell wall measurements revealed that the NE810 cell wall (19 ± 2.5nm) was thinner than the JE2 cell wall (26 ± 1.8nm), although cell size (mean cell diameter) was unaffected (Table S2).
Quantitative peptidoglycan compositional analysis was performed using UPLC analysis of muraminidase-digested muropeptide fragments extracted from exponential phase cultures of JE2 and NE810 grown for 220 mins in TSB media (Fig. S3). The PG profile of the cycA mutant revealed a significant accumulation of tripeptides compared to wild-type JE2 (Fig. 4A, B), which was associated with a significant reduction in crosslinking (Fig. 4C). In NE810, the dimer, trimer and tetramer fractions were decreased, which was accompanied by a concomitant increase in the monomer fraction (Fig. 4D). Consistent with this data, exposure of JE2 to DCS 8μg/ml was also associated with a similar accumulation in muropeptides with tripeptide stems (Fig. 4A, B), reduced cross-linking (Fig. 4C), increased muropeptide monomers and reduced dimers, trimers and tetramers (Fig. 4D). DCS had a strong dose-dependent effect, and at concentrations of 20 and 32 μg/ml, led to amplified effects on the accumulation of muropeptides with tripeptide stems, reduced cross-linking and accumulation of monomers (Fig. 4 A-D). These data are consistent with the previously reported impacts of sub-inhibitory (0.25× MIC) and 4× MIC DCS concentrations, which were accompanied by incorporation of an incomplete stem peptide (tripeptide) [20] and a 4-fold reduction in D-Ala-D-Ala levels [21], respectively. Collectively these data indicate that impaired D-Ala incorporation into the PG stem peptide in the cycA mutant or following exposure to DCS is accompanied by reduced PG cross-linking and increased β-lactam susceptibility.
Mutation of cycA or exposure to D-cycloserine increases the susceptibility of MRSA to β-lactam antibiotics
Previously reported synergy between DCS and β-lactam antibiotics [20, 22] suggests that impaired alanine uptake in the cycA mutant may have the same impact on cell wall biosynthesis as DCS-mediated inhibition of alanine racemase and D-Alanine ligase activity. To further investigate this, we compared the activity of DCS and β-lactam antibiotics, alone and in combination, against JE2 and NE810. Checkerboard microdilution assay fractional inhibitory concentration indices (ΣFICs ≤0.5) revealed synergy between DCS and several licensed β-lactam antibiotics with different PBP selectivity against JE2 and USA300 FPR3757 (Table 1). Oxacillin and nafcillin were not included in checkerboard assays because measurement of their MICs involves supplementing the media with 2% NaCl, which distorts the MIC of DCS.
Using the MRSA strains JE2, USA300, DAR173, DAR22, DAR169 and their corresponding cycA mutants, the kinetics of killing by DCS, oxacillin and cefoxitin, alone and in combination was measured over 24h using antibiotic concentrations corresponding to 0.125×, 0.25× and 0.5× MICs. Overnight cultures were adjusted to approximately 107 CFU/ml before being exposed to oxacillin, cefoxitin and DCS alone and in combination. Recovery of growth in media supplemented with oxacillin or cefoxitin alone was evident after 8 h (Fig. 5A-E), reflecting the selection and expansion of HoR mutants as described previously [4, 21, 23]. Using combinations of DCS and oxacillin or cefoxitin at 0.125× MIC did not achieve a ≥2 log10 reduction in the number of CFU/ml (data not shown). However, at 0.5× MIC for strains JE2, USA300, DAR73 and DAR22, DCS (16 μg/ml)/oxacillin (32 μg/ml) and DCS (16 μg/ml)/cefoxitin (32 μg/ml) combinations achieved a ≥5 log10 reduction in the number of CFU/ml compared to oxacillin, cefoxitin or DCS alone (Fig. 5A-D). For strain DAR169, DCS/β-lactam combinations at 0.25× MIC was sufficient to achieve a ≥5 log10 reduction in CFUs recovered compared to the individual antibiotics (Fig. 5E). DCS/β-lactam combinations at 0.5× MIC were also able to achieve ≥5 log10 reduction in the number of CFU/ml against the methicillin resistant S. epidermidis (MRSE) strain RP62A [24] compared to either antibiotic alone (Fig. S4). Checkerboard experiments with fourteen MRSA strains and MRSE strain RP62A further revealed synergy (ΣFICs ≤0.5) between DCS and a range of β-lactam antibiotics with different penicillin binding protein (PBP) specificity, namely cefoxitin (PBP4), cefaclor (PBP3), cefotaxime (PBP2), piperacillin-tazobactin (PBP3/β-lactamase inhibitor) and imipenem (PBP1) (Table 1).
