Increased whiB7 expression and antibiotic resistance in Mycobacterium chelonae carrying two prophages

Background The global rise in the incidence of non-tuberculosis mycobacterial infections is of increasing concern due their high levels of intrinsic antibiotic resistance. Although integrated viral genomes, called prophage, are linked to increased antibiotic resistance in some bacterial species, we know little of their role in mycobacterial drug resistance. Results We present here for the first time, evidence of increased antibiotic resistance and expression of intrinsic antibiotic resistance genes in a strain of Mycobacterium chelonae carrying prophage. Strains carrying the prophage McProf demonstrated increased resistance to amikacin. Resistance in these strains was further enhanced by exposure to sub-inhibitory concentrations of the antibiotic, acivicin, or by the presence of a second prophage, BPs. Increased expression of the virulence gene, whiB7, was observed in strains carrying both prophages, BPs and McProf, relative to strains carrying a single prophage or no prophages. Conclusions This study provides evidence that prophage alter expression of important mycobacterial intrinsic antibiotic resistance genes and additionally offers insight into the role prophage may play in mycobacterial adaptation to stress.


Background
Prophage (integrated viral genomes) are major drivers of bacterial virulence and antibiotic resistance in bacteria, yet the mechanisms of prophage-mediated antibiotic resistance are unknown [1,2] . Prophages are common in mycobacteria, including clinical isolates of emerging non-tuberculosis pathogenic mycobacteria, Mycobacterium avium and M. abscessus [3,4]. M. abscessus is of significant concern as it is considered one of the most antibiotic resistant pathogens [5,6]. Extensively resistant isolates share increased expression of conserved mycobacterial intrinsic antibiotic resistance genes such as whiB7, making drug treatment challenging [5][6][7]. Understanding how intrinsic antibiotic resistance genes are regulated in pathogenic mycobacteria would provide opportunities to develop novel and more effective treatment approaches [7,8].
Clarithromycin (CLA) combined with amikacin (AMK) is the treatment of choice for M. abscessus infections but with the emergence of resistance to both of these drugs, treatment is becoming increasingly difficult [7,9]. Recent efforts in sequencing of clinical M. abscessus isolates determined that mutations in the 23S rRNA gene, rrl, and in the erm gene, which confers macrolide (e.g. clarithromycin) resistance, are typically associated with elevated CLA resistant phenotypes but did not account for all the clarithromycin resistant phenotypes [7]. What was consistent in all clarithromycin-resistant isolates was elevated expression of genes in the whiB7 regulon, including the transcription factor whiB7 and its target genes, the drug efflux pumps plus other antibiotic resistance genes including erm and eis [7,[10][11][12]. These latter genes confer macrolide and aminoglycoside resistance, respectively. Likewise, in AMK-resistant isolates of M. abscessus there is typically a 10-fold increase in whiB7 expression relative to AMK susceptible strains [13]. whiB7 is conserved across all mycobacteria, including pathogenic and non-pathogenic species such as M. tuberculosis, M. abscessus, M. chelonae and M. smegmatis [14,15]. Characterizing the pathways that lead to increased whiB7 expression and intrinsic drug resistance in pathogenic mycobacteria will be important for identifying new targets for novel drug development [9,16,17].
The majority of bacterial pathogens carry prophage that are known to contribute to bacterial virulence and fitness [2,18,19]. Prophage introduce novel genes into bacterial genomes that can result in phenotypes that are more competitive in bacterial populations [2,18]. Prophage also contribute to antibiotic resistance and persistence. Nine cryptic prophages (transcriptionally active prophage that cannot carry out lytic infections) in E. coli significantly increase resistance to quinolones and betalactam antibiotics compared to strains in which all or combinations of prophages had been cured, although the mechanism by which these prophage affected resistance was not reported [1]. Toxin/antitoxin (TA) systems encoded by prophage are also known to increase resistance and persistence in the presence of antibiotics. In E. coli, prophage-encoded TA pair RalR/RalA increases resistance to broad-spectrum fosfomycin and the RelE toxin of prophage Qin leads to persistence in the presence of ciprofloxacin, ampicillin and tobramycin [20][21][22]. The majority of mycobacterial pathogens also carry prophage and they are hypothesized to play a role in virulence, yet remain largely uninvestigated [3,23,24].
In this study we examine the impact of two mycobacteriophages on intrinsic antibiotic resistance and whiB7 expression in the non-tuberculosis mycobacterial pathogen, M. chelonae, a member of the M. abscessus/chelonae complex. The disease caused by M. chelonae, mainly soft tissue and disseminating infections, differs from that of M. abscessus; however, the whiB7-dependent mechanisms of intrinsic resistance are conserved across all mycobacteria, including M. chelonae [15,25]. We identified a naturally occurring prophage in M. chelonae that also occurs in the sequenced genomes of at least 25 clinical isolates of M. abscessus. We characterized the genome of the M. chelonae prophage, McProf, and created a cured strain that lacks prophage. Antibiotic resistance and gene expression of this strain was compared to that of M. chelonae carrying a single or multiple prophages.

