Conjugative transfer of a derivative of the IncP-1α plasmid RP4 and establishment of transconjugants in the indigenous bacterial community of poplar plants.

The persistence of traits introduced into the indigenous bacterial community of poplar plants was investigated using bioluminescence mediated by the luc gene. Three endophytic bacterial strains provided with the IncP-1α plasmid RP4-Tn-luc were used to inoculate poplar cuttings at different phenological stages. Screening of isolates by bioluminescence and real-time PCR detection of the luc gene revealed stable persistence for at least 10 weeks. Although the inoculated strains became established with a high population density after inoculation at leaf development (April) and senescence (October), the strains were suppressed by the indigenous bacteria at stem elongation (June). Transconjugants could be detected only at this phenological stage. Indigenous bacteria harbouring RP4-Tn-luc became established with densities ranging from 2 × 10(5) to 9 × 10(6) CFU g(-1) fresh weight 3 and 10 weeks after inoculation. The increased colonization of the cuttings by indigenous bacteria at stem elongation seemed to strongly compete with the introduced strains. Otherwise, the phenological stage of the plants as well as the density of the indigenous recipients could serve as the driver for a more frequent conjugative plasmid transfer. A phylogenetic assignment of transconjugants indicated the transfer of RP4-Tn-luc into six genera of Proteobacteria, mainly Sphingomonas, Stenotrophomonas and Xanthomonas. © FEMS 2015.

INTRODUCTION

Plant-associated bacteria have numerous mechanisms by which they can promote plant growth and health. The synthesis of phytohormones and siderophores, or the ability to degrade pollutants, indicate that the employment of endophytic bacteria in techniques for improving plant production and in phytoremediation of contaminated soil is promising (Glick 2012; Sessitsch et al. 2013). For example, several studies have shown that the performance of trees and crops grown on polluted soils can be essentially improved by inoculation with bacteria that mediate the degradation of xenobiotic compounds or heavy metal resistance (Weyens et al. 2011; Hechmi et al. 2012). However, if these applications are to become practical, the long-term establishment of the introduced traits in the plant-associated microflora is necessary. Accessory bacterial metabolic functions, such as xenobiotic resistance, are known to be plasmid encoded (Schlüter et al. 2007). Accordingly, bacterial functions essential for success can be lost if the plasmid is lost. On the other hand, the spread of plasmids by conjugal transfer to and between the indigenous bacteria can contribute considerably to the maintenance and enhancement of the desired traits (Taghavi et al. 2005; Weyens et al. 2009; Wang et al. 2010). In particular, the broad-host-range plasmids belonging to the IncP-1 group are known to promote gene transfer because they are stably replicated in almost all Proteobacteria and effectively transferred by conjugation in a wide range of bacterial taxa (Adamczyk and Jagura-Burdzy 2003; Popowska and Krawczyk-Balska 2013; Klümper et al. 2015). Horizontal gene transfer has been documented in diverse natural environments, such as bulk soil (Filonov et al. 2010; Heuer and Smalla 2012); aquatic habitats (Dahlberg, Bergstrom and Hermansson 1998); animal systems (Hoffmann et al. 1998); the rhizo-, phyllo- and endospheres of plants (Normander et al. 1998; van Elsas et al. 1998; van Elsas, Turner and Bailey 2003; Wang et al. 2010); and biogas plant digestates (Wolters et al. 2015).

In general, transfer events require an adequate bacterial density and metabolic activity. These conditions are specifically present in microbial biofilms that occur in many habitats. Nevertheless, even under favourable conditions, transfer events are not easy to detect because they depend on environmental factors, the plasmid type and the host strain (Sørensen et al. 2005; Król et al. 2011). Although much fundamental research has been conducted, studies on plasmid transfer in the phytosphere are, up to now, rare and the establishment of transconjugants is not fully understood.

In this study, we analysed the persistence of endophytic bacterial strains harbouring a recombinant IncP plasmid that was introduced back into poplar cuttings. Moreover, the conjugative transfer and the subsequent establishment of the plasmid in the indigenous bacterial community were studied. For this purpose, cuttings obtained from field-grown poplars were inoculated with endophytic isolates carrying an RP4 derivative. The survival of the introduced bacteria and the presence of the plasmid within the bacterial community were assessed at three phenological stages of the plants to gain insight into the factors influencing the persistence and transfer of genes in the poplar phytosphere.

