Site-specific integration of Streptomyces ΦC31 integrase-based vectors in the chromosome of Rhodococcus equi

Abstract

Streptomyces ΦC31-based site-specific integration was used to transform the facultative intracellular pathogen Rhodococcus equi. The transformation efficiency of vectors incorporating the ΦC31 integrase and attP sites was comparable to that of replication plasmids using the same electroporation procedure. A single attB integration site was identified within an ORF encoding a pirin-like protein, which deviates slightly from the consensus sequence of Streptomyces attB sites. Vector integration was stably maintained in the R. equi chromosome for as many as 100 generations during unselected passage in vitro. In addition, integration does not appear to affect the replication of bacteria inside macrophages. Finally, this integration system was also used to successfully complement an R. equi mutant.

Introduction

Rhodococcus equi is a gram-positive opportunistic actinomycete of worldwide distribution which causes an often fatal pyogranulomatous pneumonia of foals and of immunocompromised persons, especially HIV-infected individuals (Harvey & Sunstrum, 1991; Prescott, 1991). One of the striking features of R. equi infection is the capacity of the bacterium to survive and replicate inside of alveolar macrophages (Hondalus & Mosser, 1994; Hondalus 1997) by means which have yet to be fully elucidated but which involve perturbation of phagosomal–lysosomal maturation (Zink et al., 1987; Hietela & Ardans, 2003; Fernandez-Mora et al., 2005). In susceptible individuals, macrophages fail to kill the organism, and intracellular replication of the bacteria eventually leads to macrophage death and dissemination of the infection.

Rhodococcus equi virulence in foals is dependent on possession of an 80.6-kb virulence plasmid and expression of Virulence-associated protein A (VapA), a virulence plasmid encoded, highly immunogenic cell wall protein (Prescott, 1991; Takai et al., 1991a, b; Tan et al., 1995; Jain et al., 2003). Rhodococcus equi strains lacking the virulence plasmid and/or vapA are avirulent and can neither multiply inside of macrophages cultured in vitro nor cause disease in vivo (Jain et al., 2003). Additional R. equi virulence determinants include aceA, a chromosomal gene encoding isocitrate lyase, an enzyme of the carboxylic shunt pathway, and two other virulence plasmid-encoded genes, virR and orf8, each of which are regulators necessary for vapA expression (Ren & Prescott, 2003; Wall et al., 2005; Byrne et al., 2007).

Until very recently, it has been difficult to experimentally address numerous questions pertaining to R. equi pathogenesis. A major hurdle to R. equi research has been the lack of molecular tools required for genetic analysis. Of late, great progress has been made in the areas of shuttle vector construction (Zheng et al., 1997; Sekizaki et al., 1998; Giguere et al., 1999; Mangan et al., 2005), transformation (Zheng et al., 1997), transposon-based random mutagenesis procedures (Mangan & Meijer, 2001; Ashour & Hondalus, 2003), and allelic exchange methodology (Navas et al., 2001; Jain et al., 2003). The study of mutant strains requires complementation analysis, which can be done using episomal shuttle vectors. However, the latter has drawbacks, such as loss of complementing plasmids under nonselective conditions, as well as incompatibility issues between the shuttle vector and the indigenous R. equi virulence plasmid. Phage integrase-mediated site-specific recombination is an alternative to produce stable transformants. Integration vectors can have several advantages over conventional episomal plasmids, including single site integration, single copy number once integrated and stability in the absence of selective pressure. This integration approach has been successfully used in genetic analysis of various bacterial species (Lauer et al., 2002; Choi et al., 2004; Murry et al., 2005). A well-studied site-specific recombination system is that of ΦC31, a bacteriophage that infects Streptomyces coelicolor, and integrates into the bacterial chromosome (Thorpe & Smith, 1998). Integration occurs via site-specific recombination between the phage attachment site, attP, and the attachment site on the bacterial chromosome, attB. ΦC31 integration is independent of host factors, requiring merely the phage integrase (recombinase) enzyme and the attP and attB sequences (Thorpe & Smith, 1998). Given that lack of requirement for specific host factors coupled with the genetic relatedness of Streptomyces spp. and Rhodococcus spp. (Goodfellow & Alderson, 1977), both of which are GC-rich actinomycetes, we reasoned that ΦC31 integrase might also mediate site-specific recombination and integration into the R. equi chromosome. Herein, we test that hypothesis.

