A rifamycin inactivating phosphotransferase familyshared by environmental and pathogenic bacteriaPeter Spanogiannopoulos, Nicholas Waglechner, Kalinka Koteva, and Gerard D. Wright1

Michael G. DeGroote Institute for Infectious Disease Research, Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, ON,Canada L8S 3Z5

Edited by Richard Losick, Harvard University, Cambridge, MA, and approved March 27, 2014 (received for review February 6, 2014)

Many environmental bacteria are multidrug-resistant and repre-sent a reservoir of ancient antibiotic resistance determinants,which have been linked to genes found in pathogens. Exploringthe environmental antibiotic resistome, therefore, reveals thediversity and evolution of antibiotic resistance and also providesinsight into the vulnerability of clinically used antibiotics. In thisstudy, we describe the identification of a highly conserved regula-tory motif, the rifampin (RIF) -associated element (RAE), which isfound upstream of genes encoding RIF-inactivating enzymes froma diverse collection of actinomycetes. Using gene expression assays,we confirmed that the RAE is involved in RIF-responsive regulation.By using the RAE as a probe for new RIF-associated genes in severalactinomycete genomes, we identified a heretofore unknown RIFresistance gene, RIF phosphotransferase (rph). The RPH enzyme isa RIF-inactivating phosphotransferase and represents a new proteinfamily in antibiotic resistance. RPH orthologs are widespread andfound in RIF-sensitive bacteria, including Bacillus cereus and thepathogen Listeria monocytogenes. Heterologous expression andin vitro enzyme assays with purified RPHs from diverse bacterialgenera show that these enzymes are capable of conferring high-level resistance to a variety of clinically used rifamycin antibiotics.This work identifies a new antibiotic resistance protein family andreinforces the fact that the study of resistance in environmentalorganisms can serve to identify resistance elements with relevanceto pathogens.

drug inactivation | phosphorylation | gene regulation | silent resistome

Our control of infectious disease is increasingly challengedbecause of the emergence and dissemination of antibiotic

resistance (1). This problem is exacerbated because of a dwin-dling number of new antimicrobials entering the clinic (2). Themajority of clinically used antibiotics originates from micro-organisms that reside in the environment (3). Recent studieshave established that antibiotic resistance is ancient, and con-temporary environmental bacteria are multidrug-resistant (4–6).There is also growing evidence that environmental bacteria, in-cluding antibiotic producers, are the progenitors of antibiotic-re-sistant determinants from pathogenic bacteria (7–9). Exploringand characterizing the antibiotic resistome, the collection of allantibiotic resistance genes and their precursors from the globalmicrobiota, will facilitate our ability to anticipate and counterantibiotic resistance before its emergence into the clinic (10). Withthis knowledge, we may begin to focus efforts on circumventingthese determinants by strategic drug development, resistance in-hibitor combinations, and prudent administration programs.Rifamycin B, the initial member of the rifamycin family of

antibiotics, was first described in the late 1950s and producedfrom the soil actinomycete Amycolatopsis mediterranei (11). Al-though the parent natural product has modest antibiotic activity,semisynthetic derivatives of this antibiotic family have experi-enced great success in the clinic. The most well-known is ri-fampin (RIF), introduced into the clinic almost 50 y ago (Fig. 1).RIF continues to be a frontline treatment of tuberculosis andother mycobacterial infections and increasingly, is as a fallbacktherapy for drug-resistant Staphylococcal infections (12). The

rifamycins are broad-spectrum antibiotics that inhibit prokaryoticRNA polymerase (RNAP) by binding to a highly conserved re-gion within the β-subunit (encoded by the gene rpoB), blockingthe exit tunnel from the growing RNA nucleotide chain, andthus, sterically hindering the extension of nascent mRNA tran-scripts (13). Recently, synthetic exploration of the rifamycinscaffold has generated a number of new analogs designed totreat a much broader repertoire of bacterial infections. Thesedrugs include rifaximin, which is administered for the treatmentof travelers’ diarrhea and irritable bowel syndrome (14, 15),rifalazil, used for chlamydial infections (16), and a number ofadditional compounds (reviewed in ref. 12). This recent exten-sion of the rifamycin family of antibiotics will likely apply pres-sure for the selection of an expanded repertoire of rifamycinresistance determinants (17).In mycobacteria, point mutations of the drug target are the

most prevalent RIF resistance mechanism (18). Rifamycin re-sistance develops quite rapidly (frequency of 10−8–10−9 perbacterium per cell division), where single amino acid changes inthe β-subunit significantly decrease binding of the antibiotic to itstarget (19). In other microbes, diverse RIF-inactivating mecha-nisms have been described from both pathogenic and non-pathogenic bacteria. Three RIF group transfer mechanisms ofresistance have been described: glycosylation, ADP ribosylation,and phosphorylation (17). These resistance enzymes chemicallymodify the ansa polyketide bridge of RIF, attenuating its affinityfor the β-subunit of RNAP. In addition, the decomposition ofRIF initiated by an RIF monooxygenase has also been described

Significance

Environmental microorganisms are a source of diverse antibi-otic resistance determinants. With the appropriate selectionpressure, these resistance genes can be mobilized to clinicallyrelevant pathogens. Identifying and characterizing elementsof the environmental antibiotic resistome provides an earlywarning of what we may expect to encounter in the clinic. Weuncover a conserved genetic element associated with variousrifamycin antibiotic-inactivating mechanisms. This element ledto the identification of a new resistance gene and associatedenzyme responsible for inactivating rifamycin antibiotics byphosphorylation. Cryptic orthologous genes are also found inpathogenic bacteria but remain susceptible to the drug. Thisstudy reveals a new antibiotic resistance protein family and theunexpected prevalence of a silent rifamycin resistome amongpathogenic bacteria.