MATERIALS AND METHODS

Bacterial cultivation

Escherichia coli strains were cultivated on Luria–Bertani medium (Sigma-Aldrich Co., St Louis, USA). Rhizobium leguminosarum and bacterial strains isolated from poplar cuttings were grown on R2A or TSA (Sigma-Aldrich). Media were supplemented with the following antibiotics (μg ml−1): tetracycline 15, kanamycin 50, rifampicin 30, streptomycin 100 and cycloheximide 100 to inhibit growth of eukaryotic microorganisms.

Construction of the recombinant IncP plasmid RP4-Tn-luc and preparation of the bacterial strains used for inoculation

The plasmid RP4-4 (IncP-1α, Tcr, Apr, Kms; Hedges and Jacob 1974) was introduced into R. leguminosarum G122Sm Tn163-luc (Ulrich and Pühler 1994) using E. coli DSM 3880 RP4-4 as the donor strain. Transconjugants were selected on TSA supplemented with tetracycline and streptomycin. A mixture of transconjugants obtained directly from the selective agar plates was used as the donor for a second mating with the rifampicin resistant strain E. coli DH5α-2 (Ulrich and Pühler 1994). Transconjugants carrying RP4-4 with the transposon Tn163-luc inserted were selected with kanamycin and rifampicin. The resulting plasmid RP4-Tn-luc conferred bioluminescence and kanamycin resistance via a 3.2 kb nptII-luc cassette within Tn163 (Ulrich and Lentzsch 1997). The strains were assessed for the bioluminescence phenotype and multiple antibiotic resistances. The insertion of the transposon did not affect the maintenance or transfer functions of the plasmid.

For the screening of endophytic isolates, four branch sections of 2 cm (Ø 5 mm) were sampled from the poplar clones Geneva and 741 (Table 1). Surface disinfection of the samples and screening of isolates were performed as described by Ulrich, Ulrich and Ewald (2008). Several isolates were subcultured on R2A and taxonomically classified by PCR and sequencing of the partial 16S rRNA gene (Ulrich, Klimke and Wirth 2008). Two Pseudomonas isolates (Q16 and E9) obtained from the poplar clone Geneva and a Stenotrophomonas strain (169-1) from the clone 741 were chosen as the strains for inoculation (Table 1). RP4-Tn-luc was transferred from DH5α-2 pBR322-sacB RP4-Tn-luc using filter matings. The E. coli donor possesses a pBR322 derivative with a 3.8 kb sacB-Kmr cassette (originated from pSUP104-sac; Simon et al. 1991), which does not allow growth on medium containing sucrose. Transconjugants were selected on TSA supplemented with kanamycin and 5% sucrose. The presence of sacB in the E. coli donor allowed selection against the donor without the presence of a selectable resistance marker or characteristic in the recipient strain.

Table 1.

Inoculated strains used in this study.

	Origin
Strain	Poplar clone	Plant material	Location	16S rRNA gene accession number
Pseudomonas sp. E9	Geneva ♀ (Populus maximowiczii ×		Tree nursery	HE652089
Pseudomonas sp. Q16	P. berolinensis)		Waldsieversdorf,	HE652090
		Branch section
Stenotrophomonas sp. 169-1	741 ♀ (P. alba × [P. davidiana + P.		Germany	HE652088
	simonii] × P. tomentosa)		52.54 N, 14.06 E

Inoculation of poplar cuttings, plant growth, sampling

The bacterial strains used as inocula were grown in R2A broth to late exponential phase, concentrated by centrifugation (6,000 g) for 2 min and resuspended in 5 ml Schenk–Hildebrand medium (Schenk and Hildebrandt 1972) to a density of approximately 108 CFU ml−1. Cuttings (Ø 5 mm, length 15 cm) were taken from the field grown poplar clone Geneva at the phenological stages of leaf development (April), stem elongation (June) and senescence (October), and inoculated with 200 μl bacterial suspension by infiltration. In total, 10 cuttings per strain and stage were inoculated; three cuttings were used to determine the abundance of the indigenous bacteria before inoculation. The inoculated cuttings were placed immediately into a mini-greenhouse in sand saturated with water and cultivated under a light regime with 16 h of white light (photon flux of 30 μmol m−2 s−1) at 23°C for 10 weeks. Three cuttings per inoculated strain were harvested at 3 and 10 weeks after inoculation.

