Differences in oxazolidinone resistance mechanisms and small colony variants emergence of Staphylococcus aureus induced in an in vitro resistance development model

ABSTRACT

Invasive Staphylococcus aureus infections are associated with a high burden of disease, case fatality rate and healthcare costs. Oxazolidinones such as linezolid and tedizolid are considered potential treatment choices for conditions involving methicillin resistance or penicillin allergies. Additionally, they are being investigated as potential inhibitors of toxins in toxin-mediated diseases. In this study, linezolid and tedizolid were evaluated in an in vitro resistance development model for induction of resistance in S. aureus. Whole genome sequencing was conducted to elucidate resistance mechanisms through the identification of causal mutations. After inducing resistance to both linezolid and tedizolid, several partially novel single nucleotide variants (SNVs) were detected in the rplC gene, which encodes the 50S ribosome protein L3 in S. aureus. These SNVs were found to decrease the binding affinity, potentially serving as the underlying cause for oxazolidinone resistance. Furthermore, in opposite to linezolid we were able to induce phenotypically small colony variants of S. aureus after induction of resistance with tedizolid for the first time in literature. In summary, even if different antibiotic concentrations were required and SNVs were detected, the principal capacity of S. aureus to develop resistance to oxazolidinones seems to differ between linezolid and tedizolid in-vivo but not in vitro. Stepwise induction of resistance seems to be a time and cost-effective tool for assessing resistance evolution. Inducted-resistant strains should be examined and documented for epidemiological reasons, if MICs start to rise or oxazolidinone-resistant S. aureus outbreaks become more frequent.

KEYWORDS:

• Staphylococcus aureus
• linezolid
• tedizolid
• resistance genes
• WGS
• in vitro
• SCV

Introduction

Staphylococcus aureus (S. aureus) belongs to the gram-positive Staphylococcaceae family and is characterized by its coagulase-positive properties. Due to its ability to produce numerous virulence factors, S. aureus can significantly contribute to infections affecting various body parts, including joints, lungs, skin, soft tissues, and medical devices like catheters and implants. Furthermore, this pathogen can cause diseases with a high case fatality rate associated with bacteraemia like endocarditis and sepsis [1,2]. Oxazolidinones play a crucial role in the treatment of S. aureus bacteraemia (SAB), particularly when therapy with β-lactam antibiotics is contraindicated due to respective allergies [3]. Further, oxazolidinones seem to offer superior efficacy compared to vancomycin as a treatment option for SAB [4,5]. An additional advantage of this treatment approach is the potential for oral administration following the initial intravenous treatment. [6] Due to persistent high mortality rates in cases of methicillin-susceptible Staphylococcus aureus (MSSA) bacteraemia (10–30%) despite targeted antimicrobial therapy based on antimicrobial susceptibility testing, continued discussions are investigating the potential use of oxazolidinones as supplementary therapy [7]. These considerations are based on the effect of oxazolidinones working as toxin suppressors like clindamycin, which could be demonstrated in various studies [8–10]. The assessment of toxin suppressors, including oxazolidinones, when used as an adjunctive treatment in SAB, is currently being examined in ongoing clinical trials [11]. The bacteriostatic effect of linezolid and tedizolid is based on binding to the 50S subunit of the bacterial 23S ribosomal RNA and therefore impeding microbial protein biosynthesis. Since efficacy and prompt initiation of antimicrobial therapy are vital in addressing SAB, rapid detection of susceptibility patterns are essential [12]. To gain more insight in the molecular background and determine possible differences regarding the antimicrobial resistance of S. aureus isolates to oxazolidinones, a stepwise in vitro induction of resistance to linezolid and tedizolid (including the prodrug tedizolid-phosphate) in MSSA isolates was performed in this study.

