A novel method to generate unmarked gene deletions in the intracellular pathogen Rhodococcus equi using 5-fluorocytosine conditional lethality.

A novel method to efficiently generate unmarked in-frame gene deletions in Rhodococcus equi was developed, exploiting the cytotoxic effect of 5-fluorocytosine (5-FC) by the action of cytosine deaminase (CD) and uracil phosphoribosyltransferase (UPRT) enzymes. The opportunistic, intracellular pathogen R. equi is resistant to high concentrations of 5-FC. Introduction of Escherichia coli genes encoding CD and UPRT conferred conditional lethality to R. equi cells incubated with 5-FC. To exemplify the use of the codA::upp cassette as counter-selectable marker, an unmarked in-frame gene deletion mutant of R. equi was constructed. The supA and supB genes, part of a putative cholesterol catabolic gene cluster, were efficiently deleted from the R. equi wild-type genome. Phenotypic analysis of the generated DeltasupAB mutant confirmed that supAB are essential for growth of R. equi on cholesterol. Macrophage survival assays revealed that the DeltasupAB mutant is able to survive and proliferate in macrophages comparable to wild type. Thus, cholesterol metabolism does not appear to be essential for macrophage survival of R. equi. The CD-UPRT based 5-FC counter-selection may become a useful asset in the generation of unmarked in-frame gene deletions in other actinobacteria as well, as actinobacteria generally appear to be 5-FC resistant and 5-FU sensitive.

INTRODUCTION

An important molecular tool in functional genomics studies is the targeted inactivation of any gene of interest. Ideally, unmarked gene deletions are constructed using a positive selection step for the rare second recombination event. The sacB gene of Bacillus subtilis is one of the most widely used suicide genes, conferring sucrose sensitivity mostly in Gram-negative bacteria (1,2). Previously, we reported the use of sacB counter-selection to efficiently generate unmarked gene deletions in Rhodococcus erythropolis (3). This method subsequently has been applied in other Rhodococcus species (4–6). Sucrose sensitivity by sacB has also been reported for other mycolic acid containing actinobacteria, i.e. Corynebacterium glutamicum and Mycobacterium sp. (7). Despite considerable efforts, we have been unable to apply sacB as a counter-selectable marker in Rhodococcus equi, due to a lack of sucrose sensitivity. Similar observations have been reported for Streptomyces lividans (7). The sacB counter-selection system generally is not applicable to other Gram-positive bacteria, like B. subtilis. In B. subtilis, mazF, encoding a toxin, has been used as a suicide marker to generate unmarked gene deletion mutants (8). This method relies on the availability of a tightly regulated, inducible promoter, limiting applicability in bacteria with less developed molecular toolboxes. The use of alternative suicide genes have been reported as well, e.g. glkA, pyrF, upp and rpsL (9–12). A major drawback of these markers, however, is that they only function in glkA, pyrF, upp null mutants, respectively, necessitating the construction, or availability, of a null mutant for every strain to be mutated. A PCR-targeted gene replacement strategy has been reported for streptomycetes that relies on the availability of ordered genomic libraries to efficiently generate mutants (13). This method, however, requires a laborious extra round of mutagenesis to remove the resistance marker integrated into the genome to generate unmarked mutants. Recently, a simplified method for marker removal in actinomycetes using Flp recombinase has been described (14).

Cytosine deaminase (CD, EC 3.5.4.5) and uracil phosphoribosyltransferase (UPRT, EC 2.4.2.9) are enzymes involved in the pyrimidine salvage pathway, converting cytosine via uracil into dUMP. CD activity has been found in certain prokaryotes and lower eukaryotes. The genes encoding these activities in Escherichia coli, codA and upp, respectively, have been cloned and characterized (15,16). Interestingly, microorganisms expressing CD convert the innocuous cytosine analog 5-fluorocytosine (5-FC) into 5-fluorouracil (5-FU), a highly toxic compound lethal to living cells. The cytotoxicity is largely exerted following UPRT mediated conversion of 5-FU into 5-fluoro-dUMP, which irreversibly inactivates thymidylate synthase inhibiting both RNA and DNA synthesis (17). Heterologous expression of E. coli codA was shown to confer 5-FC sensitivity to mammalian cells, ordinarily not producing CD (18). Concomitant expression of E. coli CD and UPRT as a fusion protein, encoded by codA::upp, was shown to further enhance the 5-FC cytotoxicity (19). A recent study on pyrimidine salvage in Streptomyces species indicated a lack of CD activity but sensitivity towards 5-FU (20). Some other actinobacteria, like Rhodococcus species, were also shown to be 5-FU sensitive (21,22). Most Nocardia species and certain Mycobacterium strains, however, were 5-FU resistant, while other mycobacteria strains were highly 5-FU sensitive (21).

