Rifaximin prophylaxis causes resistance to the last-resort antibiotic daptomycin

Daptomycin resistance in Australian VREfm is polygenic and does not correlate with known resistance determinants.

Daptomycin susceptibility testing was performed on VREfm isolated during two unbiased state-wide ‘snapshot’ studies undertaken for month-long periods in 2015 (n=294) and 2018 (n=423) in Victoria, Australia. The proportion of isolates resistant to daptomycin was 16.6% (n=49) in 2015 and 15.3% (n=65) in 2018. Given the unexpectedly high rate of resistance observed, we expanded the study to include additional VREfm isolated in 2017 (n=108) and 2018 (n=173) as part of the ‘Controlling Superbugs’ flagship study^{20, 21}, with 28.4% (n=80) of these isolates being resistant to daptomycin. Overall, we observed 189 (18.9%) daptomycin- resistant VREfm isolates, indicating a very high prevalence of daptomycin resistance in Victoria, Australia.

To investigate the relationship between daptomycin-resistant and daptomycin-susceptible VREfm, whole-genome comparisons were made for the 998 study isolates plus two additional finished VREfm genomes (one daptomycin-susceptible and one daptomycin-resistant). A maximum-likelihood phylogeny was inferred from an alignment of 6,574 core genome single nucleotide polymorphisms (SNPs) (Supplementary Figure 1). In silico multi-locus sequence typing identified 36 sequence types (STs) within the 1000 isolates; 30 of these STs included at least one of the 189 daptomycin-resistant VREfm. Daptomycin resistance was interspersed throughout the tree (i.e. mostly polyphyletic), with several distinct clades. The largest clade (ST203) of daptomycin-resistant strains, accounted for 42.3% of resistant isolates (n=80 of 189) and consisted of a single clone predominant during our sampling timeframe (2015 to 2018), suggestive of an expanding daptomycin-resistant lineage. The other predominant STs (ST80, ST796, ST1421, and ST1424) consisted of several groups of resistant isolates that did not cluster based on tree structure. The presence of daptomycin-resistant isolates in distinct genetic backgrounds suggested daptomycin resistance has arisen independently within this VREfm population on multiple occasions.

Given the high prevalence of daptomycin resistance in Australian VREfm isolates, we sought to determine the genetic determinants leading to resistance. Only seven daptomycin- resistant isolates (3.7%) contained the dual LiaR W73C and LiaS T120A mutations. This finding is of note since current literature suggests these mutations are the most important mechanism of daptomycin resistance in VREfm^{9,12, 22}. In addition, no daptomycin-resistant isolates contained the H215R or R218Q mutations in Cls or the Q75K mutation in the septum site determining protein (DivIVA), which have all previously been linked with daptomycin resistance in VREfm²³. The mechanism of daptomycin resistance in the majority (n=182) of our resistant study isolates was therefore left largely unexplained by previously characterised mutations.

The S491F mutation in RpoB is a novel mediator of daptomycin resistance in VREfm

To identify the mutations associated with daptomycin resistance in our study isolates, we performed a genome-wide association study (GWAS) approach on the collection of 1,000 VREfm isolates with known daptomycin MIC (Figure 1A). To account for the clonal population structure of the collection, we first removed non-homoplastic variants from the list of core genome mutations to reduce the data set to variants that were acquired at least twice across the phylogeny. We then applied a linear mixed model using a kinship matrix as a random effect. After correcting for multiple testing, the analysis identified 142 mutations (in 73 genes) significantly (P < 1x10^-10) associated with daptomycin resistance (as a binary variable with a breakpoint of 8 mg L^-1). The top five most significant mutations were (i) I274S in an uncharacterised ABC efflux protein (P = 7.44x10^-15), (ii) G71S in an uncharacterised permease protein (P = 7.77x10^-14), (iii) V288L in a mannitol dehydrogenase protein (P = 6.08x10^-12), (iv) S491F in RpoB, which is the RNA polymerase ß subunit (P=1.57x10^-13), and (v) T634K in RpoC, which is the RNA polymerase ß’ subunit (P=4.40x10^-11).

• Download figure
• Open in new tab

Figure 1. A.

Manhattan plot of 10,530 variants, displayed by position on the reference genome and significance association with daptomycin resistance (univariate analysis using a linear mixed model). The dashed line shows the Bonferroni-corrected significance threshold. B. Percentage of daptomycin-resistant strains for each mutation within RpoB. The mutations within the rifampicin resistance determining region (RRDR) are shown in bold. Mutations coloured in red were associated with daptomycin resistance. The total number of strains containing each mutation is shown above each bar. C. Maximum-likelihood core-SNP-based phylogeny of clinical VREfm (n=1000) inferred from 6,574 SNPs, demonstrating the presence of RpoB mutations in daptomycin-resistant isolates. Overlaid are the results of in silico multi- locus sequence type (MLST), daptomycin phenotypic testing, and mutations associated with daptomycin resistance in RpoB. In the first circle, ST is not shown for uncommon STs (n≤5). The scale bar indicates number of nucleotide substitutions per site (top), with an approximation of SNP distance (in parentheses). D. Rifampicin susceptibility testing results for the WT, isogenic rpoB and liaRS isogenic mutants and complement strains (designated with -C) (n=3). E. Daptomycin susceptibility testing results for the WT, rpoB and liaRS isogenic mutants and complement strains (designated with -C) (n=3). The median MIC for each strain is shown.

Deletion of the genes encoding the ABC efflux protein, permease protein, or mannitol dehydrogenase protein had no impact on daptomycin susceptibility in a clinical daptomycin- susceptible strain of VREfm from the ST796 genetic background. Similarly, introduction of the T634K mutation within RpoC did not lead to increased levels of daptomycin resistance. However, introduction of the S491F substitution within RpoB, resulted in a 4-fold increase in daptomycin MIC, from 2 mg L^-1 to 8 mg L^-1, and a daptomycin-resistant phenotype. Since mutations within rpoB have previously been associated with daptomycin resistance in Staphylococcus aureus^{24, 25}, we focused further experimental investigations on the RpoB S491F mutation.

