Editor: Julian Marchesi
Evidence for particle-induced horizontal gene transfer and serial transduction between bacteria
Article first published online: 23 MAR 2011
© 2011 Federation of European Microbiological Societies. Published by Blackwell Publishing Ltd. All rights reserved
FEMS Microbiology Ecology
Volume 76, Issue 3, pages 576–591, June 2011
How to Cite
Chiura, H. X., Kogure, K., Hagemann, S., Ellinger, A. and Velimirov, B. (2011), Evidence for particle-induced horizontal gene transfer and serial transduction between bacteria. FEMS Microbiology Ecology, 76: 576–591. doi: 10.1111/j.1574-6941.2011.01077.x
- Issue published online: 6 MAY 2011
- Article first published online: 23 MAR 2011
- Accepted manuscript online: 1 MAR 2011 06:09AM EST
- Received 9 November 2010; revised 7 February 2011; accepted 14 February 2011., Final version published online 23 March 2011.
- virus-like particles;
- membrane vesicles;
- marine bacteria;
- horizontal gene transfer (HGT)
Incubation of the amino acid-deficient strain Escherichia coli AB1157 with particles harvested from an oligotrophic environment revealed evidence of horizontal gene transfer (HGT) with restoration of all deficiencies in revertant cells with frequencies up to 1.94 × 10−5. None of the markers were preferentially transferred, indicating that the DNA transfer is performed by generalized transduction. The highest gene transfer frequencies were obtained for single markers, with values up to 1.04 × 10−2. All revertants were able to produce particles of comparable size, appearing at the beginning of the stationary phase. Examination of the revertants using electron microscopy showed bud-like structures with electron-dense bodies. The particles that display the structural features of membrane vesicles were again infectious to E. coli AB1157, producing new infectious particles able to transduce genetic information, a phenomenon termed serial transduction. Thus, the <0.2-μm particle fraction from seawater contains a particle size fraction with high potential for gene transfer. Biased sinusoidal field gel electrophoresis indicated a DNA content for the particles of 370 kbp, which was higher than that of known membrane vesicles. These findings provide evidence of a new method of HGT, in which mobilizable DNA is trafficked from donor to recipient cells via particles.
It is well documented that viruses in aquatic environments play an important role in regulating bacterial biomass and transferring genetic elements between bacteria (Fuhrman, 1999). Considering narrow host specificity, the latter contribution may be meaningful only among closely related strains or species (Schicklmaier & Schmieger, 1995). However, there are many indications that gene transfer may also occur among more phylogenetically divergent bacteria (Jiang & Paul, 1998; DeLong et al., 2006), which significantly increases the magnitude of the transferred genetic information within the prokaryotes. Prokaryotes are known to be unique in their ability to react to environmental changes by the rapid acquisition of the necessary genetic traits for continued survival. This genetic flexibility is, besides their very short generation time, mainly due to the natural existence of seemingly efficient means for horizontal gene transfer (HGT) among bacteria, as well as between bacteria and other organisms (Lengeler et al., 1999). Virus-mediated transfer of genetic elements between bacteria has become a major research topic in the last two decades, whereby conjugation, transduction and transformation are well-investigated mechanisms resulting in HGT between prokaryotic organisms. However, considering viral host specificity, the relatively short length of transferable DNA segments and the lack of de novo particle production from the transductants (Ochman et al., 2000; Brüssow et al., 2004), one may conclude that viruses are not always the major players in gene transfer, given that recent developments in microbial genomics and metagenomic approaches revealed the existence of reminiscent genes from exotic origins (Lawrence et al., 2002). Novel modes of lateral gene transfer have been described in the past decades. The particle sizes of gene transfer agents (GTAs) range between 30 and 80 nm (Marrs, 1974; Wall et al., 1975; Barbian & Minnick, 2000; Lang & Beatty, 2000, 2002, 2007; Biers et al., 2008; Zhao et al., 2009), and a report of McDaniel et al. (2010) states that GTA gene transfer frequencies are a thousand to a hundred million fold higher than prior estimated for HGT in the oceans. It was presumed that some 47% of the culturable natural microbial community might be seen as gene recipients. Similarly, vesicle trafficking was mentioned as a possible mechanism for the transfer of DNA between prokaryotes. However, this was rather considered as a sideline of investigations dealing with interspecies communication (Kadurugamuwa & Beveridge, 1999; Marketon et al., 2002; Mashburn & Whiteley, 2005; Mashburn-Warren & Whiteley, 2006), delivery of toxins (Kuehn & Kesty, 2005; McBroom & Kuehn, 2005) or antibiotic resistance determinants (Ciofu et al., 2000; Wai et al., 2003). It is therefore plausible that other unknown, but dynamic mechanisms function in aquatic systems to promote lateral gene exchange. Within the framework of a series of studies on HGT between bacteria in marine (Chiura, 1997) and thermal environments (Chiura, 2002; Chiura & Umitsu, 2004), it was shown that virus-like particles (VLPs) that were infectious to Escherichia coli are spontaneously produced by budding and released in the stationary phase from the following: Alphaproteobacteria such as Ahrensia kieliensis and Flavobacterium spp. 16-04, ubiquinone-possessing marine bacteria (Q10MB) and Aquificales cells (Chiura et al., 2000, 2002). A number of these experiments led to the observation that the recipient E. coli cells acquire the ability to produce new particles in the size range between 100 and 130 nm, and molecular biological analyses revealed that the nucleic acid species encapsulated in such particles was DNA (Chiura, 2004). In the context of the above observations, it was postulated that such particle production may also take place in the ocean's water column and would possibly reinforce the currently known mechanisms of HGT between bacteria. Because it is impossible to identify these particles in water samples by normal epifluorescence microscopy, we decided to work with a concentrated VLP fraction (<0.2 μm diameter particles) and test whether one would obtain recipient bacteria that could produce new particles. Particle sizes between 100 and 130 nm were selected from this size fraction in order to eliminate possible interferences with large and tailless GTA particle types such as those derived from Silicibacter pomeroyi DSS-3 cultures (Biers et al., 2008) from Bartonella spp. (Barbian & Minnick, 2000) or from Methanococcus voltae (Bertani, 1999), which all have particle diameters in the size range of 80 nm. Furthermore, the size class chosen was consistent with the previously investigated particles obtained from Aquificales (Chiura, 2002). It should be emphasized that particle production in recipient bacterial cells is a feature that is characteristic of infections by bacteriophages, but not for VLPs in the abovementioned size range, which strengthens the assumption that a nonviral transfer of genetic material may occur.
The purpose of the present investigation was to test whether particles within the mentioned size fraction from natural seawater trigger the abovementioned features in recipient cells. Furthermore, we attempted to ultrastructurally characterize the previously mentioned particles and subsequently quantify the postulated induced gene transfer using the auxotrophic enteric bacterial mutant E. coli AB1157 strain as the recipient cell. The VLP fraction was harvested from samples of the oligotrophic waters of the Western Mediterranean Sea (i.e. distant from punctual or diffuse sources of pollution).
