Regulation, Integrase-Dependent Excision, and Horizontal Transfer of Genomic Islands in Legionella pneumophila

Abstract

Legionella pneumophila is a Gram-negative freshwater agent which multiplies in specialized nutrient-rich vacuoles of amoebae. When replicating in human alveolar macrophages, Legionella can cause Legionnaires' disease. Recently, we identified a new type of conjugation/type IVA secretion system (T4ASS) in L. pneumophila Corby (named trb-tra). Analogous versions of trb-tra are localized on the genomic islands Trb-1 and Trb-2. Both can exist as an episomal circular form, and Trb-1 can be transferred horizontally to other Legionella strains by conjugation. In our current work, we discovered the importance of a site-specific integrase (Int-1, lpc2818) for the excision and conjugation process of Trb-1. Furthermore, we identified the genes lvrRABC (lpc2813 to lpc2816) to be involved in the regulation of Trb-1 excision. In addition, we demonstrated for the first time that a Legionella genomic island (LGI) of L. pneumophila Corby (LpcGI-2) encodes a functional type IV secretion system. The island can be transferred horizontally by conjugation and is integrated site specifically into the genome of the transconjugants. LpcGI-2 generates three different episomal forms. The predominant episomal form, form A, is generated integrase dependently (Lpc1833) and transferred by conjugation in a pilT-dependent manner. Therefore, the genomic islands Trb-1 and LpcGI-2 should be classified as integrative and conjugative elements (ICEs). Coculture studies of L. pneumophila wild-type and mutant strains revealed that the int-1 and lvrRABC genes (located on Trb-1) as well as lpc1833 and pilT (located on LpcGI-2) do not influence the in vivo fitness of L. pneumophila in Acanthamoeba castellanii.

INTRODUCTION

Legionella pneumophila is a Gram-negative bacterium found ubiquitously in freshwater habitats (1). It resides in biofilms or invades free-living protozoa such as Acanthamoeba castellanii (2, 3). Furthermore, Legionella is able to infect human lung alveolar macrophages. When inhaled with contaminated aerosols, Legionella can cause a severe, life-threatening pneumonia, Legionnaires' disease (4). L. pneumophila strain Corby (serogroup 1 [Sg1], monoclonal antibody type Knoxville) is a highly virulent human isolate (5). In aerosol-infected guinea pigs, the strain multiplies very rapidly within the lung, and the bacteria spread to the blood, liver, spleen, and kidney (5, 6). In host cells, Legionella avoids killing by the phagolysosomal pathway and establishes a specialized Legionella-containing vacuole (LCV) for replication (7, 8). When nutrition becomes limiting, the bacterium switches to the virulent phase, evades the vacuole, and destroys the host cell. Legionella manipulates the host cell by introducing virulence factors via specialized secretion systems. This is crucial for intracellular survival and the establishment of the replication-permissive LCV in amoebae and macrophages (9–12).

Type IV secretion systems (T4SSs) are needed for conjugation and for transport of proteins and nucleic acids into the host cell during infection (13–15). They are widespread and grouped into the IVA and IVB families (16). The Legionella type IVB secretion system (T4BSS) dot (defect in organelle trafficking)-icm (intracellular multiplication) is similar to the tra-trb system of IncI plasmids (16, 17). It enables intracellular multiplication by translocating effector proteins into the host cell (9, 11, 18–21). Type IVA secretion systems (T4ASSs) are similar to the tra system of IncN plasmids (17, 22). The T4ASS lvh is dispensable for intracellular growth of Legionella at 37°C but is involved in host cell infection at lower temperatures (23–25). In L. pneumophila Corby, further T4ASSs are encoded by the trb-tra genes on the genomic islands Trb-1 and Trb-2. Trb-1 and Trb-2 are integrated within the tRNA^Pro gene (lpc2778) and the tmRNA gene, respectively. Both islands exhibit an origin of transfer (oriT) region and are excised from the chromosome, forming episomal circles (ci's). The episomal form of Trb-1 can be transferred horizontally to another L. pneumophila strain by conjugation and is then integrated site specifically into the genome of the transconjugants (26). This finding may explain the observed horizontal transfer of chromosomal DNA in Legionella (27, 28). Recently, two further genomic islands, Trb-3 (L. pneumophila strain Lorraine) and Trb-4 (L. longbeachae NSW150), were identified. So far, nothing is known about excision of these elements (27).

In another classification system, the T4SSs are grouped into three distinct clusters, due to their homology: F-like (IncF, plasmid F), P-like (IncP, plasmid RP4), and I-like (IncI, plasmid R64) (14, 29). Juhas and colleagues described a further class of T4SSs, named the genomic island T4SS (GI-like) (30, 31). For L. pneumophila, two new GI-like islands (Legionella genomic island 1 [LGI-1] and LGI-2) were identified by genome sequence analysis of strain 130b (32). The authors speculate that these islands may be new T4SSs belonging to the integrative and conjugative elements (ICEs) and could contribute to mobilization of genomic islands in L. pneumophila. However, no experimental data were given.

The intention of our present work was to screen the L. pneumophila Corby genome for further genomic islands and to gain more insight into the horizontal transfer process. To further analyze the excision of genomic islands from the chromosome and to verify if LGI-2 is a functional ICE, we generated and analyzed integrase mutants of the genomic islands Trb-1 and LGI-2 of L. pneumophila Corby (LpcGI-2). We could verify our hypothesis (26) that a defined integrase located on the genomic island itself is indispensable for the site-specific excision of Trb-1 from the chromosome and expand this conclusion to the process of LpcGI-2 excision. Furthermore, we were able to demonstrate that the circularization of Trb-1 is regulated by the lvrRABC gene cluster, which we assumed because lvrR is predicted to be a transcriptional regulator and lvrC encodes a paralog of CsrA. CsrA is known to be involved in gene regulation of L. pneumophila (33).

MATERIALS AND METHODS

Bacterial strains, amoeba, and cell lines.

Experiments were done with L. pneumophila Sg1 strain Corby (5), L. pneumophila Sg1 strain Philadelphia I (ATCC 33152) (34), and L. oakridgensis ATCC 33761. The L. pneumophila Corby wild-type (WT) strains WT° and WT* were used as positive controls in mating experiments. L. pneumophila Corby WT° contains a kanamycin resistance (Km^r) cassette between the genes lpc2816 and lpc2817, and in L. pneumophila Corby WT*, a Km resistance cassette was introduced between the genes lpc1856 and lpc1857. Mutant strains of L. pneumophila Corby used in this study contained the mutations Δint-1, ΔlvrRABC, ΔlvrR, ΔpilT, Δlpc1833, Δlpc1884, and Δlpc2123. All strains are listed in Table 1. Escherichia coli strain DH5α was used as the host for recombinant plasmids (35). Acanthamoeba castellanii ATCC 30010 (36) and the U937 cell line (ATCC CRL-1593.2) were used for infection assays.

Table 1.

Legionella strains and plasmids used in this study

Strain or plasmid	Characteristics	Reference or source
Strains
L. pneumophila Corby	WT	Jepras et al. (5)
L. pneumophila Philadelphia I JR32	Restriction-deficient strain of L. pneumophila Philadelphia I (Smr)	Marra and Shuman (34)
L. oakridgensis	ATCC 33761	C. Lück, Dresden, Germany
L. pneumophila Corby WT°	Kmr cassette between lpc2816 and lpc2817	This study
L. pneumophila Corby WT*	Kmr cassette between lpc1856 and lpc1857	This study
L. pneumophila Corby Δint-1	lpc2818::Kmr	This study
L. pneumophila Corby ΔlvrR	lpc2816::Kmr	This study
L. pneumophila Corby ΔlvrRABC	lpc2813tolpc2816::Kmr	This study
L. pneumophila Corby Δlpc1833	lpc1833::Kmr	This study
L. pneumophila Corby Δlpc1884	lpc1884::Kmr	This study
L. pneumophila Corby Δlpc2123	lpc2123::Kmr	This study
L. pneumophila Corby ΔpilT	lpc1876::Kmr	This study
Plasmidsa
pML2	2,990-bp PCR fragment (primers Trb1-Km-U/Trb1-Km-R)	This study
pML3	pML2 after inverse PCR (primers Trb1-Km-MU/Trb1-Km-MR)	This study
pML4	pML3 with Kmr cassette cloned into XbaI site between lpc2816 and lpc2817	This study
pML9	3,208-bp PCR fragment (primers Int-U/Int-R) in pGEM-T Easy	This study
pML11	pML9 after inverse PCR (primers Int-MU/Int-MR)	This study
pML12	pML11 with Kmr cassette cloned into XbaI site instead of lpc2818	This study
pML14	3,418-bp PCR fragment (primers pilT-1876-U/pilT-1876-R)	This study
pML15	pML14 after inverse PCR (primers pilT-1876-MU/pilT-1876-MR)	This study
pML16	pML15 with Kmr cassette cloned into XbaI site instead of lpc1876	This study
pML17	2,816-bp PCR fragment (primers Int-1884-U/Int-1884-R)	This study
pML18	pML17 after inverse PCR (primers Int-1884-MU/Int-1884-MR)	This study
pML19	pML18 with Kmr cassette cloned into XbaI site instead of lpc1884	This study
pML20	2,198-bp PCR fragment (primers LpcGI-2-Km-U/LpcGI-2-Km-R)	This study
pML21	pML20 after inverse PCR (primers LpcGI-2-Km-MU/LpcGI-2-Km-MR)	This study
pML22B	pML21 with Kmr cassette cloned into XbaI site between lpc1856 and lpc2857	This study
pML23	3,103-bp PCR fragment (primers Int-1833-U/Int-1833-R)	This study
pML24	pML23 after inverse PCR (primers Int-1833-MU/Int-1833-MR)	This study
pML25A	pML24 with Kmr cassette cloned into XbaI site instead of lpc1833	This study
pML26	2,377-bp PCR fragment (primers Int-2123-U/Int-2123-R)	This study
pML27	pML26 after inverse PCR (primers Int-2123-MU/Int-2123-MR)	This study
pML28	pML27 with Kmr cassette cloned into XbaI site instead of lpc2123	This study
pML54	2,530-bp PCR fragment (primers LvrR-2816-U/lvrR-2816-R)	This study
pML56	pML54 after inverse PCR (primers LvrR-2816-MU/LvrR-2816-MR)	This study
pML58	pML56 with Kmr cassette cloned into XbaI site instead of lpc2816	This study
pVH7	1,130-bp PCR fragment (primers LvrRABC-1U/LvrRABC-1R)	This study
pVH8	1,090-bp PCR fragment (primers LvrRABC-2U/LvrRABC-2R)	This study
pVH9	Insert of pVH8 cloned into XbaI and XhoI site in pVH7	This study
pVH10	pVH9 with Kmr cassette cloned into XbaI site instead of lpc2816tolpc2813	This study