Author contributions: P.S. and G.D.W. designed research; P.S., N.W., and K.K. performedresearch; P.S. and G.D.W. analyzed data; P.S. and N.W. performed bioinformatic/phylo-genetic analysis; and P.S. and G.D.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The DNA sequence reported in this paper has been deposited in the NCBIdatabase (accession no. KJ151292).1To whom correspondence should be addressed. E-mail: wrightge@mcmaster.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1402358111/-/DCSupplemental.

7102–7107 | PNAS | May 13, 2014 | vol. 111 | no. 19 www.pnas.org/cgi/doi/10.1073/pnas.1402358111

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0

(20, 21). Despite the fact that RIF phosphorylation has beenknown for 20 y, the gene encoding the RIF phosphotransferase(rph) has yet to be identified (22).Here, we describe the discovery of a conserved upstream nu-

cleotide element associated with various genes encoding RIF-inactivating enzymes from actinomycetes. By exploiting thisconservation, we successfully identified the heretofore unrecognizedrph gene. Through a combination of phylogenetics, heterologousprotein expression, and in vitro enzyme assays, we establish thatRPH orthologs are prevalent in both environmental and patho-genic bacteria, conferring high-level and broad resistance againstclinically used rifamycin antibiotics. This effort highlights thepower of screening genome data for noncoding elements in thediscovery of new resistance elements.

ResultsA Conserved Upstream Nucleotide Element Is Associated with RIF-Inactivating Genes from Actinomycetes. A diverse collection ofRIF-inactivating mechanisms has been described from variouspathogenic and nonpathogenic actinomycetes. We recentlyreported the identification and characterization of a RIF-inactivating glycosyltransferase (rgt1438) from a screen of soilactinomycetes (23). Examination of the intergenic DNA se-quence upstream of rgt1438 identified a 19-bp inverted repeatmotif (9-bp repeats separated by a single nucleotide spacer) (Fig.2A). We hypothesized that this element could be involved ingene regulation. The conservation of this putative RIF-associ-ated element (RAE) was examined by scanning the upstreamintergenic regions from various characterized genes encodingRIF-inactivating enzymes (20, 21, 23, 24). This collection in-cluded three mechanisms of RIF inactivation (glycosylation,ADP ribosylation, and monooxygenation) from four divergentgenera of actinomycetes (Streptomyces, Mycobacteria, Rhodo-coccus, and Nocardia) (Fig. 2A). The RAE was conserved up-stream of all these genes with near-perfect nucleotide identity(Fig. 2A). A BLASTn search of the RAE was conducted fromintergenic regions of the nucleotide collection and whole-genome shotgun contigs from the National Center of Bio-technology Information (25). The RAE was found over 400 timesand predominantly associated with actinobacterial genomes(Dataset S1). The RAE was not associated with rifamycin-pro-ducing bacteria. A sequence logo with this RAE collection showsthe remarkable conservation of this element (Fig. 2B). To de-termine if there is a correlation between the RAE and specificgenes, we compiled a list of available divergent ORFs flankingeach RAE and loosely binned orthologous proteins (Dataset S2).The RAE was strongly associated with ORFs resembling pre-viously characterized RIF-inactivating enzymes (Fig. 2C). TheRAE is observed accompanying an ORF encoding a RIF ADPribosyltransferase (Arr), RIF glycosyltransferase (Rgt), or RIFmonooxygenase (Iri/Rox) enzyme at a frequency of 0.33. Further-more, the RAE is strongly associated with a number of unchar-acterized ORFs. The RAE is found upstream of ORFs annotatedas phosphoenolpyruvate (PEP) synthases and two distinct helicases

(Fig. 2C). This survey revealed that the RAE is highly conservedacross many actinobacterial genomes and has coevolved witha repertoire of genes, including RIF-inactivating genes and severalother genes that are presumably expressed in response to RIF.

The rph Gene Is Responsible for Phosphorylating RIF. A number ofactinomycetes capable of inactivating RIF by phosphorylationhave been described; however, the gene responsible for thisphenotype, rph, has yet to be identified (22, 23). The strongcorrelation of the RAE with genes encoding RIF-inactivatingenzymes led us to apply this link to identify the rph gene. Ina previous phenotypic screen, we identified a collection ofRIF-phosphorylating soil actinomycetes, and thus, we selected

Fig. 1. Rifamycin antibiotics.

Fig. 2. The RAE is associated with RIF-inactivating genes. (A) The RAE isconserved upstream of various characterized RIF-inactivating genes fromdiverse actinobacteria. arr, RIF ADP ribosyltransferase; iri and rox, RIFmonooxygenase; rgt1438, RIF glycosyltransferase; rph-Ss and rph4747, RIFphosphotransferase. The red line indicates the position of the RAE. Analignment of the RAE identified a conserved 19-nt inverted repeat boxed ingray. (B) Sequence logo from over 400 RAEs found in the nucleotide col-lection and whole-genome shotgun contig databases. Sequence logos weregenerated using WebLogo. (C) RAE gene associations.