Materials and methods
Seawater samples (170 L total) were obtained at 42°36′N, 8°56′E, from a 5 m depth (19.8 °C) near the marine station STARESO at Calvi, Corsica, France. Samples were collected during the summer and autumn and transported to the laboratory for further treatment within 1 h.
Preparation of VLPs
The sea water was passed through a 0.2-μm Durapore membrane filter (Millipore, Billerica, MA) and concentrated to c. 20 mL by successively using a Pelicon cassette (Millipore) system and a Minitan Ultrafiltration System (Millipore) with a 30-kDa cut-off filter. The final average concentration rate was 2774-fold. Concentrated VLPs were treated with 10 μg mL−1 each of DNase I and RNase A at 25 °C overnight with 100 μM phenylmethylsulphonyl fluoride (Sigma) to exclude the possibility of gene transfer by transformation. The concentrate was filtered again through 0.45- and 0.22-μm membrane filter, and then it was centrifuged at 80 000 g for 30 min using a Beckman Preparative Ultracentrifuge L8M with a 55.2Ti rotor to pellet VLPs. The supernatant of this ‘final concentrate’ was recovered and used as a negative control in the subsequent experiments. The pellet was resuspended overnight at 4 °C in 200 μL TBT buffer (100 mM Tris-HCl, 100 mM NaCl and 10 mM MgCl2) by gentle rotation with a Slow Rotator (TAAB, UK). VLPs were purified by CsCl-density equilibrium ultracentrifugation at 174 400 g for 18 h. Purified particles were harvested after ultracentrifugation from the density fraction 1.2–1.6 g cm−3 (Chiura, 1997), and the resulting bands were separately recovered by the side puncture technique using 2.5 mL syringes (Terumo, Japan). CsCl was removed via dialysis using Spectra/Por 4 tubing (Spectrum, molecular weight cut-off, MCWO=12 000–14 000 Da) against five changes of 100 volumes of TBT buffer. VLP abundance and particle size distribution in the respective bands were examined using electron microscopy as described below. The protein and nucleic acid concentrations were determined photometrically by reading A260 nm and A280 nm using a Shimadzu Spectrophotometer Type UV260 (Shimadzu Corp., Kyoto, Japan). Additionally, the protein content was also determined using the Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA) following the specifications for microassays.
Enumeration of cells and VLPs
Viable cell counts were determined on Luria–Bertani (LB) solid medium incubated at 30 °C. The number of VLPs was determined according to Børsheim et al. (1990). After staining for 30 s with 2% uranyl acetate, grids were examined at × 75 000 magnification at a voltage of 80 kV using a Zeiss EM902 (Zeiss Inc., Germany) or a JEM-1200EX transmission electron microscope (Jeol Inc., Japan). At least 50 fields were selected for counting. The average burst was derived from the number of VLPs observed in a cell (n=2000).
Transmission electron microscopy (TEM)
For electron microscopic evaluation, cells were pelleted as described above and fixed in 3% glutaraldehyde (electron microscopy grade; Serva) in 0.1 M sodium cacodylate buffer for 60 min, osmificated in veronal-acetate-buffered OsO4 for 60 min, dehydrated in a sequential series of ethanol solutions and embedded in Epon. Thin sections (80 nm) were stained with uranyl acetate and lead citrate, and they were examined using a Tecnai-20 electron microscope (FEI Company, Eindhoven, the Netherlands) at 80 kV. Pictures were taken using a slow-scan CCD camera (Gatan, MSC 794).
Gene transfer assay
An auxotrophic mutant strain, E. coli AB1157 (F−; thr-1 leuB6 thi-1 lacY1 galK2 ara-14 xyl-5 mtl-1 proA2 his-4 argE3 rpsL31 tsx-33 supE44), was used as the recipient bacterium for testing VLP-mediated gene transfer (Chiura, 1997). We focused on the following markers: leucine (Leu), proline (Pro), histidine (His) and arginine (Arg). Threonine deficiency was not considered to be a useful marker due to its considerably high spontaneous reversion frequency of ∼10−7 per cell. For the other four markers, spontaneous reversion frequencies were below the level of detection (Chiura, 1997). The strain was obtained from the National Institute of Genetics (Shizuoka, Japan). To ensure reproducible physiological conditions of the recipient E. coli AB1157 for all subsequent experiments, the bacterium was cultured at 30 °C by shaking (120 r.p.m.) until a cell density of ∼4 × 108 CFU mL−1 was reached. After the addition of glycerine (7% final concentration), the culture was dispensed in 2-mL aliquot, frozen in liquid nitrogen and stored at −85 °C until further use. For the gene transfer assays, 2-mL aliquot of the frozen seed culture were mixed with 3 mL of a fresh LB broth in L-shaped test tubes and grown to the mid-exponential phase at 30 °C. Then, they were centrifuged at 5000 g in a refrigerated centrifuge (Kubota TR 20000); the pellet was resuspended in 5 mL TBT buffer. This suspension yielded a viable count of ∼2 × 108 CFU mL−1. One millilitre of this suspension was mixed with an aliquot volume of the harvested VLP fraction to obtain the various multiplicities of infection (MOIs) of 0.12, 1, 5.12, 20 and 200. The tubes were left undisturbed at 30 °C for 15 min. After incubation, cells were washed with a Davis salt solution and finally suspended in 1 mL of the same solution. The mixture was plated in triplicate on appropriate selection media and incubated for 2 days at 30 °C. The following controls were included in the assays: (1) UV-irradiated VLPs added to recipient cells; (2) recipient cells with Davis buffer instead of VLPs to determine spontaneous reversion rates; (3) autoclave-inactivated VLPs added to recipient cells; (4) recipient cells with the supernatant of the respective final VLP concentrate; (5) UV-irradiated VLPs without recipient cells to control for possible colony formation within the VLP fraction; and (6) untreated VLPs without recipient cells to control for possible colony formation from within the VLP fraction.
Escherichia coli AB1157 colonies that reverted for one or more amino acid markers were designated as revertants. All generated revertants were also subjected to examination for unselected marker transfer. For this purpose, positive colonies for one selection medium were transferred to the remaining three selection media to test whether colony formation would occur.
Lethal effect of particles on E. coli AB1157
Recipient E. coli AB1157 cells from frozen seed cultures were grown at 30 °C in LB broth, harvested in the mid-exponential phase (5000 g) and resuspended in 5 mL TBT. The viable count of the cell suspension corresponded to 2 × 108 CFU mL−1. Harvested particles were added to the cell suspensions to obtain MOIs of approximately 0.1, 1, 5, 20 and 200 and left undisturbed for 15 min at 30 °C. Subsequently, cells were washed with Davis solution and then resuspended in 1 mL of the same solution. After dilution with Davis salt solution, the cells were plated in triplicate on LB medium and incubated for 2 days at 30 °C.
The following controls were included: (1) recipient cells plus TBT buffer, but without particles, to determine the viable cell count; (2) UV-treated particles (15-min irradiation with a 15 W sterilizing lamp; Phillips, the Netherlands) plus recipient cells; (3) autoclaved particles (121 °C, 200 kPa, 20 min) plus recipient cells; and (4) recipient cells with the supernatant of the respective final concentrate of particles.