Preliminary studies using disk diffusion assays also revealed synergy between DCS and imipenem (IMP) or piperacillin-tazobactin (Pip/Taz) against several ESKAPE pathogens. Using checkerboard assays synergy (ΣFIC values ≤0.5) was measured between DCS and Pip/Taz or IMP against Pseudomonas aeruginosa LES431, Acinetobacter baumannii DSM 30007, K. pneumoniae 511232.1 (ESBL+), and two vancomycin resistant Enterococcus strains (Table 2).
In MRSA, these data suggest that mutation of cycA has similar effects to DCS exposure, identifying the CycA permease as a new therapeutic target to potentially increase the susceptibility of these pathogens to β-lactam antibiotics. The DCS concentrations used in these experiments with wild-type MRSA strains were within the therapeutic range administered orally to patients (10-15 mg/kg/day to maintain a blood concentration of 25-30μg/ml), and significantly lower for cycA mutants. The therapeutic concentrations of DCS could be further reduced if used in combination with an agent inhibiting CycA activity. This synergy appears to be specific to β-lactams and no synergy (ΣFICs > 0.5) was measured between DCS and several antibiotics that are used topically or systemically for the decolonization or treatment of patients colonized/infected with S. aureus or MRSA (clindamycin, trimethoprim, mupirocin, ciprofloxacin), or several antibiotics to which S. aureus isolates commonly exhibit resistance (tobramycin, kanamycin and spectinomycin) (Table S1). Furthermore the NE810 cycA transposon mutation had no impact on susceptibility to any of these non-β-lactam antibiotics (apart from clindamycin resistance which is encoded by the ermB gene on the transposon).
Combination therapy with DCS and oxacillin significantly reduces the bacterial burden in the kidneys and spleen of mice infected with MRSA
The virulence of the NE810 mutant and the therapeutic potential of oxacillin in combination with DCS in the treatment of MRSA infections were assessed in mice. Bacteraemia was established via tail vein injection with 5 × 106 CFU of JE2 or NE810, and the infection allowed to proceed for 16 hours. Mice were treated with 75 mg of oxacillin/kg/12 h, 30 mg of DCS/kg/12 h, a combination of both oxacillin and DCS, or left untreated (sham). The antibiotics were administered subcutaneously every 12 hours and the infection terminated after 5 days. Mice infected with NE810 displayed more piloerection than those infected with wild-type JE2 suggestive of more severe infection, and one mouse infected with NE810 required euthanization during the experiment.
The average number of CFUs recovered from the kidneys of mice infected with NE810 (1.5 × 108 ± 1.49 × 108) was higher than for mice infected with JE2 (5.7 × 107 ± 6.3 × 107), but did not reach significance (p=0.07). Interestingly, NE810 colonies exhibited visibly increased haemolytic activity on sheep blood agar (Fig. S5A). We previously reported that increased haemolytic activity in MSSA strains was associated with increased expression of the cell-density controlled, global accessory gene regulator (agr) system [4, 21, 25]. Comparison of agr (RNAIII) temporal expression using RT-qPCR revealed a more significant activation of the agr system after 6 hrs growth in NE810 compared to JE2 (Fig. S5B). This difference in agr activation was not related to any change in growth of NE810 compared to JE2 (Fig. S5C). Taken together these data indicate that increased β-lactam susceptibility in the NE810 cycA mutant is accompanied by both increased agr expression and haemolytic activity, and a trend towards increased virulence in this model.