Spanogiannopoulos et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7103

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0

one of these strains for additional investigation (23). The draftgenome of WAC4747 (most closely resembling Streptomycesviridochromogenes based on 16S rRNA analysis) was deter-mined. We also biochemically screened our in-house collectionof genome-sequenced actinomycetes for their ability to phosphor-ylate RIF and identified Streptomyces sviceus ATCC 29083 as acandidate (Fig. S1). Putative rph genes were identified by probingthe genome sequences of WAC4747 and S. sviceus for the RAEand examining associated downstream genes. The RAE was notfound to be flanking any genes resembling known antibioticphosphotransferases (26). The RAE was found on two contigswithin the WAC4747 genome: on a 3-kb contig upstream of anORF encoding a putative PEP synthase (ORF1) and an 8.5-kbcontig upstream of a putative helicase (ORF2) (Table S1).Similarly, the RAE was found upstream of predicted PEP syn-thase and helicase ORFs within the S. sviceus genome in additionto a candidate RIF monooxygenase (Table S1). PEP synthaseand helicase proteins have not been previously associated withantibiotic resistance. Helicases harness the energy of ATP hy-drolysis for separation of nucleotide strands, whereas PEP syn-thases catalyze the conversion of ATP and pyruvate to PEP,AMP, and Pi (27, 28). Orthologous genes of ORF1 and ORF2are conserved in other actinomycete genomes and colocalizedwith the RAE (Fig. 2C).We individually disrupted ORF1 and ORF2 from WAC4747

and evaluated the ability of the mutant strains to phosphorylateRIF. Analysis of conditioned media from WAC4747 grown inthe presence of RIF showed complete inactivation of RIFwithin 48 h (Fig. 3). In contrast, a mutant strain of WAC4747with a disrupted ORF1 (putative PEP synthase) failed to in-activate RIF (Fig. 3). Disruption of ORF2 (putative helicase) inWAC4747 had no effect on RIF inactivation. These findingssuggest that ORF1 (named rph4747) encodes the RPH enzyme.Inactivation of the orthologous putative PEP synthase gene fromS. sviceus (rph-Ss) generated identical results (Fig. S2). Mutantstrains of WAC4747 and S. sviceus with a disrupted rph genedisplayed a modest twofold decrease in RIF minimum inhibitoryconcentrations (MICs) compared with the WT strains (Table S2).

This result suggests, similar to other actinomycetes, that WAC4747and S. sviceus harbor multiple RIF resistance mechanisms (21, 23).We assessed the ability of the rph gene to contribute to RIF

resistance by expression in a heterologous host. Introduction ofrph4747 or rph-Ss into RIF-sensitive Escherichia coli conferreda 64-fold increase in RIF MIC compared with the control (Table1). E. coli-expressing rph displayed high-level resistance up to orover 512 μg/mL to a panel of natural product and clinically usedsemisynthetic rifamycin antibiotics (Table 1). This result high-lights the ability of rph to confer broad high-level resistance torifamycin antibiotics.

RAE Responds to RIF. Because of the strong association of theRAE and genes encoding RIF-inactivating enzymes, we hy-pothesized that this DNA motif is likely involved in gene ex-pression in response to RIF. Changes in transcript levels ofrph4747 expression were monitored using quantitative RT-PCRwith growth of WAC4747 in the presence or absence of RIF. Inthe presence of RIF, the rph4747 transcript was up-regulated63- and 91-fold after 1 and 2 h of growth, respectively, com-pared with the control, DMSO (Fig. 4A). This finding confirmsthat genes paired with the RAE are up-regulated in responseto RIF.To validate the role of the RAE in RIF-specific gene regula-

tion, we used a kanamycin (KAN) resistance reporter assay. Thereporter was constructed by fusing the intergenic DNA upstreamof the rph4747 gene ahead of a promoter-less KAN resistancecassette followed by integration into the genome of WAC4747,creating the strain WAC4747-PRAE/neo-pSET152. This strainwas not KAN-resistant; showing gene expression downstreamof the RAE is tightly regulated. However, exposure to sub-inhibitory concentrations of rifamycin antibiotics (RIF andthe parent natural product rifamycin SV) induced growth onKAN-containing media, confirming that RAE is involved inrifamycin-responsive gene regulation (Fig. 4B). A number ofdiverse antibiotics, from natural and synthetic origins, inhibitingvarious biological targets failed to induce KAN resistance, con-firming that gene regulation from the RAE is rifamycin-specific.

Fig. 3. The rph4747 gene from WAC4747 is responsible for the inactivationof RIF. Strains of WAC4747 were grown in the presence of RIF for 48 h, andconditioned media were analyzed for residual RIF activity using an RIF-sen-sitive indicator organism (B. subtilis). Samples were additionally analyzedusing HPLC. WAC4747 and the control, WAC4747-pSET152 (empty vector),completely inactivate RIF within 48 h. The rph4747 mutant strain, WAC4747rph4747::pSUIC, does not inactivate RIF and is comparable with the controlof media and drug (Streptomyces isolation media (SIM)+RIF). Chromato-grams are displayed at an absorbance of 343 nm.