The supernatant was obtained by ultracentrifugation at 55 000 g for 45 min to precipitate particles and subsequently filtered through a 0.2-μm membrane. Control experiments were performed using 20 μL coliphage T4 under the above conditions.
The lethal effect of particles on recipient cells was expressed as efficiency of plating (EOP) calculated as % of CFU formed as compared with the number of CFU in TBT control plates.
Detection of VLP production during incubation of revertants
A revertant grown on a minimal agar plate (no amino acid-supplemented minimal agar after Davis) at an MOI of 5.1 was incubated at 30 °C for 370 h in 3.4 L of leu+ medium (leucine-supplemented minimal broth after Davis) to examine its growth profile and compare it with that of the recipient E. coli AB1157. For this purpose, 0.5-mL sample were intermittently withdrawn to determine the cell numbers and the number of free particles using epifluorescence microscopy. Samples filtered through Anodisc filters were stained with SYBR Gold (Invitrogen Corp., Carlsbad, CA) and mounted according to Noble & Fuhrman (1988). The proportion of particle-bearing cells and the number of particles per cell (burst size) were determined from 0.1-mL sample using electron microscopy.
Free particles produced during prolonged incubation were harvested, purified and tested for gene transfer capability using the same method described above.
Purified particles that were newly generated from the 370-h culture of the first revertant were subjected to a second gene transfer experiment with E. coli AB1157 in the same manner as described above. The revertants obtained from the second transduction experiment were again cultured in the same medium (leu+) at 30 °C for 167 h to determine the growth profiles, particle production and burst size.
Nucleic acid extraction from particles produced by revertants in agarose gel plugs (in situ lysis gel)
The nucleic acid species of purified particles from a leu+ revertant were examined electrophoretically using in situ lysis. For this purpose, samples were embedded in an agarose gel block. The procedures for the preparation of agarose gel plugs followed the recommendations of the agarose manufacturer (InCert Agarose).
For nucleic acid molecular type estimation, in situ lysis gel plugs were embedded in 1.0% agarose ME (SeaKem, UK). Before embedding, the gel plug was sliced using a microscope cover slip to obtain a 3.0 × 3.0 × 1.0 mm block. The separating agarose gel, liquefied 1.0% agarose in TBE buffer (89 mM Tris-HCl, 89 mM boric acid and 0.5 M EDTA, pH 8.0), was poured into a casting mould to prepare a gel that was 6 mm thick, to which the sliced gel plug (attached to a slot-making comb) was applied. The cast gel was incubated at ambient temperature for 30 min, and then it was placed in a refrigerator overnight to solidify completely.
Biased sinusoidal field gel electrophoresis (BSFGE) was used to run gels in 0.5 × TBE buffer at room temperature for 30 h under the following conditions: 1.6 V cm−1 DC, 9.6 V cm−1 AC, a start frequency of 0.01 Hz, an end frequency of 0.3 Hz and a logarithmic ramp. The molecular mass standards used were Lambda ladder (48.5 kb–1.2 Mbp, FMC) and λ/HindIII (0.13–23.13 kbp, Nippon Gene, Japan). Nucleic acid species were visualized by staining the gel with a 1/10 000 SYBR Green II stock solution (Molecular probe) at room temperature for 1 h, followed by illumination with a Spectroline Trans Illuminator Model TC-365A (Spectronics) at 360 nm. The results were recorded using a Canon PowerShot Pro 70 digital camera. The concentration of nucleic acid was determined from the pictures obtained using the public domain nih image program (developed at the U.S. National Institute for Health and available online at http://rsb.info.nih.gov/nih-image/).
Determination of nucleic acid type content in the particle
After BSFGE, bands from the leu+ revertant-produced particles were excised from the gel, and each gel slice was placed in a Falcon tube. DNase I (10 mg mL−1, Sigma) and RNase A (10 mg mL−1, Sigma) were then added to the tubes. Four gel slices from the particles were treated with DNase I, RNase A, distilled water and TM buffer overnight at 37 °C. The gel slices with and without treatment were again examined with BSFGE for 16 h and the results were recorded as described above.
DNA-antibody labelling of TEM thin sections
For DNA detection, bacterial samples were fixed and embedded as indicated above. Epon sections (80 nm) were etched with 3% sodium metaperiodate for 40 min, rinsed in A. dest. and preincubated with blocking solution B1 (1% bovine serum albumin and 1% goat serum in phosphate-buffered saline–Tween 20, pH 7.4), followed by a 2-h incubation with a monoclonal anti-DNA antibody (Ac-30-10; Progen, Heidelberg, Germany) in B1. Detection was performed using an anti-mouse colloidal gold-labelled secondary antibody. Control incubations were performed by excluding the anti-DNA antibody.
Size distribution of VLPs and ultracentrifugation
The average number of bacteria and VLPs in the three samples were 1.07 × 106 cells mL−1 (SD=0.43) and 1.17 × 108 particles mL−1, respectively. The size distribution of VLPs in the untreated oligotrophic sample and the 0.2-μm filtered seawater (Table 1) revealed that there was no significant difference in particle number or size. The concentration of the particles in the 0.2-μm filtrate by tangential flow filtration yielded a particle recovery ranging from 44% to 78%, while the use of ultracentrifugation resulted in the complete recovery of particles.
|Sample||Size classes (nm)|
|Proportion ± SD (%)||57.2 ± 17.4||36.0 ± 12.5||6.8 ± 7.6|
|Size ± SD (nm)||49.8 ± 8.7||71.3 ± 11.0||111.8 ± 25.4|
|0.2-μm filtrate (n=15)|
|Proportion ± SD (%)||65.0 ± 16.6||30.0 ± 11.9||5.0 ± 2.7|
|Size ± SD (nm)||46.9 ± 8.9||72.9 ± 10.8||107.7 ± 24.6|
Three bands were obtained (Table 2) by CsCl-density equilibrium ultracentrifugation of the 0.2-μm filtrate; the middle band with the highest protein/nucleic acid ratio and a corresponding particle diameter size range from 118 to 128 nm was used for all of the following assays (Chiura, 2002).
|Particles ± SD × 1012||0.891 ± 0.615||4.15 ± 1.64||4.36 ± 0.73|
|Particle proportion (%)||9.2 ± 6.4||42.9 ± 17.0||45.1 ± 7.5|
|Nucleic acids (μg)||20.53||12.38||10.39|
Gene transfer experiments
Exposing the amino acid-deficient E. coli AB1157 cells to the harvested particle fraction resulted in the restoration of the genetic deficiencies for all MOIs (Table 3). In the assays with MOIs of 5.1, 20 and 200, we obtained a restoration of all four amino acid synthesis deficiencies. The lowest transfer frequency of 1.39 × 10−7 was noted for MOI=5.1, and the highest frequency was 1.94 × 10−5 for MOI=200. Exposure of E. coli AB1157 at MOIs of 0.12 and 1.0 yielded no comparable results. The reversion of either Leu, Pro, His or Arg synthesis in E. coli AB1157 was observed for all MOIs (Table 3). The highest gene transfer frequency for Leu and His was 1.04 ± 1.62 × 10−2 and 9.83 ± 3.88 × 10−3, respectively, and both were at MOI=20. However, the transfer frequency for the synthesis of Pro and Arg was always the highest at MOI=5.1, with 8.34 ± 2.38 × 10−3 and 9.73 ± 1.90 × 10−3, respectively. The overall trend of the data from Table 3 indicates an increase in the transfer frequency of selected markers with increasing MOI and maximum values at MOIs of 5.1 and 20. At MOI=200, the frequencies again decreased. However, none of the markers was preferentially transferred, indicating that the DNA transfer was performed by generalized transduction.