Based on amino acid alignment, the S491F mutation is located within the predicted RRDR of E. faecium RpoB, which spans amino acids 467 to 493 (inclusive). The majority of study isolates (n=829, 82.9%) contained a WT RRDR; however, 169 (16.9%) VREfm contained at least one mutation within this region (Figure 1B), with the S491F mutation being the most common (n=105). This was followed by H486Y (n=16), G482D (n=12), G482V (n=10), Q473L (n=6), H486R (n=5), and other uncommon (n=3) mutations.

Different mutations within the RRDR of RpoB result in daptomycin resistance in VREfm

Given the association of the S491F mutation in RpoB with daptomycin resistance, we hypothesised that other mutations within RpoB may also alter daptomycin susceptibility (Figure 1B). In keeping with this hypothesis, we observed a putative correlation between strains carrying individual G482D (n=12) and H486Y (n=16) mutations in RpoB and daptomycin resistance, with 10 daptomycin-resistant isolates containing G482D (83.3% resistant) and 13 containing H486Y (81.3% resistant). Clinical isolates containing the G482D, H486Y, or S491F mutations were interspersed throughout the phylogenetic tree and identified in distinct STs, highly suggestive of multiple independent acquisitions (Figure 1C). Ongoing expansion was also observed for one dominant, daptomycin-resistant clone (ST203) containing the S491F mutation throughout the isolate collection period (2015-2018).

The G482D and H486Y mutations were also located within the predicted RRDR region, suggesting a potential correlation between rifamycin and daptomycin resistance. Therefore, rifampicin susceptibility testing (with rifampicin being a marker of rifamycin resistance²⁶) was performed on all clinical VREfm containing a RpoB mutation within the predicted RRDR (n=169). A randomly selected collection (n=169) of isolates containing a WT RpoB RRDR were used as a control group in this analysis. Mutations located within the predicted RRDR correlated with high-level rifampicin resistance (median MIC 256 mg L^-1), while control isolates containing the WT region displayed a median MIC of 8 mg L^-1 (Supplementary Figure 2). The correlation between rifampicin and daptomycin resistance in clinical strains containing the G482D, H486Y, and S491F mutations, suggested a novel link between rifamycin and daptomycin resistance in VREfm.

To confirm the G482D and H486Y mutations in RpoB also led to rifamycin and daptomycin resistance we constructed isogenic mutants carrying these mutations in the same rifamycin- susceptible, daptomycin-susceptible, clinical strain of VREfm (ST796), used to make the S491F RpoB isogenic mutant above. Introduction of the G482D, H486Y, or S491F RpoB mutations resulted in a 7-fold decrease in rifampicin susceptibility compared to the WT strain, leading to high-level rifampicin resistance (>512 mg L^-1) (Figure 1D). To confirm the changes in rifampicin susceptibility were due to the introduction of each specific mutation and not unknown secondary mutations, we reverted each RpoB mutation to WT by restoration of the chromosomal allele. Complementation to the WT rpoB allele in each of the G482D, H486Y, or S491F mutants resulted in reversion of rifampicin MIC to the WT level (8 mg L^-1), indicating the RpoB mutations were responsible for the heightened levels of rifampicin resistance observed (Figure 1D). To then determine if the G482D and H486Y mutations could also cause daptomycin resistance, daptomycin susceptibility testing was performed. Similar to the S491F isogenic strain, introduction of the G482D or H486Y mutation resulted in a 4-fold increase in daptomycin MIC, from 2 mg L^-1 to 8 mg L^-1, and a daptomycin-resistant phenotype (Figure 1E). Complementation with the WT rpoB allele onto the chromosome resulted in reversion of daptomycin MIC to the WT level, indicating the G482D, H486Y, and S491F RpoB mutations resulted in cross-resistance to rifamycins and daptomycin in VREfm.

Given the reported importance of liaRS mutations in daptomycin-resistant VREfm, we also introduced the well-characterised LiaR W73C and LiaS T120A mutations into the same clinical strain of VREfm (Figure 1D). No difference in rifampicin MIC was observed after introduction of the liaRS mutations, compared to the WT strain, indicating the cross-resistance observed with the G482D, H486Y, and S491F mutations is unique to these RpoB substitutions. However, compared to the WT, introduction of the liaRS mutations decreased daptomycin susceptibility 2-fold, with a reversion to the WT daptomycin MIC after complementation with the WT liaRS alleles. The introduction of the liaRS mutations had less of an impact on daptomycin susceptibility than the G482D, H486Y, and S491F RpoB mutations and did not result in a daptomycin-resistant phenotype (MIC ≥8 mg L^-1). When considered in conjunction with the genomic epidemiology data associated with this study, where liaRS mutations were less common than the rpoB mutations in daptomycin-resistant VREfm (n=7 versus n=141, respectively), our data suggests that mutations within RpoB might represent a previously uncharacterised, yet important mechanism of daptomycin resistance in VREfm.

The G482D, H486Y and S491F RpoB mutations are common in international VREfm strains

To determine if the rpoB mutations associated with daptomycin resistance observed in Australian VREfm were representative of other VREfm isolates globally, we performed a large- scale analysis using publicly available VREfm sequence data from healthcare-associated strains (n=4,476; n=3,476 international and n=1,000 Australian) (Figure 2A). Of the isolates analysed, 630 (14.3%) carried an amino acid substitution in the RRDR of RpoB, occurring at 16 positions (Figure 2B). The S491F mutation was the most common, being present in 461 isolates, and accounting for 77.9% of the RpoB mutations observed. Isolates carrying this mutation originated from 20 countries, and in silico MLST showed the mutations were spread across 21 different STs, with ST203 (44.7%), ST80 (30.2%), and ST117 (11.5%) accounting for the majority (86.4%) of isolates carrying this substitution. Importantly, five VREfm harbouring the S491F mutation also contained the cfr(B) (n=4) or poxtA (n=1) genes that confer resistance to linezolid, suggesting near pan-resistant strains of VREfm have already emerged.