Treatment with oxacillin or DCS alone significantly reduced the number of CFUs recovered from the kidneys of mice infected with JE2 (Fig. 6A) or NE810 (Fig. 6B). The therapeutic effectiveness of oxacillin or DCS was significantly better for NE810 infections (p≤0.0001) (Fig. 6B) than infections caused by wild type JE2 (p≤0.01) (Fig. 6A), which is consistent with the increased susceptibility of the cycA mutant to both antibiotics in vitro. The oxacillin/DCS combination was significantly more effective than either antibiotic alone and the combination was equally effective in reducing the bacterial burden in the kidneys of animals infected with JE2 or NE810 when compared to no treatment (p≤0.0001) (Fig. 6A and B). These data support the in vitro observations by demonstrating i) the increased susceptibility of NE810 to oxacillin or DCS, ii) the anti-MRSA activity of DCS and iii) the capacity of DCS to significantly potentiate the activity of β-lactam antibiotics against MRSA under in vivo conditions.
Discussion
The exploitation of antibiotic re-purposing as part of concerted efforts to address the antimicrobial resistance crisis has been hampered by a lack of mechanistic data to explain demonstrated therapeutic potential and the perception that studies attempting to identify new uses for existing drugs are not hypothesis-driven. However, basic research discoveries that advance our understanding of bacterial virulence and antibiotic resistance mechanisms can also be used to inform targeted antibiotic re-purposing, in which the mechanism of action is understood. In this study, we identified an alanine permease (CycA) required for full expression of resistance to β-lactam antibiotics and DCS. Loss of function of this alanine transporter significantly increased the susceptibility of MRSA to β-lactam antibiotics, an outcome that could be reproduced through exposure to DCS. DCS targets the early steps in cell wall biosynthesis by inhibiting the activity of Alr and Ddl.
The alanine permease identified in this study, CycA, shares 46% and 53% identity with CycA in M. tuberculosis and E. coli, respectively. Mutation of cycA increases the susceptibility of MRSA to β-lactam antibiotics and results in hyper-susceptibility to D-cycloserine, whereas a cycA point mutation in M. bovis contributes, in part, to increased DCS resistance presumably by interfering with DCS transport into the cell [19]. In E. coli, cycA null mutations can also result in increased resistance or have no effect in DCS susceptibility depending on the growth media [26–30], suggesting that CycA is primarily important for DCS resistance under conditions when its contribution to amino acid transport is also important. The opposite effects of cycA mutation on DCS susceptibility in S. aureus and E. coli or Mycobacteria indicates that DCS uptake is not significantly blocked by the cycA mutation in S. aureus. We have identified several predicted membrane-associated transport proteins involved in DCS susceptibility. However, mutations in these genes, alone or in combination only resulted in a 2-fold increase in DCS resistance, suggesting that DCS is transported into S. aureus cells via multiple routes.
Using chemically defined media supplemented with glucose, transport of alanine was impaired in the cycA mutant, which correlated with increased susceptibility to β-lactam antibiotics, suggesting that alanine utilisation via CycA is important for cell wall integrity and consequently resistance to β-lactam antibiotics. Mutation of cycA or DCS-exposure have similar effects on the structure of S. aureus peptidoglycan. Consistent with previous studies in S. aureus [20] and in M. tuberculosis [31], our studies showed a dose-dependent accumulation of muropeptides with a tripeptide stem in MRSA exposed to DCS. The cycA mutation was also associated with the increased accumulation of muropeptides with a tripeptide stem. These data indicate that a reduced intracellular alanine pool or inhibition of the D-Alanine pathway is associated with reduced D-Ala-D-Ala incorporation into the PG stem peptide. The increased accumulation of tripeptides in turn interferes with normal PBP transpeptidase activity and offers a plausible explanation for increased susceptibility to β-lactam antibiotics. This conclusion is reminiscent of a previous study, which demonstrated that growth of a HoR MRSA strain in media supplemented with high concentrations of glycine resulted in decreased methicillin resistance [32]. Peptidoglycan composition analysis revealed that under these growth conditions the D-Ala-D-Ala in the stem peptide was replaced by D-Ala-Gly indicating a central role of the terminal D-Ala-D-Ala in β-lactam resistance [32].