R. equi is a facultative intracellular pathogen that causes fatal pyogranulomatous bronchopneumonia in young foals. It is also an emerging opportunistic pathogen of immuno-compromised humans, particularly HIV infected patients (23,24). In addition to its pathogenic life-style, R. equi is a common soil-dwelling microorganism capable of rapid growth in soil and manure, using plant and animal sterols as sole carbon and energy sources (25,26). Knowledge on sterol metabolism in R. equi is extremely limited. So far, only the gene encoding the proposed first step in cholesterol degradation, cholesterol oxidase (choE), has been identified and inactivated in R. equi (27,28). A cholesterol catabolic gene cluster has been identified in the closely related Rhodococcus jostii RHA1 (6,29). This cluster also encodes a putative cholesterol uptake system, designated mce4 operon. ‘Mammalian cell entry’ (mce) genes are critical virulence factors of the intracellular pathogen Mycobacterium tuberculosis (30) and, interestingly, mce1− and mce4− strains of M. tuberculosis H37Rv showed attenuated survival in mice (31). The supAB genes are part of the mce4 operon and may encode the permease subunits of the cholesterol uptake system (6). The supAB and mce4 genes were shown to be essential for growth of R. jostii RHA1 on cholesterol (6). Pandey and Sassetti (32) recently confirmed that cholesterol is used as a carbon and energy source by M. tuberculosis H37Rv, and that the mce4 cluster in H37Rv is essential for growth on cholesterol.

The R. equi strain 103S genome sequence recently has become available (http://www.sanger.ac.uk/Projects/R_equi/). Proper genome annotation and identification of pathogenicity genes requires simple methods for gene deletion mutagenesis. A method for the isolation of gene deletion mutants of R. equi by the double homologous recombination strategy was first reported by Navas et al. (27). An improved method, using lacZ as counter-selectable marker, was subsequently reported by Jain et al. (33). These methods, however, often involve screening and handling of large numbers of colonies to select for the rare second recombination event (34,35). Here we show that introduction of the codA::upp cassette confers 5-FC sensitivity to R. equi allowing positive selection of the targeted gene deletion mutants. A simple and efficient procedure to generate unmarked in-frame gene deletions in R. equi is reported, exemplified by the construction and characterization of a ΔsupAB mutant.

MATERIALS AND METHODS

Culture media and growth conditions for R. equi strain RE1

Virulent R. equi wild type strain RE1 was isolated from a foal with pyogranulomatous pneumonia in September 2007 in The Netherlands. R. equi strains were grown at 30°C (200 r.p.m.) in Luria-Bertani (LB) medium or mineral acetate medium (MM-Ac). MM-Ac contained K2HPO4 (4.65 g/l), NaH2PO4 · H2O (1.5 g/l), sodium-acetate (2 g/l), NH4Cl (3 g/l), MgSO4 · 7H2O (1 g/l), thiamine (40 mg/l, filter sterile) and Vishniac stock solution (1 ml/l). Vishniac stock solution was prepared as follows [modified from (36)]: EDTA (10 g/l) and ZnSO4 · 7H2O (4.4 g/l) were dissolved in distilled water (pH 8 using 2 M KOH). Then, CaCl2ċ2H2O (1.47 g/l), MnCl2ċ7 H2O (1 g/l), FeSO4ċ7H2O (1 g/l), (NH4)6 Mo7O24ċ4 H2O (0.22 g/l), CuSO4ċ5H2O (0.315 g/l) and CoCl2ċ6 H2O (0.32 g/l) were added in that order at pH 6 and finally stored at pH 4. For growth on solid media Bacto-agar (15 g/l) was added. 5-Fluorocytosine (Sigma-Aldrich Chemie, Zwijndrecht, The Netherlands) and 5-fluorouracil (Sigma-Aldrich) stock solutions (10 mg/ml) were prepared in distilled water, dissolved by heating to 50°C, filter-sterilized and added to autoclaved media.