Results
Double lysogens of M. chelonae have increased resistance to aminoglycosides amikacin and tobramycin The wild type M. chelonae (ATCC 35752) carries a naturally occurring 67,657-bp prophage that we have named McProf. To determine how prophages impact gene expression and the antibiotic resistance phenotype of M. chelonae we added a second prophage. We identified three mycobacteriophages capable of infecting M. chelonae, Muddy, WildCat and BPs, of which only BPs is known to be temperate [26][27][28]. A double lysogen of M. chelonae was created from the WT M. chelonae strain using the cluster G mycobacteriophage, BPs [28]. BPs integrates into an attB site located within the 3′ end of the host tRNA-Arg gene (BB28_RS01100), that is similar to the BPs attB site in M. smegmatis (Msmeg_6349) [28]. BPs lysogens of the M. chelonae WT strain (BPs, McProf) appear to be more stable than BPs lysogens of M. smegmatis. Lysogens form at a higher efficiency in M. chelonae WT (25%) compared to that in M. smegmatis (5%) and release fewer particles into cell culture supernatant (10 4 -10 5 PFUs ml − 1 compared to 10 10 PFUs ml − 1 ) [28,29].
To determine if the presence of a second prophage in M. chelonae alters susceptibility to antibiotics, we determined the minimum inhibitory concentrations (MIC) for the double M. chelonae lysogen (BPs, McProf) relative to the WT strain (McProf) in the presence of varying levels of the aminoglycosides, amikacin (AMK) and tobramycin (TOB), and tetracycline (TET) ( Table 1). The presence of the second prophage, BPs, significantly increased resistance to both aminoglycosides. There was not a consistent significant difference in resistance to TET. As a positive control, we exposed the M. chelonae (McProf) strain to sub-inhibitory concentrations of acivicin (ACI), a known inducer of intrinsic resistance in mycobacteria [12]. As expected, ACI significantly increased the resistance of M. chelonae (McProf) in both the AMK and TOB assays (Table 1). ACI treatment did not significantly alter TET resistance of M. chelonae (McProf).
We also determined the viability of the strains after antibiotic treatment by adding AlamarBlue to the wells to detect metabolic activity (Fig. 1). There was a statistically significant difference in viability between the double lysogen and WT strains treated with 64 μg ml − 1 AMK and 8 μg ml − 1 TOB. We noted a slight increase in viability of the WT strain at 8 μg ml − 1 TOB; however, there was no evidence of growth at this concentration of TOB. The WT strain treated with ACI in both the AMK and TOB assay had the highest viability at these drug concentrations. Some background reduction of Alamar-Blue was observed for both strains at doses higher than the observed MIC; however, there was no growth detected in those wells.

Isolation of a non-lysogen and single BPs lysogen of M. chelonae
To better understand how the presence of the second prophage increases antibiotic resistance, we generated a strain of M. chelonae that contains no prophage (M. chelonae (ΔMcProf)) and from that a single BPs lysogen of M. chelonae. To remove the McProf prophage we created a recombinant strain of M. chelonae (McProf) that overexpresses the McProf excise gene, gp5, from an inducible mycobacterial expression plasmid (Table 2) [30]. Using sets of PCR primers that amplify either the bacterial attachment site (attB) and the phage attachment site (attP) or the hybrid prophage attachment sites, attL and attR, we identified ATc-induced bacterial colonies that had an intact attB site, indicating that the McProf prophage had been lost and that McProf has an active integrase system ( The non-lysogen strain of M. chelonae (ΔMcProf) was used to isolate single lysogens of BPs. Although we were able to isolate BPs lysogens in the non-lysogen strain of M. chelonae, they are less stable than BPs lysogens formed in the WT strain (McProf) and comparable to lysogens formed in M. smegmatis [28,29]. Lysogens formed at an efficiency of~5%, and the titer of BPs in lysogen culture supernatants was 10 10 PFUs ml − 1 , several orders of magnitude higher than that of the double lysogen (10 5 PFUs ml − 1 ).