The similar effects of a cycA mutation and DCS treatment on the structure of peptidoglycan and β-lactam susceptibility identify the CycA permease as a new therapeutic target. New drug leads that interfere with CycA activity, perhaps based on analogues of substrates transported by the permease, should increase susceptibility of MRSA to both β-lactams antibiotics and DCS. Because β-lactam resistance in S. aureus is linked mainly to changes in the cell wall and because production of the D-Ala-D-Ala terminus of the peptidoglycan stem peptide is dependent on alanine transport and Alr/Ddl activity, we propose that combination of DCS, β-lactams and/or new drugs targeting CycA have therapeutic potential for the treatment of infections caused by MRSA and potentially other ESKAPE pathogens (Fig. 7).
The therapeutic potential of β-lactam / DCS combinations in the treatment of infections caused by MRSA is particularly important given that S. aureus isolates resistant to all licensed anti-staphylococcal drugs have been reported, and in light of the regulatory barriers to the introduction of new antimicrobial drugs. The excellent safety profile of β-lactam antibiotics makes these drugs particularly attractive as components of combination antimicrobial therapies. When used in the treatment of tuberculosis DCS is typically administered orally in 250 mg tablets twice daily for up to two years. At this dosage, the DCS concentration in blood serum is 25-30 μg/ml. The known neurological side effects associated with DCS therapy mean that this antibiotic is unlikely to be considered for the treatment of MRSA infections unless alternative therapeutic approaches have been exhausted. DCS is a co-agonist of the N-methyl-D-aspartic acid receptor in the central nervous system can include seizures and peripheral neuropathy [33, 34], and the use of this antibiotic is limited to the treatment of recalcitrant, multi-drug resistant tuberculosis infections, for which long-term therapy is required [35]. However in the context of MRSA infections, our in vivo data with a mouse bacteraemia model demonstrated the significant effectiveness of DCS over a 5-day therapeutic window. Administration of DCS alone at a therapeutically relevant dose significantly enhanced the eradication of MRSA infections. Infections caused by the cycA mutant were eradicated more effectively by oxacillin or DCS alone than wild type infections further suggesting that the CycA permease may also represent a novel drug target as part of strategies to overcome β-lactam resistance. Oxacillin/DCS combination therapy was significantly more effective than DCS or oxacillin alone against wild type and cycA mutant infections. The anti-MRSA activity of DCS and its ability to significantly potentiate the active of oxacillin under in vivo conditions, together with its therapeutic effectiveness over a very short time frame may outweigh concerns about DCS toxicity and suggest that DCS/β-lactam combination therapy as a treatment option for recalcitrant staphylococcal infections merits further consideration. DCS may also serve as a scaffold from which to synthesise more active and less toxic derivatives, while DCS concentrations may be further reduced if used in combination with an agent that inhibits CycA activity.
Materials and Methods
Bacterial strains, growth conditions and antimicrobial susceptibility testing
Bacterial strains used in this study (Table S3) were routinely grown in Luria Bertoni (LB) (Sigma), brain heart infusion (BHI) (Oxoid), Mueller Hinton (MH) (Oxoid), nutrient (Oxoid) or sheep blood MH media.
The minimum inhibitory concentrations (MICs) of oxacillin, cefaclor, cefataxime, nafcillin, cefoxitin, imipenem, piperacillin-tazobactam (Pip/Taz), clindamycin, tobramycin, trimethoprim, mupirocin, ciprofloxacin, spectinomycin, kanamycin and D-cycloserine were determined in accordance with the Clinical Laboratory Standards Institute (CLSI) guidelines using plate and broth dilution assays in MH or MH containing 2% NaCl for oxacillin and nafcillin. Oxacillin MICs were also measured using E-tests strips from Oxoid on MH agar containing 2% NaCl. Quality control strains ATCC29213 and ATCC25923 were used for broth dilution and disk diffusion oxacillin and cefoxitin MIC assays, respectively.