• Download figure
• Open in new tab

Figure 2. A.

Map of 4,476 VREfm genomes included in this study. Circle size corresponds to total number of genomes (binned into categories), and colour corresponds to region of isolation. Country coordinates used are the country centroid position. Map was generated in R using ggplot2. B. The frequency of various RpoB mutations within the rifampicin resistance determining region (RRDR) in 4,476 VREfm genomes, sampled from 43 MLSTs. Bars are coloured by the number of isolates from each region of isolation containing the mutation. The identified daptomycin resistance associated mutations are coloured in red.

The H486Y mutation was the second most common, and was identified in 73 isolates, equivalent to 11.6% of strains with mutations within the RRDR of RpoB. These isolates were collected from 10 countries and in silico MLST identified 22 STs, with the most common being ST203 (23.2%), ST796 (21.9%), and ST80 (20.5%). The G482D mutation was the third most common, being identified in 43 isolates, and accounting for 6.8% of strains carrying RpoB mutations. It was present in nine distinct STs, although mostly commonly in ST796 (65.1%) and ST1421 (11.6%), and was identified in isolates from seven countries. Collectively, these data indicated the G482D, H486Y, and S491F mutations, which confer cross-resistance to rifamycins and daptomycin, are not restricted to Australian VREfm but are globally prevalent in healthcare-associated VREfm strains.

To determine whether the identified RpoB mutations were enriched within healthcare- associated VREfm (established as clade A1), we interrogated publicly available genomic sequences of VREfm from clade A2 (n=98), known for being animal-associated²⁷. In keeping with previous findings, the maximum-likelihood tree clustered isolates into two main clades: one healthcare-associated and one associated with VREfm from animals. No mutations were identified in the RpoB RRDR region of strains isolated from animals, with our analyses showing that RRDR RpoB mutations were significantly (P<0.001; Fisher’s exact test) associated with healthcare-associated VREfm. This suggests the identified RpoB mutations are primarily enriched within the healthcare setting (Supplementary Figure 3).

Phylodynamics indicate the S491F RpoB mutation emerged within the VREfm population following the clinical approval of rifaximin

Given the predominance of the S491F mutation in globally distributed VREfm populations, we used evolutionary phylodynamic analyses to understand its emergence. Within our Australian isolates, we observed the expansion of a dominant ST203 clone from 2015 to 2018 that carried the S491F mutation (Figure 1C). Since this clone comprised VREfm carrying the vanA resistance cluster or operon, we sequenced all “historical” vanA-VREfm from our public health laboratory (n=229), which consisted of every vanA-VREfm isolate collected from 2003 to 2014, to increase temporal signal. We then contextualised all Australian isolates (n=1,229) with the international (n=3,389) VREfm in a maximum-likelihood phylogeny inferred from an alignment of 9,277 SNPs and used clustering with core-genome MLST (cgMLST) to identify three clusters containing the RpoB S491F mutation (Figure 3A). The same ST203 clone (Cluster 1) formed the largest cluster (n=219 taxa), consisting of isolates from Australia and United Kingdom. Cluster 2 (n=85 taxa) consisted of ST80 and ST78 isolates from Australia, Europe, South America, the United Kingdom, and the United States of America while Cluster 3 (n=68 taxa) consisted of ST80 isolates from Australia, Europe, and the United Kingdom.

• Download figure
• Open in new tab

Figure 3. A.

Maximum-likelihood, core-SNP-based phylogeny for 4,476 VREfm inferred from 9,277 core-genome SNPs, demonstrating the presence of the S491F RpoB mutation in international VREfm. Overlaid is the region of isolation for each strain and the presence of the S491F mutation. The coloured branches indicate the three VREfm clusters identified with cgMLST used as input for Bayesian evolutionary analyses. The scale bar indicates number of nucleotide substitutions per site (top), with an approximation of SNP distance (in parentheses). B. Bayesian phylodynamic analyses showing the maximum-clade credibility (MCC) trees of the three VREfm clusters with the timing of emergence for each lineage. Cluster 1 (n=219) was inferred from a core alignment of 1,869,554 bp containing 329 SNP sites; Cluster 2 (n=85) was inferred from an alignment of 1,524,024 bp containing 541 SNP sites; and Cluster 3 (n=68) was inferred from an alignment of 1,860,780 bp containing 764 SNP sites. The time in years is given on the x axis. The presence of the S491F RpoB trait for each isolate is shown in purple. Overlaid onto the MCC trees is the first instance of FDA approval for rifaximin (2004) and for hepatic encephalopathy (HE) (2010). C. Violin plots for the most recent common ancestor (MCRA) for each cluster, representing when the RpoB S491F mutation first emerged, with 95% HPD intervals. Overlaid onto violin plots is the FDA approval date for rifaximin (2004).

To model the evolutionary trajectories of these three VREfm clusters, we used core-genome SNP diversity and year of isolation (Figure 3B). Bayesian phylodynamic analyses were conducted using the core-genome SNP alignments for each cluster/lineage with a discrete trait model and constant coalescent tree prior. We assessed for temporal signal within each cluster using a root-to-tip regression (Supplementary Figure 4). The substitution rate (the number of expected substitutions per site per year) was consistent with other estimates for healthcare-associated VREfm, otherwise referred to as clade A1^27–30. The median substitution rate was similar for Cluster 1 and Cluster 2, at 9.7x10^-7 [95% highest posterior density (HPD) 6.88x10^-7 – 1.24x10^-6] and 1.25x10^-6 (95% HPD 7.68x10^-7 – 1.74x10^-6) respectively, but slightly faster for Cluster 3 at 3.86x10^-6 (95% HPD 2.23x10^-6 – 5.69x10^-6). The year of emergence for the most recent common ancestor (MRCA) was estimated for each cluster to indicate when the RpoB S491F mutation was first acquired. The MRCAs for the clusters were similar, with 2006 (HPD 1993 – 2012) for Cluster 1, 2000 (HPD 1989 – 2008) for Cluster 2, and 2004 (HPD 2001 – 2010) for Cluster 3, suggesting the lineages emerged at similar times (Figure 3C).