Single and double lysogens carrying McProf have higher AMK resistance than strains that lack McProf
To determine the roles of prophages BPs and McProf in the increased resistance observed in the double lysogen, we determined the MIC and viability of double (BPs, McProf) and single (BPs or McProf) M. chelonae lysogens relative to non-lysogen cells (ΔMcProf) in the presence of varying levels of AMK, TOB, TET and CLA (Table 1, Fig. 2). The presence of the naturally occurring prophage, McProf, significantly contributes to AMK resistance in M. chelonae in the presence and absence of ACI treatment (Table 1). The WT strain carrying McProf alone had a higher MIC for AMK (64 μg mL − 1 ) relative to the non-lysogen strain (ΔMcProf) (MIC of 32 μg mL -1) [10,12]. Treatment of these two strains with ACI increased the MIC for both strains; however, the MIC for ACI-treated WT (McProf) strain was two-fold higher than that of the ACI-treated non-lysogen (ΔMcProf) strain (Table 1.) The presence of a second prophage, BPs, also increases resistance to AMK, with bacterial growth at doses as high as 64 μg mL − 1 . Although the double lysogen had the same MIC as the ACI-treated WT strain (128 μg mL − 1 ) (Table 1), the cell viability of the ACI-treated WT strain was statistically higher than that of the double lysogen (Fig. 2a). BPs alone had no effect on AMK resistance suggesting that BPs only increases AMK resistance through an interaction with the naturally occurring prophage, McProf.
The presence of the prophages McProf and BPs also altered TOB resistance in M. chelonae (Table 1 and Fig. 2b). The presence of prophages had little effect on TET and CLA resistance in M. chelonae (Table 1). ACI treatment induced significant increases in CLA resistance but not TET resistance in the WT (McProf) and nonlysogen (ΔMcProf) strains.

Prophage McProf enhances AMK resistance in response to sub-inhibitory concentrations of antibiotics
Because the M. chelonae (McProf) strain treated with ACI had higher AMK resistance than the non-lysogen strain treated with ACI, we wondered if the presence of prophage McProf enhances the effect of sub-inhibitory concentrations of antibiotics on AMK resistance. To determine the interaction between ACI and the presence of one or both prophages, we treated all four lysogen and non-lysogen strains with sub-inhibitory concentrations of ACI and repeated the AMK resistance assay. The presence of McProf increases the effect of ACI on AMK resistance compared to the non-lysogen whereas the BPs prophage alone does not (Table 1  The majority of the top-ranked genes in the double lysogen belonged to the whiB7 regulon, genes in M. tuberculosis with functions related to antibiotic resistance and increased survival in macrophage (Tables 4 and 5) [15]. The transcription factor, identified as whiB7 (BB28_ RS17590), was the fifth most highly upregulated gene in the double lysogen with a fold change of 26.5 (Log2FC = 4.7, FDR = 1.3 − 73 ) ( Table 4). The WhiB7 peptide sequence shares 95% identity with the M. abscessus WhiB7 peptide (MAB_3508c) and has all the conserved residues To determine if the presence of each prophage interacts with sub-inhibitory concentrations of antibiotics, each strain was treated or not treated with 75 μM acivicin (ACI). Graphs represent average values ± SE of the mean with n = 3. The optical density of was measured at 570-and 600 nm after the addition of 2 μl of AlamarBlue and the percent difference in reduction between antibiotic-treated cells and untreated cells was calculated. Data is representative of two independent experiments that form the iron sulfur cluster binding domain. It also has the glycine-rich motif of the signature WhiB7 Cterminal "A/T Hook" DNA binding domain, which binds to AT-rich sequences adjacent to target gene promoters [11,12]. The M. chelonae genome contains a large whiB7 regulon like that of M. abscessus, with 103 of the 128 whiB7 regulon genes found in M. abscessus [12]. We identified a total of 30 upregulated genes that belong to the whiB7 regulon, many of which are known to contribute to drug resistance such as GNAT acetyltransferases, eis1 (BB28_RS05390) and eis2 (BB28_RS22650), multidrug efflux transporter tap (BB28_RS06750), and the tetV efflux pump (BB28_RS13560). Also included in this regulon are additional GNAT acetyltransferases (BB28_ RS23100 and BB28_RS01940) and ABC transporters with ATP binding domains that likely function in drug resistance (Tables 4 and 5). In M. abscessus and M. tuberculosis, erm is part of the whiB7 regulon and provides macrolide resistance but the gene is not present in the M. chelonae genome [31]. M. chelonae lacks this whiB7 regulon gene, but it does encode a newly discovered gene in the whiB7 regulon, the ribosome splitting factor hflX (MAB_3042c), which is reported to contribute to macrolide resistance in M. abscessus [31,32]. Expression of the M. chelonae hflX was slightly elevated in double lysogens relative to the WT strain (McProf) (BB28_ RS14985; Log2FC = 1.5, FDR = 8.4 − 33 ) but this did not result in significant changes in CLA resistance in the double lysogen (Table 5). An additional 25 whiB7 regulon genes were upregulated but had fold changes of less than 2.