For growth on cholesterol, R. equi strains were inoculated in 25 ml MM-Ac liquid medium and grown for 24 h at 200 r.p.m. and 30°C. The pre-culture (0.5 ml) was used to inoculate 50 ml of MM medium containing 0.5 g/l cholesterol (Sigma-Aldrich) as sole carbon and energy source which had been finely dispersed by sonication. Regular turbidity (OD600nm) measurements of cholesterol grown cultures were not possible due to high background of the cholesterol suspension. Protein content of the culture was used as a measure for biomass formation and was determined as follows. A sample (0.5 ml) of the culture was pelleted by centrifugation (5 min at 12 000g) and thoroughly resuspended in 0.1 ml Bacterial Protein Extraction Reagent (B-PER, Pierce, PerBio Science Nederland B.V., Etten-Leur, The Netherlands). Then, 0.4 ml distilled water was added and the suspension was vortexed and incubated at room temperature for 5–10 min. An aliquot of 160 μl was mixed with 640 μl of distilled water and 200 μl of protein assay reagent (BioRad) was added. Protein content of the sample was determined using bovine serum albumin (BSA) as a standard as described by the manufacturer.

Cloning, PCR and chromosomal DNA isolation

E. coli DH5α was used as host for all cloning procedures. Restriction enzymes were obtained from Fermentas GmbH (St Leon-Rot, Germany). PCR was performed in a reaction mixture (25 μl) consisting of Tris–HCl (10 mM, pH 8), 10 × High-Fidelity polymerase buffer (Fermentas), dNTPs (0.2 mM), DMSO (2%), PCR primers (10 ng/μl each, Table 1) and High-Fidelity polymerase enzyme (1–2 U, Fermentas). For colony PCR, cell material was mixed with 100 µl of chloroform and 100 µl of 10 mM Tris–HCl pH 8, vortexed vigorously and centrifuged (2 min at 14 000g). A sample of the upper water phase (1 μl) was subsequently used as template for PCR. Chromosomal DNA of R. equi cell cultures was isolated using the GenElute Bacterial Genomics DNA Kit (Sigma-Aldrich) according to the instructions of the manufacturer.

Table 1.

Oligonucleotides used in this study

No	PCR amplicon	Size (bp)	Oligonucleotide sequence (nr.)
1	Upstream region supAB (deletion construct)	1592	supABequiUP-F ATCGCGAGGTCAGCTTGGAG (1) supABequiUP-R CCCGGGCGCCGACATCGCGAAGAATC (2)
2	Downstream region supAB (deletion construct)	1470	supABequiDOWN-F CCCGGGCTCATCCACACCTACTACGG (3) supABequiDOWN-R ACTAGTGAGCTGCTGAATCTGAACTGG (4)
3	Upstream region supAB (confirmation deletion mutant)	1786 (wt: 3192)	supABequiContrUP-F CGGGAGTGCGTAGATGAGTGCA (5) supABequiContr-R TTATCCCGAAAGGTTGAAGTTG (6)
4	Downstream region supAB (confirmation deletion mutant)	1570 (wt: 2976)	supABequiContr-F GTGGTCGACCTCCTCGAGGTAC (7) supABequiContrDOWN-R GGACTTGAGCCCGGAGGCATCG (8)
5	supAB genes (confirmation deletion mutant)	231 (wt: 1637)	supABequiContr-F GTGGTCGACCTCCTCGAGGTAC (7) supABequiContr-R TTATCCCGAAAGGTTGAAGTTG (6)
6	supAB genes (complementation ΔsupAB mutant)	3163	supABequiUP-F ATCGCGAGGTCAGCTTGGAG (1) supABequiContr-R TTATCCCGAAAGGTTGAAGTTG (6)
7	aphII promoter region	367	Pkan-F AGCTTCACGCTGCCGCAAGCACT Pkan-E5-R GATATCATGCGAAACGATCCTCATCCTG
8	Pkan-codA cassette	1733	Pkan-F AGCTTCACGCTGCCGCAAGCACT codA-R2 GTCAACGTTTGTAATCGATGGCTTCT
9	vapA	408	vapA-F GCAGCAGTGCGATTCTCAATAG (9) vapA-R TAACTCCACCGGACTGGATATG (10)

Primer numbering used in the text is shown between brackets following the nucleotide sequence.