Screening of isolates for inoculated strains and putative transconjugants

Branch sections of approximately 2 cm were sampled, washed three times with sterile tap water and ground in 4 ml of 0.3% NaCl with a mortar and pestle. Tissue extracts were serially diluted, plated on R2A supplemented with kanamycin and incubated for 3–6 days at 26°C. For each inoculation and sampling date, 300 representative isolates were screened for bioluminescence after transferring the colonies onto nylon membranes (Hybond-N). The membranes were merged with a few drops of luciferin (1 mg ml−1 in 100 mM sodium citrate pH 5.6) and exposed to an X-ray film (Kodak T-Mat G/RA film) for 2 h at 26°C. Bioluminescent colonies appeared as dark spots on the X-ray film.

Identification of isolated bacteria

To distinguish the inoculated strains from the transconjugants, bioluminescent isolates were grouped based on phenotypic characteristics and restriction analysis of the 16S rRNA gene using MspI (Ulrich, Klimke and Wirth 2008). Five to eight representative isolates from each group were identified by sequencing the partial 16S rRNA gene (Ulrich, Klimke and Wirth 2008). The assignment of the isolates at the genus level was based on NCBI-BLAST searches and on the taxonomic assignment by the CLASSIFIER program of the RDP. A phylogenetic tree was generated by the neighbor-joining algorithm using the original dataset as well as 1000 bootstrap datasets to evaluate its topology (PHYLIP vers. 3.6; Felsenstein (1993)).

Quantitative detection of RP4-Tn-luc in the indigenous community of poplar plants

For the isolation of total DNA from poplar, branch sections were washed three times with sterile tap water and cut into 1–2 mm pieces (200 mg in total). The plant material was mixed with 1 g silica beads (Ø 1 mm) and ground using a bead beater (FastPrep 24 Instrument, MP Biomedical) for 60 s at 6 m s−1. The DNA was isolated using a modified CTAB-Protokoll (Dumolin, Demesure and Petit 1995). The presence of the plasmid was determined in community DNA by real-time PCR with the primers (5′-GGGCGCACCTCTTTCGA-3′; 5′-CCTTGTCGTATCCCTGGAAGAT-3′) and the probe (5′-TTGCAACCGCTTCCCCGACTTC-3′) specific for the luc gene (position 945–1016, AB261984). The probe was labelled at the 5′ end with the reporter dye FAM (6-carboxy-fluorescein) and at the 3′ end with the quencher dye TAMRA (6-carboxy-tetramethylrhodamine). PCR was conducted in 20 μl reactions with each primer at 0.9 μM, 0.25 μM probe and TaqMan Universal PCR Master Mix (Life Technologies, Germany) in a 7500 Fast real-time PCR system (Life Technologies) using a standard protocol. The standard curve was generated using serial dilutions of genomic DNA from Stenotrophomonas 169-1 RP4-Tn-luc containing between 106 and 2 copies of the luc gene. The PCR efficiency was approximately 94%. All samples were run in duplicate. The inhibitory effects of the extracted DNA on PCR performance was tested by the quantification of serial dilutions.

Statistics

Differences between the experimental variants were tested by a three-factorial ANOVA using a block design with phenological stage as block and inoculated strain and sampling date as factors. Subsequently, differences within the stage were proven with a two-factorial ANOVA and the post-hoc test Tukey (R software, R-3.1.3, package multcomp; R Core Team 2015).

RESULTS

Establishment of inoculated endophytic bacteria and RP4-4-Tn-luc transconjugants

Cuttings of the poplar clone Geneva obtained at three growth stages were inoculated with endophytic isolates harbouring the recombinant IncP plasmid RP4-Tn-luc. Two of the strains (Pseudomons E9 and Q16) were previously isolated from the clone Geneva but the other strain (Stenotrophomonas 169-1) was obtained from the distinctly related poplar clone 741.