Material and methods

Sampling

The MSSA reference strain ATCC 29213 was used for stepwise in vitro induction of antimicrobial resistance to linezolid (LZD) and tedizolid (TZD). To identify all isolates as S. aureus, the samples were grown on Columbia agar plates supplemented with 5% sheep blood (Becton Dickinson GmbH, Heidelberg, Germany). Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF, Bruker Daltonics, Bremen, Germany) was used to confirm each sample as S. aureus on species level. The methicillin-sensitivity of S. aureus was confirmed for all isolates using MRSA ChromAgar (Oxoid Deutschland GmbH, Wesel, Germany) and cefoxitin disc diffusion test (Liofilchem, Roseto deli Abruzzi, Italy) with a cefoxitin inhibition zone diameter bigger than 22 mm according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) [13].

Stepwise selection for linezolid and tedizolid resistance

The resistance induction experiments were conducted twice for both LZD and TZD. For LZD the formulation for intravenous use application (Fresenius Kabi Austria GmbH, Graz, Austria) was used twice for the induction. For TZD once the prodrug tedizolid-phosphate (TZP) for intravenous application (Merck Sharp & Dohme Ltd, Hertfordshire, United Kingdom) and additionally the antimicrobial active TZD (Sigma-Aldrich, St. Louis, MO, USA) was used. The reference strain used was S. aureus ATCC 29213. We decided on initial concentrations of the antibiotics based on the anticipated minimum inhibitory concentration (MIC) for ATCC 29213 with respect to LZD (2 mg/L; ranging from 1-4 mg/L). Therefore, the initial LZD concentrations were 1 and 2 mg/L. The selection of initial concentrations for inducing TZD resistance occurred following the same approach of 0.125 and 1 mg/L, for both TZP and TZD. LZD and TZD/TZP were then separately added to tryptone soya broth (TSB) medium (Oxoid Ltd, Basingstoke, United Kingdom), as well as a colony of an overnight culture of S. aureus ATCC 29213. In sequence incubation for 24 h at 37°C (±1°C) with shaking at 150 rpm was performed. During the first antimicrobial resistance induction phase the prodrug TZP was used, which resulted in the inability to induce antimicrobial resistance for 17 generations until a contamination with Micrococcus luteus occurred. The resulting compound with the highest LZD and TZD concentration, in which turbidity was still visible, was further used and transferred to fresh TSB medium in order to prepare bacterial media with higher LZD and TZD concentrations (Figure 1). In order to observe resistance development MICs for linezolid and tedizolid were determined by using MIC test strips (Liofilchem, Roseto deli Abruzzi, Italy) on Mueller Hinton Agar (Becton Dickinson GmbH, Heidelberg, Germany). We determined MICs using test strips instead of the dilution method due to practical considerations. This approach provides a satisfactory level of measurement quality and is more practical. Whole genome sequencing (WGS) was performed with the resulting strains which showed resistance to linezolid and tedizolid.

Figure 1. Stepwise selection for linezolid and tedizolid resistance in S. aureus (ATCC 29213). S. aureus was grown overnight at 37°C on Columbia agar plates supplemented with 5% sheep blood. The grown colonies were added to TSB medium with linezolid, starting at concentrations of 1 and 2 mg/L, and tedizolid, starting at concentrations of 0.125 and 1 mg/L, followed by an incubation step for 24 h. The incubated mixture with the highest concentration, which showed turbidity, was used for further selection steps (as seen in Generations 1-3). During the second selection trial, induction for linezolid was repeated over 20 generations with final concentrations of 128 and 256 mg/L and induction for tedizolid was repeated over 16 generations with final concentrations of 4 and 8 mg/L, respectively.

