A Cassette Containing Thiostrepton, Gentamicin Resistance Genes, and dif sequences Is Effective in Construction of Recombinant Mycobacteria

The genetic manipulation of Mycobacterium tuberculosis genome is limited by the availability of selection markers. Spontaneous resistance mutation rate of M. tuberculosis to the widely used kanamycin is relatively high which often leads to some false positive transformants. Due to the few available markers, we have created a cassette containing thiostrepton resistance gene (tsr) for selection in M. tuberculosis and M. bovis BCG, and gentamicin resistance gene (aacC1) for Escherichia coli and M. smegmatis mc2155, flanked with dif sequences recognized by the Xer system of mycobacteria. This cassette adds to the limited available selection markers for mycobacteria.


INTRODUCTION
Gene manipulation in mycobacteria is performed using a limited number of selection markers. Mycobacteria are naturally resistant to many antibiotics and requires use of stable drugs with low frequency of spontaneous resistance for selection, hence limiting the alternative choices (Parish and Brown, 2008). The combined use of multiple markers enables more versatile genetic modifications, including the stable maintenance of multiple plasmids and inactivation of multiple genes (Wada et al., 2016).
Aminoglycoside phosphotransferase (aph) genes, conferring resistance to kanamycin (KAN), were the first to be used as selection markers in mycobacteria (Snapper et al., 1988) owing to their stability over the extended periods of incubation for slow-growing mycobacteria. However, their utility is limited by emergence of spontaneous resistance, albeit at low frequencies (Hatfull, 1996). Unlike fast-growing bacteria, slow-growing mycobacteria have a single rRNA operon (Suzuki et al., 1987) which is more prone to mutations conferring resistance to agents such as KAN (Bottger, 1994). Besides, selection using KAN in Mycobacterium w and Mycobacterium vaccae has not been achieved. Radford and Hodgson (1991) first reported the use of hygromycin (HYG) resistance gene (hyg) as a selection marker in M. smegmatis and M. bovis BCG in 1991. Since then, it has been used in other mycobacteria. The use of hyg provides a marker gene which does not provide cross resistance to clinically useful drugs (Garbe et al., 1994). It offers an improved transformation frequency over KAN and is probably more efficiently expressed in mycobacteria than the Escherichia coli-derived aminoglycoside phosphotransferase genes conferring KAN resistance (Garbe et al., 1994). The use of apramycin as a selection marker in both slow-and fast-growing mycobacteria was first reported by Paget and Davies in 1996, following its disapproval for clinical use in humans. However, its utility is limited by the acetylation of other closely related aminoglycoside such as KAN (Davies and O'Connor, 1978;Consaul and Pavelka, 2004), and low transformation efficiencies. β-lactam-based selection markers such as the ampicillin resistance gene amp r are not useful in mycobacteria since they contain endogenous β-lactamases that confers natural resistance to penicillins (Hatfull, 1996).
Other past explorations have included resistance to chloramphenicol (Das Gupta et al., 1993), but have been limited due to poor stability and high rates of spontaneous mutations, hence unsuitable for slow-growing mycobacteria. Streptomycin, sulfonamide (Gormley and Davies, 1991) and mercury salts (Baulard et al., 1995) have also been explored as possible selectable markers, but to date, KAN, HYG, and GEN resistance genes remain the often widely exploited selectable markers in mycobacterial genetics.
Owing to the limited number of markers and their disadvantages, we hence sought to explore use of methylaccepting chemotaxis protein I, a serine sensor receptor tsr gene conferring resistance to thiostrepton (TSR), as a selectable marker in M. tuberculosis and M. bovis BCG. TSR, a thiazole antibiotic, was first isolated and characterized from Streptomyces azureus (Cundliffe, 1971) in 1954 at the Squibb Institute (Pagano et al., 1956) and is used in veterinary medicine to treat mastitis, and as a topical agent for dogs. However, it has only found limited applications due to its poor solubility and toxicity (Kuiper and Conn, 2014).
Thiostrepton inhibits protein translation by firmly binding to the complex formed by 23S rRNA and ribosomal protein L11 in bacterial ribosomes (Cannon and Burns, 1971;Cundliffe, 1971). The tsr gene encodes an RNA methyltransferase that prevents TSR from binding to ribosomes by 23S rRNA methylation (Thompson et al., 1982). The tsr confer total resistance to TSR and thus has been the selection marker of choice in many of the Streptomyces spp. cloning vectors (Thompson et al., 1980).
In a recent study on ovarian cancer cell lines, Westhoff et al. (2014) demonstrated that when TSR is used in combination with the standard paclitaxel/cisplatin chemotherapy, it decreases Forkhead box M1 (FOXM1) gene expression besides showing an enhanced synergistic cytotoxicity in ascites cells from platinum-resistant patients. In addition, Wada et al. (2016) also demonstrated tsr as a viable selection marker for the thermophilic Geobacillus kaustophilus besides demonstrating accurate selection as a single copy in Streptomyces strains.
However, only scanty data showed that TSR is active against M. tuberculosis (Vermeulen and Wu, 2004;Lougheed et al., 2009) in drug testing.
Thiostrepton and GEN were purchased from Sigma-Aldrich (China) and dissolved in dimethyl sulfoxide (DMSO) and double distilled water, respectively. GEN 20 and 5 µg/mL was used for selection of E. coli and M. smegmatis mc 2 155, respectively, and TSR 5 and 10 µg/mL of both M. tuberculosis H37Rv and M. bovis BCG Tice. LB broth was augmented with 170 µg/mL chloramphenicol Sigma-Aldrich (China).