Given the association of the S491F mutation in RpoB and rifamycin resistance, we hypothesised rifamycin use might be driving the emergence of this mutation. The most commonly used rifamycin antibiotics in clinical practice are rifampicin and rifaximin. Since rifampicin was approved for clinical use by the United States Federal Drug Administration (FDA) in 1971, several decades before the estimated emergence of the MRCAs, it is unlikely to have played a major role in the emergence of the S491F mutation within this VREfm population. However, the MRCA for all three clusters is predicted to have emerged around the same time as the first clinical introduction of rifaximin, in 2004. The acquisition of the same mutation in three genetically distinct lineages at a similar time, provides support for our hypothesis that rifamycin use might be driving selection of the S491F mutation within VREfm. Further, the structure of the three maximum clade credibility (MCC) trees (Figure 3B) suggested the lineages have expanded over time, which may be correlated with the approval and subsequent widespread, international use of rifaximin for the prevention of hepatic encephalopathy since 2010 (Figure 3B).

The S491F mutation in Cluster 1 and Cluster 2 was found to be stably maintained within each lineage after its acquisition, with few high probability events of reversion to the WT rpoB allele, indicative of maintenance after emergence (Supplementary Figure 5A-B). For Cluster 3, the RpoB S491F mutation emerged a single time, with high (∼0.95) posterior probability of a single acquisition within the cluster and subsequent maintenance (Supplementary Figure 5C). There was also an equal posterior probability of either a second acquisition within the cluster or loss of the RpoB S491F mutation within Cluster 3. Overall, the Markov jumps for all three clusters suggests that the S491F mutation emerged and was then subsequently maintained within these globally prevalent lineages. Taken together these data show the S491F mutation has emerged within the VREfm population on several occasions since the early 2000s, with the predicted dates of emergence being closely correlated with the clinical introduction of rifaximin.

Rifaximin drives the emergence of daptomycin-resistant VREfm in a murine model of gastrointestinal colonisation

Rifaximin is a non-absorbable oral agent with direct antimicrobial activity in the gastrointestinal tract. It is predominately used to prevent recurrent hepatic encephalopathy in patients with liver cirrhosis^{18, 31}. Importantly, this patient cohort is high-risk for VREfm colonisation within the gastrointestinal tract³². Since the Bayesian phylodynamic analyses highlighted a putative correlation between the S491F RpoB mutation in VREfm and use of rifaximin, we hypothesised rifaximin use may be driving the emergence of this mutation and therefore, daptomycin-resistant VREfm within the gastrointestinal tract of patients receiving this antibiotic. To test this hypothesis, mice were colonised with a clinical, daptomycin- sensitive (MIC 2 mg L^-1) VREfm strain (Aus0233) containing a WT rpoB gene before being administered a human-equivalent dose of rifaximin, rifampicin, daptomycin, or vehicle (Figure 4A and Supplementary Figure 6). Rifampicin was chosen as a comparison since it is also a commonly used rifamycin in clinical practice. After 7 days of rifamycin treatment, we observed rifamycin-resistant VREfm in significantly more mice receiving rifaximin (90% of mice) or rifampicin (80% of mice) than in mice that received daptomycin (0% of mice) (P < 0.0001 and P < 0.0001; unpaired t-test) or vehicle (0% of mice) (P < 0.0001 and P < 0.0001; unpaired t-test) (Figure 4B).

• Download figure
• Open in new tab

Figure 4. A.

Timeline of the mouse experiment. VREfm-colonised mice (n=5 for vehicle, n=10 for rifampicin, n=10 for rifaximin, n=10 for daptomycin) received a human-equivalent dose of vehicle, rifampicin, or rifaximin (twice per day for rifaximin) for 7 days by oral gavage or subcutaneous injection with daptomycin for 7 days. CRO=ceftriaxone; DAP=daptomycin; RIFAX=rifaximin; RIF=rifampicin. Figure to scale. B. Percentage of total mice with rifampicin- resistant VREfm strains C. or daptomycin-resistant VREfm strains after 7 days of antibiotic treatment. D. Percentage of VREfm from each mouse (n=50 colonies per mouse) that were resistant to rifampicin E. or daptomycin after 7 days of antibiotic treatment. Points represent an individual mouse. Percentage was calculated from rifampicin or daptomycin MIC values (either resistant or susceptible) from 50 VREfm colonies isolated from each mouse. Boxes represent the median and interquartile range for each group. F. Overview of the RpoB mutations identified in the rifampicin-resistant colonies. Each point represents a single colony. Isolates are separated by each RpoB mutation and grouped into either daptomycin- susceptible or daptomycin-resistant. The RpoB mutations coloured in red had an association with daptomycin resistance. n values represent the number of isolates containing each mutation for rifaximin and rifampicin, respectively. *P < 0.05, **P < 0.01, ***P < 0.001, ****P<0.0001; unpaired t-test (vehicle versus rifampicin or vehicle versus rifaximin and rifampicin verses daptomycin or rifaximin verses daptomycin).