Table 1. Rifamycin MIC determinations of various rph genesexpressed in E. coli

Construct*/IPTG

MIC (μg/mL)

RIF Rifamycin SV Rifabutin Rifaximin

pET22b− 8 64 8 16+ 8 64 8 16

pET22b-rph4747− 512 512 512 64+ >512 >512 512 512

pET28b− 8 64 8 16+ 8 64 8 16

pET28b-rph-Ss− 512 >512 512 512+ >512 >512 512 512

pET19Tb− 8 64 8 8+ 8 64 8 8

pET19Tb-rph-Bc− 128 256 64 64+ >512 >512 512 512

pET19Tb-rph-Lm− 128 128 32 32+ >512 >512 512 512

IPTG, isopropyl β-D-1-thiogalactopyranoside.*E. coli Rosetta(DE3)pLysS host.

7104 | www.pnas.org/cgi/doi/10.1073/pnas.1402358111 Spanogiannopoulos et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0

Additionally, a reporter construct was made where the RAE wasreplaced with a random sequence of nucleotides. This constructfailed to respond to RIF (Fig. S3) and established that the RAEnucleotide sequence is critical for gene expression.In some instances, antibiotic resistance gene expression is

dependent on the antibiotic successfully binding and inhibitingits target (29, 30). We generated RIF-resistant rpoB mutants ofWAC4747-PRAE/neo-pSET152 to determine if inhibition of RNAPis a prerequisite for RAE-mediated gene induction. This mu-tant strain gained KAN resistance in response to RIF (Fig. S3),indicating that gene expression mediated by the RAE is in-dependent of RNAP inhibition.

Diversity of the RPH Protein Family. A BLASTp analysis of theRPH enzyme reveals that it is related to enzymes annotatedas PEP synthases (EC 2.7.9.2). The mechanism of PEP syn-thase proceeds through a phosphoenzyme intermediate, wherethe β-phosphate of ATP is transferred to a histidine (His) resi-due, which in turn, is transferred to pyruvate (31). PEP syn-thase is composed of three domains: the N-terminal domainis responsible for nucleotide binding, the C-terminal domainbinds pyruvate, and the central segment contains the catalyticHis domain.A global alignment of RPH4747 and PEP synthase from

E. coli indicates low overall amino acid similarity but relativelyhigh similarity near the N terminus. A search of the ConservedDomain Database (32) with RPH4747 showed that the N terminusconsists of an ATP binding domain, similar to PEP synthase.In contrast to PEP synthase, the catalytic His domain of RPHis located at the C terminus (Fig. S4). The central region ofRPH4747 shows no similarity to any previously characterizedproteins. Therefore, RPH4747 shares some similarities with PEPsynthase, but the overall architectures of these two enzymesdiffer. A search of the RPH using the Conserved Domain Ar-chitecture Retrieval Tool (33) showed that this domain archi-tecture is found in over 1,600 proteins and primarily restricted tobacterial genomes. Although the majority of these enzymes isannotated as PEP synthases (or PEP binding proteins), there isno biochemical evidence confirming this activity, and therefore,it represents an unexplored protein family.

We performed a phylogenetic analysis of the RPH proteinfamily to explore the relationship of the sequences in this family(Dataset S3). The RPH cladogram is shown in Fig. 5. We canidentify several clades with high bootstrap support that corre-spond to well-defined taxonomic groups, even if resolutionwithin and between these groups remains unclear (Fig. S5).RPH proteins are widely distributed, predominately in Gram-positive bacteria. Although many RPH orthologs are foundwithin Actinobacteria, the majority is associated with Bacilli andClostridia. We examined the intergenic DNA sequence upstreamof RPH ORFs and discovered that the majority of ORFs fromActinobacteria is associated with the RAE (Fig. 5). This findingis consistent with our previous survey of the RAE and geneassociations (Fig. 2C). The RAE was not associated with RPHsoutside Actinobacteria. It is, therefore, uncertain whether theseenzymes possess RPH activity.A previous biochemical screen by Dabbs et al. (34) showed

that a number of bacteria from the genus Bacillus inactivatedRIF by phosphorylation. This report was intriguing consid-ering that Bacillus spp. are highly susceptible to RIF (35). Ourphylogenetic analysis of the RPH protein family revealedorthologous enzymes from a number of Bacilli. This list alsoencompassed a number of pathogenic bacteria, including Bacilluscereus, Bacillus anthracis, and Listeria monocytogenes. Thesepathogens have been previously reported to be highly susceptibleto RIF (36, 37). Indeed, this finding was confirmed by per-forming RIF susceptibility testing with B. cereus and L. mono-cytogenes, which showed a RIF MIC of 62.5 ng/mL. RPHorthologs from B. cereus (RPH-Bc) and L. monocytogenes (RPH-Lm) displayed 55% and 52% identity and 74% and 71% simi-larity, respectively, compared with RPH4747. We selected thesetwo RPH orthologs as representatives from our phylogeneticanalysis and evaluated their ability to confer RIF resistance whenexpressed in a heterologous host. E. coli expressing rph-Bc or