Drug Susceptibility Testing
We first tested the potential of TSR as a selection antibiotic for M. tuberculosis up to a final concentration of 10 µg/mL in liquid culture of autoluminescent M. tuberculosis H37Ra (AUlRa) (Table 1) as previously described (Zhang et al., 2012). Briefly, 2 mL of AUlRa was inoculated in 50 mL 7H9 plus OADC and tween80 with shaking at 37 • C to mid log phase (OD 600 = 0.6-0.8) in a flask and then diluted to appropriate concentrations. Drugs (5 µL/drug) were added into the 1.5 mL vial, mixed with 195 µL AUlRa and incubated at 37 • C. Controls using 195 µL AUlRa and DMSO (5 µL) or 195 µL AUlRa and water (5 µL) tubes were included. Relative light measurements (RLUs) were monitored starting day 0, day 1, day 3, and day 5 using GloMax 20/20 Luminometer (Promega).
Minimum inhibition concentration (MIC) was defined as the lowest concentration of a drug inhibiting 99% of bacterial growth (Zhang et al., 2010). The MIC values for wild-type and recombinant mycobacteria were detected on Middle Brook 7H11 agar plates containing different concentrations of TSR (0-160 µg/mL) and GEN (0-100 µg/mL).

General DNA Techniques
Polymerase chain reaction (PCR) amplification reactions were performed with pfu DNA polymerase (Takara). The PCR products and plasmids were analyzed by electrophoresis in agarose gels and purified using a DNA gel extraction kit (Magen, China). Plasmids were also extracted and purified using kits from the same company. Purified PCR products and plasmids were sequenced (BGI, Shenzhen, China). The aacC1 gene (0.543 kb)  was amplified from plasmid pPR27 ( Table 1) using primers Gm-f and Gm-r (Table 2) while the 0.8 kb tsr gene was amplified from plasmid pIJ6902 ( Table 1) using primers Tsr-f and Tsr-r ( Table 2).

Construction of Shuttle Vector Containing tsr + aacC1 Resistance Genes
To construct a vector bearing tsr+aacC1, we arranged the genes into a cassette under the control of the M. tuberculosis hsp60 promoter (Figure 1) in plasmid p60LuxN  intending the aacC1 gene to be used for selection in E. coli and M. smegmatis mc 2 155 and the tsr gene to be used in M. tuberculosis and M. bovis BCG. The aacC1 was cloned adjacent to the hsp60 promoter into the NdeI-PstI sites of p60LuxN resulting in plasmid p60Gm. The tsr gene was cloned into the PstI-HindIII sites of plasmid p60Gm to get E. coli-mycobacteria shuttle plasmid p60GTE bearing hsp60-aacC1-tsr cassette.

Construction of dif-hsp60-aacC1-tsr-dif Cassette
The hsp60-aacC1-tsr cassette was excised with XbaI from plasmid p60GTE and cloned into the XbaI sites of E. coli pUCDHmke derived from pTYdHm (Yang et al., 2014) plasmid (Table 1) bearing a dif -HYG-dif cassette replacing hyg gene and creating plasmid pUCDGT. The dif-hsp60-aacC1-tsr-dif cassette (Figure 1) was excised by HindIII from pUCDGT and cloned into the integrative plasmid pMH94 (Table 1) replacing the KAN resistance gene and creating plasmid p60GTI.