Upregulation of whiB7 only occurs in double lysogens of M. chelonae
To determine how the presence and absence of each prophage impacts whiB7 expression, whiB7 mRNA levels were measured by qPCR in the BPs single lysogen, double  Enoyl-(acyl-carrier-protein) reductase (NADH) Rv1484 NADH-dependent enoyl-ACP reductase lysogen (BPs, McProf), and non-lysogen (ΔMcProf) and compared to that of the WT strain (McProf) (Fig. 5). Although whiB7 expression was slightly elevated in the nonlysogen (2-fold) and BPs single lysogen (4-fold) strains relative to the WT strain (McProf), the dramatic increase in whiB7 expression (~40-fold) only occurred in M. chelonae carrying both prophages (BPs, McProf). The elevated whiB7 expression occurred in the absence of known inducers of whiB7, such as ACI, which suggests BPs interacts with prophage McProf, resulting in whiB7 induction. The elevated expression of whiB7 in the double lysogen likely explains the increased resistance to AMK and TOB in the absence of ACI treatment (Table 1 and Fig. 2).
Sub-lethal concentrations of ACI but not AMK induce whiB7 expression in the double lysogen of M. chelonae We were surprised that the M. chelonae (McProf) strain had the lowest expression of whiB7 expression among the four strains given that it had higher AMK resistance, both in the presence and absence of ACI, than the two strains that lack McProf. We reasoned this may be due to whiB7-independent intrinsic resistance, such as cell wall permeability, and/or whiB7 induction in the presence of AMK, which is a more potent inducer of whiB7 than ACI [12]. Likewise, we wondered if the heightened viability observed in the single and double McProf lysogen strains in the presence of AMK and ACI was due to increased whiB7   (Fig. 6). ACI treatment resulted in increased whiB7 expression in all four strains relative to untreated strains (Fig. 6a). Expression of whiB7 was highest in the double lysogen strain (BPs, McProf) treated with ACI which correlates with the observed AMK resistance of this strain. whiB7 expression in the single and non-lysogen strain increased with ACI treatment; however, the relative levels of whiB7 expression did not correlate with AMK resistance (Fig. 2a). Although the fold-increase of whiB7 in ACItreated strains relative to control strains was highest in the WT strain (McProf) (9.5-fold) whiB7 was lower than that of the ACI-treated BPs single lysogen, which demonstrated lower AMK resistance.
To determine if exposure to AMK also contributes to whiB7 expression in each of the four strains, whiB7 expression was determined in each of the strains in the presence and absence of sub-lethal concentrations of AMK (16.7 μM). Strains that lack the McProf prophage had the greatest increase in whiB7 expression in response to AMK treatment. The non-lysogen and BPs single lysogen had 28-and 7-fold increases in whiB7 expression in response to AMK treatment, respectively (Fig. 6b) The right attachment site, attR, overlaps a leftward oriented tRNA-Lys (BB28_RS07905). Located adjacent to the left attachment site, attL, is a rightward transcribed tyrosine integrase (gp1), one gene of unknown function (gp2) and a leftward transcribed gene, gp3, that is likely to be the immunity repressor, as it shares high amino acid sequence similarity with the immunity repressors of singleton mycobacteriophage DS6A (66%) and cluster K2 mycobacteriophages (70%) DismalFunk, DismalStressor, Findley, Marcoliusprime and Milly [35]. Gp4 and gp5 both have helixturn-helix DNA binding motifs and encode Cro (control of repressor's operator) and excise, respectively.
Located between attR and the structural genes (gp51-82) are genes that are typically expressed during lysogeny [36]. We were unable to predict a function for the majority of these genes; however, we were able to identify an ADP-riboysl glycosylhydrolase (gp86), a helixturn-helix DNA binding protein (gp89), a membrane protein (gp90), and an AAA-ATPase (gp91). Most intriguing is the leftward transcribed gene cassette immediately adjacent to attR, which encodes proteins that may be secreted by the mycobacterial Type 7 secretion system (T7SS) (Esx-3 or Esx4) (Fig. 7b and c). Gp98 encodes a 105-amino acid gene product that forms four HHpred predicted helical domains with high probability matches to WXG-100 family motifs of T7SS proteins. The gp98 sequence contains a SAG motif, which strays slightly from the conserved WXG motif that is characteristic of T7SS secreted substrates [37]. Gp97 encodes a 732-residue polymorphic toxin that has a WXG-100  motif in the N-terminus and a possible T7SS secretion signal (YxxxD/E) in the C-terminus [38]. The Cterminus also includes a toxin_43 motif (PF15604.6) and high sequence similarity to the C-terminus of Type 6 secretion system (T6SS) polymorphic toxin, TdeI (Atu4350), found in Agrobacterium tumefaciens [39]. This family of proteins has DNAse activity and shares a conserved HXXD catalytic domain located in the Cterminus (Fig. 7b) [39]. Tde toxins are typically paired with a Tdi immunity protein and a likely immunity protein, gp96, was identified downstream of McProf gp97.
The immunity repressors from both the BPs (gp33) and McProf (gp3) genomes are highly expressed during lysogeny of M. chelonae (Fig. 8a). The BPs genome also expresses, gp58, a gene of unknown function that is part of a mycobacteriophage mobile element (MPME1) (Fig.  8a) [28]. There are an additional 15 genes expressed at varying levels from the McProf genome (Fig. 8b). The integrase (gp1) is expressed at low levels and is adjacent to a moderately expressed genes of no known function (gp2). There are three reverse oriented genes, gp48-50, located between the HNH endonuclease (gp47) and the small subunit terminase (gp51) and a small reverse oriented gene (gp56) adjacent to the scaffolding protein with moderately and low expression, respectively. We were not able to determine functions for these genes; however, gp48 and gp49 do have predicted membrane domains. The remaining genes expressed from the McProf prophage genome are located between the structural genes and attR and many do not have predicted gene functions, including the most highly expressed McProf gene, gp84. There is also strong expression from the gene cassette containing the putative WXG-100 family polymorphic toxin and immunity protein (gp96-98). None of the expressed McProf genes were significantly differentially expressed in the presence of the BPs prophage (FC > 1.99 and FDR < 0.05).