As a result of inoculation at leaf development in April and senescence in October, all three strains became established within the poplar twigs for at least 10 weeks (Table 2). The number of culturable inoculated strains and quantification of the luc gene showed high and seasonal stable population densities with significant differences between the two developmental stages (P < 0.01) but a rather marginal variance of the single strains and the two sampling dates. A transfer of the recombinant plasmid to the indigenous bacteria could not be detected. Depending on the abundance of the inoculated strains, the limit of detection for transconjugants ranged from 8 × 103 to 1 × 102 g−1 fresh weight. The population densities of indigenous bacteria were similarly low in both spring and autumn (Table 2).

Table 2.

Establishment of the recombinant plasmid RP4-Tn-luc in the indigenous microflora of poplar cuttings inoculated with Stenotrophomonas sp. 169-1, Pseudomonas sp. E9 and Q16, each harbouring RP4-Tn-luc.

	Population densities (CFU g−1 fresh weight)c
	Inoculated strain	Transconjugants	Real-time PCR detection of RP4-Tn-luc
			(genomes g−1 fresh weight)[c,d]
Inoculation at leaf development in Aprila
Indigenous bacteriab	7.0 × 104 ± 5.3 × 103 A
3 weeks after inoculation
Stenotrophomonas sp. 169-1	2.3 × 106 ± 5.7 × 105 C	n.d.	1.2 × 107 ± 6.5 × 106 A
Pseudomonas sp. E9	1.7 × 106 ± 2.2 × 105 B,C	n.d.	3.7 × 107 ± 3.4 × 106 B
Pseudomonas sp. Q16	1.5 × 106 ± 2.7 × 105 B	n.d.	9.6 × 107 ± 1.3 × 107 C
10 weeks after inoculation
Stenotrophomonas sp. 169-1	1.6 × 106 ± 7.1 × 104 B,C	n.d.	2.0 × 107 ± 1.2 × 107 A,B
Pseudomonas sp. E9	1.2 × 106 ± 1.0 × 105 A,B	n.d.	9.8 × 106 ± 1.1 × 106 A
Pseudomonas sp. Q16	6.4 × 105 ± 2.2 × 105 A	n.d.	2.1 × 107 ± 3.6 × 106 A,B
Inoculation at stem elongation in June
Indigenous bacteria	3.1 × 106 ± 1.5 × 105 B
3 weeks after inoculation
Stenotrophomonas sp. 169-1	n.d.	1.6 × 105 ± 6.3 × 103 A	4.4 × 106 ± 8.5 × 105 A
Pseudomonas sp. E9	3.8 × 105 ± 3.2 × 105	5.3 × 105 ± 4.5 × 105 A	1.0 × 107 ± 1.0 × 106 A
Pseudomonas sp. Q16	4.1 × 106 ± 3.3 × 106	2.5 × 106 ± 2.0 × 106 A	5.1 × 106 ± 1.8 × 106 A
10 weeks after inoculation
Stenotrophomonas sp. 169-1e	n.d.	9.3 × 106 ± 8.8 × 106 A	5.2 × 107 ± 2.4 × 107 B
Pseudomonas sp. E9	n.d.	n.d.	9.2 × 105 ± 5.5 × 105 A
Pseudomonas sp. Q16e	n.d.	6.8 × 105 ± 7.5 × 105 A	8.7 × 106 ± 7.3 × 105 A
Inoculation at senescence in October
Indigenous bacteria	8.6 × 103 ± 1.2 × 103 A
3 weeks after inoculation
Stenotrophomonas sp. 169-1	3.1 × 105 ± 8.2 × 104 C	n.d.	3.8 × 106 ± 4.6 × 105 A
Pseudomonas sp. E9	2.3 × 105 ± 5.3 × 104 B,C	n.d.	2.3 × 106 ± 7.9 × 104 A
Pseudomonas sp. Q16	5.9 × 105 ± 1.2 × 105 D	n.d.	9.8 × 106± 4.3 × 106 B
10 weeks after inoculation
Stenotrophomonas sp. 169-1	7.5 × 104 ± 1.4 × 104 A,B	n.d.	1.2 × 105 ± 3.9 × 104 A
Pseudomonas sp. E9	1.2 × 104 ± 3.9 × 103 A	n.d.	3.8 × 105 ± 4.8 × 104 A
Pseudomonas sp. Q16	5.0 × 104 ± 8.7 × 102 A	n.d.	6.5 × 105 ± 5.9 × 104 A