^aFor all plasmids, pGEM-T Easy (Promega) was used as the vector.

Media and growth conditions.

Legionella was grown in AYE medium {1% yeast extract supplemented with 1% ACES [N-(2-acetamido)-2-aminoethanesulfonic acid], 0.025% ferric PP_i, 0.04% l-cysteine} or on buffered charcoal-yeast extract (BCYE) agar at 37°C. Antibiotics used for L. pneumophila were kanamycin (Km) at a concentration of 12.5 mg ml⁻¹ and streptomycin (Sm) at a concentration of 50 mg ml⁻¹. Bacterial growth in broth was monitored by determining the optical density at 600 nm (OD₆₀₀) with a Thermo Scientific GENESYS 10 Bio spectrophotometer (VWR, Darmstadt, Germany). Growth phases were defined as follows: an OD₆₀₀ of ∼1.0 corresponded to the exponential growth phase (E), additional growth for 8 h and an OD₆₀₀ of ∼1.7 corresponded to the late exponential growth phase (LE), additional growth for 4 h and an OD₆₀₀ of ∼1.8 corresponded to the postexponential growth phase (PE), and additional growth for 8 h and an OD₆₀₀ of ∼2.0 corresponded to the stationary growth phase (S). E. coli was cultivated in Luria-Bertani (LB) medium or on LB agar. The antibiotics used for E. coli were ampicillin (Ap) at a concentration of 100 mg ml⁻¹ and Km at a concentrations of 40 mg ml⁻¹. Acanthamoeba castellanii ATCC 30010 was cultured in PYG 712 medium [2% proteose peptone, 0.1% yeast extract, 0.1 M glucose, 4 mM MgSO₄, 0.4 M CaCl₂, 0.1% sodium citrate dihydrate, 0.05 mM Fe(NH₄)₂(SO₄)₂ · 6 H₂O, 2.5 mM NaH₂PO₄, 2.5 mM K₂HPO₄] at 20°C. U937 cells were cultured in RPMI with 10% fetal calf serum (FCS) at 37°C in 5% CO₂.

DNA techniques and sequence analysis.

Genomic DNA for PCR was prepared with a Generation capture column kit (Qiagen, Hilden, Germany), and genomic DNA for real-time PCR was prepared with a DNeasy blood and tissue kit (Qiagen, Hilden, Germany). The preparation of plasmid DNA was done with an Invisorb spin plasmid mini-two kit (Stratec, Berlin, Germany). Plasmid DNA was introduced into E. coli by electroporation with a gene pulser (Bio-Rad, Munich, Germany) at 1.7 kV, 100 Ω, and 25 μF. Both strands of plasmid DNA or a PCR product were sequenced with infrared dye-labeled primers by using an automated DNA sequencer (LI-COR-DNA4000; MWG-Biotech, Ebersberg, Germany). Oligonucleotides were obtained from Eurofins MWG Operon (Ebersberg, Germany). Restriction enzymes were purchased from New England BioLabs (Frankfurt am Main, Germany).

Legionella mutant construction.

The int-1 (lpc2818) gene-encoding DNA region was amplified by PCR with the primers Int-U and Int-R and cloned into the pGEM-T Easy vector. The resulting plasmid (pML9) was used as a template in an inverse PCR, using KAPAHiFi DNA polymerase (Peqlab, Erlangen, Germany) for reaction. The PCR product was amplified with the primer pair Int-MU/Int-MR, with one of the primers containing an XbaI restriction site and enabling religation. The resulting plasmid (pML11) and a Km^r cassette were restricted with XbaI and ligated. To generate the mutant, the insert of the plasmid (pML12) containing the Km^r cassette and the flanking DNA sequences was amplified with the primer pair Int-U/Int-R. To generate other mutants of this study, plasmids containing the Km^r cassette and the flanking sequences were used as follows: pML56 for the ΔlvrR mutant, pVH10 for the ΔlvrRABC mutant, pML25A for the Δlpc1833 mutant, pML19 for the Δlpc1884 mutant, pML27 for the Δlpc2123 mutant, and pML16 for the ΔpilT mutant. The following plasmids were used to clone a Km^r cassette in genomic islands: pML4 for Trb-1 and pML22B for LpcGI-2. Natural transformation of L. pneumophila Corby was done as described before with modifications (37). In brief, 2 ml of a culture exponentially grown overnight at 30°C was transferred to a plastic tube and incubated with the PCR product for 3 days at 30°C without agitation. Subsequently, bacteria were grown on antibiotic selective medium for 4 more days at 37°C. All mutant strains of L. pneumophila Corby generated in this study were produced by a method analogous to the method used for the Δint-1 mutant. Mutant strains and plasmids are given in Table 1; specific primers used for mutant construction are listed in Table S1 in the supplemental material.

PCR analysis.

PCR was carried out using a TRIO-Thermoblock thermocycler (Biometra, Göttingen, Germany) and HotStar Taq DNA polymerase (Qiagen, Hilden, Germany). The characterization of Trb-1 was done as described before (26). PCR analysis of the genomic islands LpcGI-2 and LpcGI-1 was done with specific primer pairs analogous to those used for Trb-1. The tRNA regions of LpcGI-2 (tRNA^Met) and LpcGI-1 (tRNA^Thr) were amplified with the primer pairs LpcGI-2-1U/LpcGI-2-4R (referred to as primers 1/4), LpcGI-2-2R/LpcGI-2-6U (primers 2/6), and LpcGI-2-4R/LpcGI-2-6U (primers 4/6). The amplification of the different circular forms of the genomic islands was done with the primer pairs LpcGI-2-2R/LpcGI-2-3U (primers 2/3), LpcGI-2-1U/LpcGI-2-5R (primers 1/5), and LpcGI-2-3U/LpcGI-2-5R (primers 3/5). Points of integration of LpcGI-2 and LpcGI-1 into the chromosome of L. pneumophila Corby were shown by the primer combinations 1/2, 3/4, and 5/6. The characterization of LpcGI-Asn and LpcGI-Phe was done by a method analogous to the method used for Trb-1. The specific primer pairs 2/3, 1/4, 1/2, and 3/4 were used to demonstrate the circular form of the genomic island, to amplify the equal tRNA-encoding region, and for the amplification of the 5′and 3′ regions of the point of integration of LpcGI-Asn or LpcGI-Phe into the genome, respectively. In general, initial denaturation was performed at 95°C for 15 min and final extension was performed at 72°C for 10 min. The cycling conditions (35 cycles) were 94°C for 1 min, 60°C for 45 s, and 72°C for 2 min. All primers specific for the genomic islands are listed in Table S1 in the supplemental material and illustrated in Fig. 2, 3, and 5.

Fig 2.

Genomic islands LpcGI-1, LpcGI-Asn, and LpcGI-Phe of L. pneumophila Corby. (A) Genomic island LpcGI-1 is integrated within the tRNA^Thr gene. Episomal forms of rings A and B were detected, and thereby, chromosomal DNA regions were generated with primers 2/3 and 1/4 and primers 1/5 and 2/6, respectively. No amplification product was detected by using primers 3/5, specific for the episomal form of LpcGI-1-AB. The A-B fusion point (attR-1) was detected with primers 1/2, and the 5′ and 3′ sites of integrated LpcGI-1 (5′i and 3′i) were detected with primers 3/4 and 5/6, respectively. (B) LpcGI-Asn is integrated within the tRNA^Asn gene (3′i, primers 1/2; 5′i, primers 3/4) and can also exist in an episomal form (ci, primers 2/3). (C) LpcGI-Phe is integrated within the tRNA^Phe gene (3′i, primers 1/2; 5′i, primers 3/4), but no episomal form could be detected (ci, primers 2/3, and tRNA, primers 1/4). Numbers with arrows, specific primers and their orientations. PCR was done for 35 or 40 (*) amplification cycles. All primers are specific for the respective islands and are named LpcGI-1, LpcGI-Asn, and LpcGI-Phe plus the following suffixes: 1U (primer 1), 2R (primer 2), 3U (primer 3), 4R (primer 4), 5R (primer 5), and 6U (primer 6) (see Table S1 in the supplemental material). Abbreviations: ci, episomal circular form; tRNA, chromosomal tRNA without integrated island; attR, chromosomal LpcGI-1 without B; i, integrated form.