Discussion
The incidence of non-tuberculosis mycobacterial disease has increased over the last 20 years [40]. M. abscessus, along with M. avium, is the major cause of bronchopulmonary infections in cystic fibrosis patients, and is of increasing concern due its high levels of intrinsic antibiotic resistance [9]. CLA and AMK are the two core drugs used to treat M. abscessus infections; however, development of resistance to these drugs is common during treatment [9]. Mycobacterial resistance to CLA and AMK are often the result of mutations in 23 s rRNA or 16 s rRNA, respectively, but mutations alone do not completely account for AMK and CLA resistant phenotypes in M. abscessus clinical isolates [7,13]. Induction of the whiB7 regulon in mycobacteria is the second major contributor to resistance to AMK and CLA [7]. Heightened expression of whiB7 is consistently observed in extensively resistant M. abscessus isolates relative to drug susceptible isolates [7,13]. Understanding the mechanisms that drive increased expression of whiB7 will be important for improving treatment of resistant mycobacterial infections. The WhiB7 response is highly conserved across all mycobacteria and our studies shows that the WhiB7 regulon in M. chelonae overlaps considerably with that of M. abscessus [12,15]. Here we describe for the first time that prophages in mycobacteria alter antibiotic resistance and expression of intrinsic antibiotic resistance genes of the whiB7 regulon. This study provides valuable insight into the role prophages play in mycobacterial antibiotic resistance and novel mechanisms of whiB7 induction.
In this report we show for the first time that prophages in mycobacteria can contribute to increased resistance to aminoglycosides AMK and TOB (Table 1; Fig. 1; and Fig. 2). This is similar to the resistance observed in E. coli carrying multiple prophages [1]. Of the two prophages investigated in this study, the naturally occurring prophage, McProf, appears to play the more important role in inducing intrinsic resistance in M. chelonae (Table 1; Fig. 1; and Fig. 2). Strains carrying the McProf prophage demonstrated increased AMK resistance relative to the non-lysogen and BPs single lysogen strains in the absence of ACI. The increased AMK resistance observed in McProf carrying strains was further increased by either ACI treatment or the presence of a second prophage, BPs. Enhanced resistance in McProfcarrying strains in response to ACI treatment was also observed in the TOB MIC assays. We know that subinhibitory concentrations of ribosome targeting antibiotics, such as ACI, induce whiB7 expression and intrinsic drug resistance and we observed this in the nonlysogen (ΔMcProf) strain (Fig. 2) [14,15]. The presence of McProf appears to enhance the effect of ACI on resistance in the AMK and TOB assays (Figs. 2 and 3).
The dramatically higher whiB7 expression in the strain carrying both prophages (BPs, McProf) likely contributes to the heightened AMK and TOB resistance observed in the double lysogen (Figs. 2 and 4; Tables 4 and 5). We observed high whiB7 expression in the double lysogen in the absence of antibiotics or other conditions known to induce whiB7, revealing prophage as a novel mechanism of inducing whiB7 expression and intrinsic antibiotic resistance. Also upregulated in the double lysogen are whiB7 regulon genes that can explain the increased resistance to aminoglycosides AMK and TOB [41]. The GNAT acetyltransferase, eis2, and tap, a multidrug efflux pump, were each upregulated~10-fold in the double lysogen and confer resistance to aminoglycosides [12,42,43]. 