Electrotransformation of R. equi strain RE1

R. equi strain RE1 cells were transformed by electroporation essentially as described (27). Briefly, cell cultures were grown in 50 ml LB at 30°C until OD600 reached 0.8–1.0. The cells were pelleted (20 min at 4500g) and washed twice with 10% ice-cold glycerol. Pelleted cells were re-suspended in 0.5–1 ml ice-cold 10% glycerol and 200 μl aliquots were put on ice. MilliQ-eluted plasmid DNA (5–10 μl; GenElute Plasmid Miniprep Kit, Sigma-Aldrich) was added to 200 μl cells in 2 mm gapped cuvettes. Electroporation was performed with a single pulse of 12.5 kV/cm, 1000 Ω and 25 μF. Electroporated cells were gently mixed with 1 ml LB medium and allowed to recover for 2 h at 37°C and 200 r.p.m. Aliquots (200 μl) were plated onto LB agar medium containing apramycin (50 μg/ml; Duchefa Biochemie, Haarlem, The Netherlands). Transformants appeared after 2–3 days of incubation at 30°C. The transformation efficiency for non-replicative plasmids integrating via homologous recombination was ∼ 10 transformants/μg plasmid DNA.

5-Fluorocytosine positive selection in R. equi

R. equi transformants were inoculated in LB liquid medium (25 ml) and grown overnight (20–24 h) at 30°C and 200 r.p.m. 5-FC selection was performed by plating 100 μl aliquots of a dilution series (10−1 to 10−3 in MM-Ac medium) of the grown culture onto MM-Ac agar plates supplemented with 5-FC (100 μg/ml). Dilution of the culture prior to plating was crucial for effective 5-FC selection. 5-FC resistant colonies appeared after 2–3 days of incubation at 30°C.

Construction of plasmids pSET-Pkan-codA, pSET-Pkan-codAupp and pSelAct

The aphII promoter region was amplified from pRESQ (37) using PCR primers Pkan-F and Pkan-E5-R (Table 1). The obtained PCR product of 367 bp was blunt-ligated into EcoRV digested pBluescript(II)KS (Stratagene), resulting in pBs-Pkan. A SalI/NotI restriction released a 431 bp fragment comprising the aphII promoter which was then cloned into SalI/NotI digested pORF-codA::upp (InvivoGen, San Diego, USA), yielding plasmid pORF-Pkan-codAupp. The Pkan-codA::upp cassette was subsequently isolated from pORF-Pkan-codAupp as a 2.4 kb SmaI/NheI fragment and ligated into EcoRV/XbaI digested pSET152 resulting in plasmid pSET-Pkan-codAupp (Figure 2). The Pkan-codA cassette (1733 bp) was amplified from pSET-Pkan-codAupp using primers Pkan-F and codA-R2 (Table 1) and ligated into EcoRV digested pSET152, resulting in pSET-Pkan-codA (Figure 2). Suicide plasmid pSelAct (Figure 3) was constructed by ligating a 2.4 kb Klenow-treated EcoRI/NheI fragment of pORF-Pkan-codAupp into SspI digested pBs-Apra-ori dephosphoryated with alkaline phosphatase. Plasmid pBs-Apra-ori was constructed from pBluescript(II)KS in which the bla cassette was removed with BspHI, followed by Klenow treatment and replaced by an apramycin-oriT cassette obtained as a 1.3 kb XbaI fragment (Klenow treated) from pIJ773 (13).

Figure 2.

The pSET152 derived integrative plasmids used to introduce the (A) codA or (B) codA::upp cassette into R. equi RE1, expressed under control of the aphII kanamycin resistance cassette promoter (P). The apramycin resistance cassette (aac(3)IV), the Streptomyces PhiC31 integrase gene (int) and the RP4 origin of transfer (oriT) are also indicated.

Figure 3.

Schematic representation of the non-replicative plasmid pSelAct used to generate unmarked in-frame gene deletions in R. equi RE1. Restriction sites indicated are unique.