For each mouse, we then determined the percentage of individual VREfm isolates that were rifamycin-resistant or daptomycin-resistant. There were significantly more rifamycin- resistant VREfm isolated from mice receiving rifaximin or rifampicin than mice receiving the vehicle control (P < 0.001 and P < 0.01; unpaired t-test) or daptomycin (P < 0.001 and P < 0.01; unpaired t-test) (Figure 4D). Similarly, there was significantly more daptomycin-resistant VREfm in mice receiving rifaximin or rifampicin than vehicle control (P < 0.05 and P < 0.05; unpaired t-test) or daptomycin (P < 0.05 and P < 0.05; unpaired t-test) (Figure 4E). We estimated that daptomycin-resistant VREfm accounted for between 0-41% of the gastrointestinal VREfm population in mice given rifaximin and 0-36% in mice given rifampicin, demonstrating conclusively that rifamycin administration can drive the emergence of VREfm with cross-resistance to rifamycins and daptomycin. Notably, no daptomycin-resistant VREfm were isolated from mice receiving daptomycin, in agreement with prior research³³.

To identify which mutations were present in the rifamycin-resistant VREfm isolates collected from mice administrated either rifaximin or rifampicin, we randomly selected 150 isolates from each antibiotic group (rifaximin or rifampicin, n=300 total) to undergo WGS, consisting of 100 rifamycin-resistant isolates collected following the last day of treatment and 50 isolates from before rifaximin or rifampicin administration. No mutations in RpoB were identified in any VREfm isolate collected prior to rifaximin or rifampicin exposure. However, following the administration of either rifaximin or rifampicin, VREfm carrying mutations within RpoB were commonly identified. The S491F mutation was most abundant (n=53 and 63, respectively), with all isolates carrying this mutation being daptomycin-resistant (Figure 4F). The H486Y mutation was also commonly identified, albeit less so than S491F, (n=12 and 28, respectively), with all isolates again being daptomycin-resistant. The G482D mutation was the third most commonly identified RpoB mutation (n=15 and 6, respectively), with 13 isolates carrying this mutation being daptomycin-resistant. Note that 2 isolates containing the G482D mutation in the rifampicin treated mice were daptomycin-sensitive, likely due to other confounding mutations within the genome. Other RpoB mutations in addition to S491F, H486Y, and G482D were also identified. These included V135F, L471V, E473L, and H486A, however all VREfm isolates carrying these mutations were daptomycin-sensitive. Importantly, the proportions of each RpoB mutation observed in VREfm collected from the gastrointestinal tract of mice administered either rifaximin or rifampicin, correlated closely the proportions of each mutation observed in our collection of human clinical VREfm isolates, with the S491F mutation most commonly identified, followed by H486Y, and then G482D, suggesting a similar selective pressure (i.e. rifamycin use) found in our mouse experiments might be driving the emergence of similar RpoB mutations within human clinical isolates. Taken together, these data demonstrate that exposure to rifaximin can drive the emergence of daptomycin resistance in colonising strains of VREfm, through the enrichment of isolates carrying select mutations in RpoB.

VREfm collected from patients receiving rifaximin are more likely to be daptomycin- resistant than VREfm collected from patients that did not receive rifaximin

To further test our hypothesis that rifaximin use might be driving the emergence of daptomycin resistance in VREfm, we performed a case-control analysis of clinical VREfm isolates collected from a retrospective cohort of patients from a single tertiary healthcare centre in Melbourne, Australia over 4 years. The VREfm isolates were stratified according to whether the patient had received rifaximin within 1 month of VREfm isolate collection or not. A total of 50 VREfm strains were identified as being from patients receiving rifaximin. As a control group, we randomly selected 50 VREfm that had been collected from patients that had not received rifaximin. The majority (88%) of these VREfm were screening samples. All isolates were collected over the same time frame and from the same institution.

A maximum-likelihood phylogeny was inferred from an alignment of 12,430 core-genome SNPs (Figure 5A). The VREfm isolates in the rifaximin and control groups were dispersed throughout the tree, with 5 different STs (ST78, ST80, ST203, ST796, ST1421, and ST1424) from rifaximin patients and 6 different STs (ST17, ST78, ST80, ST203, ST796, ST1421, and ST1424) from the control patients. There were 28 VREfm from the rifaximin group that contained a mutation within the RRDR of RpoB, with 17 containing the S491F mutation, 5 G482D, 4 H486Y, 1 E473L, and 1 G482V mutation. All isolates containing the S491F, H486Y, and G482D mutations were resistant to daptomycin, while isolates with the E473L and G482V mutations were daptomycin-susceptible. The control group had 2 isolates with RRDR RpoB mutations, with 1 susceptible isolate containing a D476Y and 1 daptomycin-resistant isolate containing the S491F substitution.

• Download figure
• Open in new tab

Figure 5. A.

Maximum-likelihood core-SNP-based phylogeny of clinical VREfm (n=100) inferred from 12,430 SNPs. Overlaid is the results of in silico MLST, rifaximin treatment group, rifampicin phenotypic testing, daptomycin phenotypic testing, and mutations in RpoB. Mutations associated with daptomycin resistance are bolded. The scale bar indicates number of nucleotide substitutions per site (top), with an approximation of SNP distance (in parentheses). B. Rifampicin susceptibility data for clinical VREfm (n=50 per group). Red bars represent percent of rifampicin-resistant isolates. A Fisher’s exact test comparing the percentage of rifampicin resistance in the control versus rifaximin, P < 0.0001. C. Daptomycin susceptibility data for clinical VREfm (n=50 per group). Red bars represent percent of daptomycin-resistant isolates. A Fisher’s exact test comparing the percentage of daptomycin resistance in the control versus rifaximin, P < 0.0001.

Rifaximin exposure in patients was significantly correlated with the isolation of a rifamycin- resistant (P = 0.00007, OR = 60.0, 95% CI = 8.8 – 2562.1; Fisher’s Exact Test) and daptomycin- resistant VREfm strain (P = 0.0006, OR = 25.0, 95% CI = 5.5 – 235.3; Fisher’s exact test) (Figure 5B and C). These data indicated that patients receiving rifaximin are significantly more likely to carry rifamycin-resistant and daptomycin-resistant strains of VREfm than patients who did not receive rifaximin, suggesting rifaximin use might be an important driver in the de novo emergence and spread of daptomycin-resistant VREfm.