Fig 3.

Circularization of LpcGI-2 of L. pneumophila Corby. (A) Mechanism for the formation of episomal rings A, B, and AB of LpcGI-2. The chromosomal form of LpcGI-2 (blue double-headed arrows) is integrated within the tRNA^Met gene (gray arrow) and bordered by the attL and attR-2 sites (black arrows). Numbers with arrows, specific primers and their orientations. (B) Episomal forms of rings A, B, and AB were detected and chromosomal DNA regions were thereby generated with primers 2/3 and 1/4, primers 1/5 and 2/6, and primers 3/5 and 4/6, respectively; the A-B fusion point (attR-1) was detected with primers 1/2, and the 5′ and 3′ sites of integrated LpcGI-2 (5′i and 3′i) were detected with primers 3/4 and 5/6, respectively. PCR was done with 35 or 40 (*) amplification cycles. Primers: 1, LpcGI-2-1U; 2, LpcGI-2-2R; 3, LpcGI-2-3U; 4, LpcGI-2-4R; 5, LpcGI-2-5R; and 6, LpcGI-2-6U. (C) Nucleotide sequence of the attP, attP′, and attP″ sites (black arrows) of episomal LpcGI-2-A, -B, and -AB, respectively. Putative IHF-binding sites (gray boxes) and identified direct repeats (NTTTN, where N is any nucleotide; underlined) and indirect repeats (green arrows) are indicated. (D) Nucleotide sequences of the attL, attR-1, and attR-2 sites (black arrows) of chromosomal LpcGI-2. The tRNA^Met gene (gray arrow) and sequence variations between attL and attR sites (marked in red) are indicated. LpcGI-2 consists of region A (dark blue) and region B (light blue). Chromosomal DNA is shown in black. Highlighted gray boxes numbered 1 to 5, putative IHF-binding sites (WATCAANNNNTTR, where W is dA or dT, R is dA or dG, and N is any nucleotide).

Fig 5.

Genomic island Trb-1 of L. pneumophila Corby. (A) The chromosomal form of Trb-1 (Trb-1_i) is integrated in the tRNA^Pro gene and bordered by the attL and attR sites. The organization of the int-1 and lvrRABC genes is given above. After excision, Trb-1 is present as an episomal circular form (Trb-1_ci) and an intact tRNA^Pro gene remains in the genome. Numbers and arrows, the specific primers and their orientations, respectively (26). (B) Determination of the int-1-dependent excision mechanism via PCR. Integrase-1 is essential for ring formation, because the episomal island (primers 2/3) was detectable in the wild type but not in the Δint-1 mutant strain (Δint-1). An intact tRNA^Pro gene (primers 1/4) was detected in the wild type but not in the Δint-1 mutant. The chromosomal form of Trb-1 was present in both strains (primers 1/2 and 3/4). The episomal form of the genomic island Trb-2 (Trb-2_ci) was detected in the wild type and the Δint-1 mutant strains (primers 6/7). (C) Determination of the circular forms of Trb-1 in ΔlvrRABC and ΔlvrR mutants. Trb-1_ci was analyzed in vitro in the wild type as well as in the ΔlvrRABC and ΔlvrR mutant strains and in vivo (after 20 h of intracellular growth) in the wild type and in the ΔlvrRABC mutant (primers 2/3). The amount of Trb-1_ci and intact tRNA^Pro gene (primers 2/3 and 1/4) was upregulated in both mutants, but the integrated form Trb-1_i was still detectable (primers 1/2 and 3/4). Primers: 1, trb-1; 2, trb-2; 3, trb-3; 4, trb-4; 6, trb-6; 7, trb-7. Abbreviations: attP, episomal integration sites; attL and attR, chromosome-genomic island junctions. Results of PCR analysis were confirmed by using two independently generated mutant strains. Lanes M, molecular markers.

L. pneumophila mating experiments.

Recipient (L. pneumophila JR32 Sm^r or L. oakridgensis ATCC 33761) and donor (L. pneumophila Corby WT°, WT*, Δint-1, ΔlvrRABC, ΔlvrR, ΔpilT, Δlpc1833, Δlpc1884, or Δlpc2123) strains were grown in AYE medium at 37°C. One milliliter of the donor strain (exponential phase) was mixed with 2 ml of the recipient strain (stationary phase) strain. Matings were performed in triplicate by incubating the mixed bacterial cultures for 24 h at 30°C on BCYE agar plates with or without the presence of DNase (1 μg/μl). After mating, L. pneumophila transconjugants were selected on BCYE plates containing kanamycin and streptomycin. For the selection of L. oakridgensis transconjugants, BCYE plates with kanamycin but without additional l-cysteine were used. In contrast to L. oakridgensis, L. pneumophila is not able to grow on these agar plates (38). Dilutions of transconjugants were plated on agar plates, and the number of transconjugants was determined by determination of the numbers of CFU. Conjugation frequencies were calculated as the number of transconjugants divided by the number of donor cells.

RNA techniques and cDNA synthesis.

For RNA preparation, an overnight culture was diluted in AYE medium to an OD₆₀₀ of ∼0.3 and cultured at 37°C to the favored growth phase. Total RNA was extracted from exponential and postexponential growth phase using a High Pure RNA isolation kit (Roche, Mannheim, Germany). Purified RNA was incubated with 100 U DNase I per ml (Qiagen, Hilden, Germany) for 30 min at room temperature. After DNase treatment, RNA was repurified with an RNeasy minikit (Qiagen, Hilden, Germany). PCR with primers specific for gyrA was done to analyze the purified RNA for the absence of genomic DNA. Synthesis of cDNA was performed with a SuperScript VILO cDNA synthesis kit (Invitrogen, Darmstadt, Germany) and started with 50 ng μl⁻¹ of total RNA. Synthesis was done according to the instructions of the manufacturer.

Real-time PCR.

Real-time quantitative PCR (qPCR) was performed using an Mx3000P thermal cycler (Stratagene) and an Express SYBR GreenER qPCR SuperMix universal kit (Invitrogen, Darmstadt, Germany) according to the instructions of the manufacturers. A standard curve was used to quantify the amount of target present in unknown samples. For a standard curve, 10 μl from each probe was mixed and diluted from 10⁷ to 10¹ in diethyl pyrocarbonate water. The primer pair RT-gyrA-U/RT-gyrA-R was used for the standard curve. Genomic DNA isolated from the exponential and stationary phases was used to determine the amount of the episomal forms of the genomic islands Trb-1 (DNA concentration, 35 ng μl⁻¹) and LpcGI-2 (DNA concentration, 50 ng μl⁻¹). Episomal forms of LpcGI-2 were amplified with the primer pairs 2/3, 1/5, and 3/5. The episomal form of Trb-1 was shown by the primer pair RT-trb-2R/RT-trb-3U. To determine the relative amounts of amplicons of the episomal forms, the chromosomal gene flaA was used as an internal standard. All specific primers for qPCR are listed in Table S1 in the supplemental material. Data analysis and calculation of quantity (gene copies) were done with the Stratagene MxPro software.

Intracellular replication in A. castellanii.

Infection assays of L. pneumophila Corby and the mutant strains in A. castellanii were performed as described previously (39). In brief, 3-day-old cultures of A. castellanii were washed in AC buffer (PYG 712 medium without proteose peptone, glucose, and yeast extract), adjusted to 1 × 10⁵ cells per ml, and incubated in 24-well plates for 2 h at 37°C in 5% CO₂. Stationary-phase Legionella bacteria grown on BCYE agar were diluted in AC buffer and mixed with A. castellanii at a multiplicity of infection (MOI) of 0.01. After invasion for 2 h at 37°C, the A. castellanii layer was washed twice with AC buffer. To determine the numbers of CFU of L. pneumophila, different dilutions of the Legionella-amoeba mix were plated on BCYE agar. Each infection was carried out in duplicate wells and was done at least three times.

Infection/survival assay in A. castellanii.

To study the intracellular multiplication and survival in A. castellanii, a protocol was used as described recently (40). After 3 days of infection, A. castellanii cells were resuspended, 100-μl aliquots were lysed, and several dilutions were plated on BCYE agar to determine the number of CFU. To study the replication rates, the infection was repeated weekly with fresh amoebae. Afterwards, the remaining solution was incubated at 37°C in 5% CO₂ for a further 4 days, diluted in AC buffer (1:1,000), and plated on BCYE agar. Of this dilution, 1 ml was used to infect fresh amoeba cultures as described above. Four rounds of infection were performed. Each infection was carried out in duplicate wells.