Macrophage survival test

The human monocyte cell line U937 (38) was used to test for survival of R. equi strains. The monocytes were grown in RPMI 1640 (Invitrogen) + NaHCO3 (1 g/l) + sodium pyruvate (0.11 g/l) + glucose medium (4.5 g/l) (RPMI 1640 medium), buffered with 10 mM HEPES (Hopax fine chemicals, Taiwan) and supplemented with penicillin (200 IU/ml), streptomycin (200 IU/ml) and 10% fetal bovine serum (FBS). The cells were grown in suspension at 37°C and 5% CO2. For the macrophage survival assay, monocytes were grown for several days as described above. The culture medium was replaced with fresh culture medium and the cells were activated overnight with phorbol 12-myristate 13-acetate (60 ng/ml, PMA, Sigma-Aldrich) to induce their differentiation to macrophages. The differentiated cells were spun down (5 min at 200g) and the pellet was re-suspended in fresh, antibiotic free RPMI 1640 medium with 10% FBS. For each strain to be tested, a tube containing 10 ml of a cell suspension (∼106 cells/ml) was inoculated with R. equi, pre-grown in nutrient broth (Difco, Detroit, MI., USA) at 37°C, at a multiplicity of infection (MOI) of approximately 10 bacteria per macrophage. The bacteria were incubated with the macrophages for 1 h at 37°C and 5% CO2. The medium was replaced with 10 ml RPMI 1640 medium supplemented with 10% FBS and 100 µg/ml gentamycin and incubated again for 1 h to kill any extra-cellular bacteria. The macrophages (with internalized R. equi) were spun down (5 min at 200g) and the pellet was re-suspended in 40 ml RPMI 1640 medium, buffered with 10 mM HEPES and supplemented with 10% FBS and gentamycin (10 µg/ml). This suspension was divided over four culture bottles (10 ml each) and incubated at 37°C and 5% CO2. After 4, 28, 52 and 76 h the macrophages (one culture bottle per strain) were spun down (5 min at 200g) and the pellet washed twice in 1 ml antibiotic free RPMI 1640 medium. Finally the pellet was lysed with 1% Triton X-100 (Sigma-Aldrich) in 0.01 M phosphate buffered saline, followed by live count determination (plate counting).

Culture media and growth conditions for 5-FC selection in actinobacteria

Actinobacterial strains (Table 2) were grown until late exponential phase as shaken liquid cultures in complex medium at 30°C, except for Amycolatopsis methanolica and Mycobacterium smegmatis which were grown at 37°C. Tryptic Soy Broth (TSB) was used for Corynebacterium glutamicum, Arthrobacter globiformis, Amycolatopsis orientalis, A. methanolica and M. smegmatis. TSB supplemented with 2.5% NaCl was used for Salinospora tropica, and YEME:TSB (1:1) (39) for all Streptomyces strains and Saccharopolyspora erythrea. LB was used for R. equi and Rhodococcus rhodochrous and LBP (3) was used for R. erythropolis and R. jostii. Minimal regeneration medium (MRM) consisted of K2SO4 (0.25 g/l), (NH4)2SO4 (2 g/l), MgCl2ċ6H20 (0.6 g/l), 2-[(hydroxyl-1, 1-bis (hydroxymethyl)ethyl)amino] ethanesulfonic acid; TES (5.73 g/l) and trace elements (39). Sucrose (10.3% w/v) and l-proline (0.3% w/v) were added to MRM medium for Streptomyces strains, but omitted for all other strains. Glucose (20 mM) was used as carbon and energy source, except for all Rhodococcus strains which were grown in the presence of acetate (2 g/l). Autoclaved MRM was supplemented with KH2PO4 (0.1 g/l), CaCl2ċ2H20 (3 g/l) and NaOH (0.28 g/l). For agar plates 2% granulated agar was added. Filter-sterilized thiamine (40 μg/ml) was added to autoclaved MRM medium for growth of Rhodococcus strains. 5-FC and 5-FU were freshly prepared as 10 mg/ml stocks in distilled water, dissolved by heating to 50°C and added to autoclaved medium.

Table 2.

Growth of actinobacteria on MRM mineral agar media supplemented with different concentrations of 5-FC or 5-FU

Strain	5-FC (μg/ml)				5-FU (μg/ml)			Reference/origin
	0	20	50	100	20	50	100
Amycolatopsis methanolica NCIB11946	++	+	−	−	–	–	–	(44)
Amycolatopsis orientalis ATCC 19795	++	++	+	+	−	−	−	ATCC
Arthrobacter globiformis DSM 20124	++	++	+	+	–	–	–	DSMZ
Corynebacterium glutamicum ATCC 13032	++	+	−	−	−	–	–	ATCC
Mycobacterium smegmatis mc2155	++	+	−	−	–	–	–	(45)
Rhodococcus rhodochrous DSM 43269	++	++	++	++	+	–	–	DSMZ
Rhodococcus erythropolis SQ1	++	++	++	++	–	–	–	(46)
Rhodococcus jostii RHA1	++	++	++	++	+	−	–	M. Fukuda, Nagaoka (Japan)
Rhodococcus equi RE1	++	++	++	++	+	–	–	Schering-Plough, Netherlands
Salinospora tropica	++	++	++	++	−	−	−	DSMZ
Saccharopolyspora erythraea	++	++	++	++	–	–	–	DSMZ
Streptomyces albus J1074	++	++	++	++	–	–	–	(47)
Streptomyces avermitilis MA-4680	++	++	++	++	++	++	++	ATCC (ATCC 31267)
Streptomyces coelicolor M145	++	++	++	++	–	–	–	(39)
Streptomyces griseus DSM40236	++	++	++	++	–	–	–	DSMZ
Streptomyces lividans TK23	++	++	++	++	–	–	–	(39)
Streptomyces scabies ISP5078	++	++	++	+	–	–	–	Wellington, Warwick (UK)
Streptomyces tendae Tü 901/8c	++	++	++	+	+	–	–	(48)