Fig. 4. The RAE is involved in gene regulation in response to RIF. (A)Quantitative RT-PCR analysis of rph4747 transcripts. WAC4747 was grown inthe presence of RIF or DMSO. Data are presented as fold difference ofrph4747 transcripts from three independent biological replicates, and errorbars represent SDs. (B) A WAC4747 KAN resistance reporter assay. The RAEfrom WAC4747 was fused to a promoterless KAN resistance cassette andintegrated in the genome of WAC4747. Growth on KAN-containing mediawere only achieved in the presence of RIF and RIF SV. Subinhibitory con-centrations of antibiotics were used: AMP, 5 μg; ciprofloxacin (CIPRO), 0.5 μg;erythromycin (ERYTH), 0.125 μg; novobiocin (NOVO), 0.05 μg; RIF, 2.5 μg; RIFSV, 5 μg; streptomycin (STREPTO), 0.05 μg; vancomycin (VANCO), 0.05 μg.

Fig. 5. Cladogram of the RPH protein family. Colored clades representgroups with over 85% bootstrap support based on maximum likelihoodanalysis using RAxML. Black circles indicate RPHs that are paired with theRAE. Red stars indicate RPHs selected for characterization: RPH4747,WAC4747; RPH-Bc, B. cereus ATCC 14579; RPH-Lm, L. monocytogenes str. 4b.F2365; RPH-Ss, S. sviceus ATCC 29083. The gray bar represents RPHs frombacteria of the Bacillales order. The light gray bar represents RPHs from theListeriaceae family. A comprehensive cladogram with labels can be found inFig. S5.

Spanogiannopoulos et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7105

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0

rph-Lm exhibited a 64-fold increase in RIF MIC and also con-ferred high-level resistance to a variety of rifamycin analogs(Table 1). This result confirms that RPH orthologs are found inRIF-sensitive pathogenic bacteria and will confer high-leveland broad rifamycin resistance when expressed in a hetero-logous host.

Biochemical Characterization of RPH Activity. Recombinant RPHenzymes were overexpressed from E. coli and purified for char-acterization. Initially, RPH activity was monitored using a TLCassay, which confirmed that ATP served as a cosubstrate (Fig.S6). RIF-phosphate product from various RPHs was purifiedand analyzed using high-resolution MS and NMR spectroscopy(Fig. S6). This analysis confirmed that RPHs share the sameregiospecificity and phosphorylate RIF at the hydroxyl groupattached to the C21 of the ansa-chain, consistent with precedentusing whole-cell inactivation (Fig. 1) (22).Liquid chromatography electrospray ionization MS coupled

with multiple reaction monitoring were used to quantify RPHactivity. Steady-state kinetics were performed for three RPHenzymes from diverse bacterial hosts; RPH-Ss from S. sviceus,RPH-Bc from B. cereus, and RPH-Lm from L. monocytogenes.The gene encoding RPH-Ss is associated with the RAE, whereasthe genes encoding RPH-Bc and RPH-Lm are not. All RPHsshowed comparable kinetic constants regardless of the hostsource and RAE association (Table S3). RPHs exhibited very lowKm values for RIF within the high-nanomolar range, indicatinga high affinity for RIF. The Km values for RIF only varied twofoldamong the different RPH enzymes. Similarly, all RPHs displayedcomparable kcat constants (catalytic rate constant).Although PEP synthase and RPH share aspects of function and

amino acid structure, it is unclear whether these two proteinfamilies have similar catalytic features. In addition to the liquidchromatography electrospray ionization MS assay, we furthermonitored RPH-Bc activity by measuring the generation of free Pi.Steady-state kinetic constants generated using this orthogonalassay were analogous to values obtained with the MS methoddescribed above (Table S3), confirming stoichiometric productionof RIF-phosphate and Pi by RPH-Bc. With this assay, RPHs werealso assayed for PEP synthase activity. All RPHs failed to phos-phorylate pyruvate and showed no PEP synthase activity.A sequence alignment of the region surrounding the catalytic

His residue from PEP synthase and structurally and mecha-nistically related enzymes identified a conserved motif T-X-X-G-G-X-X-X-H; the conserved His is required for formation ofa phosphoenzyme intermediate (Fig. S7). A similar motif wasfound in the RPH family of enzymes fewer than 50 aa awayfrom the C terminus of the protein. An His-827-Ala mutant ofRPH-Bc failed to confer RIF resistance in E. coli, and the en-zyme had no in vitro enzymatic activity (Fig. S7). Together, theseresults suggest that RPH is mechanistically related to PEPsynthase, likely using a phosphohistidine intermediate duringcatalysis, despite having architectural differences with themetabolic enzyme.