Materials and methods

Bacterial strains and growth conditions

RapidTrans™ -competent Escherichia coli (Active Motif, Carlsbad, CA) were used for all cloning procedures. Virulent R. equi 103+ was kindly provided by J. Prescott, University of Guelph, Canada. Rhodococcus equi was cultured in brain heart infusion (BHI) broth with shaking at 200 r.p.m. or on BHI agar. When required, antibiotics were added at the following concentrations: apramycin, 80 μg mL⁻¹; hygromycin, 180 μg mL⁻¹. Electrocompetent R. equi cells were prepared and used as described (Jain et al., 2003). The R. equi vapA-mutant strain was constructed in a previous study (Jain et al., 2003). The pSET152-integrant strains were generated by transforming pSET152 into R. equi 103+ competent cells and were selected on BHI agar with apramycin. The integrase-based vapA-complemented strain was made by transforming vector pSET152-vapA into the R. equi vapA mutant and selecting on BHI plates with hygromycin. The integration of either of the vectors on the R. equi chromosome was confirmed by specific amplification of the right hybrid formed in the integration using primer set Int-rh-F 5′-CTCTATGGCCCGTACTGACG-3′ and Int-rh-R 5′-CGCATCGAAGCCGTGGATCC-3′ (Fig. 1). A specific 310-bp PCR amplicon was generated where recombination occurred on the chromosome. Rhodococcus equi 103+ was also transformed with pMV261-hyg to generate a hygromycin-resistant strain, strain 103+hyg, used in the stability analysis.

Fig. 1.

Recombination event mediated by ΦC31 integrase between attP and attB. The short vertical line in the middle of the attP and attB sites indicates the core region (TT) where recombination occurs. The arrow indicates the position of primers used in sequencing analysis and PCR amplification.

DNA manipulation

All restriction enzymes, Klenow fragment, and T4 DNA ligase were purchased from New England Biolabs (NEB, Beverly, MA). DNA digestion and ligation were performed according to the manufacturer's instructions. Rhodococcus equi genomic DNA was prepared as described (Navas et al., 2001). Plasmid DNA was purified from E. coli using Wizard Plus Miniprep DNA Purification System (Promega Corporation, Madison, WI). A gel purification kit (Qiagen, Valencia, CA) was used to purify PCR products or restricted DNA fragments. All PCR were conducted with a routine PCR program: (1) 94 °C for 3 min, one cycle; (2) 94 °C for 20 s; 45–55 °C (depending on the T_m of primers), 30 s; 68 °C for 1 min, 35 cycles; (3) 72 °C for 10 min, one cycle. Platinum^® High fidelity Taq polymerase (Invitrogen Corporation, Carlsbad, CA) was used to ensure high-fidelity DNA synthesis. Oligonucleotides were synthesized by IDT technologies (Coralville, IA). DNA sequencing was carried out by the sequencing and synthesis facility at the University of Georgia.

Plasmids and vector construction

pSET152 carrying a copy of the Streptomyces ΦC31 integrase gene, the attP site, and an apramycin-resistant marker was a kind gift from Janet Westpheling at the University of Georgia. pSET152-d is a derivative of pSET152 in which the integrase gene and attP sequence were deleted. To construct pSET152-d, pSET152 was digested with PvuI and SgrAI and the resultant 3006-bp fragment lacking the int gene and the attP site was self ligated. pMV261-hyg is a Mycobacterium spp.–E. coli shuttle vector carrying the pAL5000 replication origin, oriM (Giguere et al., 1999), an E. coli replication origin, oriE, along with a hygromycin resistance cassette. pSJ35, also having a pAL5000 origin, is a plasmid from a previous study (Jain et al., 2003) carrying both apramycin and hygromycin cassettes.