|Experiment||All amino acids reverted||0||0||1.39 × 10−7 (2)||1.46 × 10−5 (1)||1.94 × 10−5 (12)|
|Leu reverted||4.41 ± 3.04 × 10−6 (10)||3.33 ± 3.22 × 10−5 (9)||9.04 ± 2.85 × 10−3 (13)||1.04 ± 1.62 × 10−2 (714)||7.34 ± 3.04 × 10−4 (443)|
|Pro reverted||5.73 ± 2.82 × 10−6 (13)||2.22 ± 2.38 × 10−5 (6)||8.34 ± 2.38 × 10−3 (12)||7.42 ± 1.12 × 10−3 (509)||7.15 ± 0.28 × 10−5 (43)|
|His reverted||3.53 ± 0.92 × 10−6 (9)||7.41 ± 1.76 × 10−5 (2)||9.04 ± 3.07 × 10−3 (13)||9.83 ± 3.88 × 10−3 (675)||7.34 ± 4.13 × 10−4 (443)|
|Arg reverted||3.09 ± 0.88 × 10−6 (7)||1.48 ± 0.52 × 10−5 (4)||9.73 ± 1.90 × 10−3 (14)||6.03 ± 0.92 × 10−3 (414)||5.83 ± 1.08 × 10−4 (352)|
|Control||VLPs+UV ± AB1157 VLPs aut.+AB1157 VLPs aut./untr. ± AB1157||0||0||0||0||0|
|Davis buffer+AB1157 or UC supernat.+AB1157||0||0||0||0||0|
To define the most favourable conditions for the gene transfer, with respect to the number of particles to which a recipient cell was exposed in our experiments, we calculated the specific gene transfer frequencies (i.e. the ratio of the average transfer frequency of the restored amino acids divided by the respective MOI). The highest transfer frequency was obtained for MOI=5.1, with a mean of 1.41 ± 2.50 × 10−3. It should be noted that the effect of particle exposure on the recipient cells (Table 4) indicated an increase in cell mortality with increasing MOI. Additionally, exposing recipient cells to UV-treated particles had a similar effect on mortality compared with the exposure to native particles. At an MOI of 5, we still obtained an EOP amounting to 48.2% and 64.9% for exposure to native particles and UV-treated particles, respectively. However, these values declined to 15.2% and 17.8% at an MOI of 200. An increase in mortality was also observed when recipient cells were exposed to the ultracentrifugation supernatant of the respective final concentrate of particles. Notably, this increase was less pronounced, and at an MOI of 200, we recorded 62.6% CFUs. Only a marginal lethal effect was caused by exposure to autoclaved VLPs; the EOP ranged from 98.3% to 94.0% for MOIs of 0.1–200, respectively.
|UV-treated VLPs||Native VLPs||Davis buffer||VLP autoclaved||UC Supernatant|
|MOIs||EOP ± SD (%)|
|0.1||100.0 ± 9.8||100.0 ± 9.9||100.0 ± 9.1||98.3 ± 0.5||98.3 ± 0.3|
|1||76.5 ± 9.8||64.8 ± 9.8||100.0 ± 9.1||98.3 ± 6.3||95.8 ± 6.1|
|5||64.9 ± 1.7||48.2 ± 0.2||98.7 ± 0.5||98.4 ± 4.4||93.5 ± 1.0|
|20||35.3 ± 1.4||20.8 ± 0.6||99.4 ± 2.7||94.0 ± 6.9||86.3 ± 0.0|
|200||17.8 ± 0.5||15.2 ± 0.0||98.2 ± 3.1||94.0 ± 6.9||62.6 ± 5.8|
Unselected marker restoration was examined for revertants obtained at MOIs of 0.12, 1.0 and 5.1, while those for MOIs of 20 and 200 were excluded. Additionally, Table 5 shows that for unselected marker restoration, the gene transfer frequencies were unexpectedly high, especially at an MOI of 5.1. Although the unselected marker transfer was low for the selected marker (His) at an MOI of 0.12 and absent at an MOI of 1.0, we obtained 84.6–92.3% of restored amino acid synthesis at an MOI of 5.1. For the other three markers, unselected marker restoration accounted for 69.2–92.9% at this same MOI. At the lower MOIs, unselected marker restoration remained low, ranging from 10.0% to 42.9% at an MOI of 0.12 and between 0.00% and 33.3% at an MOI of 1.0. This implies that the presence of five particles per cell results in unselected marker restoration that is three times greater than one potentially infective particle per cell.
|Unselected marker||Selected marker|
|MOI: 0.12, total colonies formed: 39|
|Leu reverted (10)||Pro reverted (13)||His reverted (9)||Arg reverted (7)|
|Leu+||–||7.7 (1)||11.1 (1)||14.3 (1)|
|Pro+||10 (1)||–||11.1 (1)||42.9 (3)|
|His+||20 (2)||15.4 (2)||–||28.6 (2)|
|Arg+||10 (1)||7.7 (1)||11.1 (1)||–|
|MOI: 1.0, total colonies formed: 22|
|Leu reverted (9)||Pro reverted (6)||His reverted (3)||Arg reverted (4)|
|Leu+||–||16.7 (1)||0 (0)||0 (0)|
|Pro+||22.2 (2)||–||0 (0)||25.0 (1)|
|His+||33.3 (3)||33.3 (2)||–||25.0 (1)|
|Arg+||22.2 (2)||16.7 (1)||0 (0)||–|
|MOI: 5.1, total colonies formed: 66|
|Leu reverted (13)||Pro reverted (12)||His reverted (13)||Arg reverted (28)|
|Leu+||–||75.0 (9)||84.6 (11)||78.6 (22)|
|Pro+||84.6 (11)||–||92.3 (12)||85.7 (24)|
|His+||84.6 (11)||91.7 (11)||–||92.9 (26)|
|Arg+||69.2 (9)||75.0 (9)||84.6 (11)||–|
To study and quantify the growth behaviour of the obtained revertants in a liquid medium and to test whether particle formation would take place, two revertants were selected, and their growth behaviours were compared with that of the parental strain.