Media and reagents

E. faecium was routinely cultured at 37°C in brain heart infusion (BHI) broth (Becton Dickson) or BHI agar (BHIA), BHI solidified with 1.5% agar (Becto Dickson). For electroporation, E. faecium was cultured in BHI supplemented with 3% glycine and 200 mM sucrose (pH 7.0). Escherichia coli was cultured in Luria broth (LB). Broth microdilution (BMD) MICs were performed in cation-adjusted Mueller Hinton with TES broth (CAMHBT) broth (Thermo Fisher). A concentration of 10 mg L^-1 chloramphenicol (Sigma Aldrich) was used for plasmid selection in E. faecium and E. coli. The following antibiotics were used at variable concentrations for susceptibility testing: rifampicin (Sigma Aldrich), rifaximin (Sigma Aldrich), and daptomycin (Cubicin).

Oligonucleotides were purchased from Integrated DNA Technologies and are listed in Supplementary Table 1. Plasmids were purified with Monarch Plasmid Miniprep Kit (NEB). PCR products and gel extractions were purified using Monarch DNA Gel Extraction Kit (NEB). Genomic DNA was purified using the Monarch Genomic DNA Purification Kit (NEB). Phusion and Phire DNA polymerase was purchased from New England Biolabs.

Bacterial isolates

Bacterial strains used in this study are listed in Supplementary Table 2. Australian bacterial strains were collected across three data projects in the Microbiological Diagnostic Unit Public Health Laboratory (MDU PHL). Two unbiased cross-sectional surveys of VREfm were conducted between 10 November and 9 December 2015 (n=331)⁴³ and between 1 November and 30 November 2018 (n=323) in the State of Victoria (referred to as the 2015 and 2018 Snapshot). During this period, all VREfm-positive isolates (including screening and clinical samples) collected by laboratories across the state were sent to the MDU PHL. In addition, this project included vanA-VREfm collected from the “Controlling Superbugs” study²⁰, a 15- month (April-June 2017 and October 2017-2018) prospective study including eight hospital sites across four hospital networks, resulting in 346 VREfm isolates (308 patients) sent for WGS at MDU PHL. The VREfm were isolated from patient samples (including screening and clinical samples) routinely collected from hospital inpatients. For the ‘historical vanA-VREfm,’ every vanA isolate collected within MDU PHL was included. This resulted in an additional 225 isolates, sampled between 2008 and 2014.

For publicly available isolates, our aim was to capture the diversity of E. faecium circulating globally by including isolates that formed part of several key studies involving hospital- associated VREfm (as of January 2021). To be included, isolates needed to have short-read data available, with geographic location (by country), year of collection, and source (human or animal). Reads were only included if they had a sequencing depth of >50x. To capture the diversity of VREfm circulating in the United States, isolates from human sources were downloaded from the PathoSystems Resource Integration Center⁴⁴. All isolates were confirmed to be E. faecium with the Kraken2 database (v.2.1.2)⁴⁵. The final number of international isolates compromised those from Africa (n=8), Asia (n=25), Europe (n=2941), North America (n=424), and South America (n=78) (Supplementary Table 2).

Antibiotic susceptibility testing

Daptomycin susceptibility testing was performed using the BMD MIC method as according to CLSI guidelines. In a 96-well plate, a two-fold dilution series (from 32 to 0.5 mg L^-1) of daptomycin was made in 100 μL volumes of CAMHBT, additionally supplemented with 50 mg L^-1 Ca²⁺. An inoculum of 100 μL E. faecium broth culture adjusted to 1 x 10⁶ CFU mL^-1 in CAMHBT was then added to each well. After 24 hours incubation, the MIC was defined as the lowest antimicrobial concentration that inhibited visible growth. All assays were performed in biological triplicate, with the median MIC reported. In accordance with recent guidelines⁴⁶, isolates with a daptomycin MIC ≥ 8 mg L^-1 were considered to be daptomycin-resistant. A daptomycin-sensitive strain (AUS0085)⁴⁷ and a daptomycin-resistant strain (DMG1700661)¹⁷ were used as a control.

Rifampicin susceptibility testing was performed using the BMD method in CAMHBT. High- level rifampicin resistance was defined with a MIC > 32 mg L^-1. All susceptibility testing was performed in triplicate.

Whole-genome sequencing

Genomic DNA was extracted from a single colony using a JANUS automated workstation (PerkinElmer) and Chemagic magnetic bead technology (PerkinElmer). Genomic DNA libraries were prepared using the Nextera XT kit according to manufacturer’s instructions (Illumina, San Diego, CA, USA). Whole-genome sequencing was performed using Illumina NextSeq platform, generating 150 bp paired-end reads.

The short reads of isolates sequenced at MDU-PHL are available on the NCBI Sequence Read Archive [BioProjects PRJNA565795 (Controlling Superbugs), PRJNA433676 (2015 Snapshot) and PRJNA856406 (2018 Snapshot), and PRJNA856406 (historical vanA isolates)].