Infection/survival assay in competition.

For intracellular multiplication in competition, the infection protocol was carried out as described recently (40). The procedure is similar to the assay described above, except that a 1:1 mixture of L. pneumophila wild type and one of its isogenic mutant strains was used for infection. The number of CFU was determined by plating serial dilutions on BCYE with and without kanamycin. The number of wild-type bacteria was calculated by subtracting the number of CFU on BCYE agar with kanamycin from the number of CFU on BCYE plates without kanamycin. Each infection was carried out in duplicate wells.

Intracellular multiplication in human macrophages.

Transformation and infection of U937 cells were done as previously described with modifications (41–43). U937 cells were adjusted to 3 × 10⁵ cells/ml and transferred to 100 ml fresh RPMI medium containing 10% FCS. For differentiation into macrophage-like cells, phorbol-12-myristate-13-acetate (PMA; stock concentration, 1 mg/ml in double-distilled H₂O [P-8139; Sigma-Aldrich]) was added at a concentration of 1:20,000, and cells were incubated for 36 h at 37°C in 5% CO₂. Afterwards, the supernatant was discarded and cells were washed once with 10 ml 0.2% EDTA in phosphate-buffered saline (PBS). Cells were removed from the flask bottom with RPMI–10% FCS, transferred to 50-ml tubes, and centrifuged at 800 × g for 10 min. To determine the cell number, 100 μl of cell solution was treated with 100 μl trypan blue. Viable cells were counted in a Neubauer counting chamber, and the concentration of the cell solution was adjusted to 1 × 10⁶ cells/ml with RPMI–10% FCS. To each well of a 24-well plate, 1 ml of the cell suspension was added and the mixture was incubated for 2 h at 37°C in 5% CO₂ to allow adhesion. Stationary-phase Legionella grown on BCYE agar was diluted in PBS and added to the macrophage cells. Infection was done with an MOI of 0.01 (time zero) for 2 h at 37°C in 5% CO₂. Thereafter, infected cells were washed 3 times with RPMI and covered with 1 ml RPMI–10% FCS. To determine the number of CFU, coincubations of U937 cells and legionellae were lysed by addition of 10 μl 10% saponin (S4521; Sigma-Aldrich) for 5 min, and different dilutions were plated on BCYE agar. Each infection was carried out in duplicate wells and was done at least three times.

RESULTS

Genomic islands of L. pneumophila Corby.

There are six genomic islands (LpcGI-1 and -2, LpcGI-Asn, LpcGI-Phe, Trb-1 and -2) present in the genome sequence of L. pneumophila Corby (Fig. 1 and 2) (26). The organization of the LpcGI-1 and -2 islands is shown in Fig. 1. The genomic island LpcGI-2 of L. pneumophila Corby exhibits a putative T4-like secretion system. It seems to belong to a class of T4SSs named genomic island-associated T4SS (GI-like), and this class of T4SSs was recently identified within the genome sequence of L. pneumophila 130b (30, 32). LpcGI-2 shares similarity with the genomic island LGI-2 of L. pneumophila strains 130b and Paris and LpcGI-1 (Fig. 1; see below for LpcGI-1). The island LpcGI-2 (64,401 bp, 39% G+C content, lpc1833 to lpc1888 and lpc2136 to lpc2121) is integrated within the tRNA^Met gene lpc1832 (Fig. 1). In contrast to Trb-1, the island exhibits two attR sites (attR-1 and -2; Fig. 1 and 3). The putative T4SS is encoded by the region lpc1857 to lpc1880 with a DNA identity of approximately 87% to the respective region of strain 130b (lpw_21631 to lpw_21861) (Fig. 1, green box, region I). However, in strain 130b, LGI-2 (LpwGI-2) is integrated within the tRNA^Arg gene, and therefore, it is not surprising that the site-specific integrase Lpc1833 of LpcGI-2 is only 59% identical to the respective putative integrase (lpw_21181) of LpwGI-2. In addition, LpcGI-2 and LpwGI-2 exhibit a divergent genomic organization, predominantly within regions II and III, whereas region IV is not present in LpwGI-2 (Fig. 1).

Fig 1.

Genetic organization of the genomic islands LpcGI-2, LppGI-2, LpwGI-2, and LpcGI-1 of L. pneumophila (Lp) strains Corby (Lpc), Paris (Lpp), and 130b (Lpw) encoding LGI-like T4SSs. LpcGI-2 and LppGI-2 are integrated within the tRNA^Met gene, whereas LpwGI-2 and LpcGI-1 are integrated within the tRNA^Arg and tRNA^Thr genes, respectively. The gene numbers are given above the genes, which are indicated by arrows. DNA regions encoding clustered homologous proteins are boxed in the same color. The colors of the genes indicate their degree of homology to the genes on LpcGI-2. Subregions (I to IV) of the islands are given below the genes.

LpcGI-2 seems to contain all genes necessary for a functional T4ASS/conjugation system and additional genes encoding putative regulatory proteins (lvrRABC, lpc1838, lpc2122), putative persistence or fitness factors (helABC, cadA, proline/betaine transport protein; Fig. 1, region III), metabolic enzymes, transposases (lpc2127, lpc2136, lpc1856), and three putative integrases (lpc1833, lpc1884, lpc2123) (Table 2 and Fig. 1). Within region IV of LpcGI-2, we identified a gene encoding a homolog of traK and close to its 5′ site a region containing a partial oriT (Fig. 1). This region exhibits a putative TraI- and TraK-binding site, including the putative nick site, but without the inverted repeat responsible for TraJ binding (data not shown). The presence of a partial oriT region indicates that this island may be transferable by conjugation (see below).

Table 2.

Structure of LpcGI-2 (65,401 bp) of L. pneumophila Corby

Gene	Name	Putative function or similar protein
	Repeat	attL site
lpc1833	int	Integrase, similar to lpp2312
lpc1834		Acetyltransferase, similar to lpp2313
lpc1835		Proline/betaine transport protein-like protein, similar to lpp2314
lpc1836		Acetyltransferase, similar to lpw21221
lpc1837		Lipolytic enzyme, similar to lp12_2062
lpc1838		Transcription regulator protein, response regulator containing CheY-like receiver domain and a helix-turn-helix DNA-binding domain, similar to lp12_2063
lpc1839		Similar to lpp2318
lpc1840		Similar to lpg1012
lpc1841		Similar to lpg1011
lpc1842		Putative cadmium efflux ATPase, similar to lp12_2067
lpc1843		Cadmium efflux ATPase, similar to lpg1010
lpc1844		Similar to lpc2269
lpc1845		Similar to lpc2267
lpc1846	cadA	Cadmium-translocating P-type ATPase CadA, similar to lp12_2070
lpc1847	helA	Cobalt/zinc/cadmium efflux RNDa transporter permease HelA, similar to lp12_2071
lpc1848	helB	Cation efflux system HelB, similar to lp12_2072
lpc1849	helC	Cobalt/zinc/cadmium efflux RND transporter outer membrane protein, similar to lp12_2073
lpc1850		Reverse transcriptase, similar to MEALZ2163 of Methylomicrobium alcaliphilum
lpc1851		Similar to serine/threonine protein kinase/putative ATPase of Moorea product 3L
lpc1852		Similar to llo0765
lpc1853		Similar to llo1727
lpc1854		Similar to hypothetical protein NH8B_0948 of Pseudogulbenkiania sp. strain NH8B
lpc1855		Similar to retron-type reverse transcriptase of Bacteroides sp. strain 1_1_14
lpc1856		Transposase IS4, similar to lpl0192
lpc1857	lvrR	Phage repressor, similar to lp12_2074
lpc1858	lvrA	Legionella vir region protein LvrA, similar to lp12_2075
lpc1859	lvrB	Legionella vir region protein LvrB, similar to lp12_2076
lpc1860	lvrC	CsrA paralog, similar to lp12_2077
lpc1861	pilL	Putative exported protein, similar to PilL of Vibrio tubiashii ATCC 19109
lpc1862		Similar to lp12_2079
lpc1863		Similar to lp12_2080, TIGR03759, integrating conjugative element protein, PFL_4693 family
lpc1864		Similar to lp12_2081
lpc1865		Similar to lp12_2082
lpc1866		Similar to lp12_2083
lpc1867		Similar to lpp2385
lpc1868		Similar to lp12_2085
lpc1869		Similar to lpp2387
lpc1870		Similar to lpp2388, integrating conjugative element protein of Gallibacterium anatis UMN179
lpc1871		Similar to lpp2389
lpc1872		Exported membrane protein, similar to lp12_2089
lpc1873		Similar to lpp2391
lpc1874	virB4	Type IV secretory protein VirB4 component, similar to lpp2392
lpc1875		Similar to lpp2393, TraU superfamily protein
lpc1876	pilT	Membrane protein, Tfp pilus assembly, pilus retraction ATPase PilT, similar to lp12_2093
lpc1877	traG	Membrane protein, TraG-like protein, N-terminal region, similar to lpp2395
lpc1878		Similar to lp12_2095
lpc1879		Similar to lp12_2096
lpc1880	traD	Conjugative coupling factor TraD, similar to lp12_2097
lpc1881		Similar to lp12_2098
lpc1882		Similar to lpp2400
lpc1883		Similar to lp12_2101
lpc1884	int	Putative integrase, similar to lp12_2102
lpc1885		Similar to lp12_2103
lpc1886		Antirestriction protein, similar to lp12_2104
lpc1887		Similar to lpp2408
lpc1888		Similar to lpp2409
	Repeat	attR-1 site
lpc2136		TnpA transposase, similar to ldg6041
lpc2135		Hypothetical protein
lpc2134		Similar to GM18_2913 of Geobactersp. strain M18
lpc2133		Similar to lpw25661
lpc2132	traK	TraK protein, similar to lpp0067
lpc2131		Similar to lpp2428
lpc2130		Similar to lpw25801, putative Dot/Icm T4SS effector
lpc2129		Similar to lpp2419
lpc2128		Similar to lpc0225, SidC (llo_p0059) homolog Legionella longbeachae NSW150
lpc2127		Transposase (IS652), similar to lpp2402
lpc2126		Similar to llo1617
lpc2125		Similar to lpc2174,lpa03424
lpc2124		Similar to lpa03424
lpc2123	int	Putative prophage CP4-6 integrase, similar to lpa03425
lpc2122		Transcriptional regulator, LysR family, similar to lpa03426
lpc2121		Similar to lpa03427
	Repeat	attR-2 site