Growth behaviour of revertants and particle production
Two revertants obtained in minimal medium at an MOI of 5.1 in this study (arbitrarily named Calvi-E-trans-F1a and Calvi-E-trans-F1b) were examined, and their growth characteristics were compared with that of the parental E. coli AB1157 cells. Because both strains displayed similar growth characteristics, the growth curves of the parental strain E. coli AB1157 and of one transductant (Calvi-E-trans-F1a) are presented in Fig. 1a, showing that there were only minor differences in the profiles of these strains. AB1157 reached the stationary phase after 50 h, while the revertant strain required nearly 80–90 h, and its density remained slightly above that of AB1157 during the 370 h of observation. The appearance of particles, although at very low quantities, was observed after only 12 h in the liquid medium, and the maximum density was observed between 50 and 60 h when the transcolony strain entered the stationary phase; the maximum density reached 0.8 × 109 VLP mL−1. Once the stationary phase of the host cell was well established (after 130–140 h), the particles range remained low and varied between 6.8 × 108 and 8.3 × 108 VLP mL−1 (Fig. 1b). No particle formation was observed in the liquid medium for the original AB1157 strain (data not shown).
Inspection of revertant cells by electron microscopy revealed that the number of infected cells, indicated by the number of particle-bearing cells, ranged from 4% to 23% during the stationary phase (Fig. 1c), with the average burst size per cell ranging from one to four particles (Fig. 1d). Particle formation was observed both in the previously mentioned strain and in all revertant strains that were retained for later use.
CsCl-density equilibration centrifugation of the harvested particles yielded five bands; of these five, we chose the middle band for the second transduction experiment. The size range of the corresponding particles was 124.91 ± 11.5 nm (n=69), which was comparable to the range of the particles that were selected from the 0.2-μm seawater filtrate for the production of the first revertants.
Molecular biological and ultrastructural characterization of VLPs produced by Calvi-E-trans-F1a
Several Calvi-E-trans-F1a cultures were separately incubated for 165 h in Minimal Medium (MM) (3.8 L) to allow the harvest of both the particles produced for further analysis and bacterial cells for TEM observation. Alternative treatment of revertant cells and VLPs with RNase and DNase, followed by BSFGE demonstrated that the particles contained DNA (Fig. 2).
Figure 3 demonstrates that revertant E. coli cells (lane T and lanes 6–12) are characterized by two distinct DNA fractions: the E. coli genome and a DNA molecule of ∼360 kbp. However, E. coli AB1157 (before and at the beginning of the infection; lanes E1, E2, and 1 and 2) is characterized only by the genomic DNA fraction. The isolated VLPs contained the 360-kbp band. Thus, even though the particles observed varied in size between 100 and 140 nm, they contained DNA with a generally uniform molecular mass of ∼370 kbp.
TEM inspection of the E. coli AB1157 revertant cells during growth over 165 h provided morphological evidence of particle production by the revertant Calvi-E-trans-F1a; it also revealed previously unseen structures in E. coli revertant cells before and during particle production. Figure 4a and b demonstrate that particle production is achieved by budding, which accounts for the observed low mortality of the revertants during growth. This phenomenon is in agreement with the fact that during the 370-h incubation of Calvi-E-trans-F1, the stationary phase was maintained without a noticeable decrease in cell density, indicating that mortality of the cultured cells was low. Also, the addition of the particles from the concentrated sample of the seawater onto lawns of AB1157 cells did not result in plaque formation compared with the addition of coliphage T4, which produced clear plaques, thus indicating virulence (data not shown).
TEM observation of various stages of revertant cells during particle formation revealed the appearance of electron-dense bodies (EDBs) in the cells (Figs 4b and 5). These bodies were often in close vicinity to a connected electron-dense network. Figure 5a shows EDBs in the recipient cells at different degrees of electron density, Fig. 5b and c indicate situations where the EDBs are in close contact with the inner layer of the cell membrane and Fig. 5d shows a double-membraned budding structure with an EDB and displays the features of a membrane vesicle. To argue that these EDBs are precursors of the observed budding particles, a basic assumption is that these bodies contain DNA. Incubation of gold-conjugated DNA-antibodies indicated discrete clusters of DNA-antibody-binding sites in the EDBs in the revertant cells (Fig. 6a–c). Revertants without EDBs were used as controls (Fig. 6d) and showed no or sparse DNA-antibody binding. Figures 5 and 6c provide evidence that the EDBs that appear in the recipient cells vary not only in shape but also in composition. While gold-conjugated DNA-antibody complexes are densely clustered on and around the dark structures within the cells (as seen in Fig. 6a and b), Fig. 6c shows a concentration of gold grains aligned along half of the circumference of the electron-dense material within the cell. Thus, the data suggest that EDBs appearing in the recipient cells vary in shape and in the composition of their constituents.
For this experiment, harvested and purified particles derived from Calvi-E-trans-F1a were added to the recipient E. coli strain at an MOI of 5.5, which was assumed to yield the most efficient gene transfer with respect to both the selected marker and unselected marker restoration, as mentioned above. Assessing the effect of particle exposure on recipient cells at an MOI of 5.5 again showed a particle-induced mortality of 33.8%, which is comparable to the mortality obtained via UV-irradiated particles (32.2%). The results of the second gene transfer experiments are presented in Table 6. The generated revertants were arbitrarily named Calvi-E-trans-F2. Using a 10−1 dilution, revertants of Leu+ (42 colonies), Pro+ (42 colonies), His+ (31 colonies) and Arg+ (46 colonies) were generated. The highest gene transfer frequency was obtained for the Leu marker (12.3 ± 18.10 × 10−6), which was two to three times higher than for the other markers. Revertants where the synthesis of all amino acids was restored were generated on MM plates at a frequency of 2.0 ± 1.3 × 10−6. The gene transfer frequency was reduced by three orders of magnitude compared with the results obtained with the particles from the Calvi-seawater concentrate at nearly the same MOI. However, the unselected marker restoration test again demonstrated that there is a strong linkage between selected and unselected markers (Table 7). The highest unselected marker transfer was observed for the His revertants, with an unselected marker transfer of 96.8% for Pro; all other restorations of amino acid synthesis ranged from 67.4% to 90.5%.
|All amino acids reverted||Leu reverted||Pro reverted||His reverted||Arg reverted|
|Transcolonies||2.0 ± 1.3 × 10−6||12.3 ± 18.1 × 10−6||8.77 ± 8.74 × 10−6||3.16 ± 0.65 × 10−6||12.0 ± 1.9 × 10−6|
|VLPs+UV ± AB1157 VLPs aut.+AB1157 VLPs aut./untr. ± AB1157||0||0||0||0||0|
|Davis buffer+AB1157 or UC supernat.+AB1157||0||0||0||0||0|
|Unselected marker||Total colonies formed: 161|
|Leu reverted (42)||Pro reverted (42)||His reverted (31)||Arg reverted (46)|
|Leu+||–||90.5 (38)||90.3 (28)||67.4 (31)|
|Pro+||72.4 (21)||–||96.8 (30)||71.7 (33)|
|His+||82.7 (24)||90.5 (38)||–||82.6 (38)|
|Arg+||72.4 (21)||85.7 (36)||90.3 (28)||–|
Isolated Calvi-E-trans-F2 colonies and parental E. coli AB1157 were cultured in MM medium supplemented with the required amino acids at 30 °C for 167 h. We found that all of the revertants acquired the ability to produce particles, and the incubation profiles of the revertants were comparable to those of the parental strain. As a typical example, the growth profile of the Leu+ revertants is presented in Fig. 7a.