Phylogenetic analysis

De novo assemblies of the genomes were constructed using Spades⁴⁸ (v3.13). In silico MLST were determined using the program mlst with the efaecium database (https://github.com/tseemann/mlst). The 1000 Australian genomes as well as the 4,612 Australian and international VREfm were mapped to the reference E. faecium genome AUS0085 isolated from a human bacteraemia infection in Victoria, Australia (NCBI accession: CP006620)⁴⁷ using snippy (https://github.com/tseemann/snippy) (v4.4.5), applying a minfrac value of 10 and mincov value of 0.9. This reference was selected as it was a publicly available complete genome collected locally and daptomycin-sensitive. A maximum likelihood phylogenetic tree was inferred using IQ-TREE (v2.1.2) with a general time-reversible (GTR + G4) substitution model, including invariable sites as a constant pattern and 1000 bootstrap replicates. Recombination masking was not performed for species maximum likelihood trees due to the small size of the resulting core alignment. All trees were mid-point rooted and visualised in R (v4.0.3, https://www.r-project.org/) using phangorn⁴⁹ (v2.5.5), ape⁵⁰ (v5.4), ggtree⁵¹ (v2.3.4), and ggplot (v3.3.2).

The genome assemblies of all isolates were screened for acquired antimicrobial resistance determinants using abriTAMR (https://github.com/MDU-PHL/abritamr).

Genome-wide association study of daptomycin resistance

A GWAS approach was applied to identify genetic variants of daptomycin resistance in E. faecium. A genotype matrix of SNPs was constructed and used as input to homoplasyFinder⁵² (v0.0.0.9) to determine the consistency index at each locus and kept mutations that had an index of ≤0.5 (indicating at least two independent acquisitions across the phylogeny). We then ran GWAS using daptomycin resistance as a binary trait, where isolates were categorised as resistant if their daptomycin MIC was ≥8 mg L^-1. To correct for population structure, we used the factored spectrally transformed linear mixed models (FaST-LMM) implemented in pyseer⁵³ (v.1.3.6), which computes a kinship matrix based on the core genome SNPs as a random effect. The Bonferroni method was used to correct P values for multiple testing.

Core genome MLST (cgMLST) and clustering

cgMLST alleles for each isolate was defined using the public E. faecium cgMLST scheme⁵⁴ and chewBBACA (v2.0.16), implemented locally in the COREugate pipeline (v2.0.4) (https://github.com/kristyhoran/Coreugate). The pipeline determines the alleles of each core gene for every isolate as defined by the specific pathogen scheme. The E. faecium cgMLST scheme contains 1,423 genes. The number of allelic differences between each isolate within this core set of genes is then determined. The cgMLST clusters were determined using single linkage clustering and a pairwise allelic difference threshold of ≤250. This threshold was chosen since it maximised diversity within clusters, to improve temporal sampling depth, while still clustering based on maximum-likelihood tree structure.

Phylodynamic analyses of the emergence of the S491F RpoB mutation in VREfm lineages

To investigate the emergence of the S491F mutation in RpoB in three different lineages, as defined with cgMLST, we undertook further analysis on these clusters/lineages. From the species-level maximum-likelihood tree (Figure 3A), three lineages/clusters were identifiable by cgMLST due to their size (n>50) and presence of the S491F mutation. The three clusters were analysed independently, such that individual core-genome SNP alignments were generated, since this increased the length of the core alignment and number of sites considered. Snippy (https://github.com/tseemann/snippy) (v4.4.5) was used to generate the alignments for each cluster to the corresponding reference genome (AUSMDU00004024 for cluster 1, AUSMDU00004055 for cluster 2, and AUSMDU00004142 for cluster 3). Each core alignment used a within ‘cluster reference’ (complete genome of the same cluster) to maximise core-SNP alignment length. The reference for each cluster was chosen since they were a locally-collected, closed genome. Recombination was removed from the final alignment using Gubbins⁵⁵ (v.2.4.1) to ensure modelling was only informed by SNPs with tree- like evolution within the core genome. Maximum-likelihood trees for each of the three clusters were inferred from the core-SNP alignments [Cluster 1: (n=219 taxa) 329 SNPs; Cluster 2: (n=85 taxa) 541 SNPs; Cluster 3: (n=68 taxa) 764 SNPs] with IQ-tree (v2.1.2)⁵⁶ with a general time-reversible (GTR+Γ) substitution model, including invariable sites as a constant pattern. Phylogenetic uncertainty was determined through 1000 nonparametric bootstrap replicates.

To investigate temporal signal in the three clusters of VREfm genomes, we first used TempEst⁵⁷ (v1.5). A root-to-tip regression analysis was performed on the root-to-tip branch distances within the three, cluster maximum-likelihood phylogenies as a function of year of collection, with the position of the root optimised according to the heuristic residual mean squared method.

The frequency of the emergence of the rpoB mutation in VREfm was inferred using a discrete trait model implemented in BEAST⁵⁸ (v1.10.4). Under this model the SNP alignments are used to infer the evolutionary process (i.e. phylogenetic tree, time, and nucleotide substitution model parameters) for the three clusters. The alignments all shared the HKY substitution model with a gamma distribution for among-site rate variation, and a constant- size coalescent population prior²⁸. To avoid ascertainment bias due to using a SNP alignment, the number of constant sites were taken into account for the likelihood calculations. The molecular clock was a relaxed clock with an underlying lognormal distribution. The molecular clock was calibrated using isolation dates for each genome by year of collection and the mean clock rate is shared between all three alignments, but the model allows for the individual alignments to have different standard deviations of the lognormal distribution and also different branch rates. The mean molecular clock rate requires an explicit prior distribution, for which we used a Γ distribution and a 0.95 quantile range of 4.9x10^-6 and 1.1x10^-4 substitutions/site/year. This informative prior means that it acts as an additional source of molecular clock calibration that can drive estimates, even in the absence of temporal signal. The presence or absence of the S491F mutation in rpoB was used as a binary trait^{59, 60}.

The trait model was shared between the three alignments, with the different Markov jumps and rewards (ie changes of trait state and time spent in each state, respectively) recorded for each of the three alignments. The posterior distribution of model parameters was sampled using a Markov chain Monte Carlo of 100,000,000 iterations, sampling every 100,000 iterations. Two independent runs were run for the models. We assessed sufficient sampling from the stationary distributions by verifying the effective sample size of key parameters was around or above 200. The final maximum-clade credibility (MCC) trees were visualised in R (v4.0.3, https://www.r-project.org/) using ggtree⁵¹ (v2.3.4). The Markov jumps for the rpoB trait for each alignment were visualised in R (v4.0.3, https://www.r-project.org/).