DiscussionThe genetic reservoirs of antibiotic resistance and their regula-tion are poorly understood. We and others have shown thatnonpathogenic environmental bacteria harbor an extensive andancient resistome that is the likely source of resistance elementscirculating in pathogens today (4–6, 9). The regulation of re-sistance genes by target antibiotics has been shown in only a fewcases (for example, the induction of the efflux protein TetA bytetracycline binding to TetR and the complex regulation ofβ-lactam resistance by these antibiotics in Staphylococcus aureus)(38, 39). In this study, we characterize a new family of rifamycinresistance enzymes along with a conserved RIF-associated reg-ulatory element, the RAE, which is shared among genesencoding various RIF-inactivating enzymes from actinomycetes.The RAE is composed of a 19-nt inverted repeat that is

essential for modulating gene expression. Such inverted re-peat DNA motifs have long been linked to gene regulation andoften interact with protein regulators. Additional work is re-quired to reveal the details of the mechanism of induction as-sociated with this RAE and determine if any additionalregulators are involved.For over two decades, a number of actinomycetes has been

reported to inactivate RIF by phosphorylation; however, thegene encoding the RPH was not known. Our observation of thestrong correlation of the RAE with various actinomycete genesknown to encode RIF-inactivating enzymes suggested a commonregulatory mechanism and an entry point to discover the rphgene. Indeed, we successfully identified and validated that a genepredicted to encode a PEP synthase was, in fact, the rph. TheRPH enzyme does not show similarity to any previously identi-fied antibiotic phosphotransferases. Instead, RPH is related tothe multidomain gluconeogenic enzyme PEP synthase. AlthoughRPH and PEP synthase differ in overall domain architectures(Fig. S4), RPH enzymes have been routinely annotated as PEPsynthases from sequenced genomes. Therefore, RPH representsa previously unexplored and misannotated protein family. Ki-netic analysis and site-directed mutagenesis confirmed that theRPH enzyme is mechanistically related to PEP synthase anddoes not use pyruvate as a substrate. A Conserved Domain Ar-chitecture Retrieval Tool search has shown that this proteinfamily is widely distributed across bacteria and that these enzymesare likely not involved in gluconeogenesis and may have diversefunctions, including antibiotic resistance. Additional work is re-quired to decipher their substrates and functions.RPH orthologs from actinomycetes are colocalized with the

RAE. However, the majority of RPHs is found in other Gram-positive genera (Fig. 5). Interestingly, many of these Gram pos-itives have an RIF-sensitive phenotype. Of particular concern arethe orthologs of rph that are found within the genomes of patho-gens, such as B. anthracis, B. cereus , and L. monocytogenes. Het-erologous expression and in vitro enzyme assays confirmed theyare bona fide RPHs and not PEP synthases and confer high-leveland broad-spectrum resistance with equivalent competence to theirStreptomycete counterparts. This unexpected diversity and preva-lence of rph genes in a number of bacterial genera are troublinggiven the renewed interest in expanding the application of theseantibiotics outside their use in mycobacterial infection (12, 14–16).Antibiotic resistance genes that do not normally confer re-

sistance for their native host but are proficient at conferringresistance when expressed in an alternate host are called silentresistance genes (40). The majority of rph genes is not fromactinomycetes and not associated with the RAE. Therefore,these bacteria cannot respond to RIF and activate gene ex-pression accordingly. We speculate that these nonactinomyceterph genes are evolutionary related to rphs from actinomycetesbut have an alternate function in their native host, which is notassociated with antibiotic resistance. Their ability to bind andphosphorylate rifamycins is possibly caused by structural simi-larities with their natural substrates. Silent resistance genes have,thus far, been largely underappreciated, likely because of theirinability to confer an antibiotic resistance phenotype in theirnative host. Recently, functional metagenomics have revealednumerous novel resistance genes, and continued research in thisarea is likely to uncover additional genes (41–43).Antibiotic resistance mediated by enzymes is of particular

clinical concern because of their potential to become mobilizedand distributed horizontally between diverse species of bacteria.Examples of this mobilization are prevalent for many families ofantibiotics, and this trend has begun to emerge with RIF-inac-tivating enzymes. The Arr enzyme, which is responsible for theinactivation of RIF by ADP ribosylation, was first described fromthe normally benign Mycobacterium smegmatis; however, thegene encoding this enzyme is now mobile and found on trans-missible plasmids and integrons carrying multiple antibiotic re-sistance determinants (24, 44, 45). These mobile elements are now

7106 | www.pnas.org/cgi/doi/10.1073/pnas.1402358111 Spanogiannopoulos et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0

found in a variety of pathogens. The dissemination of Arr andpotentially other RIF-inactivating enzymes in the near futurethreatens the efficacy of the entire rifamycin family of antibiotics.This work has established that RPH enzymes do not dis-

criminate between various rifamycins and effectively inactivatenatural product and semisynthetic derivatives of this family.Understanding the molecular interactions between RPH andrifamycin substrates will provide the basis for generating rifa-mycin analogs that are not susceptible to these enzymes. Thepossibility of resistance-proof rifamycins has been exemplifiedwith the recent development of C25 carbamate rifamycin deriv-atives that retain activity against Arr-containing bacteria (46).

Materials and MethodsAdditional details are available in SI Text.

Inactivation of the rph Gene from WAC4747 and S. sviceus. Disruption of therph gene was accomplished by insertional inactivation using a previouslydescribed strategy (23). A partial fragment (∼1,000 bp) of the rph4747 genefrom WAC4747 was amplified by PCR using primers rph4747-part-F andrph4747-part-R and cloned into pSUIC (a suicide version of pSET152 thatlacks attP and int). The rph-Ss gene from S. sviceus ATCC 29083 was ampli-fied using primers rph-Ss-part-F and rph-Ss-part-R and cloned as described

above. The helicase gene from WAC4747 was amplified using primershel4747-part-F and hel4747-part-R. Conjugations were performed as pre-viously described (23). A list of primers can be found in Table S4.