Leu+ Calvi-E-trans-F2 revertants showed a growth profile comparable to that of the parental E. coli AB1157 strain. Both the revertant and the parental strains reached values of 2.9 × 109 cells mL−1 in the stationary phase, and free particles in the MM medium reached 2.7 × 107 particles mL−1, a value that is much lower than for the particles produced by the F1 particle generation (Fig. 7b). The number of infected cells over 167 h showed a different time-course profile than that of Calvi-E-trans-F1a, with several values<5% for particle-bearing cells (Fig. 7c). The number of visible particles per cell again ranged between one and five, but the mean decreased to 2.6 particles per cell (Fig. 7d). Purified particles from Leu+ revertants (Pt2) were found to be of similar mean size and mode compared with the Pt1 particles of the F1 generation (Table 8). However, the most frequently occurring particle diameter of the Pt2 particles (128 nm) was nearly twice as high as for Pt1 at 24.2%. Furthermore, the range values indicate a more constrained diameter distribution, favouring particle production with a lower size limit of 70 nm (instead of 36 nm as for Pt1) and an upper size diameter of 162 nm (instead of 190 nm).
|Particle type||Source||n||Diameter (mean ± SD)||Mode (frequency)||Range|
|Pt0||Seawater||317||78.5 ± 35.8||60 (13.2)||32–186|
|Pt1||Calvi-E-trans-F1||1526||115.6 ± 43.5||130 (12.7)||36–190|
|Pt2||Calvi-E-trans-F2||598||110.5 ± 28.7||128 (24.2)||70–162|
Because cell mortality increases with increasing MOI and because UV treatment of particles primarily leads to DNA damage, but does not seem to impact the envelope composition of the particles (Table 4), we assume that ‘lysis from without’ (Heldal & Bratbak, 1991; Proctor & Fuhrman, 1992; Weinbauer, 2004) is the main process, in which multiple infections lead to the leakage of ions from the cytosol of the recipient cell, resulting in cell death (Dreiseikelmann, 1994; Kiveläet al., 2004). This assumption is partially supported by the fact that autoclaved particles induce low mortality in the recipient cells. Surface molecules of the envelope, which may act as potential contact receptors (e.g. proteins or lipoproteins), experience conformation changes via heat treatment. Therefore, successful binding to the cells after the contact phase is rather restricted. In the actual state of analysis, we have difficulties finding an explanation for the cell mortality resulting from exposure to the various ultracentrifugation supernatants. The material in the supernatant (>30 kDa, but <0.2 μm) is assumed to consist of cell lysing substances, such as holins, endolysins or bacteriocins, which could act in complement to the postulated lysis and enhance cell mortality. A comparable observation was recorded by Chiura (2006) for particles from thermophiles that were shown to contain peptidase and glycosidase. Thus, our experiments clearly indicate that the control of the bacterial population is not only due to phages but also concentration-dependent nonviral particles. In contrast, the appearance of revertants at MOIs of 5, 20 and 200, where all marker deficiencies were repaired (i.e. all amino acid synthesis deficiencies reverted), can be taken as an indication that multiple infections occurred. This is also supported by the coordinated marker transfer experiments, where the percentage of revertants with unselected markers increased from an MOI of 0.12–5.1, as shown in Table 5. It should be recalled that the coordinated marker transfer data in Table 5 are due to transducing particles from the water column, where their donor bacteria are unknown. However, for the second set of transduction experiments (Tables 6 and 7), the particle DNA was provided by E. coli revertants. Therefore, the data obtained from the coordinated marker transfer allow a number of considerations: Leu and Pro (90.5% linkage) should have the highest percentage of coordinated marker transfer because the distance between the loci of the biochemical markers amounts to only 185 kbp (Dewitt & Adelberg, 1962; Rudd, 1998), which is shorter than the length of the particle-bearing DNA (∼370 kbp). Regardless, the highest percentage of coordinated marker transfer was recorded for His and Pro (96.8%), which have a mutual loci distance that exceeds 1762 kbp (far greater than the length of the transferred particle DNA fragment). Also, we recorded a 90.3% coordinated marker transfer for His and Arg, for which the loci distance is 2134 kbp. Such a high marker linkage can be explained by successful multiple particle infection, providing enough DNA matrix for mismatch repair.
Because we have no information on the relative abundance of the described particles within the viral fraction of the water column, one may only speculate about the potential impact of the observed gene transfer on the bacterial population. Burst size, which can be equated with ‘budding size’ in our study, indicates that the particle abundance may be well below that of marine bacteriophages. Events where 20–200 particles are available for one recipient cell would imply an abundance of ∼2 × 107–2 × 108 particles mL−1 in the water column, but this could not be verified experimentally. Experiments with MOIs of 20 and 200 are probably not representative because they enhance multiple infections in the recipient cell, which may have led to the increase in gene transfer frequencies with increasing MOI (Table 3). It is therefore assumed that lower MOIs may be a better reflection of the in situ situation with episodic events, where several bacterial strains of the population that simultaneously produce budding particles lead to higher MOIs.
In the above investigation, we reconfirmed (Chiura, 2001; Rohwer & Thurber, 2009) that there are particles among the VLPs harvested from seawater that are able to induce intergenic HGT to the enterobacterium E. coli. The recorded gene transfer frequencies ranging from 10−2 to 10−6 are among the highest data published thus far. Comparison with published information reveals that the observed frequencies range between 1.5 × 10−8 and 4.7 × 10−7 (Amin & Day, 1988; Saye et al., 1990; Jiang & Paul, 1998; Hertwig et al., 1999). Only the frequencies recorded by McDaniel et al. (2010) were higher than those obtained in the present study.
The high frequencies of gene transfer obtained in the present investigation suggest that this type of particle is highly efficient in transferring genes. All of the above experiments indicate that we observed nonspecific, VLP-induced gene transfer. One of the most surprising features of this study is that the results of this gene transfer are revertant cells, which are able to produce new particles. These particles produced by the first cell generation (F1) are again infectious and also possess the capability to transfer genes. The resulting second-generation (F2) cells were observed to manifest particle production again. These particles had a mean diameter and mode that were similar to those produced by the F1 cells (Table 8), were again infectious and also induced the production of an F3 revertant population. It is noteworthy that a difference in the diameter size range was recorded between the two particle generations, Pt1 and Pt2. One possible interpretation of this observation, along with the fact that the most frequent particle diameter was nearly twice as abundant for Pt2 (Table 8), is a trend towards a higher frequency of packaging of larger DNA molecules into the particles as serial reinfection processes increase.