2′-N-acetyltransferase AAC (2′) also contributes to resistance to TOB and AMK, however this gene (BB28_ RS22055) was not significantly upregulated in our data set [44]. It is possible that other N-acetyltransferases (BB28_RS23100, BB28_RS01940, BB28_RS14560), aminoglycoside phosphotransferases (BB28_RS12685), or potential efflux pumps (BB28_11540) upregulated in our dataset contributed to AMK and TOB resistance of the double lysogen (Table 5). Tap, along with TetV (BB28_ RS13560) target tetracycline efflux, however, we did not consistently see an increase in TET resistance across all our MIC trials [45,46]. Although M. chelonae lacks an erm gene, there was a slight elevation in the whiB7 regulon gene, hflx, but this did not result in a significant change in CLA resistance [32].
Other mechanisms of intrinsic resistance also likely contribute to the AMK and TOB resistance observed in the McProf-carrying strains. The McProf-carrying strains demonstrated the highest AMK resistance in the presence and absence of ACI treatment. Although whiB7 expression was highest in the double lysogen (BPs, McProf) treated with ACI among the four strains, whiB7 expression levels in the single McProf strain did not correlate with its relative AMK resistance among the four strains. AMK is also a potent inducer of whiB7 expression in mycobacteria; however, AMK had little to no effect on whiB7 expression in strains carrying McProf [12]. The strong induction of whiB7 with AMK treatment in strains that lack McProf suggests that the presence of the McProf prophage may affect cell wall permeability. A decrease in cell wall permeability in McProf-carrying strains would likely contribute to the observed AMK resistance in these strains. Investigating differences in cell wall permeability in the presence and absence of McProf will be important for understanding the effect of prophages on drug resistance.
We do not yet know how the two prophages, BPs and McProf, interact to alter whiB7 expression in the double lysogen strain. The McProf genome appears to only express genes through lysogenic infection of M. chelonae, whereas BPs can carry out lysogenic and lytic infection (via induction) in a population of double lysogen cells. It's therefore possible that either lytic or lysogenic gene expression from BPs interacts with any of the 16 genes products expressed from the McProf genome through an unknown mechanism to alter whiB7 expression. Each of the 16 expressed McProf genes will be investigated for a potential role in altered whiB7 expression; however, the genes in the WXG-100 family polymorphic toxin cassette are strong candidates. Activated toxin systems could potentially act as a trigger to the WhiB7 stress response. Sub-inhibitory concentrations of antibiotics and BPs phage infection both enhanced antibiotic resistance in M. chelonae in the presence of prophage McProf (Figs. 1 and 4), conditions also known to activate toxin/ antitoxin systems [47][48][49][50][51]. Toxin/antitoxin systems are also known to function as stress response modules and regulators of adaptive responses to stresses associated with host environment and drug treatment [49]. Further, toxin/antitoxin systems are abundant in pathogenic mycobacteria and are more highly expressed in the most virulent strains of M. tuberculosis [49]. In comparison there are relatively few toxin/antitoxin systems in nonpathogenic mycobacteria [47]. Toxin/antitoxin systems also stabilize replicative elements (e.g. plasmids and prophage) and defend against phage lytic infection [52,53]. The increased stability of the BPs prophage in the presence of McProf compared to BPs lysogens in the cured M. chelonae (ΔMcProf) and its reported instability in M. smegmatis strains [28,29] suggests that such system encoded by McProf is active.