^aRND, resistance-nodulation-cell division.

The genomic island LpcGI-1 (120,190 bp, 40.5% G+C content) exhibits a region (lpc2190 to lpc2314) encoding another putative LGI T4-like secretion system (LpcGI-1 [30, 32]), two attR sites, and three integrases and is inserted into the tRNA^Thr gene (Fig. 1 and Fig. 2A, primers 3/4 and 5/6). Like LpcGI-2, LpcGI-1 exhibits an lvrRABC region, several genes encoding putative persistence and fitness factors, metabolic proteins, and resistance factors (Fig. 1). The genomic islands LpcGI-Asn (6,066 bp, 37.4% G+C content) and LpcGI-Phe (11,555 bp, 37.3% G+C content), depicted in Fig. 2B and C, do not encode T4SSs (data not shown). LpcGI-Asn (lpc0085 to lpc0092) is integrated within the tRNA^Asn gene and exhibits one attR site, only the lvrA paralog of the lvrRABC region, and a putative integrase gene, lpc0085 (Fig. 2B, primers 3/4 and 1/2). LpcGI-Phe (lpc1383 to lpc1395) is integrated within the tRNA^Phe gene (Fig. 2C, primers 3/4 and 1/2) and exhibits one attR site, several putative transposases, and two putative integrase genes but no lvr paralog (data not shown).

We analyzed L. pneumophila Corby for the presence of episomal forms of LpcGI-1, LpcGI-2, LpcGI-Asn, and LpcGI-Phe. Therefore, we employed PCR analysis to investigate whether an episomal ring is generated. For LpcGI-1, two episomal forms (A and B) were detectable, indicating that this island is excised from the genome of L. pneumophila Corby (Fig. 2A, primers 2/3 and 1/5). In contrast to LpcGI-2 (see below), the AB form, exhibiting the complete genomic island, does not seem to be generated (primers 3/5). However, the PCR product obtained using primers 4/6 indicated that both islands could be in the episomal state at the same time (Fig. 2A). While we were able to identify an episomal form of LpcGI-Asn (Fig. 2B, primers 2/3), no episomal form could be detected for LpcGI-Phe (Fig. 2C, primers 2/3). PCR with primers 1/4 revealed no PCR product, corroborating the finding that LpcGI-Phe is not able to be excised from the genome. For LpcGI-2, PCR analysis for the detection of episomal forms revealed the presence of three different episomal forms of LpcGI-2, the medium-sized A ring (primers 2/3), the small-sized B ring (primers 1/5), and the complete AB ring (primers 3/5) (Fig. 3A and B), thereby generating a chromosomal region without part A (primers 1/4), part B (primers 2/6), and parts A and B (primers 4/6), respectively. This indicates that the island is excised without leaving a copy within the genome. The generation of episomal forms of genomic islands was then analyzed for LpcGI-2 and Trb-1 in detail.

Analysis of mechanism of LpcGI-2 excision.

We chose LpcGI-2 to further analyze the excision of the island and to investigate if the island-associated new T4SS encodes a functional conjugation system. First, we amplified the attP sites of the three different episomal forms of LpcGI-2 by PCR and determined the DNA sequences (Fig. 3C). Sequence analysis of the attP sites revealed a site-specific excision between attL and attR-1 or attL and attR-2, generating the episomal forms A and AB, respectively (Fig. 3D). It is obvious that the attP sites of the episomal forms exhibit the typical format of mobile elements using an integrase-dependent excision/integration mechanism (5, 10, 44). The attP site is >200 bp long and exhibits arm sites (short repeats; Fig. 3C, underlined), a core site (crossover segment attP), and two putative integration host factor (IHF)-binding sites (Fig. 3C, marked in gray). We numbered the different IHF-binding sites (sites 1 to 5) as they appear in the integrated form within LpcGI-2 (Fig. 3D). IHF-binding site 5 is located within the chromosomal DNA and therefore is not part of LpcGI-2. A similar structure, but without IHF-binding sites and only one attR site, was identified for the attP site of Trb-1 (26).

To investigate which of the three integrases present on LpcGI-2 is responsible for the excision of the island, we then replaced the integrase genes lpc1833, lpc1884, and lpc2123 with a Km^r cassette. The Δlpc1833, Δlpc1884, and Δlpc2123 mutants obtained were verified by PCR analysis (data not shown). Next, we demonstrated that growth of the mutant strains was similar to that of the wild type in AYE medium (data not shown). We then performed qPCR analysis to quantify the number of episomal forms of LpcGI-2-A, LpcGI-2-B, and LpcGI-2-AB within the wild type and the three integrase mutant strains. Results are given in Fig. 4A and Table 3. The ratio of the episomal form of LpcGI-2-A to the chromosomal flaA gene was 2.43 × 10⁻³:1, for LpcGI-2-B the ratio was 7.57 × 10⁻⁷:1, and for LpcGI-2-AB the ratio was 4.65 × 10⁻⁷:1 in E phase. Furthermore, the quantities of the circular forms were nearly equal in S and E phases, and differences showed no or only a low significance (Table 3). Therefore, the A ring was the predominant form (Fig. 4A, WT LpcGI-2-A) and the AB and B rings were 3,352-fold (ratio of LpcGI-2-AB to LpcGI-2-A) and 1,826-fold less present than the A ring, respectively, in E phase (Fig. 4A, WT LpcGI-2-AB and GI-3B; Table 3). However, the amount of the A ring in E phase was reduced 3,945-fold in the Δlpc1833 (integrase) mutant strain, whereas the amounts of the AB and B rings were not significantly influenced (Table 4, Δlpc1833 mutant). This indicated that the excision of LpcGI-2-A is an lpc1833-dependent process. The integrase genes lpc1884 and lpc2123 both slightly increased the presence of the episomal form of LpcGI-2-B in exponential growth phase (Fig. 4A and Table 4, Δlpc1884 and Δlpc2123 mutants).

Fig 4.

Relative quantification of the three episomal forms of LpcGI-2 and conjugation frequency. (A) SYBR green quantitative PCR was done with chromosomal DNA from L. pneumophila wild-type, Δlpc1833, Δlpc1884, and Δlpc2123 strain in exponential (E) and stationary (S) growth phases. Circularization of LpcGI-2 was detected with primers RT-LpcGI-2-2R/RT-LpcGI-2-3U for GI-2-A, RT-LpcGI-2-1U/RT-LpcGI-2-5R for GI-2-B, and RT-LpcGI-2-3U/RT-LpcGI-2-5R for GI-2-AB. The flaA gene served as a chromosomal control, and the relative amount of copies was calculated in relation to a standard curve. Results are means of three independent experiments. Statistical significance is characterized by symbols above the columns: ∞, comparison of GI-2-B and GI-2-AB versus GI-2-A of the wild type; ∼, comparison of the episomal forms of the mutant strains versus the wild type. (B) For conjugation experiments, L. pneumophila Corby wild type (WT*) or the Δlpc1833 and ΔpilT mutant strains were used as the donor and the L. pneumophila Philadelphia I JR32 strain served as the acceptor. Conjugation was done at 30°C on BCYE agar plates in the presence of DNase I. The transconjugation rates (ratios of transconjugants/donor) were 7.3 × 10⁻³ for the wild-type strain, 5 × 10⁻⁵ for the Δlpc1833 mutant strain, and 4.9 × 10⁻⁵ for the ΔpilT mutant strain. Results of the conjugation experiments are means of two independent experiments.

Table 3.