Identification of cefoxitin susceptible MRSA mutant NE810
The Nebraska transposon mutant library (NTML) comprising 1,952 strains harbouring transposon insertions in all non-essential genes was constructed in the USA300 derivative JE2, which has been cured of plasmids p01 and p03 [15]. Cefoxitin susceptibility of individual mutants was determined using the disk diffusion method in accordance with the CLSI guidelines using cefoxitin (30μg) disks (Oxoid) on Mueller Hinton agar (Oxoid). Briefly, each library mutant was revived from the freezer on BHI agar supplemented with 10μg/ml erythromycin and grown overnight for 18 h at 37°C. The wild-type parent strain of the library JE2 and the library mutant containing a transposon insertion in mecA (NE1868) were used as positive (cefoxitin resistant) and negative (cefoxitin susceptible) controls, respectively. The cefoxitin susceptible S. aureus strain ATCC25923 was also used as a negative control. Several colonies were used to prepare a bacterial suspension in sterile phosphate-buffered saline (PBS), which was standardised to 0.5 McFarland and used to inoculate a MH agar plate before the application of the cefoxitin-impregnated disc. The MH agar plates were then incubated at 35°C for 18 h and the zone sizes measured. The zone diameter for JE2 was 18mm and >35mm for NE1868 (mecA). The cefoxitin susceptible mutant NE810 exhibited a zone diameter = 22mm. The entire library was screened in duplicate and putative cefoxitin susceptible mutants were tested at least five times. This susceptibility of cefoxitin sensitive mutants to oxacillin was also determined by E-test, as well as broth and plate dilution assays.
PCR was used to verify the presence of the transposon insertion NE810 using the primers NE810_Fwd and NE810_Rev (Table S4). Phage 80α was used to transduce the transposon mutation from NE810 back into JE2 and USA300 to ensure that secondary mutations were not involved in increased cefoxitin susceptibility phenotype. DNA extracted from transductants selected on BHI agar supplemented with erythromycin 10μg/ml were verified by PCR using primers NE810_Fwd and NE810_Rev. Comparative whole genome sequencing of JE2 and NE810 (MicrobesNG, UK) using the USA300_FPR3757 genome as a reference revealed the absence of other mutations in NE810.
To complement NE810, the cycA gene was amplified from JE2 on a 1608 bp PCR product using primers NE810F1_Fwd and NE810F1_Rev (Table S4) and cloned into the E. coli-Staphylococcus shuttle plasmid pLI50 using the Clontech In-fusion HD cloning kit. The primers were designed using the CloneAmp In-fusion tool with 15 bp extensions that were complementary to the ends of the linearized vector. The recombinant plasmid generated was transformed into E. coli HST08 and verified by Sanger sequencing (Source Biosciences) before being transformed by electroporation into the restriction-deficient laboratory strain RN4220, and subsequently NE810. All plasmid-harbouring strains were cultured in medium supplemented with 100 µg/ml ampicillin (E. coli) or 10 μg/ml chloramphenicol (S. aureus) to maintain plasmid selection.
Growth measurements of JE2 strains in chemically defined media and amino acid analysis
Overnight cultures of JE2 and NE810 were grown at 37°C in TSB, washed in PBS and inoculated at a starting cell density of A600 = 0.05 into chemically defined medium supplemented with 14mM glucose (CDMG) [36]. The bacteria were grown aerobically (250 rpm; 10:1 flask to volume ratio, 37°C) and the A600 was determined every two hours for 8 hours total. One ml of the bacterial culture was collected every two hours and pelleted by centrifugation at 14,000x g for 3 minutes. The spent media was filtered using an Amicon Ultra centrifugal filter (Millipore; 3,000 molecular weight cutoff [MWCO]). Amino acid analysis was performed by the Protein Structure Core Facility, University of Nebraska Medical Center, using a Hitachi L-8800 amino acid analyzer as described previously [37].
mecA gene expression analysis
Quantitative reverse transcription PCR (RT-qPCR) was used to measure mecA transcription on the Roche LightCycler 480 instrument using the LightCycler 480 Sybr Green Kit (Roche) with primers mecA1_Fwd and mecA1_Rev (Table S4). The following cycling conditions were used: 95°C for 5 minutes and followed by 45 cycles of 95°C for 10 seconds, 58°C for 20 seconds and 72°C for 20 seconds. Melt curve analysis was performed at 95°C for 5 seconds followed by 65°C for one minute up to 97 C at a ramp rate of 0.11c/sec with five readings taken for every degree of temperature increase. The gyrB gene amplified with primers gyrB_Fwd and gyrB_Rev (Table S4) served as an internal standard for all reactions. For each reaction, the ratio of mecA and gyrB transcript number was calculated as follows: 2(Ct gyrB - Ct mecA). Each RT-qPCR experiment was performed three times and presented as average data with standard errors.