Whole genome sequencing

Bacterial DNA was extracted using a modified phenol–chloroform method. Overnight colonies were transferred to 1.5 mL tubes with 400 µL phenol–chloroform-isoamylalcohol (25:24:1). The suspensions were then transferred to lysing tubes (Lysing Matrix E 2 mL Tube, MP Biomedicals Germany GmbH, Eschwege, Germany), which were vortexed on the Precellys® 24 homogeniser (PEQLAB Biotechnologies GmbH, Polling, Austria) for 2x 15 s with 10 s pause in between. The supernatant was transferred to a clean 1.5 mL tube, which was centrifuged for 5 min at 18,000 g. The upper aqueous phase was transferred to a new tube and 300 µL chloroform was added before centrifugation for 5 min at 18,000 g. This step was repeated once. Afterwards, the upper aqueous phase was transferred to a new 2 mL tube and 225 µL 1M ammonium acetate was added to help with precipitation. The suspension was vortexed for 5 s and the tube was filled to 1.8 mL with ice-cold 100% ethanol and mixed by inversion. The tube was incubated for 30 min on ice before centrifugation for 15 min with 18,000 g at 4°C. The supernatant was discarded and the DNA pellet was washed twice with 70% ethanol. Purity was measured using the Nanodrop 2000c and the concentrations were determined with the Qubit 4.0 fluorometer (both Thermo Fisher Scientific, Waltham, Massachusetts) using the Qubit dsDNA BR Assay Kit. An 260/280 absorbance ratio >1.8 was assumed to be a pure sample free of proteins. The 260/230 absorbance ratio >2.0 was assumed to be a pure sample, free of other contaminants such as chloroform or free nucleotides. DNA library prep was done according to manufacturer using the Nextera DNA Library Preparation Kit (Illumina, San Diego, California). Sequencing was performed on the MiSeq platform using a V3 Flowcell (both Illumina, San Diego, California).

Bioinformatics analysis

Bioinformatics analysis was performed to detect resistance mutations in the resistance-induced strains compared to the original susceptible isolate. To remove lower quality bases and ensure a read length of at least 90 bp, Trim Galore.v0.6.5 was used. SPAdes v3.15.2 was used for assembling the first isolate of each series de novo and QUAST was used for quality check [14,15]. Bowtie2 v2.4.2 was used for mapping the reads of all subsequent isolates to its susceptible parent strains and variant calling was performed using VarScan v2.4.4 [16,17]. The Comprehensive Antibiotic Resistance Database (CARD) and several tools offered by the Centre for Genomic Epidemiology of the National Food Institute of the Technical University of Denmark were used for detection of resistance genes and characterization of the isolates [18–25].

Results

Stepwise selection for linezolid resistance

During our first induction phase with linezolid concentrations of 16 and 32 mg/L the isolate showed an MIC of 16 mg/L (isolate LZD16). These results indicate resistance to linezolid according to current EUCAST guidelines [13]. Unfortunately, his isolate has not been tested for tedizolid resistance. After our second induction phase, we were able to identify three isolates with oxazolidinone resistance and assigned names based on the generation in which they emerged: to LZD were- LZD15 (64 and 128 mg/L), LZD18 (64 and 128 mg/L) and LZD20 (128 and 256 mg/L).

Stepwise selection of tedizolid resistance

During our first trail of induction, the passage with tedizolid-phosphate had to be stopped due to contamination after with Micrococcus luteus. No development of resistance could be identified, probably due to usage of the prodrug of tedizolid, which therefore was not present in its antimicrobial active form. During our second induction trail with the antimicrobial active tedizolid formulation, three isolates with resistance named after their generation in which they appeared, were identified – TZD13, TZD15 and TZD16 (all at 4 and 8 mg/L).

The extracted and diluted DNA of all resistant isolates showed concentrations between 27.8 and 35.7 ng/μL (arithmetic mean ∼33 ng/μL), thus being optimal concentrations for performing WGS. All samples showed high purity with 260/280 ratios >1.9 and 260/230 ratios >2.2. The Q30 of the raw sequencing output was 92.1% indicating a quality of at least Q30 for 92.1% of the sequenced bases with an error rate less than 1:1000. Nevertheless, the average coverage of the de novo assemblies was 150–300x and thus suitable for detailed downstream analysis (Table 1).

Table 1. Identification of linezolid resistant (LRSA) and tedizolid resistant (TRSA) strains after induction with their respective MIC against either LZD and TZD as well as the identified underlying mutations being the cause of resistance.