RIF Inactivation Assay. Fresh spores were used to inoculate 5 mL SIM media (5).Cultures were grown for 4 d at 30 °C with shaking at 250 rpm; 0.5 mL culturewas then used to inoculate 4.5 mL fresh SIM containing RIF at 20 μg/mL andrepresented the zero time point. At 48 h, conditioned media samples weretaken and analyzed for residual RIF activity using the Kirby–Bauer disk dif-fusion assay with Bacillus subtilis as the RIF-sensitive indicator microorgan-ism. Additionally, supernatant samples were extracted with an equal volumeof methanol, clarified by centrifugation, and analyzed using a Waters e2695HPLC system equipped with an XSelect CSH C18 5 μM column (4.6 × 100 mm)with the following method: linear gradient of 37–55% (vol/vol) acetonitrilein water with 0.05% TFA at 1 mL/min over 12 min.

ACKNOWLEDGMENTS. We thank Andrew G. McArthur for assistance withthe RPH phylogenetic analysis and Christine King for help with sequencing oftheWAC4747 genome. We thank Michael Fischbach for providing Streptomycessviceus ATCC 20983 and Lori Burrows for the gift of Listeria monocytogenesstr. 4b. F2365. We also thank Georgina Cox and Grace Yim for helpfuldiscussions. This research was funded by Canadian Institutes of HealthResearch Grant MT-13536 and a Canada Research Chair in AntibioticBiochemistry (G.D.W.).

1. Fischbach MA, Walsh CT (2009) Antibiotics for emerging pathogens. Science 325(5944):1089–1093.

2. Cooper MA, Shlaes D (2011) Fix the antibiotics pipeline. Nature 472(7341):32.3. Bérdy J (2012) Thoughts and facts about antibiotics: Where we are now and where we

are heading. J Antibiot (Tokyo) 65(8):441.4. D’Costa VM, et al. (2011) Antibiotic resistance is ancient. Nature 477(7365):457–461.5. D’Costa VM, McGrann KM, Hughes DW, Wright GD (2006) Sampling the antibiotic

resistome. Science 311(5759):374–377.6. Bhullar K, et al. (2012) Antibiotic resistance is prevalent in an isolated cave micro-

biome. PLoS ONE 7(4):e34953.7. Benveniste R, Davies J (1973) Aminoglycoside antibiotic-inactivating enzymes in ac-

tinomycetes similar to those present in clinical isolates of antibiotic-resistant bacteria.Proc Natl Acad Sci USA 70(8):2276–2280.

8. Marshall CG, Broadhead G, Leskiw BK, Wright GD (1997) D-Ala-D-Ala ligases fromglycopeptide antibiotic-producing organisms are highly homologous to the entero-coccal vancomycin-resistance ligases VanA and VanB. Proc Natl Acad Sci USA 94(12):6480–6483.

9. Forsberg KJ, et al. (2012) The shared antibiotic resistome of soil bacteria and humanpathogens. Science 337(6098):1107–1111.

10. Wright GD (2007) The antibiotic resistome: The nexus of chemical and genetic di-versity. Nat Rev Microbiol 5(3):175–186.

11. Sensi P, Margalith P, Timbal MT (1959) Rifomycin, a new antibiotic; preliminary re-port. Farmaco Sci 14(2):146–147.

12. Aristoff PA, Garcia GA, Kirchhoff PD, Hollis Showalter HD (2010) Rifamycins—obstacles and opportunities. Tuberculosis (Edinb) 90(2):94–118.

13. Campbell EA, et al. (2001) Structural mechanism for rifampicin inhibition of bacterialrna polymerase. Cell 104(6):901–912.

14. Huang DB, DuPont HL (2005) Rifaximin—a novel antimicrobial for enteric infections.J Infect 50(2):97–106.

15. Cremonini F, Lembo A (2012) Rifaximin for the treatment of irritable bowel syn-drome. Expert Opin Pharmacother 13(3):433–440.

16. Rothstein DM, van Duzer J, Sternlicht A, Gilman SC (2007) Rifalazil and other ben-zoxazinorifamycins in the treatment of chlamydia-based persistent infections. ArchPharm (Weinheim) 340(10):517–529.

17. Tupin A, et al. (2010) Resistance to rifampicin: At the crossroads between ecological,genomic and medical concerns. Int J Antimicrob Agents 35(6):519–523.

18. Ramaswamy S, Musser JM (1998) Molecular genetic basis of antimicrobial agent re-sistance in Mycobacterium tuberculosis: 1998 update. Tuber Lung Dis 79(1):3–29.

19. Floss HG, Yu TW (2005) Rifamycin-mode of action, resistance, and biosynthesis. ChemRev 105(2):621–632.

20. Andersen SJ, Quan S, Gowan B, Dabbs ER (1997) Monooxygenase-like sequence ofa Rhodococcus equi gene conferring increased resistance to rifampin by inactivatingthis antibiotic. Antimicrob Agents Chemother 41(1):218–221.

21. Hoshino Y, et al. (2010) Monooxygenation of rifampicin catalyzed by the rox geneproduct of Nocardia farcinica: Structure elucidation, gene identification and role indrug resistance. J Antibiot (Tokyo) 63(1):23–28.