aPrincipal growth stages. Leaf development: green leaf tips 10 mm above the bud scales, first leafs unfolded; stem elongation: stem approximately 80% of final length; senescence: leaf discoloration (Finn, Straszewski and Peterson 2007). Inoculation of the freshly harvested cuttings was performed at these three dates.

bCulturable indigenous bacteria were determined directly before inoculation by plating on R2A. The abundance of the inoculated strains and transconjugants was estimated by plating on R2A supplemented with kanamycin and subsequently testing for bioluminescence and phylogenetic characterisation 3 and 10 weeks after inoculation.

cNumbers are means and standard deviations of triplicates. Different capital letters indicate significant differences (P < 0.05). Limit of detection was approximately 1 × 102 CFU g−1 fresh weight for the indigenous bacteria and the inoculated strains. Due to screening of 300 kanamycin resistant isolates per sample, limit of the detection of transconjugants was possibly increased to 1/300 of the population density of the inoculated strains of the respective sample. Limit of detection for the real-time PCR quantification was approximately 5 × 102 genomes g−1 fresh weight.

dThe values were calculated based on the whole-genome size of Stenotrophomonas maltophilia (5.2 fg).

eTransconjugants were detected in two of the replicates.

n.d. = not detectable

Different results were obtained by inoculating the cuttings at the stem elongation stage in June. Three weeks after inoculation, only the two clone-specific Pseudomonas strains persisted. After 10 weeks, none of the inoculated strains was detectable. However, transconjugants were found at both sampling dates. Indigenous bacteria harbouring the recombinant plasmid were detected with population densities ranging from 2 × 105 to 9 × 106 CFU g−1 fresh weight (Table 2). Except for inoculation with Pseudomonas E9, the numbers of culturable bacteria possessing RP4-Tn-luc were similarly high at both sampling dates. This indicates the stable presence of bacteria containing the plasmid. Corresponding to the observed conjugation events, the density of the indigenous bacteria was considerably higher than at the two other growth stages (Table 2).

Real-time PCR verified the presence of bacteria carrying RP4-Tn-luc and indicated an increasing trend from 3 to 10 weeks which was significant for the Stenotrophomonas inoculation. The PCR detection of the recombinant plasmid was consistent with the abundance of the inoculated strains and the indigenous bacteria harbouring the plasmid, except for strain E9 10 weeks after inoculation. Similar to the other phenological stages, quantification of the luc gene resulted mostly in 10-fold higher population densities compared with the cultivation approach.

Community composition of indigenous bacteria harbouring the plasmid RP4-Tn-luc

Taxonomic assignment of RP4-Tn-luc-harbouring isolates based on phenotypic characteristics and sequencing of the 16S rRNA gene revealed 10 16S rRNA genotypes (Table 3). At the first sampling date, transconjugants of eight taxa of the α- and γ-Proteobacteria were detected. In cuttings inoculated with Stenotrophomonas 169-1 RP4-Tn-luc and Pseudomonas E9 RP4-Tn-luc, the plasmid had been transferred to members of the genera Sphingomonas, Xanthomonas, Rhizobium, Methylobacterium and Stenotrophomonas. In contrast, conjugation with the donor Pseudomonas Q16 RP4-Tn-luc predominantly resulted in the occurrence of transconjugants of the genus Stenotrophomonas. Four weeks later, only two genera—Achromobacter (β-Proteobacteria) and Stenotrophomonas—were found in the indigenous bacteria carrying the plasmid (Table 3). Thus, for the entire observation period, gene transfer in the genus Stenotrophomonas contributed obviously to the maintenance of the plasmid in the bacterial community of the poplar cuttings.

Table 3.

Taxonomic classification of isolates carrying RP4-Tn-luc.