	LZD16	LZD15	LZD18	LZD20	TZD13	TZD15	TZD16
LRSA	X	X	X	X	X	X	X
TRSA		X	X	X	X	X	X
LZD MIC (in mg/L)	16	8	32	96	4	12	24
TZD MIC (in mg/L)		2	8	6	1	1.5	2
rplC mutations	p.G155R p.M169I	p.F147V	p.F147V	p.F147V	p.G152D p.G155R	p.G152D p.G155R	p.G152D p.G155R p.F147L

Mutations in LZD16, LZD15, LZD18, LZD20

After stepwise selection for linezolid resistance with ATCC 29213, a linezolid-resistant isolate (LZD16) was shown to have a MIC of 16 mg/L, which is considered resistant according to current EUCAST guidelines [13]. The single nucleotide variants (SNV) p.R159H in splF, a gene encoding a serine protease, was detected, as well as the two SNVs p.G155R and p.M169I in rplC, which encodes the 50S ribosomal protein L3. LZD15 had a MIC of 8 mg/l with a SNV p.F147V in rplC. LZD18 had a MIC of 32 mg/L with a SNV p.F147V in rplC. LZD20 had a MIC of 96 mg/l with a SNV p.F147V in rplC. All three isolates had a SNV p.V40I coding for a hypothetical protein.

Mutations in TZD13, TZD15 and TZD16

TZD13 had a MIC of 8 mg/l supposedly due to SNVs p.G152D and p.G155R in rplC. TZD15 had a MIC of 32 mg/L supposedly due to SNVs p.G152D and p.G155R in rplC. TZD16 had a MIC of 96 mg/l supposedly due to SNVs p.G152D, p.G155R as well as p.F147L in rplC. TZD16 had a SNV p.R308P coding for a hypothetical protein. All S. aureus strain has been shown to be phenotypically small colony variants (SCV) after TZD resistance induction (Figure 1).

Discussion

Linezolid

While recent studies have shown that there has been no substantial rise in resistances to linezolid, there have been multiple reports of a gradual increase in the MIC for this antimicrobial agent [26–30]. To mitigate the emergence of linezolid-resistant S. aureus (LRSA) [19,20], it is imperative to promptly investigate potential mechanisms of resistance. This can be achieved through the implementation of WGS [26–30]. In this study by a stepwise in vitro induction of linezolid resistance, we were able to demonstrate such mutations. Through WGS in LZD16 two missense mutations were detected in the rplC gene, which codes for the 50S ribosomal protein L3, being the substitutions p.G155R and p.M169I. These mutations lead to a rearrangement of a loop link with Helix 90, prorating a structural change causing a swing of the nucleotide U2504 into the binding site of the antibiotic and therefore causes linezolid resistance. Variants at the same position as p.M169I have also been reported, albeit with other amino acid residues p.M169L [31]. Recently, a study has been published, showing the same G155R/M169I double mutation we have discovered. Interestingly, MRSA but not MSSA as well as different types of broth and techniques of selection with different concentrations of linezolid were used, but lead to the same mutations. Besides Perlaza-Jiménez et al. we were able to induce these mutations (G155R/M169I) as well. Furthermore, this is the first study to show these mutations in MSSA, thus the induction of these mutations is independent of the methicillin-susceptibility status of S. aureus. The authors demonstrated, that strains with a M169I mutation in ribosomal protein L3 had a deficit in fitness resulting in slower growth rates [32]. This indicates a possible explanation for why LRSA reports have been rare despite the frequent use of linezolid and why M169I/L mutations have only been reported during in vitro studies at all [31,32]. Both detected substitutions, p.G155R and p.M169I in the rplC gene in LZD16 were found to cause linezolid resistance (Figure 2).