Scanning Electron Microscopy (SEM)
Cell shape and cell size analysis were carried out using SEM imaging. Overnight BHI cultures were diluted 1:200 in BHI broth supplemented with 1% glucose and incubated for 24h at 37°C. Subsequently, the slides were rinsed three times with distilled water, dried at 60°C for 1h, before being rinsed with 0.1M phosphate buffer, fixed in 2.5% glutaraldehyde for 2h, rinsed again with 0.1M phosphate buffer, dehydrated in ethanol (20%, 30%, and 50% for 10 min each step; 70% ethanol with 0.5% uranyl acetate for 30 min; and then 90%, 96%, and 100% ethanol), soaked in hexamethyldisilazane (HMDS) for 30 minutes and finally dried overnight. The slides were fixed to metal stubs before being coated in gold and imaged using a Hitachi S2600N Variable Pressure Scanning Electron Microscope.
Transmission Electron Microscopy (TEM)
Overnight cultures were diluted 1:200 in 10ml BHI media and grown at 37°C to A600 = 1. The cells were pelleted at 8,000 × g before being re-suspended in fixation solution (2.5% glutaraldehyde in 0.1 M cacodylate buffer [pH 7.4]) and incubated overnight at 4°C. The fixed cells were further treated with 2% osmium tetroxide, followed by 0.25% uranyl acetate for contrast enhancement. The cell pellets were then dehydrated in increasing concentrations of ethanol as described above for the SEM cell preparation, followed by pure propylene oxide, and transferred to a series of resin and propylene oxide mixtures (50:50, 75:25, pure resin) before being embedded in Epon resin. Thin sections were cut on an ultramicrotome. Images were analysed using a Hitachi H7000 instrument. At least 3 to 5 measurements of cell wall thickness were performed on each cell and approximately 30 cells were measured for each sample.
Analysis of peptidoglycan composition in NE810 and JE2 treated with D-cycloserine
Using overnight cultures of JE2 and NE810, independent quadruplicate 50ml broth cultures were set up at a starting cell density of A600=0.05 and grown for approximately 2 hours to ≈A600=0.5, before being left untreated or dosed with DCS at a final concentration of 8, 20 or 32 μg/ml for a further 100 mins before the cells in the entire 50 ml culture were harvested by centrifugation and resuspended in 5ml PBS (Fig. S3). The cell samples were slowly dropped into an equal volume of boiling 10% (wt/vol) SDS, vigorously stirred for 2h in a boiling water bath and left stirring overnight at room temperature. The insoluble fractions were recovered by high speed centrifugation (150,000 x g, 15 min, 25°C) and washed until the fractions were free from SDS. The insoluble fractions were purified for peptidoglycan according to the protocol described previously [38]. Purified peptidoglycan was re-suspended in 50 mM NaPO4 buffer, pH 4.9 and treated with 80 µg/mL muramidase (Cellosyl) for 16 h at 37°C. Muramidase digestion was stopped by incubation in a boiling water bath, and coagulated proteins were removed by centrifugation (21,000 x g, 10 min). The supernatants were adjusted to pH 9.0 with sodium borate and reduced with sodium borohydride for 30 min at room temperature. Finally, samples were adjusted to pH 3.5 with orthophosphoric acid and filtered prior to UPLC analysis with a Waters Acquity UPLC H-class system (Waters) equipped with a Waters Xevo G2-XS QTof MS (Waters MS Technologies). Chromatographic separation was achieved on acquity UPLC® BEH C18, 1.7 µm, 150 x 2.1 mm id (Waters) column. The mobile consisted of solvent A: 0.1% formic acid in Milli-Q water and B: 0.1% formic acid in acetonitrile. The gradient was set as follows: 0-3 min 2-5% B, 3-6 min 5-6.8% B, 6-7.5 min 6.8-9% B, 7.5-9 min 9-14% B, 9-11 min 14-20% B, 11-12 min hold at 20% B keeping the constant flow of 0.250µl/min, 12-12.1 min 20-90% B with a flow rate ramp of 0.250µl/min to 0.300µl/min further holding the same flow rate and % B till 13.5 min, 13.5-13.6 min 90-2% B, 13.6-16 min hold at 2% B. The flow rate was ramped back again to 0.250µl/min at 16.1 min and column was equilibrated form next analysis till 18min. The chromatographic separation was recorded at 204nm.