Quantitative analysis (qPCR) of episomal forms of LpcGI-2 of L. pneumophila Corby WT

Genes compareda	Growth phase	No. of copies		P valueb
		Mean	SEM
LpcGI-2-A vs flaA	E	2.436 × 10−3	1.447 × 10−3	<0.001
	S	1.283 × 10−3	5.819 × 10−4	<0.001
LpcGI-2-B vs flaA	E	7.578 × 10−7	4.007 × 10−8	<0.001
	S	8.174 × 10−7	3.702 × 10−7	<0.001
LpcGI-2-AB vs flaA	E	4.655 × 10−7	1.043 × 10−7	<0.001
	S	6.639 × 10−7	2.383 × 10−7	<0.001
LpcGI-2-B vs LpcGI-2-A	E	5.475 × 10−4	2.036 × 10−4	<0.01
	S	9.465 × 10−4	5.518 × 10−4	<0.01
LpcGI-2-AB vs LpcGI-2-A	E	2.983 × 10−4	1.037 × 10−4	<0.01
	S	7.756 ×10−4	3.829 × 10−4	<0.01

^aThe ratios of the amount in S phase to that in E phase were 0.968 ± 0.304 (P, not significant), 1.278 ± 0.141 (P, not significant), and 2.064 ± 0.594 (P < 0.05)for LpcGI-2-A, LpcGI-2-B, and LpcGI-2-AB, respectively.

^bStatistical significance was determined by Student's t test. NS, not significant (P > 0.05).

Table 4.

Quantitative analysis (qPCR) of episomal forms of LpcGI-2 of L. pneumophila Corby Δlpc1833, Δlpc1884, and Δlpc2123

Circular form of LpcGI-2	Growth phase	Δlpc1833 mutant vs WT			Δlpc1884 mutant vs WT			Δlpc2123 mutant vs WT
		No. of copies		P valuea	No. of copies		P value	No. of copies		P value
		Mean	SEM		Mean	SEM		Mean	SEM
A	E	2.535 × 10−4	7.198 × 10−5	<0.01	1.227	0.448	NS	0.831	0.503	NS
	S	6.318 × 10−4	2.456 × 10−4	<0.01	1.277	0.308	NS	0.674	0.340	NS
B	E	1.137	0.120	NS	2.190	0.495	<0.001	2.693	0.860	<0.001
	S	0.926	0.192	NS	1.234	0.262	NS	1.534	0.578	NS
AB	E	0.868	0.179	NS	1.375	0.313	NS	0.922	0.314	NS
	S	0.886	0.117	NS	1.121	0.265	NS	0.813	0.168	<0.05

^aStatistical significance was determined by Student's t test. NS, not significant (P > 0.05).

To investigate a putative role of the episomal form of LpcGI-2 for intracellular replication of L. pneumophila Corby within A. castellanii, we performed infection assays. However, the intracellular replication rate was not influenced in any of the three integrase mutant strains investigated (data not shown).

LpcGI-2 encodes a functional conjugation system.

Since we identified a partial oriT region within LpcGI-2 (see above), we performed conjugation assays with the L. pneumophila Corby wild type as the donor strain and L. pneumophila JR32 as the acceptor strain. We could demonstrate for the first time that a genomic island without a complete classical oriT region can be transferred horizontally to another Legionella strain by conjugation (Fig. 4B, WT*). In addition, we analyzed 10 transconjugants by PCR using specific primers, and we could corroborate that LpcGI-2-A is the predominant episomal form transferred by conjugation (Table 5, circular form of LpcGI-2). This was surprising, since the partial oriT region is not present on LpcGI-2-A (Fig. 1 and 3A). Next, we did the same conjugation experiment using the Δlpc1833 mutant as the donor. The conjugation rate was reduced ∼148 times compared to that for the wild-type strain (Fig. 4B, Δlpc1833 mutant). In contrast to the experiment using the wild-type strain as the donor, all 10 transconjugants analyzed were positive for LpcGI-2-AB and LpcGI-2-B but not for LpcGI-2-A (Table 5, circular form of LpcGI-2).

Table 5.

PCR analysis of transconjugants, recipient L. pneumophila JR32, and donor strains L. pneumophila Corby WT*, ΔpilT, Δlpc1833, and Δlpc2123^h

Region used in analysis	PCR result
	Transconjugants				L. pneumophila JR32 Smr recipient	L. pneumophila Corby donor
	WT*a	ΔpilTa	Δlpc1833a	Δlpc2123b		WT*	ΔpilT	Δlpc1833	Δlpc2123
Genes
lpg0402c	+	+	+	+	+	−	−	−	−
lpc1850c	+	+	+	+	−	+	+	+	+
lpc2123d	−	+	+	ni	−	+	+	+	ni
Circular forms of LpcGI-2d (primers)
A (2/3)	+	+	−	+	−	+	+	−	+
B (1/5)	−	+	+	+	−	+	+	+	+
AB (3/5)	−	+	+	+	−	+	+	+	+
Primers for analysis of integration in tRNAMete
2/6	−	−	−	−	−	+	+	+	+
2/6f	+	+	+	+	−	−	−	−	−
3/4	−	−	−	−	−	+	+	+	+
3/4g	+	+	+	+	−	+	+	+	+
5/6	−	−	−	−	−	+	+	+	+
5/6f	−	+	+	+	−	−	−	−	−
4g/6f	+	+	+	+	+	−	−	−	−

^aTransconjugant selection only for LpcGI-2-A and LpcGI-2-AB and not for LpcGI-2-B.

^bTransconjugant selection only for LpcGI-2-B and LpcGI-2-AB and not for LpcGI-2-A.

^cThirty transconjugants were analyzed.

^dTen transconjugants were analyzed.

^eThree transconjugants were analyzed. tRNA-Met gene lpc1832 or lpg2362.

^fPrimer specific for strain JR32 (trnM-lpg2362-U).

^gPrimer specific for strain JR32 (trnM-lpg2362-R).

^hAll specific primers are listed in Table S1 in the supplemental material. +, detected by PCR; −, not detected by PCR; ni, not investigated; Sm^r streptomycin-resistant strain.

To verify if the new T4SS, encoded by LpcGI-2, is functional and necessary for horizontal transfer, we generated a ΔpilT mutant strain. We performed conjugation assays with the ΔpilT mutant strain as the donor strain and L. pneumophila JR32 as the acceptor strain. The transconjugation rate was reduced and comparable to that for the Δlpc1833 mutant strain (Fig. 4B, ΔpilT). Surprisingly, all 10 transconjugants investigated were positive for all three episomal forms of LpcGI-2, indicating that LpcGI-2-A is, in contrast to the ring formation in the wild-type strain, not the predominant form transferred by the ΔpilT mutant. The same results were obtained for transconjugants using the Δlpc2123 mutant as the donor strain. However, in these experiments, the marker for the selection of transconjugants was present in region IV; thus, transconjugants which received only LpcGI-2-A could not be selected. Three transconjugants of each experiment were also analyzed for the integration of the island within the tRNA^Met gene (Table 5, integration in tRNA^Met). The experiments revealed that the received island was integrated into the tRNA^Met gene of each transconjugant. Altogether, the results prove that the new T4SS of LpcGI-2 encodes a functional conjugation system.

First evidence for genes involved in the regulatory process of genomic island excision.

We assumed that the lvrRABC genes may be involved in the regulation of genomic island excision, since lvrC encodes a paralog of CsrA which is known to be involved in gene regulation in L. pneumophila (33, 45). In addition, LvrR is a putative transcriptional regulator and the lvrRABC gene region is often found in association with T4ASSs in Legionella (26, 27, 32). Since the genomic island Trb-1 also exhibits an lvrRABC gene locus, only one episomal form is generated, and because the island is relatively small and composed primarily of trb and tra genes (26), we decided to use this island to analyze the regulatory mechanisms of island excision. In addition, within the genomic island Trb-1, only one site-specific putative integrase (int-1, lpc2818) was identified, and it was hypothesized that this enzyme may be necessary for the excision of Trb-1 (26). To verify that this integrase is responsible for the excision of Trb-1 from the genome, we constructed a specific Δint-1 (lpc2818) mutant strain of L. pneumophila Corby. To study the role of the putative regulatory elements, we constructed L. pneumophila Corby ΔlvrRABC (lpc2816 to lpc2813) and ΔlvrR (lpc2816) deletion mutants by replacing the respective genes with a kanamycin resistance cassette (see Materials and Methods). The three mutants obtained were verified by PCR analysis (data not shown). Next, we investigated the mutant strains for the excision of Trb-1 from the genome by PCR and qPCR analyses (Fig. 5 and 6A). The PCR and qPCR analyses revealed that int-1 is necessary for the excision of Trb-1 from the genome. In contrast to the wild type, the episomal circular form of Trb-1 (Trb-1_ci) and the intact chromosomal form of the tRNA^Pro gene (without the integrated island) thereby generated were not detectable in the Δint-1 strain (Fig. 5B, Δint-1 with primers 2/3 and 1/4). As expected, the integrated form of Trb-1 (Trb-1_i) was present in both strains (Fig. 5B, primers 1/2 and 3/4). In a control experiment, we showed that the episomal form of Trb-2 (Trb-2_ci) was still present in the Δint-1 mutant (Fig. 5B, primers 6/7). Our data revealed that Int-1 is specifically necessary for the excision of Trb-1 from the genome of L. pneumophila Corby but does not influence the excision of the second trb-tra gene containing genomic island Trb-2.

Fig 6.