Genetic exchange plays a key role in the evolution of prokaryotes. In the last few years, additional mechanisms responsible for horizontal DNA transfer (in addition to the well-known mechanisms, e.g., conjugation, transduction and transformation) were discovered, including GTAs (Marrs, 1974; Wall et al., 1975; Lang & Beatty, 2002; Stanton, 2007) and the transfer of DNA via membrane vesicles (Kolling & Matthews, 1999; Yaron et al., 2000; Renelli et al., 2004; Mashburn-Warren & Whiteley, 2006). GTAs, occurring among Alphaproteobacteria, resemble tailed phages with heads of 30 nm diameter, short spikes and a tail varying between 30 and 50 nm. They are smaller than any morphologically similar virus and package random fragments of chromosomal DNA from the donor bacterium and have incomplete copies of their own genome, ranging between 4.5 and 13.6 kbp. The small DNA fragments can be taken up by recipient cells and recombine with chromosomal DNA. Concerning gene transfer via formation and shedding of membrane vesicles during the growth of Gram-negative bacteria, it was shown that these membrane vesicles are proficient in mediating the transfer of DNA from one strain to another. It is also worth mentioning that the formation and shedding of membrane vesicles in Gram-negative bacteria is a well-known and common phenomenon. Membrane vesicles usually contain only small DNA molecules between 3.3 and 36 kbp (Dorward et al., 1989; Yaron et al., 2000). Notably, none of the characteristics of these additional mechanisms are comparable to the traits of the particles in the above investigation.
The present study allows us to assume that in addition to the previously described mechanisms responsible for horizontal DNA transfer, a thus far undescribed vector enforces the flow of genetic information between prokaryotes. The particle type described shows an analogy to membrane vesicles, but their DNA content is a multiple (>350 kbp), the host range is broader and the transferred material induces the production of new particles in the recipient cells (i.e. serial transduction). The particles studied are larger than any known GTA particle, tailless, are released by budding from the transduced cell and characterized by a double membrane (Fig. 5d). Their DNA is more than 77 × larger than in RcGTA from Rhodobacter capsulatus (Lang & Beatty, 2000), which has the smallest DNA content, and 26 × larger than in Dd1 from Desulfovibrio desulfuricans (Rapp & Wall, 1987) or BLP from Bartonella spp. (Barbian & Minnick, 2000), which has the largest known DNA content.
Thus, we have evidence of a new HGT type in which we observed the packaging of mobilizable DNA into particles and its trafficking from donor to recipient cells. This gives rise to a number of pertinent questions that are related to the peculiarity of the observed particles and that concern their potential as GTAs.
Although we have no information about the original particle type, which was harvested from the sea and obtained by ultracentrifugation, it is highly plausible that in analogy to the observations from the reinfection experiments, we are faced with a naturally occurring particle that cannot be clearly assigned to viruses or so far described membrane vesicles. The experimentally induced particles are larger than the classical membrane vesicles, and the DNA, which is larger than for all reported membrane vesicles, is associated with an unidentified moiety and forms an electron-dense particle that distinguishes itself from all other particles. Furthermore, none of the investigated membrane vesicles (Dorward et al., 1989; Beveridge, 1999; Kolling & Matthews, 1999; Yaron et al., 2000) or viral particles (Lengeler et al., 1999) are known to trigger new particle production after cell infection has taken place.
All of these specific features [especially particle production and subsequent release via budding during the stationary phase, particle-related transfer of large amounts of DNA (up to 370 kbp) to recipient cells and the potential of these particles for serial gene transfer] are an indication that we are confronted with a new gene transferring particle that induces horizontal gene flow between prokaryotic cells. Although we lack information about the as yet undetected original particle that induced the observed gene transfer to the first recipient cell, speculation on the evolutionary importance of particle formation by prokaryotes should continue. Experimental observations indicated that the recipient cells always undergo particle formation in the stationary phase. In aquatic environments, most bacteria are assumed to live under conditions of a nutrient shortage (Haller et al., 1999; Denner et al., 2002), except for the periods of planktonic blooms or seasons of allochthonous organic inputs from various sources. The starvation survival reaction (Kjelleberg, 1993) is one possibility to bridge the periods of limitation between two events of nutrient input. However, each bacterial species does not have the same capability to survive over long periods of nutrient limitations, and any mechanism leading to the conservation of specific genetic information may be an advantage for this specific bacterial population. One such strategy to ensure a backup of genetic information is the transfer of bacterial DNA to recipient cells; the possibility of perpetuating this information is by a sustained cell-to-cell transfer.
Because transfected cells produce particles with roughly 370 kbp of DNA in the stationary phase, ∼13 particles (each carrying a unique piece of the E. coli chromosome) would be sufficient to save the entire bacterial genome. It seems obvious that one may expect species-specific processes of this particle formation after the stationary phase is induced, but it is also highly probable that the overlying mechanisms that determine the formation of these backup particles are similar.
We thank the STARESO S.A. for providing laboratory facilities and the requested ship service for in situ sampling. Thanks are also due to H. Naito, D. Koketsu, H. Soma, T. Takemura, M. Imada, Y. Suzuki and other Lab members at ICU, as well as I. Gerstl, M. Fliesser, B. Mallinger, E. Scherzer and R. Wegscheider from the Medical University of Vienna for their excellent assistance during the experiments and electron microscopic analysis. We sincerely thank R.W. Ridge, S. Suzuki, K. Matsumoto and L. Stöger for critical reading, advice, useful comments and suggestions on the manuscript. This research was supported in part by the International Joint Research Programme ‘Horizontal gene transfer by marine virus-like particles’ Grant No. 10044199, Grant-in-Aid for Scientific Research No. 10490012 and No. 12490009 from the Ministry of Culture, Science, Sports and Education, Tokyo, Japan to K.K. The study was also supported by the Fonds zur Förderung wissenschaftlicher Forschung, Wien, Austria (Project No. P11936-MOB and No. and P17246-BO3) to B.V. This research was additionally funded in part by the Japan Society of the Promotion of Science, Grant-in-Aid for Scientific Research No. 16310031, and Donations to Encourage Research by Kyowa Hakko Kogyo and S.I.C. to H.X.C.