Conclusions
We have established that dramatic increases in whiB7 expression and AMK resistance only occurs in M. chelonae strains carrying a type of prophage that is also found naturally in M. abscessus strains. The observed AMK resistance in the presence of prophage McProf is further enhanced by exposure to sub-inhibitory concentrations or by the presence of a second prophage, BPs. Pathogenic mycobacteria typically carry one or more prophages that are capable of induction and in infected tissues are likely exposed to lytic phage infection and sub-inhibitory concentrations of antibiotics during treatment. Our novel research findings indicate that prophage could be drivers of important intrinsic antibiotic resistance genes in response to such stresses. To determine the mechanism by which phage alter intrinsic antibiotic resistance in mycobacteria, we are exploring the function and impact of specific phage genes on expression of whiB7 in the presence of various environmental stressors.  Table 2). Cloning was carried out in chemically competent Escherichia coli DH5α (New England Biolabs (NEB), Ipswich, MA). Kanamycin was used for selection of the expression vector pST-KT at 250 μg ml − 1 . Strains used in this study are listed in Table 2.

Bacterial and viral strains
Bacteriophage BPs was obtained from the Hatfull Laboratory [28]. Phage lysates were propagated through plaque assays in either M. smegmatis MC 2 155 or M. chelonae (McProf), or a cured strain of M. chelonae that we refer to as the non-lysogen M. chelonae strain (ΔMcProf) ( Table 2) [54]. Briefly, 0.5-ml aliquots of latelog phase bacteria were incubated with serially-diluted phage samples for 15 min before plating in 4.5 ml of 7H9 top agar containing 0.45% agar onto 7H10 agar plates. Phage stocks were created by flooding plates with nearly confluent bacterial lysis with phage buffer (10 mM Tris/HCl pH 7.5, 10 mM MgSO 4 , 1 mM CaCl 2 , 68.5 mM NaCl).  [30]. Recombinant plasmids were sequenced to verify the presence of the xis sequence prior to electroporating into competent WT M. chelonae (McProf) [55]. Cultures of recombinant M. chelonae carrying pST-KT_xis were grown in 10-mL volumes for 48 h at 30°C with shaking. Optical density was measured at a wavelength of 600 nm, and samples sub-cultured to an optical density of 0.05. Cultures were then grown to an optical density of 0.6 and treated with 500 μg mL − 1 of anhydrotetracycline (ATc) prepared in dimethylsulfoxide (DMSO) or an equivalent volume of DMSO. Cultures were incubated at 30°C with shaking for an additional 72 h. A 0.5-mL sample of each culture was harvested and serially diluted in 7H9-OAD. Dilutions were plated onto 7H10-OAD supplemented with 250 μg mL − 1 of kanamycin in 100-μL volumes and incubated at 30°C for 5 d. Resulting colonies were PCR screened for the loss of prophage McProf using a set of four primers that amplify either the bacterial attB site, indicating loss of the prophage, or the attachment junctions attL and attR, indicating the presence of the prophage ( Table 3). The attB PCR product was sequenced to confirm the clean excision of the prophage.

Isolation of lysogenic strains
Lysogens were isolated by plating serially diluted M. chelonae strains in 4.5 ml of 7H9 top agar onto 7H10 agar treated or not treated with 10 9 PFUs of BPs. After 6 d of incubation at 30°C, colonies were picked and screened for properties indicative of lysogens, including release of phage particles into culture supernatant, superinfection immunity to BPs infection and PCR detection of prophage attachment sites, attL and attR ( Table 3). The efficiency of lysogeny was determined by dividing the number of colonies present on virus-treated plates by the number of colonies present on un-treated plates and multiplying by 100. Genomic DNA from BPs lysogens of the WT M. chelonae strain (referred to as M. chelonae double lysogen (BPs, McProf)) ( Table 2) were sequenced to confirm the presence of the BPs genome in the M. chelonae genome. Whole genome libraries were generated by Genome Technologies at Jackson Laboratory (Bar Harbor, ME) and sequenced on one 2X150-bp MiSeq sequence run. Sequence reads were assembled by aligning reads to a reference genome and reads that did not map to the host genome were assembled de novo.

RNA isolations
Total RNA was isolated from six replicates of 4-ml samples of M. chelonae grown to an OD 600 of 1.0. Cultures were treated with RNAProtect Bacteria Reagent (Qiagen, Germantown, MD) before centrifuging at 5000 x g for 10 min. Cell pellets were resuspended in 100 μl of TE containing 100 μg ml − 1 lysozyme and incubated at room temperature for 40 min. After adding 700 μl of RLT buffer (Qiagen), cells were transferred to 2-ml Lysing Matrix B tubes (MP Biomedicals, Irvine, CA) and homogenized in ice-cold adaptors in the TissueLyser LT (Qiagen) set for 8 min at 50 Hz. RNA extractions were carried out on the lysates using the RNeasy Mini Kit (Qiagen) with DNAse treatment (Qiagen) on the column according to the manufacturer's recommendations. After elution of RNA in 50 μl of water, samples were treated with a second application of DNAse using the Turbo DNA-free Kit (Thermo Scientific, Waltham, MA) according to the manufacturer's recommendations. The quantity of RNA was determined with the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Montchanin, DE, USA). The quality of RNA was determined by gel electrophoresis using the FlashGel RNA system (Lonza, Rockland, ME).
In the ACI and AMK induction experiments, RNA was isolated as described above with the following modifications. Cultures were grown to an OD 600 of 0.7 or 0.9 and treated with 75 μM ACI for 3 h or with 16.7 μM AMK for 1 h, respectively, before harvesting cells for RNA. Untreated control cultures were incubated for equivalent amount of time prior to harvesting cells.