Relative quantification of episomal form of Trb-1 (Trb-1_ci) and conjugation frequency. (A) Quantitative PCR for circularization frequency of Trb-1. SYBR green real-time PCR was done with chromosomal DNA from L. pneumophila Corby wild type and the Δint-1, ΔlvrRABC, and ΔlvrR mutant strains in exponential (E) and stationary (S) growth phases. Trb-1_ci formation was a rare, int-1-dependent event and occurred more often in the ΔlvrRABC and ΔlvrR mutants. The circularization frequencies in E and S growth phases (Trb-1_ci-E and -S) are given. The flaA gene served as a chromosomal control, and the relative amount of copies was calculated in relation to a standard curve. Results are means of three independent experiments. Statistical significance (Student's t test) is characterized by symbols above the columns: ∞, comparison of Trb-1_ci of mutant strain versus wild type; ∼, comparison of Trb-1_ci of ΔlvrRABC mutant versus ΔlvrR mutant. (B) Conjugation frequency of Trb-1. L. pneumophila Corby wild type (WT°) or the Δint-1 mutant strain was used as the donor, and L. pneumophila Philadelphia I JR32 served as the acceptor. Conjugation was done at 30°C on BCYE agar plates in the presence of DNase I. The transconjugation rates (ratios of transconjugants/donor) were 2.6 × 10⁻⁴ for the wild-type strain and 3.8 × 10⁻⁶ for the Δint-1 mutant strain. Results are means of two independent experiments.

In addition, the results of PCR analyses suggested that the amount of Trb-1_ci is upregulated in the ΔlvrRABC and ΔlvrR mutant strains (Fig. 5C, in vitro, primers 2/3), and thus, the amount of the tRNA^Pro gene without integrated Trb-1 was elevated compared to that in the wild type (Fig. 5C, in vitro, primers 1/4). However, the integrated form of Trb-1 was still detectable (Fig. 5C, in vitro, primers 1/2 and 3/4). Subsequently, we performed a PCR analysis with intracellularly grown wild-type and ΔlvrRABC mutant strains and demonstrated that the upregulation of Trb-1_ci in the ΔlvrRABC mutant strain also occurs during replication within A. castellanii (Fig. 5C, in vivo, primers 2/3).

To further investigate (quantitatively) if the generation of Trb-1_ci is negatively regulated by the lvrRABC gene cluster and the lvrR gene, we performed qPCR analysis using chromosomal DNA as the template (Fig. 6A). The flaA gene was used as the chromosomal control, and the value for the flaA gene was used as the reference value for qPCR analysis. The qPCR results demonstrated that Trb-1_i is the predominant form in L. pneumophila, since the ratio of Trb-1_ci to the chromosomal flaA gene was approximately 1 × 10⁻⁴:1 in E phase in the wild-type strain (Fig. 6A, WT Trb-1_ci-E, and Table 6). Compared to the wild-type strain, the amount of Trb-1_ci was reduced ∼30-fold in the Δint-1 strain (Fig. 6A, Δint-1 Trb-1_ci, and Table 6). In contrast, the amount of Trb-1_ci was ∼147-fold higher in the ΔlvrRABC strain than in the wild-type strain (Fig. 6A, ΔlvrRABC Trb-1_ci, and Table 6). Similar results were obtained with the ΔlvrR mutant strain (Fig. 6A, ΔlvrR Trb-1_ci, and Table 6). The results corroborate the finding that the excision of Trb-1 is an Int-1-dependent but rare event. In addition, the excision of Trb-1_i is negatively regulated by the lvrRABC gene cluster and the lvrR gene. The experiments revealed a minor but significant difference between ΔlvrRABC and ΔlvrR mutant strains in both growth phases, indicating a further regulatory influence of lvrABC on the excision of Trb-1 (Fig. 6A). The influence of the growth phase on the presence of Trb-1_ci is low, since Trb-1_ci was induced only ∼2-fold from E to S phase (Table 6).

Table 6.

Quantitative analysis (qPCR) of Trb-1_ci of L. pneumophila Corby WT andL. pneumophila Corby Δint-1, ΔlvrRABC, and ΔlvrR mutant strains^a

Growth phase	WT (Trb-1ci vs flaA)			Δint-1 vs WT			ΔlvrRABC vs WT			ΔlvrR vs WT
	No. of copies		P value	No. of copies		P value	No. of copies		P value	No. of copies		P value
	Mean	SEM		Mean	SEM		Mean	SEM		Mean	SEM
E	1.065 × 10−4	4.865 × 10−06	<0.001	0.034	0.009	<0.001	146.80	11.50	<0.001	68.12	9.66	<0.001
S	2.122 × 10−4	2.451 × 10−05	<0.001	0.029	0.014	<0.001	74.24	27.99	<0.001	125.09	35.61	<0.001

^aStatistical significance was determined by Student's t test. NS, not significant (P > 0.05). The mean ± SEM ratios of the number of copies for S phase versus number of copies for E phase were 1.867 ± 0.653 (P < 0.05) for the wild type, 1.691 ± 0.868 (P was not significant) for Δint-1, 0.839 ± 0.241 (P was not significant) for ΔlvrRABC, and 2.916 ± 0.627 (P < 0.001) for ΔlvrR.

We recently demonstrated that Trb-1 can be transferred horizontally between Legionella strains by conjugation (26). Therefore, we performed conjugation assays to investigate the frequency of Trb-1 conjugation using the L. pneumophila Corby wild type (WT^o) or the Δint-1 mutant strain as the donor and the L. pneumophila JR32 strain as the acceptor. The conjugation rates were 2.6 × 10⁻⁴ for the wild-type strain and 3.8 × 10⁻⁶ for the Δint-1 mutant strain; thus, the conjugation rate of Trb-1 from the Δint-1 mutant was reduced ∼68-fold in comparison to that from the wild-type strain (Fig. 6B).

Trb-1_ci is not necessary for intracellular replication of L. pneumophila within host cells.

To round up Trb-1 analysis, we then investigated if Trb-1_ci is involved in the in vivo fitness of L. pneumophila, since Trb-1_ci was shown to be present during replication in A. castellanii cells (see above; Fig. 5C). The growth of the Δint-1 and the ΔlvrRABC mutant strains in AYE medium was similar to that of the wild-type strain (data not shown). We then performed infection assays and infection/survival assays with and without competition of the wild type and the respective mutant strain with A. castellanii. The infection experiments revealed no differences between the wild type and the Δint-1 or ΔlvrRABC mutant strain (data not shown). The results demonstrated that the Δint-1 and ΔlvrRABC genes do not influence the intracellular replication or fitness of L. pneumophila in A. castellanii.

To further analyze a putative role of Trb-1 for intracellular replication, we transferred Trb-1_ci to L. oakridgensis by conjugation. L. oakridgensis is a less virulent strain and is negative for Trb-1 and other genomic islands, with the exception of the lvh system (unpublished results; see Fig. S1A in the supplemental material, Lvh_ci). In the transconjugants of L. oakridgensis, Trb-1_i and Trb-1_ci were present and also detectable after passage (10 times) (see Fig. S1A in the supplemental material, TC3 and TC3/10). Since L. oakridgensis was described not to replicate in amoebae (46), we investigated the L. oakridgensis wild-type strain and the Trb-1-positive transconjugants for their ability to replicate within the human macrophage-like cell line U937. The replication of the transconjugants was similar to the replication of L. oakridgensis wild type (see Fig. S1B in the supplemental material). These results demonstrate that Trb-1_ci has no supporting effect on the ability of L. oakridgensis to replicate within human macrophages.

DISCUSSION

Recently, we identified and described two genomic islands in L. pneumophila Corby (Trb-1 and Trb-2) which can be excised from the chromosome, forming episomal plasmid-like forms. Both genomic islands exhibit an oriT region, and the whole Trb-1 island can be transferred to other Legionella strains by conjugation. After conjugation, the island is integrated site specifically within the genome of the transconjugants (26). Among further genes, Trb-1 contains the putative integrase Int-1 and the lvrRABC region, which shows homology to regulatory proteins. So, the questions were if genomic islands in L. pneumophila are excised integrase dependently and if the lvrABC region is involved in this process. In addition, L. pneumophila Corby exhibits four further genomic islands within its genome (LpcGI-1 and -2, LpcGI-Asn, and LpcGI-Phe). Genomic islands LpcGI-1 and LpcGI-2 exhibit GI-like secretion systems which are similar to those of LGI-2 of L. pneumophila strains 130b and Paris (27, 32). Both islands exhibit gene loci with genes involved in the metal ion resistance, persistence, and fitness of L. pneumophila. LpcGI-1 and -2 are integrated within the tRNA^Thr and the tRNA^Met genes, respectively. In contrast to Trb-1, both islands exhibit two attR sites. Surprisingly, only the two episomal forms LpcGI-1-A and LpcGI-1-B were detectable, whereas for LpcGI-2, the three episomal forms A, B, and AB were present. It remains to be elucidated why the episomal LpcGI-1-AB form is absent. Like Trb-1, both islands exhibit a region encoding paralogs of the lvrRABC genes. The genomic islands LpcGI-Asn and LpcGI-Phe do not encode a T4ASS but are also integrated within tRNA genes. In contrast to LpcGI-Phe, LpcGI-Asn also exists as an episomal form. While LpcGI-Asn exhibits at least an lvrA paralog, LpcGI-Phe lacks the complete lvr region. Nothing is yet known about the putative horizontal transfer of these islands.