- 1988) Influence of pH value on viability and transduction frequency of Pseudomonas aeruginosa phage F116. Lett Appl Microbiol 6: 93–96. & (
- 2000) A bacteriophage-like particle from Bartonella bacilliformis. Microbiology 146: 599–609. & (
- 1999) Transduction-like gene transfer in the methanogen Methanococcus voltae. J Bacteriol 181: 2992–3002. (
- 1999) Structures of gram-negative cell walls and their derived membrane vesicles. J Bacteriol 181: 4725–4733. (
- 2008) Occurrence and expression of gene transfer agent genes in marine bacterioplankton. Appl Environ Microb 74: 2933–2939. , , , , & (
- 1990) Enumeration and biomass estimation of planktonic bacteria and viruses by transmission electron microscopy. Appl Environ Microb 56: 352–356. , & (
- 2004) Phages and evolution of bacterial pathogens: from genomic rearrangements to lysogenic conversion. Microbiol Mol Biol R 68: 560–602. , & (
- 1997) Generalized gene transfer by virus-like particles from marine bacteria. Aquat Microb Ecol 13: 74–85. (
- 2001) Broad host range xenotrophic gene transfer by virus-like particles from a hot spring. Microbes Environ 17: 53–58. (
- 2002) Broad host range xenotrophic gene transfer by virus like particles from a hot spring. Microb Environ 17: 53–58. (
- 2004) Novel broad-host range gene transfer particles in nature. Microb Environ 19: 249–264. (
- 2006) Novel gene mediator originating from thermophiles in the environment. J Jpn Soc Extrem 5: 75–79. (
- 2004) Isolation and characterization of broad-host range gene transporter particles from geo-thermal vent of Toyoha mine. Microb Environ 19: 20–30. & (
- 2000) Virus-like particles in microbial population control and horizontal gene transfer in aquatic environments. Microbial Biosystems: New Frontiers (BellCR, BrylinskyM & Johnson-GreenP, eds), pp. 167–173. Atlantic Society for Microbial Ecology, Halifax, Canada. , & (
- 2002) Virus-like particle derived from a bacterium belonging to the oldest lineage of the domain Bacteria. Microb Environ 17: 48–52. , , , & (
- 2000) Chromosomal beta-lactamase is packaged into membrane vesicles and secreted from Pseudomonas aeruginosa. J Antimicrob Chemoth 45: 9–13. , , , & (
- 2006) Community genomics among stratified microbial assemblages in the ocean's interior. Science 311: 496–503. , , , , , & (
- 2002) Vibrio calviensis sp. nov., a halophilic, facultatively oligotrophic 0.2 μm-filterable marine bacterium. Int J Syst Evol Micr 52: 549–553. , , , & (
- 1962) The occurrence of a genetic transposition in a strain of Escherichia coli. Genetics 47: 577–585. & (
- 1989) Export and intercellular transfer of DNA via membrane blebs of Neisseria gonorrhoeae. J Bacteriol 171: 2499–2505. , & (
- 1994) Translocation of DNA across bacterial membranes. Microbiol Rev 58: 293–316. (
- 1999) Marine viruses and their biogeochemical and ecological effects. Nature 399: 541–548. (
- 1999) Investigation of 0.2 μm filterable bacteria from the Western Mediterranean Sea using a molecular approach: dominance of potential starvation forms. FEMS Microbiol Ecol 31: 153–161. , , , & (
- 1991) Production and decay of viruses in aquatic environments. Mar Ecol-Prog Ser 72: 205–212. & (
- 1999) Generalized transduction of small Yersinia enterocolitica plasmids. Appl Environ Microb 65: 3862–3866. , , , & (
- 1998) Gene transfer by transduction in the marine environment. Appl Environ Microb 64: 2780–2787. & (
- 1999) Membrane vesicles derived from Pseudomonas aeruginosa and Shigella flexneri can be integrated into the surfaces of other Gram-negative bacteria. Microbiology 145: 2051–2060. & (
- 2004) Penetration of membrane-containing double-stranded-DNA bacteriophage PM2 into Pseudoalteromonas hosts. J Bacteriol 186: 5342–5354. , , , & (
- 1993) Starvation in Bacteria. Plenum, New York. (
- 1999) Export of virulence genes and Shiga toxin by membrane vesicles of Escherichia coli O157:H7. Appl Environ Microb 65: 1843–1848. & (
- 2005) Bacterial outer membrane vesicles and the host–pathogen interaction. Genes Dev 19: 2645–2655. & (
- 2000) Genetic analysis of a bacterial genetic exchange element: the gene transfer agent of Rhodopseudomonas capsulata. P Natl Acad Sci USA 97: 859–864. & (
- 2002) A bacterial signal transduction system controls genetic exchange and motility. J Bacteriol 184: 913–918. & (
- 2007) Importance of widespread gene transfer agent genes in alpha-proteobacteria. Trends Microbiol 15: 54–62. & (
- 2002) Imbroglios of viral taxonomy: genetic exchange and motility. J Bacteriol 184: 4891–4905. , & (
- 1999) Biology of the Prokaryotes. Thieme Verlag, Stuttgart. , & (
- 2002) Characterization of the Sinorhizobium meliloti sin R/sinI locus and the production of novel N-acyl homoserine lactones. J Bacteriol 184: 5686–5695. , , & (
- 1974) Genetic recombination in Rhodopseudomonas capsulata. P Natl Acad Sci USA 71: 971–973. (
- 2005) Membrane vesicles traffic signals and facilitate group activities in a prokaryote. Nature 437: 422–425. & (
- 2006) Special delivery: vesicle trafficking in prokaryotes. Mol Microbiol 61: 839–846. & (
- 2005) Chapter 2.2.4. Outer membrane vesicles. EcoSal – Escherichia coli and Salmonella: Cellular and Molecular Biology (CurtissRIII, ed). http://www.ecosal.org. ASM Press, Washington, DC. & (
- 2010) High frequency of horizontal gene transfer in the oceans. Science 330: 50. , , , , & (
- 1988) Use of SYBR Green I for rapids epifluorescence counts of marine viruses and bacteria. Aquat Microb Ecol 14: 113–118. & (
- 2000) Lateral gene transfer and the nature of bacterial innovation. Nature 405: 299–304. , & (
- 1992) Mortality of marine bacteria in response to enrichments of the virus size fraction from seawater. Mar Ecol-Prog Ser 87: 283–293. & (
- 1987) Genetic transfer in Desulfovibrio desulfuricans. P Natl Acad Sci USA 84: 9128–9130. & (
- 2004) DNA-containing membrane vesicles of Pseudomonas aeruginosa PAO1 and their genetic transformation potential. Microbiology 150: 2161–2169. , , & (
- 2009) Viruses manipulate the marine environment. Nature 459: 207–212. & (
- 1998) Linkage map of Escherichia coli K-12, edition 10: the physical map. Microbiol Mol Biol R 62: 985–1019. (
- 1990) Transduction of linked chromosomal genes between Pseudomonas aeruginosa strains during incubation in situ in a freshwater habitat. Appl Environ Microb 56: 140–145. , , & (
- 1995) Frequency of generalised transducing phages in natural isolates of the Salmonella typhimurium complex. Appl Environ Microb 61: 1637–1640. & (
- 2007) Prophage-like gene transfer agents – novel mechanisms of gene exchange for Methanococcus, Desulfovibrio, Brachyspira, and Rhodobacter species. Ananerobe 13: 43–49. (
- 2003) Vesicle-mediated export and assembly of pore-forming oligomers of enterobacterial ClyA cytotoxin. Cell 115: 25–35. , , , , , , , , & (
- 1975) Genetic transfer of nitrogenase-hydrogenase activity in Rhodopseudomonas capsulata. Nature 258: 630–631. , & (
- 2004) Ecology of prokaryotic viruses. FEMS Microbiol Rev 28: 127–181. (
- 2000) Vesicle mediated transfer of virulence genes from Escherichia coli O157:H7 to other enteric bacteria. Appl Environ Microb 66: 4414–4420. , , & (
- 2009) Gene transfer agent (GTA) genes reveal diverse and dynamic Roseobacter and Rhodobacter populations in the Chesapeake Bay. ISME J 3: 364–373. , , , , , & (