Transformation
Plasmids p60GTE and p60GTI were used. M. smegmatis was transformed as previously described (Snapper et al., 1990), while M. tuberculosis and M. bovis BCG were transformed as previously described (Wards and Collins, 1996;Yang et al., 2015) with some modifications. The competent M. tuberculosis and M. bovis BCG cells were first incubated at 37 • C for 10 min before electroporation and transformation was performed at room temperature. Transformants were selected on plates containing FIGURE 1 | Scheme of vector construction. The Escherichia coli-mycobacterial plasmid p60GTE was derived by inserting aacC1 and tsr fragments next to the mycobacterial hsp60 promoter. The hsp60-aacC1-tsr cassette was excised and inserted on to the XbaI sites of pUCDHmke bearing the dif sequences. The dif-hsp60-aacC1-tsr-dif cassette was excised and inserted on the Hind III sites of plasmid pMH94. attP, mycobacteriophage L5 attachment site; int, integrase gene; oriE, origin region of E. coli; oriM, thermosensitive origin region of mycobacteria; KanR, KAN resistance gene, dif: the putative MTB dif sequence. Useful enzyme sites: NdeI; HindIII; PstI and XbaI.

M. tuberculosis and M. bovis BCG Transformants
Unmarked recombinant transformants were analyzed according to Yang et al. (2014). Briefly, PCR verified TSR-resistant single  p60GTI colonies were individually cultured in 7H9 media to late log phase (OD 600 = 0.8-1.0) without selection to allow excision of the dif-hsp60-aacC1-tsr-dif cassette by the endogenous mycobacteria XerC and XerD. Ten-fold serial dilutions of bacterial culture were spread on plain agar plates. The colonies were picked and replica streaked on both plain and 10 µg/mL TSR-containing 7H11 plates. The TSR-sensitive colonies were verified further by PCR amplification of the 1.9 kb cassette using primers Tsr-f1 and Tsr-r1 (Table 2) and the shorter PCR products (∼0.5 kb) bearing one single dif sequence were confirmed by sequencing.

TSR as a Potential Selection Antibiotic against Mycobacteria
We first tested the potential use of TSR as a selective antibiotic against mycobacteria. Using liquid culture autoluminescent M. tuberculosis H37Ra, we tested different TSR concentrations up to 10 µg/mL and the relative light units (RLUs) declined sharply within 2 days and continuously till the end of the assay, while those of blank control rose steadily (MIC lux = 0.05 µM, ∼ = 0.08 µg/mL

Construction of Plasmids p60GTE and p60GTI, Their Transformation Frequencies and MICs in Respective Recombinant Strains
We set out to construct two plasmids expressing tsr and aacC1 genes in both E. coli and mycobacteria. We constructed episomal and integrative E. coli-mycobacterial shuttle plasmids bearing the mycobacterial hsp60 promoter, aacC1 and the tsr gene flanked by dif sequences (Figure 1). Both antibiotic resistance markers, the streptomyces TSR resistance gene, tsr, and the Pseudomonas aeruginosa GEN resistance gene, aacC1, worked in mycobacterial transformants. TSR resistance is not a selectable marker in E. coli due to outer membrane exclusion of TSR by gram-negative bacteria (Gale et al., 1981). To circumvent this, we used GEN for selection in E. coli and supplemented the media with chloramphenicol 170 µg/mL to increase the plasmid copy number. The transformation frequency for H37Rv and M. bovis BCG overexpressed with the episomal plasmid p60GTE were 1.26 × 10 4 and 4.3 × 10 3 CFUs and 3.5 × 10 3 and 2 × 10 2 CFUs, respectively, with the integrative plasmid p60GTI on TSR 5 µg/mL (Table 4). Both H37Rv and M. bovis BCG recombinant strains increased the MICs by >300-fold (Table 3) while M. smegmatis mc 2 155 strains increased the MICs by 40-fold ( Table 5).
The loss of the tsr marker gene verified by PCR, yielded ∼0.5 kb products confirmed by sequencing to bear one dif sequence as expected, from 12 and 20 randomly selected recombinant p60GTI containing M. tuberculosis H37Rv and M. bovis BCG colonies. We found that five of each recombinant strain had lost the tsr gene which should be excised by the endogenous mycobacterial recombinase XerCD system expressed by XerC and XerD genes recognizing the dif cassette (Cascioferro et al., 2010;Yang et al., 2014), resulting in selectable marker-free colonies.
Our TSR MICs results concurs with the antimicrobial bactericidal activity reported by others (Vermeulen and Wu, 2004;Lougheed et al., 2009), and to the best of our knowledge this is the first report showing the use of TSR resistance as putative selective marker for gene transfer in mycobacteria.

CONCLUSION
We have successfully constructed a cassette containing tsr and aacC1 genes flanked by dif sequences for selection in mycobacteria and demonstrated the potential of this cassette for use as a mycobacteria selection marker in M. tuberculosis and M. bovis BCG. The novelty of this work is the introduction and expression of genes in a new cassette and verified by raising of resistance in the corresponding host cells. The new reliable selection marker comes in handy for M. tuberculosis genetic manipulation studies and is a new tool for efficient construction of selection-marker free recombinant strains.