Reconstruction of clinical mutations by allelic exchange

The liaR^W73C, liaS^T120A, rpoC^T634K, rpoB^G482D, rpoB^H486Y, or rpoB^S491F mutations were recombined into the chromosomal copy of each gene in ST796 VREfm (Ef_aus0233) by allelic exchange. The liaR^W73C and liaS^T120A mutations were first introduced individually, then together. The region encompassing each gene was amplified by SOE-PCR and recombined into pIMAY-Z⁶¹ by the seamless ligation cloning extract (SLiCE)⁶² method and transformed into Escherichia coli IM08B⁶¹. The construct was transformed into electrocompetent VREfm⁶², with allelic exchange performed as described previously⁶³. Allelic exchange was performed as described⁶³. Reversion of liaR^W73C and liaS^T120A, rpoB^G482D, rpoB^H486Y, or rpoB^S491F mutations were completed using allelic exchange with a construct containing the respective wild-type allele. To delete the ABC transporter protein (EFAU085_02633), permease protein (EFAU02892), or mannitol dehydrogenase (EFAU02627) from the chromosome, deletion constructs were PCR-amplified from Ef_aus0233 genomic DNA and allelic exchange performed as described above. Genome sequencing and analysis of all mutants was conducted as described, with resulting reads mapped to the Ef_aus0233 reference genome and mutations identified using Snippy (https://github.com/tseemann/snippy) (v4.4.5).

VREfm in vivo gastrointestinal colonisation experiments

Female C57BL/6 mice at 6-8 weeks of age were purchased from WEHI and maintained in a specific-pathogen-free facility at the Peter Doherty Institute for Infection and Immunity. All animal handling and procedures were performed in a biosafety class 2 cabinet. Animal procedures were performed in compliance with the University of Melbourne guidelines and approved by the University’s Animal Ethics Committee The dose for each antibiotic was calculated using the FDA human conversion formula to ensure each mouse was given a human-equivalent dose⁶⁴. To establish gastrointestinal colonisation of VREfm, mice were administered ceftriaxone (410 mg kg^-1 day^-1; AFT Pharmaceuticals) via subcutaneous injection once daily for 7 days, followed by an antibiotic wean period of 24 hours. Mice were then inoculated with 10⁶ VREfm in 100 μl PBS by oral gavage. Three days after VREfm inoculation, single-housed mice were administered either rifaximin (113 mg kg^-1 administered twice daily; Sigma Aldrich), rifampicin (123 mg kg^-1 administered once day; Sigma Aldrich), or vehicle (Corn oil with 10% DMSO) via oral gavage; or daptomycin (50 mg kg^-1 administered once daily; Cubicin) via subcutaneous injection [this results in similar exposure (AUC0-24) to that observed in humans receiving 8 mg kg^-1 of intravenous daptomycin⁶⁵]. The above antibiotic dosing protocol was followed for 7 days. Faecal samples were collected at specific time points throughout the experiment to determine VREfm gut colonisation and for downstream rifamycin and daptomycin resistance analysis. Faecal samples were resuspended in PBS to a normalised concentration (100 mg ml^{- 1}). Serial dilutions were performed, and samples were plated onto Brilliance VRE agar (Oxiod) for VREfm CFU enumeration.

For rifamycin and daptomycin analysis, VREfm colonies (n=50 per mouse) from the Brilliance VRE agar plates were replica plated onto BHIA with and without rifampicin 20 mg mL^-1 to determine the proportion of rifampicin-resistant VREfm in each mouse. All colonies (n=50 per mouse) were then screened for daptomycin resistance using a daptomycin screen. Of which, a single colony was resuspended in PBS, then diluted 1/100 into CAMHBT containing 50 mg L^{- 1} Ca²⁺, and 1/100 in MH containing 50 mg L^-1 Ca²⁺, and 8 mg L^-1 daptomycin. All suspected daptomycin-resistant colonies were confirmed using a daptomycin BMD MIC as before.

To determine which mutations were present in the rifamycin-resistant isolates, a random selection of 300 colonies, 150 from rifaximin treated mice and 150 from rifampicin treated mice, were sampled for WGS as described above.

Analysis of VREfm isolates from patients receiving rifaximin

To examine the potential association of rifaximin use in humans and the presence of daptomycin-resistant VREfm strains, we analysed VREfm collected between 2018 and 2021 from a single hospital institution in Melbourne. These isolates underwent WGS and daptomycin and rifampicin susceptibility testing as before. Isolates were stratified according to whether the patient received rifaximin within 1 month of VREfm isolation or remained rifaximin free prior to isolate collection. The majority (80%) of isolates were routine screening samples. This included 50 VREfm isolates from patients receiving or previously receiving rifaximin at time of isolation, which were randomly matched to 50 VREfm isolates from patients not receiving rifaximin administration. All isolates were collected within the same time frame. The VREfm isolates were visualised in a maximum-likelihood phylogenetic tree as before, using a core-SNP alignment of 12,430 sites. Isolate MLST was defined with the mlst tool and mutations in RpoB were determined using Snippy (https://github.com/tseemann/snippy) (v4.4.5) as described.

Data visualisation and statistics

All figures were generated in R (v4.0.3, https://www.r-project.org/) using tidyverse (v.1.3.1), patchwork (v.1.1.1), and ggnewscale (v.0.4.5). Statistical analyses were performed using R (v4.0.3, https://www.r-project.org/) and GraphPad Prism (v9.3.1) software packages. Specific tests are given together with each result in the text.

Ethics approval

Ethical approval for analysis of patient data was received from the University of Melbourne Human Research Ethics Committee (study number 1954615.3).