RNAseq
RNA used in RNAseq experiments was sent to the Delaware DNA Sequencing and Genotyping Center (Newark, DE) for quality control analysis, library preparation and paired end sequencing on the Illumina HiSeq 2500. Read length was set to 51 bases with the samples run on two separate lanes. Raw sequencing data files were uploaded to the public Galaxy server at usegalaxy.org [56]. Read files from the two lanes were concatenated and read quality was determined using FastQC [57]. The reads were processed using the Trim Galore! with the FastQC output as a guide [58]. Retained reads had a quality score minimum of 30, and with the first 9 bases on the 5′ end and the last base on the 3′ end removed. Though rRNA was depleted prior to sequencing, we discovered that depletion of rRNA was incomplete. The rRNA reads were computationally removed by alignment to the M. chelonae rRNA operon using BowTie2 and saving the non-aligned reads [59]. Average number of reads per sample was 2,280,954 reads for the M. chelonae  orientation to the M. chelonae (CCUG 47445) transcriptome using a coding transcript fasta (GenBank). The alignment was adjusted for the high GC content of the mycobacterial genome. Mate pair 1 was specified as coming from the reverse strand (SR). Strand specificity was necessary because reads from two convergent genes often overlapped. Output from Salmon quantification was used for pairwise comparisons of expression, using the R statistical package DESeq2 [61] and the NumReads values as described by the authors. Genes with low expression levels (reads < 10) were removed. Genes were considered significantly regulated if Log2 fold change (Log2FC) was greater than 1.0 and the False Discovery Rate (FDR) was less than 0.05. Although the M. chelonae genome is sequenced and has an annotation, the gene functions are poorly characterized. We therefore generated a table of M. chelonae genes with orthologs in M. tuberculosis, M. abscessus and M. smegmatis using the OrthoDB pipeline, a series of scripts from OrthoDB [62]. This gave us the best alignment between the three genomes and together with blastP on MycoBrowser and HHpred, helped identify numerous significantly upregulated M. chelonae genes with potential virulence functions [62][63][64]. The RNAseq data set was validated in two independent RNA isolation experiments using qPCR assays that quantified expression of upregulated (whiB7 and tap) and downregulated genes (glycerol kinase (glpK)) from the RNAseq data set (data not shown).
RTqPCR cDNA was synthesized from 500 ng of total RNA in 20μl reactions containing qScript cDNA Supermix (Quantabio, Beverly, MA) according to the manufacturer's recommendations. Reactions were incubated for 5 min at 25°C, 20 min at 42°C and heat inactivated at 85°C for 5 min. cDNA was diluted 1:6 in 10 mM Tris and stored at − 20°C.
Real-time PCR assays were performed using the Bio-Rad CFX96 Real-Time system (Bio-Rad Laboratories, Hercules, CA). Using Primer3 software, primer sets were designed to amplify a 100-bp sequence in the gene of interest (Table 6). Quantitative PCR (qPCR) was carried out in triplicate 25-μL reactions containing 200 nM gene-specific primers (Table 6), 1 μl diluted cDNA (1:5) and PerfeCTa SYBR Green Supermix (Quantabio), according to manufacturer's instructions. Reactions were incubated at 95°C for 3 min, followed by 40 cycles of 95°C for 10 s and 60°C for 30 s. A melt curve analysis was performed to confirm that only one amplicon was created by each primer set. The change in abundance of gene-specific RNA was normalized to M. chelonae 16 s rRNA and calculated using the 2 -ΔΔCT method [65].
Positive and no-template controls were included in realtime PCR analysis.

Minimum inhibitory concentration determination
Minimum inhibitory concentration (MIC) assays were performed according to  and Ramon-Garcia et al. (2013) [14,66]. Briefly, cultures were grown for 2 d in 7H9 supplemented with OAD and then sub-cultured such that overnight incubation at 30°C with shaking allowed cultures to reach an OD 600 of 0.1-0.3. Cultures were diluted to a density of 10 5 cells ml − 1 and applied in 50-μl volumes to wells of a 96-well plate containing 50 μl of 7H9 media with antibiotic concentrations that varied by 2-fold dilutions across the plate. Because clarithromycin was prepared in DMSO, an equivalent amount of DMSO was included in all wells. Each strain was tested at each antibiotic concentration in replicates of six and no-antibiotic controls were performed in replicates of 16. Inoculated plates were sealed with porous adhesive culture plate films (VWR International, Radnor, PA), wrapped with parafilm and incubated at 30°C for two d before adding 1 μl (assays presented in Figs. 1 and 2) or 2 μl (assays presented in Fig. 3) of Ala-marBlue (BioRad, Hercules, CA) and 25 μl of 25% Tween80 to each well. After incubation at 30°C for 1 d, the MIC was determined as the lowest drug concentration that completely inhibited growth. Viability was also determined by measuring the optical density at 570-and 600 nm and the percent viability of cells was calculated as the percent difference in reduction between antibiotic-treated cells and untreated cells according to the manufacturer's instructions. Each assay was replicated in at least three independent experiments.

McProf genome analysis
The McProf genome was detected in the M. chelonae genome using Phaster [67]. The genome ends were defined as attL and attR and the sequence was annotated using DNA Master (http://cobamide2.bio.pitt.edu) and PECAAN (https://pecaan.kbrinsgd.org/index.html). Genes were identified and gene start coordinates determined first by auto-annotation using Glimmer and GeneMark, then by manual inspection of each predicted gene [68,69]. Gene functions were predicted using HHPRED and BLAST [70,71]. Genome map representations were created in Phamerator using database McProf_DB [34].