Symbols indicate growth (++), moderate growth (+), slight/minor growth (−) or no growth (–).

ATCC, American Type Culture Collection; DSMZ, Deutsche Sammlung von Mikroorganismen und Zellkulturen.

RESULTS AND DISCUSSION

CD-UPRT mediated 5-fluorocytosine sensitivity of R. equi strain RE1

An obvious prerequisite for the applicability of a 5-FC based conditionally lethal positive selection system is natural resistance of R. equi strain RE1 towards 5-FC. To test this, R. equi RE1 was streaked onto acetate mineral (MM-Ac) agar plates supplemented with 5-FC or 5-FU and incubated for 3 days at 30°C. Examination of the plates revealed that R. equi was resistant to high concentrations of 5-FC (100 μg/ml), but highly sensitive to a lower concentration of 5-FU (50 μg/ml), indicating the feasibility of developing a 5-FC based positive selection system for R. equi (Figure 1). Next, we examined whether expression of the E. coli genes codA and upp in R. equi would confer sensitivity to 5-FC and thus could act as a suicide marker. The integrative E. coli-Streptomyces shuttle vector pSET152 provided a stable and convenient vehicle to introduce codA or a functional codA::upp fusion into the R. equi genome (40,41). The aphII promoter region of the kanamycin resistance cassette (Pkan) was used to drive expression of codA or codA::upp, respectively. The two resulting integrative plasmids, pSET-Pkan-codA and pSET-Pkan-codAupp (Figure 2), were mobilized separately by electroporation to R. equi RE1. Transformants harboring either pSET-Pkan-codA or pSET-Pkan-codAupp were streaked onto MM-Ac agar plates (Figure 1A), MM-Ac agar plates supplemented with 100 μg/ml 5-FC (Figure 1B), or MM-Ac agar plates supplemented with 50 μg/ml 5-FU (Figure 1C) and incubated for 3 days at 30°C. The presence of the codA::upp cassette rendered R. equi sensitive to 5-FC, whereas expression of codA alone did not result in 5-FC sensitivity (Figure 1B). Growth of the R. equi strain carrying the codA::upp cassette under nonselective conditions was similar to that of wild type (Figure 1A), indicating that expression is conditionally lethal and not detrimental to cells when 5-FC is omitted from the medium. Importantly, no 5-FC selection was observed when complex agar media (LB) was used (Figure 1D). This may be due to repression of the pyrimidine salvage pathway affecting 5-FC uptake.

Figure 1.

Introduction of E. coli codA::upp cassette, encoding CD and UPRT, confers 5-FC sensitivity to R. equi strain RE1. Panels show wild type strain RE1, recombinant strain RE1 containing pSET152, recombinant strain RE1 containing plasmid pSET-codA::upp streaked on mineral acetate agar (MM-Ac) containing (A) no addition, (B) 5-FC (100 μg/ml) or (C) 5-FU (50 μg/ml) and on (D) LB agar containing 5-FC (100 μg/ml).

To enable the generation of unmarked in-frame gene deletions in R. equi using CD-UPRT based 5-FC conditional lethality, we next developed a non-replicative suicide vector designated pSelAct (Figure 3; see Materials and Methods section). Vector pSelAct is a small (5.6 kb) conjugative plasmid based on pBlueScript(II)KS, harboring the aac(3)IV cassette for apramycin resistance that can be used both in E. coli and R. equi. The vector contains a MCS with several unique restriction sites for cloning flanking regions of the gene of interest using lacZ dependent blue-white screening in E. coli.