Figure 2. ATCC 29213 colonies before (A, E) and after (B-D, F-H) induction of antimicrobial resistance. Linezolid resistant isolates LZD20 (B), LZD15 (C) and LZD18 (D) showing barely any difference in growth compared to the reference strain (A). Being small colony variants (SCV) TZD16 (F), TZD13 (G) and TZD15 (H) showing difference in growth as means of elevated resistance.

Regarding LZD15, LZD18 and LZD20 the SNV in the rplC gene being p.F147V, to the best of our knowledge these mutations have not been described in this context in literature previously.

Tedizolid

Introduction with the prodrug tedizolid-phosphate was not successful, probably due to being present in the non-active form with exerting no antimicrobial activity. Three resistant isolates could be found during induction by using tedizolid, while other studies failed to obtain isolates with an MIC >2 mg/L in vitro. Furthermore, changes in the rplC gene with these distinct mutations could have only been induced during induction with LZD so far [31]. Due to its slightly different structure, tedizolid binding to the 50S subunit seems to be more potent, especially in strains being resistant to linezolid showing mutations in cfr, resulting in better efficacy in terms of MIC [33,34]. In contrast to linezolid, development of resistance against tedizolid seems to be more difficult to induce [31,35]. No definitive answer to explain this phenomenon could be found in the current literature. Furthermore, no report in literature has been found regarding SNVs p.G152D, p.G155R and p.F147L being responsible for resistance against tedizolid as well as SNVs p.G152D and p.G155R in rplC probably being cause of SCV formation in S aureus which is linked to recurrent infection and higher mortality in the clinical setting [36,37]. These mutations code for rplC gene, encoding for 50s ribosomal protein, being the target of oxazolidinones. All isolates, except LZD16 (due to not testing for resistance against TZD), with resistance to one of the tested oxazolidinones were to be found to be resistant to the other one as well.

Conclusion

Linezolid and tedizolid are antimicrobials known for their relatively low resistance rates in S. aureus isolates. However, cases of increasing MICs and outbreaks of resistance, similar to those observed with vancomycin, are rising. These events have been associated with unfavorable clinical outcomes. In this analysis, we aimed to investigate the development of resistance against linezolid and tedizolid by performing an in vitro antimicrobial resistance induction and analysis of in vivo resistance. Using stepwise increasing induction, a linezolid-resistant S. aureus has been induced and consecutively sequenced. Few variants in multiple genes compared to the original ATCC 29213 isolate could be demonstrated, for whom no distinct function in literature has been described. WGS revealed various missense variants in the rplC gene, which may be causative for resistance development, with the SNV p.F147V coding for rplC being described for the first time. Recently the M169I mutation has been described in literature, during in vitro studies in MRSA. Both mutations have been associated with resistance to linezolid [32]. In our in vitro experiments no induction of resistance against tedizolid occurred in the first trial, due to the fact that the oral prodrug tedizolid-phosphate was used which highlights its lack of influence on the development of antimicrobial resistance in its inactive form. During the second induction trail, we found various SNVs in the rlpC gene caused by linezolid as well as tedizolid. All of them were described for the first time as the cause of tedizolid resistance after in vitro induction as well as being caused for the first time by use of tedizolid. This analysis demonstrates that oxazolidinone (cross-)resistance does occur in vitro. Since oxazolidinones have proven to be an effective therapeutic option in SAB, measures to understand and counteract on development of resistance should be taken into action. Identification of SNVs p.G152D and p.G155R should lead to suspicion since these might be associated with the formation of SCV. Since to date, there is little evolvement of resistance against tedizolid, especially in short-term treatment in vivo and during in vitro studies, its effectiveness, good safety profile and tolerability call for further evaluation as a therapeutic option in clinical studies especially regarding the formation of resistance and rates of recurrent infection due to SCV.

Institutional review board statement

The study has been performed in accordance with the Declaration of Helsinki (1964), Good Clinical Practice guidelines of the European Commission and the Good Scientific Practice guidelines of the Medical University of Vienna.

Informed consent statement

Informed consent was obtained from all subjects involved in the study.

Disclosure statement

No potential conflict of interest was reported by the author(s).