ΦC31 integrase-based vector pSET152-vapA was constructed to demonstrate complementation of the R. equi vapA mutant. vapA was amplified with Hsp60-vapA-F, 5′-CGC TGG CCA CTC TTC ACA AGA CG-3′, and Hsp60-vapA-R, 5′-CTA TGG CCA CTA GGC GTT GTG CCA-3′. The 563-bp PCR product was cloned downstream of a heat shock promoter, groEL at an MscI site on pMV261-hyg to yield pMV261-vapA. The plasmid was then cleaved with NotI, and a fragment containing the heat shock promoter, vapA, the hygromycin cassette, and the E. coli replication origin, oriE, was then circularized by self ligation. Next, the ΦC31 integrase and attP attachment sites were amplified from pSET152 using primers int-KpnI-F, 5′-CCTCTAGATCGACAGACGTAGATCAG-3′, and int-XbaI-R, 5′-CTGGTACCATGCAGCGGAAAAGATCC-3′. The amplified PCR product was digested with KpnI and XbaI and ligated into the circulized pMV261 NotI fragment cut with the same enzymes. Vector pSET152-vapA was used to transform the R. equi vapA mutant.

Determination of integrant stability

The pSET152-integrant strain and the hygromycin-resistant strain 103+hyg were cultured in BHI broth to OD_{600 nm} 1.0. One microliter of cell culture was then inoculated into 20 mL of BHI broth without antibiotic and then cultured in a 30 °C incubator with a constant shaking at 200 r.p.m. The number of generations reached during the growth was calculated using the formula n = 3.3 logb B⁻¹, for which B represents the number of bacteria at the beginning of a time interval and b stands for the number of bacteria at the end of the time interval. The bacterial number in the culture was determined by a R. equi conversion factor which is defined as 1 × 10⁸ CFU mL⁻¹ at OD_{600 nm}. At approximately the 20th, 40th, 60th, 80th, and 100th generations, dilutions of bacteria were spread in duplicate on BHI plates with and without antibiotic. Bacterial colonies were counted after 48 h of incubation at 30 °C. The percentage of R. equi maintaining the integration vector pSET152, or the episomal plasmid pMV261-hyg, was determined by comparison of CFU arising on selective media to that of CFU obtained on plates lacking antibiotic.

Macrophage infection

The intracellular survival of R. equi strains in RAW 264 macrophages was investigated using an assay described previously (Hondalus & Mosser, 1994). Briefly, R. equi was grown at 37 °C to OD_{600 nm} 1.0. Macrophage monolayers in 24-well tissue culture-treated plates were infected at a multiplicity of infection of 10:1. Following washing to remove extracellular bacteria, and the addition of amikacin (final concentration 20 μg mL⁻¹) to prevent extracellular growth of noninternalized bacteria, macrophage monolayers were lysed at several times post infection using 400 μL of distilled water. Subsequently, lysates were serially diluted 10-fold with 1 × PBS, plated on BHI agar, and bacterial colonies arising from the plates were counted after incubation at 37 °C. Each time point was done in triplet for each strain analyzed.

Results

Plasmid integration in R. equi

To test if Streptomyces integrase could mediate site-specific integration in R. equi (Fig. 1), pSET152 harboring the integrase gene, an attP site, and an apramycin cassette was used to transform wild-type R. equi 103+ cells. The Mycobacterium spp.–E. coli shuttle vector pMV261-hyg, pSJ35 and pSET152-d, a derivative of pSET152, lacking the integrase gene and attP site, were similarly used and served as controls. Bacterial colonies were counted after incubation at 30 °C and the transformation efficiency was calculated (CFU μg⁻¹ of DNA). A transformation efficiency of 8.5 × 10⁴ CFU μg⁻¹ was determined for pSET152 which is comparable to the transformation efficiency of the episomal plasmid pMV261-hyg (Table 1). A 50-fold lower transformation rate was obtained when plasmid pSJ35 was transformed following the same procedure. This reduced transformation efficiency can be explained by the larger size of pSJ35 (11.2 kb) compared with pSET152 (5.7 kb) and pMV261-hyg (4.6 kb). Transformation with pSET152-d yielded no colonies, indicating the essentiality of the ΦC31 integrase and attP site for transformation.

Table 1.