Unmarked in-frame gene deletion of supAB in R. equi

A BLAST search with the supA (ro04696), supB (ro04697) and mce4 (ro04698-ro04703) genes of R. jostii RHA1, predicted to encode the cholesterol uptake machinery (6), revealed that these genes and their apparent operonic organization were highly conserved in the R. equi 103S genome (Figure 4; http://www.sanger.ac.uk/Projects/R_equi/). Highest amino acid identities were observed for SupA (94%) and SupB (89%), whereas the mce4 encoded proteins Mce4A-Mce4F showed lower identities (57–71%) (Figure 4).

Figure 4.

(A) Schematic overview of the molecular organization of the mce4 gene clusters in R. jostii RHA1 and R. equi 103S. Percentages indicate amino acid sequence identities between the Mce4 proteins of R. jostii RHA1 and R. equi RE1. (B) Construction of mutagenic plasmid pSelAct-ΔsupAB used to generate an unmarked supAB gene deletion mutant of R. equi RE1. Small black arrows with numbers indicate the PCR oligonucleotides used in this study (Table 1) and the site of their annealing. (C) Molecular organization of the mce4 locus of R. equi RE1 following integration of pSelAct-ΔsupAB by single homologous recombination, and second homologous recombinant event after 5-FC counter-selection, resulting in supAB gene deletion.

The supA and supB genes are the first two genes of the mce4 operon, and may encode the cholesterol permease subunits (Figure 4). To demonstrate a specific role of supAB in cholesterol catabolism, we aimed for their selective inactivation, without affecting expression of the downstream genes. This prompted us to develop the unmarked in-frame gene deletion system for R. equi based on counter-selection with 5-FC. The upstream and downstream flanking regions (∼1.5 kb) of the supAB genes were amplified by PCR and ligated into pSelAct generating pSelAct-ΔsupAB (Figure 4; Table 1 for primers). This construct was used to generate a ΔsupAB mutant, by reducing the wild type supA (765 nt) and supB (852 nt) genes, encompassing 1637 bp, to a single in-frame open-reading-frame of 231 bp, encoding the first 20 amino acid of SupA and the last 54 amino acids of SupB, separated by a 6 bp SmaI restriction site. The first and last part of supA and supB, respectively, were left intact to ensure proper expression of downstream genes in the mce4 operon. Introduction of plasmid pSelAct-ΔsupAB into R. equi RE1, selecting for apramycin resistance, resulted in 29 transformants of which four were selected and grown nonselectively overnight for 20 h in LB medium (i.e. lacking apramycin and 5-FC). Aliquots (100 μl) of the overnight cultures were plated in 10−1 to 10−3 dilutions onto MM-Ac agar plates supplemented with 100 μg/ml 5-FC and incubated for 3 days at 30°C. Confluent growth on 5-FC selection plates was obtained for one of the transformants. Conceivably, the codA::upp suicide cassette had been inactivated in this transformant by spontaneous mutation or transposon insertion, as has been reported to occur during sacB counter-selection (3,42). Typically, >102 colonies grew on the 10−2 plate from which 50 colonies were replica picked onto LB agar with and without apramycin to select for 5-FCR/ApraS colonies. The frequency of 5-FCR/ApraS colonies amongst the other three transformants varied between 70–90%, indicating that the suicide cassette had been inactivated in 10–30% of the 5-FCR/ApraS colonies. For one transformant, eighteen 5-FCR/ApraS colonies were checked by colony PCR for the presence of the mutant ΔsupAB genotype using PCR with oligonucleotides 6 and 7 amplifying the supAB genes (Table 1, Figure 4). Two of the eighteen FCR/ApraS colonies gave a PCR product of the expected size (231 bp) and were selected for further characterization. Genomic DNA was isolated from these two ΔsupAB mutants and subjected to PCR analysis of the supAB locus and the up- and downstream flanking regions of supAB (Figure 5, see Table 1 for oligonucleotides used). This confirmed the presence of a genuine supAB gene deletion in both cases and revealed no aberrant genomic rearrangements at the supAB locus (Figure 5). One mutant strain was chosen, designated R. equi RE1ΔsupAB, and was used for further characterization.

Figure 5.

PCR analysis of wild type R. equi strain RE1 (wt) and two ΔsupAB gene deletion mutants of R. equi strain RE1 (1 and 2). Panels show amplification of (I) the supAB locus [6 + 7], (II) the upstream region of supAB [5 + 6], (III) the downstream region of supAB [7 + 8], (IV) the vapA virulence gene [9 + 10]. The oligonucleotides used are mentioned between brackets and described in Table 1 and/or Figure 4.