	3 weeks after inoculation			10 weeks after inoculation
16S rRNA genotype	169-1	E9	Q16	169-1	E9	Q16
Donor strains (RP4-Tn-luc)
Stenotrophomonmas 169-1	–			–
Pseudomonas E9		42			–
Pseudomonas Q16			78			–
Transconjugants
α-Proteobacteria
Rhizobium 1		1
Rhizobium 2a	11	8
Methylobacterium		2
Sphingomonas 1		2
Sphingomonas 2	42	11	3
β-Proteobacteria
Achromobacter				3
γ-Proteobacteria
Stenotrophomonas 1			43	1		3
Stenotrophomonas 2		1
Xanthomonas	19	32
Sum of isolates	72	99	124	4	–	3

aThis group of isolates is represented by two similar 16S rRNA genotypes. Because the isolates of the group were not distinguishable based on phenotypic characteristics, both groups were considered as one group.

As shown in Fig. 1, two of the Rhizobium genotypes were phylogenetically related to R. radiobacter (formerly Agrobacterium tumefaciens), whereas the third genotype showed a high similarity to R. soli. A phylogenetic differentiation was also found within the genus Sphingomonas. Most isolates of this genus showed homology to S. yanoikuyae, and two isolates were related to S. faenia. The genus Methylobacterium was represented by two isolates with a high similarity to M. mesophilicum. Betaproteobacteria represented by the genus Achromobacter were detected based on three isolates with nearest relationship to A. piechaudii. The remaining 16S rRNA genotypes were γ-Protebacteria of the genera Stenotrophomonas and Xanthomonas. Isolates identified as members of Xanthomonas and Stenotrophomonas 1 displayed identical RFLP patterns, but clearly differed in colony colour. Xanthomonas transconjugants detected after two different inoculations had identical 16S rRNA gene sequences and were most nearly related to X. campestris und X. populi. By contrast, transconjugants of the genus Stenotrophomonas represented two genotypes that were similar to S. rhizophilia or S. chelatiphaga, and each was distinguishable from the donor stain Stenotrophomonas 169-1 RP4-4-Tn-luc.

Figure 1.

Dendrogram showing the similarity between the transconjugants and related bacterial species. The phylogenetic tree was generated using the neighbor-joining algorithm. The 16S rRNA gene sequence of E. coli (J01695) was used as an outgroup. Type strains of the bacterial species were used for the comparison; the sequence accession numbers are provided in brackets. Numbers at nodes indicate levels of bootstrap support >75%. Strains used for inoculation are underlined. 16S rRNA genotypes Rhizobium 1 were represented by S169-III-8 and S169-34, Rhizobium 2 by P9-10, Sphingomonas 1 by P9-50, Sphingomonas 2 by S169-9, Stenotrophomonas 1 by S169-III-5 and Stenotrophomonas 2 by P9-23. The 16S rRNA gene sequences of the transconjugants written in bold letters were deposited in the EMBL database (HE652088–HE652100).

The donor strains Pseudomonas Q16 and E9 differed noticeably in colony morphology and were found to be related to P. graminis or P. rhizosphaerae.

DISCUSSION

We introduced the broad-host-range plasmid RP4-Tn-luc into the indigenous microflora of poplars by inoculating cuttings with three endophytic bacterial strains as donors. The luc gene served as a model for genes conferring growth promoting traits and was used to trace the plasmid by bioluminescence and real-time quantification. Inoculation at all three phenological stages revealed the nearly stable establishment of the plasmid and the corresponding bioluminescence trait. The culture-independent approach always gave higher population densities of recombinant bacteria in comparison to cultivation, which is well known from studying bacterial communities in different habitats (Yang et al. 2001; Kent and Triplett 2002). This discrepancy can be explained by limitations of bacterial culture and by the detection of dormant and dead cells as well as free bacterial DNA with PCR (Blagodatskaya and Kuzyakov 2013). Furthermore, these factors can vary during plant growth (Saito et al. 2007). To assess viable and metabolically active bacteria, data from the cultivation-based approach should be taken into account.