In the present work, we could demonstrate for the first time that the site-specific integrases Int-1 and Lpc1833 are responsible for the generation of the episomal forms of Trb-1 and LpcGI-2-A, respectively. In the Δint-1 and Δlpc1833 mutant strains, the rate of Trb-1 and LpcGI-2-A conjugation was drastically reduced. The results indicate that the excision of the genomic islands Trb-1 and LpcGI-2-A in L. pneumophila Corby depends on a functional site-specific integrase and that both islands are mobilizable via conjugation. The generation of the episomal Trb-1_ci and LpcGI-2-A is an even rarer event, since the ratios of Trb-1_ci and LpcGI-2-A to chromosomal flaA were shown to be 1 × 10⁻⁴:1 and 2 × 10⁻³:1, respectively. In contrast to our findings, the lvh island of L. pneumophila Paris, encoding another T4ASS, was described to be present as a multicopy plasmid (47). The authors also published findings that the episomal form of lvh is more frequently generated in the exponential growth phase. However, this is not the case for Trb-1 or LpcGI-2-A.

We then investigated the mechanism of genomic island excision. We demonstrated that the island LpcGI-2 can exist in three different episomal forms and that integrase lpc1833 is necessary for the excision of the predominant episomal form LpcGI-2-A (see above). Furthermore, the excision of LpcGI-2-A was independent from the integrase genes lpc1884 and lpc2123, but in the respective mutant strains, the episomal form of LpcGI-2-B was slightly increased. We do not yet know which (additional) proteins (e.g., an excisionase) are involved in the excision of LpcGI-2-AB and LpcGI-2-B. It was shown that an excisionase helps the site-specific recombinases in the direction of excision (48). On Trb-1 and Trb-2, putative excisionase-like proteins are present (lpc2780 [65 amino acids] and lpc0198 [68 amino acids], respectively). Nevertheless, we could not identify an excisionase-like protein on LpcGI-2. Low cross activity of other integrases is probably involved, since the relative quantity of LpcGI-2-AB and LpcGI-2-B was similar to the amount of episomal LpcGI-2-A in the Δlpc1833 integrase mutant strain (Fig. 4). On the other hand, sequencing of the attP′ and attP″ sites of LpcGI-2-B and LpcGI-2-AB revealed site-specific recombination for the excision of both islands between attR-1 and attR-2 or attL and attR-2, respectively. In addition, conjugation experiments using Δlpc1833 as the donor revealed that 10 out of 10 analyzed transconjugants were negative for the episomal form of LpcGI-2-A (Table 5). However, three of these transconjugants were analyzed for LpcGI-2 integration into the tRNA^Met gene, revealing that they were positive for integrated LpcGI-2-A and the episomal form of LpcGI-2-B. These results confirmed that the generation of the episomal form of LpcGI-2-A is an lpc1833-dependent process, whereas this is not the case for LpcGI-2-B. Remarkably, LpcGI-2-AB was integrated into the tRNA^Met gene within the transconjugants, although Lpc1833 was not present. In this case, integration into the genome of the recipient may be due to the action of RecA, as shown recently for the high-pathogenicity island of Yersinia pseudotuberculosis (49). Furthermore, it is likely that an integration host factor (IHF) is involved in excision/integration of LpcGI-2, since we found four putative IHF sites on the genomic island directly associated with the attP/attR sites. The presence of putative IHF-binding sites and the role of the IHFs for the function of integrative and conjugative elements (ICEs) are known (50–54).

It was discussed that oriT-negative islands have been acquired horizontally by Legionella, but it has not been shown experimentally (27). In this work, we could demonstrate for the first time that LpcGI-2-A is transferred horizontally by conjugation using L. pneumophila Corby as the donor strain and L. pneumophila Philadelphia I as the acceptor strain. The island was integrated site specifically into the genome of the transconjugants (Table 5). Since we were not able to identify a classical oriT region on LpcGI-2-A, even classical oriT-negative islands of L. pneumophila can be transferred horizontally by conjugation. Further experiments are needed to identify the mechanism of this transfer.

In addition, we investigated if LpcGI-2 encodes a functional conjugation system by analyzing the transfer of LpcGI-2 using a ΔpilT mutant strain as the donor. pilT is located in region I and therefore present on the DNA, forming rings A and AB. In the ΔpilT mutant strain, the rate of conjugation of LpcGI-2 or LpcGI-2-A was reduced 149-fold, and the conjugation rate was comparable to the conjugation rate obtained using the Δlpc1833 mutant as the donor. Therefore, LpcGI-2 is transferred horizontally in a pilT-dependent process. Surprisingly, LpcGI-2-AB and not LpcGI-2-A was mainly transferred from the Δlpc1833 or ΔpilT donor strain. It seems as if LpcGI-2-AB could be transferred by using another conjugation system, probably the oriT-dependent system present on Trb-1 and Trb-2, since a partial oriT region is present on LpcGI-2-AB. However, we can demonstrate for the first time that LpcGI-2 encodes a functional conjugation system in L. pneumophila which is integrated site specifically into the genome of the transconjugants and that the excision of this new GI-like secretion system from the genome depends on a site-specific integrase.

We then analyzed lvrRABC and lvrR for their putative roles in the regulatory process of genomic island excision. Most of the genomic islands identified so far in Legionella exhibit an lvr region (26, 27, 55), encoding the putative phage-repressor LvrR; LvrA and LvrB, two proteins with unknown function; and LvrC, a CsrA paralog. In Legionella, CsrA is a regulatory protein acting on the mRNA level of its target genes (33, 45, 56). It negatively regulates the switch from the replicative to the transmissive phase in L. pneumophila (33). LvrR exhibits Pfam_HTH_XRE and S24-LexA-like peptidase motifs. These motifs are found in proteins involved in bacterial plasmid copy control, repressors of the SOS system, and other DNA-binding proteins (57, 58). To analyze the role of the Trb-1 lvr region, we first deleted lvrRABC. We found that the excision of Trb-1 is negatively regulated by the lvr region, since the episomal form was upregulated 147 times in the mutant strain. Results obtained using the generated ΔlvrR mutant strain confirmed that the phage repressor-like protein LvrR is involved in this repression. In addition, the lvr gene (lpc2273) of LpcGI-1 encodes a putative LvrR protein of only 84 amino acids in length. The protein exhibits the HTH-Xre motif, but it lacks the S24_LexA-like motif found within LvrR of Trb-1, suggesting that this protein may be nonfunctional. This assumption is supported by the finding that the amount of LpcGI-1-A and LpcGI-1-B was similar to the quantity of Trb-1_ci in the ΔlvrR (lpc2816) mutant strain (Fig. 2A and Fig. 5C). Furthermore, qPCR analyses revealed that the expression of lpc2819 and int-1 is induced in the ΔlvrRABC mutant strains (data not shown). The observed induction of int-1 expression is in line with the upregulation of Trb-1_ci in the ΔlvrRABC mutant, since we demonstrated that the generation of Trb-1_ci is Int-1 dependent. The influence of lvrRABC on genomic island gene expression and conjugation will be analyzed in detail in a further study.

We then asked if the islands may be involved in the in vivo fitness of L. pneumophila, since Trb-1_ci was also present during the replication within A. castellanii (26). Neither the mutant strains exhibiting an elevated level of Trb-1_ci nor the int-1 mutant showed a modified rate of intracellular multiplication. In addition, the conjugation of Trb-1_ci into the less virulent strain L. oakridgensis did not support the ability of the transconjugants to replicate within U937 cells. Although we could confirm that Trb-1_ci is generated in vivo, its relevance for the life cycle of L. pneumophila remains unknown. As demonstrated for Trb-1, none of the three integrase mutant strains of LpcGI-2 showed an effect on the intracellular replication of L. pneumophila in A. castellanii, suggesting that the excision of LpcGI-2 does not influence the in vivo fitness of L. pneumophila. Accordingly, Kim and colleagues published findings that a metal efflux island of L. pneumophila (similar to LpcGI-1) is not required for survival in macrophages and amoeba (59).

In conclusion, we could demonstrate that the excision and conjugation of the genomic islands Trb-1 and LpcGI-2-A are site-specific integrase-dependent events in L. pneumophila. The elements are integrated into the genome of the transconjugants. The attP and attL/attR sites and probably the IHF-binding sites of LpcGI-2 are involved in these processes. Therefore, the genomic islands should be classified as ICEs. ICEs are defined to be elements that excise site specifically from the chromosomal DNA, leading to an episomal circular form that is generally transient. After conjugation, the element is integrated into the recipient chromosome. The target site for the integrase- or recombinase-dependent integration is often a tRNA gene (44, 60, 61). The recently defined new class of GI-associated T4SSs (GI-like) identified in Haemophilus influenzae (14, 30, 31) has also been identified in genome sequences of L. pneumophila (32). We can now demonstrate that the GI-like element of L. pneumophila (LpcGI-2) encodes a functional conjugation system localized on an ICE. Furthermore, we present first experimental data for the involvement of the lvrRABC gene cluster in the regulation of the process of excision of ICEs (Trb-1) in L. pneumophila.