Transformation efficiency of integration and episomal vectors

Plasmids	Transformation rate (CFU μg−1 of DNA)
pSET152	8.5 × 104
pMV261-hyg*	7.9 × 104
pSJ35*	1.6 × 103
pSET152-d	0
No DNA control	0

^*Plasmid containing an origin based on the pAL5000 plasmid of Mycobacterium fortuitum.

Identification of the R. equi integration site

In order to determine the integration site, pSET152 was rescued from the R. equi 103+ transformants. To do this, 12 independent R. equi transformants were selected and genomic DNA was purified from each cultured clone. Each genomic DNA was digested with PspXI, an enzyme that does not cleave within pSET152, and the fragments were self ligated. Each ligation reaction was then independently transformed into E. coli and plasmids were recovered. Plasmids isolated from E. coli transformants were predicted to contain the pSET152 plasmid and the two fragments of R. equi chromosomal DNA extending from the left and right flanking arms of the insertion site. Totally, 30 plasmid preparations representing at least two from each of the 12 independent E. coli transformations were obtained. When analyzed by HindIII restriction, four distinct restriction profiles were identified. Plasmids representing each of the distinct restriction profiles were submitted for sequencing analysis using primer Int-rh-F. One of the aforementioned four unique restriction digest patterns of the plasmids recovered from the R. equi pSET152 transformants was identical to that of pSET152, which implied that no chromosomal DNA was present in those particular plasmids. Sequencing analysis confirmed the latter and further revealed that the three other distinct patterns represented varied partial HindIII restriction surrounding a common integration site. The obtained sequences covered the left half of the attP site and the right half of the attB site in the right hybrid plus additional chromosomal sequence downstream (Fig. 1). Using the R. equi genome sequences accessible from the Sanger Institute website coupled with the NCBI blast search program, we determined the chromosomal location wherein the attB site resides. All integration events occurred at a common recombination site located within ORF06158 in the R. equi genome. This putative gene encodes a pirin-like protein. By searching the right half of the attB site against the whole R. equi genome, the left half of the attB sequence was deduced and the entire sequence was aligned with the published attB sites of other bacterial species phylogenetically related to R. equi (Fig. 2). A thymine dinucleotide present at the core of the attB sites was found to be conserved among the various Actinomycetales sequences aligned (Fig 2).

Fig. 2.

Sequences of the attachment sites for ΦC31 integrase in Rhodococcus equi and closely related species. The core sequence is TT, in bold at the center of each attB. Conserved nucleotide bases in each attB site flanking the core sequence are also in bold.

Stability of integrated vector

To be a useful molecular tool, plasmid integration must be stable. We examined the stability of the integrated vectors by serial passage of pSET152-integrant strains in BHI broth in the absence of antibiotic selection. For comparison, the stability of a strain (103+hyg) carrying the episomal plasmid pMV261-hyg was similarly analyzed. As shown in Fig. 3, pSET152 was stably maintained in the R. equi chromosome, providing apramycin resistance even after the cells had been cultured in nonselective media for up to 100 generations. In contrast, nearly half of the 103+hyg cells had lost the episomal plasmid and become hygromycin sensitive following the same number of replications.

Fig. 3.

Stability of integrated pSET152 and the episomal plasmid pMV261-hyg. Rhodococcus equi pSET152-integrated strain int1 and 103+hyg were cultured in BHI without the addition of antibiotics for up to 100 generations. At approximately the 20th, 40th, 60th, 80th, and 100th generation, the samples of cultures were diluted and plated on BHI plates with or without antibiotics and the number of colonies arising was counted. Bacterial numbers obtained from the BHI plates without antibiotic were used as controls to calculate the percentage of control CFU arising from growth on plates with antibiotic supplementation. This experiment is representative of three independent experiments.