Against this background, the results obtained by both approaches are consistent with other studies demonstrating the plant colonisation by endophytic bacteria or the persistence of their plasmids (Taghavi et al. 2005; Compant et al. 2008; Sun et al. 2014). We showed that the survival of the introduced strains and the transfer of the plasmid into the indigenous community differed at the various phenological stages of the plants. After inoculation during leaf development and senescence, all strains survived by maintaining nearly stable densities. Both at the beginning and the end of vegetation, the plants were only weakly colonized, which obviously favoured the establishment of the strains. In contrast, the abundance of the introduced bacteria finally decreased below the detection limit when inoculating cuttings at the stem elongation stage. During this period, bacterial colonization was about two orders of magnitude higher than in the other two stages. Adapted indigenous bacteria living in complex interactions with their hosts (van der Lelie et al. 2009; Hibbing et al. 2010) and known as plant genotype specific (Ulrich, Ulrich and Ewald 2008) seemed to be competing with the inoculated strains. Accordingly, 3 weeks after inoculation, the Stenotrophomonas donor strain—originating from a distantly related poplar clone—could not be recovered. The two Pseudomonas strains—once isolated from the clone used for inoculation—were still present at a similarly high level at least 3 weeks after inoculation and therefore demonstrated greater competitiveness. In general, a decrease in the inoculated strains at the stem elongation stage was accompanied by the occurrence of transconjugants. Similar results were reported by Taghavi et al. (2005), who inoculated poplar cuttings with two Burkholderia strains containing a plasmid conferring toluene degradation. After 10 weeks, none of the strains became established, but the plasmid was shown to be transferred into the indigenous community.

In our study, gene transfer was associated with an increased abundance of potential plasmid recipients. However, it is safe to assume that factors related to the plant development and metabolism as well as the plant areas used as microhabitats for the bacteria represent a stronger driver for transfer events than a suitable density of recipients (Normander et al. 1998; Mølbak et al. 2003). For example, Lilley and Bailey (1997) detected transconjugants only within a narrow temporal window coincident with the midseason maturation of sugar beets. Moreover, Mølbak et al. (2003) revealed that the hypocotyl and the root areas were metabolically active regions that enabled plasmid transfer frequencies at detectable levels. In general, a correlation between the numbers of transconjugants and the density of potential plasmid recipients was not observed (Björklöf et al. 1995; Lilley and Bailey 1997). In this study, synergetic effects of an optimal plant growth stage providing conditions as known for protected biofilms (Sørensen et al. 2005) and a strong physical contact due to the high abundance of donor cells and putative recipients might contribute to the frequent plasmid transfer and establishment of the transconjugants after inoculation in June.

The plasmid RP4-Tn-luc was transferred into six genera of Proteobacteria. All these genera were previously described as endophytes of poplar (Ulrich, Ulrich and Ewald 2008; van der Lelie et al. 2009). Two of the genera belong to γ-Proteobacteria that were identified as the most frequent recipients for IncP plasmids in natural environments (Sørensen et al. 2005). Plasmid transfer events observed in the phytosphere are rare. Taghavi et al. (2005) showed a predominant transfer of the TOM plasmid (IncP-1) in different genera of γ-Proteobacteria as well.

Interestingly, plasmid transfer to the genus Pseudomonas was not observed in this study, although two of the donor strains belong to this genus and Pseudomonas is known to be a predominant genus of endophytic bacteria in poplar (Ulrich, Ulrich and Ewald 2008). Altogether, the detection of transconjugants within the three classes of Proteobacteria demonstrated the wide host range of the plasmid within the indigenous bacteria of poplar. A transfer beyond the Proteobacteria, e.g. to the phyla Actinobacteria and Firmicutes, could only be shown in more complex habitats such as the rhizosphere or bulk soil, and has only been demonstrated by a culture-independent approach; this is likely restricted to the plasmid transfer and did not include a stable replication (Musovic et al. 2006; Shintani et al. 2014; Klümper et al. 2015). Against this background, we found a widespread horizontal gene transfer in the poplar phytosphere.

The conjugative transfer of plasmids allows the possibility of introducing desired traits regardless of the competitiveness of the inoculated strains. This could change our understanding of the use of plasmids to improve the plant-promoting bacterial community, from the risk of plasmid loss to possible advantages of the establishment of desired traits.