Mass spectrometry was performed on a Xevo G2-XS QTof quadrupole time-of-flight mass spectrometer (Waters MS Technologies). The scan range was from 10 to 2000 m/z with a scan rate of 0.25s. Instrument was operated in positive electro-spray ionization mode. The capillary and sample cone voltages were 3kV and 40 V, respectively. Gas flows were set at 100 and 500 l/hr for cone gas and desolvation gas, respectively. The source temperature was 120°C and desolvation temperature was 350°C. For MSE low collision energy was set at 6 eV while ramping the high collision energy from 15-40 eV. Leucine-enkephalin reference was used as the lockmass at a concentration of 200 pg/mL with a continuous flow of 5μL/min and 0.25s scan time, to maintain the accuracy of the analysis. All the acquisition and analysis of data were controlled by Waters UNIFI software. Structural characterization of muropeptides was performed based on their MS data and MS/MS fragmentation pattern, which matched with the PG composition and structure reported in previous publications [32, 39–41].
Antibiotic synergy analysis using the microdilution checkerboard assay
The synergistic activity of D-cycloserine and other antibiotics was measured using the checkerboard microdilution method in 96-well plates. The final inoculum in each well was 5 × 105 CFU/ml, and the results (growth or no growth) were determined after 24 h incubation at 37°C. Un-inoculated MH broth supplemented with the antibiotics was used as a negative control. Inoculated MH broth with no antibiotics was used as a positive control. The fractional inhibitory concentration index (ΣFIC) was calculated for each drug combination. An FIC index of ≤0.5 was considered synergistic, one of >0.5 to <2 was considered indifferent, and one of >2 was considered antagonistic. All experiments were performed in triplicate.
Kill curve assays
Suspensions of bacterial cells from overnight cultures adjusted to 107 CFU/ml were exposed to increasing concentrations (0.125×, 0.25×, and 0.5× MIC) of oxacillin, cefoxitin and DCS alone or in combination, and the number of colony forming units (CFU)/ml enumerated at 0, 2, 4, 8 and 24 h. Data is presented at the antibiotic concentrations where synergy was measured i.e. 0.5× MIC for JE2, USA300, DAR173, DAR22, DAR113, BH1CC, and RP62A, and 0.25× MIC for DAR169. Antibiotic synergism was defined as a ≥2 log10 decrease in the number of CFU/ml in cell suspensions exposed to DCS/β-lactam combinations compared to the most effective individual drug after 8 h.
Mouse infection experiments
6- to 8-week-old, age matched, outbred CD1 female mice (Charles River, UK) were used in a non-lethal model of bacteremia. JE2 and NE810 cultures were grown to A600=0.5 in BHI, washed in PBS, adjusted to 1 × 108 CFU/ml. Mice were infected intravenously (via the tail vein) with 5 × 106 CFU (n = 10 mice per group). The infections were left untreated (PBS control) or treated with either 75 mg oxacillin/Kg/12 hours, 30 mg DCS/Kg/12 hours or a combination of both (first antibiotic dose administered 16 hours post infection), before being sacrificed after 5 days. Bacteria present in homogenised spleens and kidneys recovered from the mice were enumerated on blood agar. Statistical significance was assessed using Mann-Whitney U tests (unpaired).
Ethics Statement
Mouse experiments were approved by the United Kingdom Home Office (Home Office Project License Number 40/3602) and the University of Liverpool Animal Welfare and Ethics Committee. This study was carried out in strict accordance with the United Kingdom Animals (Scientific Procedures) Act 1986. All efforts were made to minimize suffering.
Statistical analysis
Two-tailed Student’s t-Tests and Mann-Whitney U tests were used to determine statistically significant differences in assays performed during this study. A p value <0.05 was deemed significant.
Acknowledgements
We thank Craig Winstanley and Kate Reddington for generously providing bacterial strains.