Survival of the integrant R. equi strain inside macrophages

To ensure that the integration of pSET152 into the R. equi chromosome did not affect the ability of R. equi to grow intracellularly, we analyzed the growth of the integrant strains in in vitro grown RAW 264 macrophages. We used a monolayer lysis and plating method as a means to follow bacterial intracellular replication (Hondalus & Mosser, 1994). As shown in Fig. 4, both R. equi wild-type strain 103+ and the integrant strain int1 showed a significant 60–70-fold increase in bacterial number over 48 h of macrophage infection. The data indicated that chromosomal integration into ORF06158 does not adversely affect the ability of R. equi to grow in macrophages, an accepted parameter of virulence. To demonstrate the utility of the vector, we used pSET152 in a complementation analysis and showed that a wild-type copy of vapA cloned into pSET152 and integrated into the chromosome of a vapA mutant restored the ability of that mutant strain to grow in macrophages. As expected, the growth of the vapA-mutant strain in macrophages was retarded, but the complemented strain vapA-/vapA followed similar intracellular growth kinetics as that of wild-type or the int1-integrant strain.

Fig. 4.

ΦC31 integrase-based integration does not alter intracellular replication. The intracellular growth of various Rhodococcus equi strains in RAW264 macrophages was determined by lysis of infected macrophages and subsequent plating of lysates. Macrophages were infected with wild-type R. equi 103+, the pSET152-integrated strain (int1), a vapA mutant and a vapA mutant complemented with wild-type vapA (strain vapA-/vapA). At various time points post infection, macrophages were lysed and the lysate was serially diluted and plated. The number of CFU was determined after 48 h of incubation at 37 °C. Each time point represents the mean ± SD for three replicate lysates.

Discussion

Streptomyces ΦC31 belongs to the large serine recombinase family with a dozen of its members derived from bacterial phages. This integrase-based system has been shown to be able to perform efficient, unidirectional, and site-specific integration into the bacterial host chromosome. Because integration is independent of specific host-derived factors, this system has been exploited experimentally and used as the basis of integrating vectors, allowing long term and robust expression of foreign genes carried on the vector. The ΦC31 integrase-based vectors have been reported to integrate at single or multiple integration sites on host chromosomes. In our study, a single integration site on the R. equi chromosome was found and determined to be located within an ORF encoding a pirin-like protein. The full R. equi attB site was found to resemble closely that of Streptomyces and Kitasatospora species in which a pirin-like homolog was determined to be the dominant chromosomal integration site (Combes et al., 2002; Choi et al., 2004). The function of pirin, a conserved protein among various biological species, remains mostly uncharacterized. In Serratia marcescens, the pirin protein was reported to play a regulatory role in dictating the direction of pyruvate metabolism either towards the TCA cycle or the fermentation pathways (Soo et al., 2007). We showed that integration in the pirin-like gene did not affect the intracellular replication of R. equi within macrophages, a prerequisite for constructing a gene complementation system. In S. coelicolor, the ΦC31 integrase could also use pseudo-attB sites to facilitate recombination (Combes et al., 2002). In Mycobacterium smegmatis, a single attB site was determined to be located within gatA, a gene encoding a Glu-tRNAGln amidotransferase while in the same study, three distinct integration sites were reported to exist on the chromosome of Mycobacterium bovis (Murry et al., 2005). The finding of a single integration site in R. equi adds to the value of this system in this bacterium.

ΦC31-based integration provides another option for gene complementation, which eliminates potential plasmid maintenance issues arising with the bacteria trying to simultaneously replicate two episomal plasmids, i.e. virulence plasmid and a complementation plasmid for example. Furthermore, in situations wherein it is necessary to add two genes to a strain, one gene can be placed onto the chromosome through integration and the second on an episomal plasmid.

Our data showed that the pSET152 vector was more stable in R. equi as compared with the pAL5000-based episomal plasmid pMV261-hyg during nonselective culture. However, we did observe that a low level of rescued plasmid did not contain R. equi chromosomal sequences and were in fact parental pSET152 vectors. Considering that the integrative vector does not have a replication origin in R. equi, we reasoned that the free circular pSET152 represented excision products of plasmid originally integrated on the chromosome, as has been described previously (Murry et al., 2005). The underlying mechanism of the excision remains to be elucidated. It is, therefore, possible that the integrated plasmid could be lost over time. Nonetheless, for most studies, we believe that the ΦC31 integrase-based vectors will be a useful tool for gene function analysis in R. equi.