Phenotypical analysis and functional complementation of R. equi RE1ΔsupAB mutant

R. equi RE1 wild type and the RE1ΔsupAB mutant strain were grown in MM-Ac liquid medium and used to inoculate MM-cholesterol liquid medium. The RE1ΔsupAB mutant was completely blocked in growth on cholesterol as sole carbon and energy source (Figure 6). Growth on acetate or the steroid substrate 4-androstene-3,17-dione (AD) was unaffected and comparable to the wild type strain (data not shown). This indicated that supAB are essential for cholesterol catabolism, probably acting as the permease subunits of the cholesterol ABC transporter. These results are fully consistent with the phenotype of the supAB mutant of strain RHA1 (6).

Figure 6.

Growth curves in mineral medium supplemented with cholesterol (0.5 g/l) of R. equi RE1 wild-type (filled diamond), RE1 ΔsupAB mutant strain (filled square), RE1 ΔsupAB + pSET-supAB complemented strain (filled triangle), and RE1 ΔsupAB + pSET152 control strain (x). Protein content (mg/l) of the culture was used as a measure for biomass formation. The data represent the averages for two independent experiments.

To ensure that the growth deficiency of the RE1ΔsupAB mutant was solely due to deletion of the supAB genes, a 3163-bp DNA fragment carrying the wild-type supAB genes was introduced by electroporation into the RE1ΔsupAB mutant. The DNA fragment was obtained by PCR on strain RE1 wild type DNA using oligonucleotides 1 and 6 (Table 1, Figure 4) and cloned into the integrative plasmid SET152, resulting in pSET-supAB. The cholesterol growth negative phenotype of the RE1ΔsupAB mutant harboring pSET-supAB was fully complemented, restoring growth on cholesterol to levels comparable to the wild type (Figure 6). We conclude that the supAB deletion did not exert any polar effects on the expression of other genes in the mce4 operon.

These data show that we have developed a novel, simple and efficient method to generate unmarked in-frame gene deletions in R. equi, based on 5-FC lethality in the presence of the E. coli codA::upp fusion gene.

Deletion of supAB does not affect R. equi RE1 survival in macrophages

Intracellular survival and proliferation of the R. equi RE1ΔsupAB mutant in the human monocyte cell line U937 was compared to those of wild type strain RE1 (Figure 7). The avirulent, plasmid free strain R. equi 103− (43) was included as a negative control for macrophage survival (Figure 7). The results revealed that the RE1ΔsupAB mutant is able to survive and proliferate in macrophages comparable to the wild-type parent strain RE1. By contrast, the avirulent strain 103− failed to proliferate, resulting in reduced numbers of intracellular bacteria in time (Figure 7). These results indicate that cholesterol metabolism is not essential for macrophage survival of R. equi RE1 and suggest that cholesterol metabolism is not important for virulence of R. equi RE1 in vivo. These observations are consistent with the finding that ChoE is not important in the virulence of R. equi (28).

Figure 7.

Survival and proliferation of R. equi strains in the human monocyte cell line U937. Macrophage cell suspensions were infected with wild type virulent strain R. equi RE1 (filled diamond), mutant strain R. equi RE1ΔsupAB (filled square) and non-virulent (control) strain R. equi 103− (filled triangle). Following a 1-h incubation to allow phagocytosis, cells were washed and treated with gentamycin to kill remaining extra-cellular bacteria. The numbers of intracellular bacteria were determined by plate counts following macrophage lysis. The data represent the averages for two independent experiments. Plate counts were carried out in duplicate.

Actinobacteria are generally 5-FC resistant

The CD-UPRT based 5-FC selection could also be a useful asset in the generation of unmarked in-frame gene deletions in other actinobacteria. To examine whether CD-UPRT selection potentially is an effective counter-selectable marker in this family of microorganisms, we tested the sensitivity of several actinobacteria towards 5-FC and 5-FU. A wide selection of actinobacterial strains were grown in complex medium until late exponential/early stationary phase and plated in 10-fold dilutions onto mineral selection media containing increasing concentrations of 5-FC or 5-FU (Table 2). The tested strains generally were resistant to high concentrations of 5-FC (100 μg/ml), but highly sensitive to lower concentrations of 5-FU (20–50 μg/ml), indicating the feasibility of developing a 5-FC based positive selection system for other actinobacteria.

FUNDING

Intervet Schering-Plough Animal Health, Boxmeer, The Netherlands. Funding for open access charge: Groningen Biomolecular Sciences and Biotechnology Institute (GBB), University of Groningen, The Netherlands.

Conflict of interest statement. None declared.