22. Yazawa K, Mikami Y, Maeda A, Morisaki N, Iwasaki S (1994) Phosphorylative in-activation of rifampicin by Nocardia otitidiscaviarum. J Antimicrob Chemother 33(6):1127–1135.

23. Spanogiannopoulos P, Thaker M, Koteva K, Waglechner N, Wright GD (2012) Char-acterization of a rifampin-inactivating glycosyltransferase from a screen of environ-mental actinomycetes. Antimicrob Agents Chemother 56(10):5061–5069.

24. Quan S, Venter H, Dabbs ER (1997) Ribosylative inactivation of rifampin by Myco-bacterium smegmatis is a principal contributor to its low susceptibility to this anti-biotic. Antimicrob Agents Chemother 41(11):2456–2460.

25. Altschul SF, et al. (1997) Gapped BLAST and PSI-BLAST: A new generation of protein

database search programs. Nucleic Acids Res 25(17):3389–3402.26. Morar M, Wright GD (2010) The genomic enzymology of antibiotic resistance. Annu

Rev Genet 44:25–51.27. Cooper RA, Kornberg HL (1967) The mechanism of the phosphoenolpyruvate syn-

thase reaction. Biochim Biophys Acta 141(1):211–213.28. Singleton MR, DillinghamMS, Wigley DB (2007) Structure and mechanism of helicases

and nucleic acid translocases. Annu Rev Biochem 76:23–50.29. Narayanan CS, Dubnau D (1985) Evidence for the translational attenuation model:

Ribosome-binding studies and structural analysis with an in vitro run-off transcript of

ermC. Nucleic Acids Res 13(20):7307–7326.30. Alexieva Z, Duvall EJ, Ambulos NP, Jr., Kim UJ, Lovett PS (1988) Chloramphenicol in-

duction of cat-86 requires ribosome stalling at a specific site in the leader. Proc Natl

Acad Sci USA 85(9):3057–3061.31. Narindrasorasak S, Bridger WA (1977) Phosphoenolypyruvate synthetase of Escher-

ichia coli: Molecular weight, subunit composition, and identification of phosphohis-

tidine in phosphoenzyme intermediate. J Biol Chem 252(10):3121–3127.32. Marchler-Bauer A, et al. (2013) CDD: Conserved domains and protein three-dimensional

structure. Nucleic Acids Res 41(Database issue):D348–D352.33. Geer LY, Domrachev M, Lipman DJ, Bryant SH (2002) CDART: Protein homology by

domain architecture. Genome Res 12(10):1619–1623.34. Dabbs ER, et al. (1995) Rifampicin inactivation by Bacillus species. J Antibiot (Tokyo)

48(8):815–819.35. Dey A, Chatterji D (2012) Tracing the variation in physiological response to rifampicin

across the microbial spectrum. J Bacteriol Virol 42(2):87–100.36. Vogler AJ, et al. (2002) Molecular analysis of rifampin resistance in Bacillus anthracis

and Bacillus cereus. Antimicrob Agents Chemother 46(2):511–513.37. Hawkins AE, Bortolussi R, Issekutz AC (1984) In vitro and in vivo activity of various

antibiotics against Listeria monocytogenes type 4b. Clin Invest Med 7(4):335–341.38. Hillen W, Berens C (1994) Mechanisms underlying expression of Tn10 encoded tet-

racycline resistance. Annu Rev Microbiol 48:345–369.39. Llarrull LI, Fisher JF, Mobashery S (2009) Molecular basis and phenotype of methicillin

resistance in Staphylococcus aureus and insights into new beta-lactams that meet the

challenge. Antimicrob Agents Chemother 53(10):4051–4063.40. Dantas G, Sommer MO (2012) Context matters - the complex interplay between re-

sistome genotypes and resistance phenotypes. Curr Opin Microbiol 15(5):577–582.41. Torres-Cortés G, et al. (2011) Characterization of novel antibiotic resistance genes

identified by functional metagenomics on soil samples. Environ Microbiol 13(4):

1101–1114.42. Pehrsson EC, Forsberg KJ, Gibson MK, Ahmadi S, Dantas G (2013) Novel resistance

functions uncovered using functional metagenomic investigations of resistance res-

ervoirs. Front Microbiol 4(145):1–11.43. McGarvey KM, Queitsch K, Fields S (2012) Wide variation in antibiotic resistance

proteins identified by functional metagenomic screening of a soil DNA library. Appl

Environ Microbiol 78(6):1708–1714.44. Arlet G, et al. (2001) Plasmid-mediated rifampin resistance encoded by an arr-2-like

gene cassette in Klebsiella pneumoniae producing an ACC-1 class C beta-lactamase.

Antimicrob Agents Chemother 45(10):2971–2972.45. Tribuddharat C, Fennewald M (1999) Integron-mediated rifampin resistance in

Pseudomonas aeruginosa. Antimicrob Agents Chemother 43(4):960–962.46. Combrink KD, et al. (2007) New C25 carbamate rifamycin derivatives are resistant to

inactivation by ADP-ribosyl transferases. Bioorg Med Chem Lett 17(2):522–526.

Spanogiannopoulos et al. PNAS | May 13, 2014 | vol. 111 | no. 19 | 7107

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 5,

202

0